Endothelial glycocalyx degradation and disease severity in Plasmodium vivax and Plasmodium knowlesi malaria

Degradation of the endothelial glycocalyx is associated with mortality in adult falciparum malaria. However, its role in the pathogenesis of non-falciparum malaria is unknown. In Malaysian patients with knowlesi (n = 200) and vivax (n = 61) malaria, and in healthy controls (n = 50), we measured glycocalyx breakdown products plasma syndecan-1 and urinary glycosaminoglycans, and evaluated correlations with biomarkers of disease severity. Urinary glycosaminoglycans were increased in patients with knowlesi and vivax malaria compared to healthy controls, and in knowlesi malaria were highest in those with severe disease. In knowlesi malaria, plasma syndecan-1 was also highest in those with severe disease, and correlated with markers of endothelial activation (angiopoietin-2, osteoprotegerin, ICAM-1), asymmetric dimethylarginine (ADMA) and impaired microvascular reactivity. Syndecan-1 also correlated with endothelial activation (ICAM-1, angiopoietin-2) and ADMA in vivax malaria. In knowlesi malaria increased syndecan-1 was associated with acute kidney injury, after controlling for age and parasitemia. In knowlesi malaria, the difference in median syndecan-1 between severe and non-severe disease was more marked in females than males. Endothelial glycocalyx degradation is increased in knowlesi and vivax malaria, and associated with disease severity and acute kidney injury in knowlesi malaria. Agents that inhibit glycocalyx breakdown may represent adjunctive therapeutics for severe non-falciparum malaria.

and adherent plasma proteins (see 19,20 for review, images and figures). The major glycosaminoglycans are heparan sulfate, chondroitin sulfate and the non-sulfated hyaluronan, while the proteoglycans include the syndecans and glycipans. When the glycocalyx degrades, a number of these components are shed into the plasma and/or urine and can be measured as biomarkers of glycocalyx breakdown. The best studied of these biomarkers include plasma syndecan-1, plasma heparin sulfate, urinary hyaluronan, and total urinary sulphated GAGs. Glycocalyx degradation has been increasingly recognized to play a key role in the pathogenesis of a number of other conditions, including sepsis 20 , dengue 21 and most recently in severe COVID-19 22 . In sepsis, breakdown is mediated by glycocalyx "sheddases" such as heparanase, metalloproteinases, and hyaluronidase, activated by reactive oxygen species and proinflammatory cytokines 20 . Another activator of heparanase is the endothelial NO synthase (eNOS) inhibitor asymmetric dithemethylarginine (ADMA) 23 . ADMA is known to be elevated in falciparum malaria 24,25 and sepsis 26 , and hence may contribute to glycocalyx breakdown in these conditions. While ADMA has not been found to be increased in vivax malaria 14 , the role of ADMA in knowlesi malaria has not been evaluated.
In this study, we measured concentrations of the glycocalyx breakdown products plasma syndecan-1 and urinary GAGs, and plasma concentrations of ADMA, in Malaysian adults with severe and non-severe knowlesi and vivax malaria, and in healthy controls. We assessed correlations between the glycocalyx breakdown products and endothelial activation (as measured by angiopoietin-2, osteoprotegerin [OPG] and intercellular adhesion molecule-1 [ICAM-1]), ADMA, microvascular function, and other biomarkers of disease severity. Plasma concentrations of the glycocalyx stabiliser sphingosine-1-phosphate (S-1-P) were also evaluated.

Patients.
A total of 261 patients with malaria were enrolled and had plasma and/or urine available for measurement of glycocalyx breakdown products, including 200 patients with knowlesi malaria (152 non-severe and 48 severe) and 61 with vivax malaria (52 non-severe, 9 severe). Plasma was available for 197 patients with knowlesi malaria (149 non-severe, 48 severe), and 58 patients with vivax malaria (50 non-severe, 8 severe). Urine was available for 126 patients with knowlesi malaria (85 non-severe, 41 severe) and 51 with vivax malaria (43 non-severe, 8 severe). Fifty healthy controls were enrolled; plasma syndecan-1 and S-1-P were measured in 20, and urinary GAGS were measured in 47. Clinical and pathophysiological data from these patients have been previously reported 10,11,13,14,27 . Baseline demographic, clinical and laboratory features are shown in Table 1. Patients were enrolled a median of 6 h (IQR 1.4-12.0 h) after commencement of antimalarial treatment.
Glycocalyx breakdown products and disease severity in knowlesi and vivax malaria. In patients with knowlesi malaria, urinary concentrations of GAGs were higher in those with severe compared to non-severe disease ( Fig. 1). However, there was no difference observed between patients with severe and non-severe vivax malaria. Plasma concentrations of syndecan-1 were again unexpectedly reduced in patients with non-severe disease compared to healthy controls (median 140 [IQR 109-205] vs. 177 [IQR 130-519 ng/ml p = 0.046). Concentrations of plasma syndecan-1 were higher in those with severe (median 180 [IQR 154-326] ng/ml) compared to non-severe disease, although numbers were small and this was not significant. In patients with vivax malaria there was no significant difference in concentrations of S-1-P between disease groups, or between either disease group and controls (Table 1 and Fig. 1).
Plasma concentrations of syndecan-1 were correlated with urinary GAGs in knowlesi malaria (r = 0.21, p = 0.019), although this did not remain significant after controlling for parasitemia. There was no correlation between plasma syndecan-1 and urinary GAGs in vivax malaria. There was also no correlation between S-1-P and either plasma syndecan-1 or urinary GAGs, in either knowlesi or vivax malaria.
Plasma concentrations of syndecan-1 were correlated with duration of fever in knowlesi malaria (r = 0.30, p < 0.0001) and in vivax malaria (r = 0.49, p < 0.001); in both species this remained significant after controlling Asymmetric dimethylarginine and disease severity in knowlesi malaria. In patients with knowlesi malaria, the median concentration of ADMA was higher in patients with severe compared to nonsevere disease (0.60 [IQR 0.48-0.75] vs. 0.49 [IQR 0.43-0.54] µM, p < 0.0001), and higher in those with severe disease compared to controls (0.52 [IQR 0.48-0.58], p = 0.014). As with syndecan-1, concentrations of ADMA were unexpectedly lower in those with non-severe disease compared to controls (p = 0.007). The concentrations of ADMA in patients with severe knowlesi malaria were comparable to those of patients with severe falciparum malaria, enrolled in the same study cohort (median 0.57 [IQR 0.52-0.65] µM, N = 22 24 ).
Correlations with age and sex. In Table 2. Correlation coefficients between glycocalyx breakdown products and markers of severity in knowlesi malaria. After applying Bonferonni's correction for the 11 variables evaluated in the correlation analyses above, correlations were considered statistically significant if p < 0.0045. GAGs glycosaminoglycans, ADMA asymmetric dimethylarginine, ICAM-1 intercellular adhesion molecule-1. www.nature.com/scientificreports/ There was a weak correlation between age and syndecan-1 in patients with knowlesi malaria (r = 0.16, p = 0.027); however, this did not remain significant after controlling for parasitemia. There was no correlation between age and syndecan-1 in patients with vivax malaria. There were also no significant correlations between age and S-1-P, in malaria patients or in controls.

Discussion
In this study, we demonstrate that urinary GAGs, markers of glycocalyx breakdown, are increased in knowlesi and vivax malaria, and in knowlesi malaria are increased in proportion to disease severity. In knowlesi malaria the plasma marker of glycocalyx degradation, syndecan-1, was also elevated in severe compared to non-severe disease, and was independently correlated with nearly all biomarkers of severity evaluated, including endothelial activation, microvascular dysfunction, creatinine, and lactate. In knowlesi malaria, in logistic regression models syndecan-1 was also independently associated with both severe disease and AKI. Taken together these findings suggest that glycocalyx breakdown is likely a key contributor to disease pathogenesis in knowlesi malaria.
The finding that urinary GAGs, but not syndecan-1, were increased in non-severe knowlesi malaria, but that syndecan-1 correlated more strongly with biomarkers of severity, suggests that shedding of negatively charged GAG sidechains, which comprise the most superficial layer of the intraluminal glycocalyx, may occur earlier in the disease process. In contrast, shedding of syndecan-1, a deeper core glycocalyx protein attached to endothelial cells through a transmembrane domain, may occur later in disease, reflecting greater glycocalyx damage with greater functional impairment. The correlation of syndecan-1, but not urinary GAGs, with fever duration, is consistent with this hypothesis.
Consistent with previous reports in falciparum malaria 15,16 , we found that in both knowlesi and vivax malaria, syndecan-1 was associated with the endothelial activation marker angiopoietin-2, likely reflecting the known role of angiopoietin-2 in mediating glycocalyx breakdown 30 . Syndecan-1 was also associated with the adhesion molecule ICAM-1 in both knowlesi and vivax malaria. ICAM-1 is contained within the glycocalyx, and the correlation between syndecan-1 and ICAM-1 likely reflects shedding of these adhesion molecules. Glycocalyx degradation may also expose adhesion molecules; while this has been shown in vitro to increase endothelial cytoadherence of P. falciparum-infected red cells 31,32 , the consequences in knowlesi and vivax malaria are less certain, given the apparent paucity of endothelial cytoadherence with these species 33,34 . Table 3. Partial correlation coefficients between glycocalyx breakdown products and markers of severity in vivax malaria. After applying Bonferonni's correction for the 11 variables evaluated in the correlation analyses above, correlations were considered statistically significant if p < 0.0045. GAGs glycosaminoglycans, ADMA asymmetric dimethylarginine, ICAM-1 intercellular adhesion molecule-1. www.nature.com/scientificreports/ We also found that syndecan-1 was associated with OPG, a soluble decoy receptor of receptor activator of NF-kB (RANK) ligand. In addition to the ability of OPG to bind to syndecan-1 35 , OPG has also been shown to inhibit RANK/RANK ligand-activation of eNOS, and may therefore contribute to glycocalyx breakdown by reducing eNOS-mediated inhibition of heparanase expression 23 . Increased heparanase expression may lead to glycocalyx degradation not just by the cleavage of heparan sulfate chains from core proteoglycans 36 , but also by activating metalloproteinases 37 , which cleave proteoglycans such as syndecan-1 directly from endothelial cells 38 .
Heparanase expression may also be induced by the eNOS inhibitor ADMA 23 . Although ADMA is known to be elevated in severe falciparum malaria, an association with glycocalyx breakdown has not previously been evaluated. In this study we found that ADMA was increased in severe knowlesi malaria, and was associated with both syndecan-1 and urinary GAGs. The findings in our study suggest that in addition to the role of angiopoietin-2 in mediating glycocalxy breakdown in malaria, agents that increase heparanase expression such as ADMA and OPG may also contribute. Importantly, ADMA and OPG are both increased in other conditions associated with glycocalyx breakdown, such as sepsis 26,39 , and may represent common mechanisms of glycocalyx degradation. It is notable that in vivax malaria, an increase in ADMA is not observed 14 , and it is possible that this may account for the lack of any observed increase in plasma syndecan-1 in these patients.
Our finding that glycocalyx breakdown was an independent risk factor for AKI in knowlesi malaria is consistent with the key role that the glycocalyx in known to play in kidney function, with the glomerular glycocalyx regulating exchange of plasma proteins across glomerular capillaries 40 . Mouse models have demonstrated that glycocalyx degradation increases glomerular permeability to albumin, leading to albuminuria 41 . We have found that microalbuminuria is common in AKI complicating knowlesi malaria (unpublished), further suggesting that glycocalyx breakdown may be a key mechanism of malaria-associated AKI. Albumin is also able to stabilise glycocalyx 42 , and hence hypoalbuminemia may in itself exacerbate glycocalyx degradation.
A notable finding is our study was the inverse correlation in knowlesi malaria between syndecan-1 and haemoglobin, with this correlation remaining significant after controlling for parasitemia and fever duration. RBCs are known to be major producers of the glycocalyx stabiliser S-1-P 28 ; however, we did not find a correlation between S-1-P and haemoglobin, and therefore reduced RBC production of S-1-P does not seem a likely explanation for the increased glycocalyx breakdown and elevated levels of syndecan-1. Alternatively, shedding of glycocalyx components from RBCs themselves may lead to increased plasma syndecan-1, given that RBCs, like endothelial cells, are known to contain a glycocalyx 43 . Shedding of glycocalyx from RBCs could further exacerbate anaemia, due to exposure of phosphatidylserine or other receptors enhancing eryptosis and/or phagocytosis.
We also found independent inverse correlations between platelets and both syndecan-1 and GAGs, in both vivax and knowlesi malaria. As with RBCs, platelets have been shown to produce S-1-P 29 , and in our study S-1-P was independently associated with platelet count. Thus, it is possible that reduced platelet-derived S-1-P could potentiate glycocalyx breakdown in malaria. Alternatively, glycocalyx breakdown may contribute directly to thrombocytopenia by mediating platelet adhesion and aggregation 44,45 .
We observed sex differences in glycocalyx degradation products. In patients with knowlesi malaria, syndecan-1 was higher and S-1-P lower, in males compared to females. This difference however was only observed in those with non-severe disease, and is consistent with a previous study in acute coronary syndrome demonstrating that males shed more glycocalyx than females 46 , and another study demonstrating that males have a thicker glycocalyx than females 47 . Consistent with this, we also found a trend towards higher syndecan-1 in male controls compared to female controls. Interestingly we found that in severe knowlesi malaria, syndecan-1 was as high in females compared to males, suggesting that severe disease in females may be associated with a greater proportionate loss of glycocalyx. In knowlesi malaria female sex is an independent risk factor for death 3 . While the cause of this increased risk is not well understood, it is possible that a thinner glycocalyx in females and a greater proportionate loss may contribute. Sex differences in disease severity and outcomes are observed in many acute infections 48,49 ; whether differences in glycocalyx degradation contribute warrants further evaluation.
Our study had some limitations. First, the small number of cases of severe vivax malaria limited our ability to examine relationships between markers of glycocalyx degradation and disease severity in vivax malaria. However, with larger numbers of cases of severe knowlesi malaria clear associations with disease severity were found. Second, the median age of the control group was higher than that of the patients with vivax malaria; however, given that we did not find significant correlations between age and markers of glycocalyx breakdown, this is unlikely to have affected our results. Third, our study did not have a non-malaria fever comparison group; however, it is well documented that glycocalyx degradation is not specific to malaria fevers, with degradation also found in a number of other conditions including sepsis 20 , dengue 21 and COVID-19 22 . Finally, although we found associations between markers of glycocalyx degradation and biomarkers of disease severity, the nature of our study does not provide definitive evidence of causation.
In conclusion, we found that glycocalyx breakdown occurs in patients with knowlesi and vivax malaria, and is associated with disease severity and AKI in knowlesi malaria. Plasma syndecan-1 appeared to be a more clinically relevant biomarker than urinary GAGs, and may reflect more severe glycocalyx damage, occurring later in disease. Several agents have been demonstrated to stabilise endothelial glycocalyx, and may be potential agents for adjunctive treatment of severe non-falciparum malaria. www.nature.com/scientificreports/ Study site and patients. Patients were enrolled as part of a prospective observational study of all malaria patients admitted to Queen Elizabeth Hospital, an adult tertiary-referral hospital in Sabah, Malaysia 11,27 . For the current study, patients enrolled between September 2010 and December 2012 were included if they had PCRconfirmed P. knowlesi or P. vivax monoinfection, were non-pregnant, ≥ 12 years old, had no major comorbidities or concurrent illness, were within 18 h of commencing antimalarial treatment, and had stored frozen urine and/ or plasma available for measurement of glycocalyx breakdown products. Severe malaria was defined according to modified WHO criteria 50 , as previously described 11 . Renal function was further assessed using the Kidney Disease: Improving Global Outcomes (KDIGO) criteria for AKI 51 . Healthy controls were visitors or relatives of malaria patients, with no history of fever in the past 48 h and with blood film negative for malaria parasites. Standardized history and physical examination were documented. Haematology, biochemistry, and lactate (by bedside blood analysis; iSTAT system) were obtained on enrolment. Parasite counts were determined by microscopy, and parasite species identified by PCR 52,53 . Patients with severe disease were treated with intravenous artesunate, while those with non-severe disease received oral artemisinin combination treatment (and 14 days of primaquine for those with vivax malaria), as previously described 27 .

Methods
Laboratory assays and measurements of microvascular function. Venous blood collected on enrolment in lithium heparin and citrate tubes was centrifuged within 30 min of collection and plasma stored at -70 °C. Urine was also collected on enrolment and stored within 30 min at -70 °C. ELISA was used to measure heparanized plasma concentrations of syndecan-1 (Abcam); sphingosine-1-phosphate (S-1-P; Echelon Biosciences); angiopoietin-2, ICAM-1 and E-selectin (Quantikine R&D Systems) and osteoprotegerin (DuoSet R&D Systems). Cell free haemoglobin was measured on citrated platelet-free plasma by ELISA (Bethyl Laboratories). Heparanized plasma concentrations of the endothelial nitric oxide synthase (eNOS)-inhibitor asymmetric dimethylarginine (ADMA) were measured by HPLC 54 . Total urinary sulfated GAGs were measured on a subset by the dimethylmethylene blue (DMMB) colorimetric assay as previously described 15,55 . Urinary creatinine was measured by the alkaline picrate method, and the urine GAG concentrations were normalised to creatinine levels. All samples were diluted as per manufacturers' recommendations but were adjusted if the resulting OD was outside the standard curve. Microvascular function was assessed using Near Infrared Spectroscopy (InSpectra 650, Hutchinson, MN), as previously described 56 . These data have been previously reported 11 , and are included here to assess correlation with glycocalyx degradation.
All methods were carried out in accordance with relevant guidelines and regulations.

Statistics. Statistical analysis was performed with STATA software (v15). Student's T-test or Wilcoxon-
Mann-Whitney tests were used for two-group comparisons, depending on distribution. Categorical variables were compared using χ 2 or Fisher's exact tests. Associations between continuous variables were assessed using Spearman's correlation coefficient. Partial correlation was used to evaluate associations between continuous variables after adjusting for parasitemia. For this analysis, non-normally distributed variables were log-transformed to a normal distribution. Since correlations between glycocalyx breakdown products (urinary glycosaminoglycans and syndecan-1) and 11 variables were assessed, Bonferroni's correction was used to adjust for multiple correlations, with correlations considered statistically significant if p < 0.0045 (p = 0.05/11). Backward stepwise regression was used to evaluate factors associated with severe malaria and AKI, with variables removed at a significance level of > 0.05; for this analysis, patients with hyperparasitemia as a sole severity criterion were reclassified as having non-severe malaria.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request and with an appropriate data sharing agreement.