Plasmodium falciparum malaria is transmitted to humans through the bite of infected female Anopheles mosquitoes. Sporozoites that have been injected into the skin migrate from the site of injection, and reach the liver, where they invade hepatocytes and change into merozoites; merozoites penetrate and replicate inside red blood cells. Heparan sulphate proteoglycans (HSPGs) may play an important role in the biology of Plasmodium through their carbohydrate chains (heparan sulphate) in both the mammalian host and the vector. Anopheles heparan sulphate has been shown to bind circumsporozoite protein (CSP), suggesting a role for the carbohydrate chains within Anopheles salivary glands for infection and transmission of the parasite 1 . CSP also interacts with the HSPGs on host liver cells, and this interaction has been shown to determine the choice between migrating through or invading the cell 2 . Highly sulphated HSPGs of hepatocytes activates the rodent malaria parasite Plasmodium berghei for invasion, whereas the parasite migrates through cells with low sulphated HSPGs in skin and endothelium 2 . In addition, a P. falciparum merozoite antigen (EBA-140) has been shown to bind to red blood cells in a heparan sulphate-manner, whereas soluble heparan sulphate and heparin inhibit the merozoite invasion into red blood cells 3 4 . Finally, heparan sulphate is thought to be a receptor for PfEMP1 expressed on infected red blood cells (iRBC), and to mediate the binding of iRBC on endothelial cells or other red blood cells 4 5 6 .

These observations suggest that the outcome of malaria infection may be influenced by variations in the biosynthesis of heparan sulphate, owing to genetic variations within genes encoding the enzymes involved. These include O-sulphotransferases, which catalyze 2-O, 6-O, or 3-O sulphation 7 . The O-sulphation steps are the last steps of the synthesis of heparan sulphate (HS); the sulphation level is a measure of the completion of this synthesis, and is thought to influence the binding properties and therefore the function of HSPGs 7 . The 3-O-sulphation is thought to be a rare event, whereas the 3-O sulphated HS has been shown to serve as an entry receptor of Herpes Simplex Virus 1 (HSV-1) 8 . Interestingly, whereas only one 2-O sulphotransferase, and one 6-O sulphotransferase are known, seven isoforms of 3-O sulphotransferases have been reported 7 9 . The genes encoding the 3-O sulphotransferases are located in different chromosomal regions except for HS3ST3A1 and HS3ST3B1, and the only significant sequence homology between these proteins occurs in the sulphotransferase domains. Nevertheless, HS3ST3A1 and HS3ST3B1, which encode the 3-O sulphotransferases 3-OST-3A1 and 3-OST-3B1 respectively, are 700 kb apart in the same chromosomal region, and show a high sequence identity 10 . 3-OST-3B1, which has a sulphotransferase domain 99.2% identical to that of 3-OST-3A1, sulphates an identical disaccharide 11 . Recently, HS3ST3A1 that encodes 3-OST-3A1 has been associated with mother-to-child transmission of human immunodeficiency virus (HIV) through a genome-wide association study 12 , whereas HSPGs promotes HIV penetration through endothelial cells in a heparan sulphate-manner 13 . This supports the hypothesis that genetic variations within the genes encoding 3-O-sulphotransferases may affect the susceptibility to infectious diseases, such as malaria.

There is a growing body of evidence for human genetic factors controlling the outcome of infection. Familial aggregation and segregation analyses showed the existence of a genetic component of phenotypes related to P. falciparum malaria resistance or susceptibility 14 15 . Several candidate genes have been associated with resistance against severe malaria 16 . Linkage or association analyses mapped various loci controlling mild malaria and/or parasitaemia in humans 15 17 18 19 . The first chromosomal regions that showed linkage to mild malaria and/or parasitaemia were 6p21.3 and 5q31-q33 15 17 20 21 . However, a limited number of confirmed alleles involved in human malaria have been identified 22 23 . In crosses between genetically defined strains of mice, chromosomal regions responsible for the genetic variance of complex traits can be mapped as quantitative trait loci (QTL) in experimental populations available for precise study under defined conditions. Linkage analyses based on experimental crosses have been done in mice, leading to the mapping of loci controlling Plasmodium chabaudi parasitaemia (Char1-10) or cerebral malaria 22 24 25 . Notably, such analyses mapped two loci on chromosomes 17 (Char3) and 11 (Char8), which show extensive conservation of synteny with human chromosomes 6p21.3 and 5q31-q33, respectively 26 27 . The 95% confidence interval of Char8 (chromosome 11 between D11Mit231 and D11Mit30) contained mostly genes, the orthologs of which are located in human chromosome 5q31-q33. Nevertheless, it also contained genes, the orthologs of which are located in human chromosome 17; these include Hs3st3a1 and Hs3st3b1, which encode 3-O sulphotransferases in mice 26 27 , suggesting that genetic variations within HS3ST3A1 and HS3ST3B1 may influence parasitaemia in humans. This hypothesis was further supported by a linkage study based on microsatellite markers in humans (P.Rihet, unpublished data). This prompted us to screen HS3ST3A1 and HS3ST3B1 in a population living in an endemic area in Burkina Faso to identify polymorphisms, to evaluate their linkage and association with parasitaemia, and gene-gene interactive effects by using the pedigree-based generalized multifactor dimensionality reduction (PGMDR) method 28 .

This is apparently the first study to investigate the association between a phenotype related to malaria susceptibility and genes involved in HS biosynthesis. Two genes 700 kb apart, which encode 3-O sulphotransferases involved in the synthesis of HS (HS3ST3A1 and HS3ST3B1), were considered as candidate genes.

First, P. falciparum parasitaemia was found to be genetically linked to HS3ST3A1 and HS3ST3B1 polymorphisms based on a multipoint linkage analysis. This result is consistent with linkage studies based on microsatellite markers in mice 26 27 and humans (P. Rihet, unpublished data). This suggests that polymorphisms within the chromosomal region may partly explain the variance of parasitaemia, and may affect resistance against malaria. In the same way, rs6503319, which is located in the human chromosomal region genetically linked to parasitaemia (P.Rihet, unpublished data), has been associated with severe malaria 39 . It should be stressed, however, that the location of the peak of linkage depends on the SNPs included in the analysis. This indicates either that the linkage analysis does not accurately locate the causal polymorphisms, or that several polymorphisms within the region may influence parasitaemia.

Second, linkage and association between parasitaemia and rs28470223 was detected. Thus, this one-locus analysis confirmed the linkage signal obtained with the multipoint analysis. Furthermore, rs28470223, which is located within the promoter of HS3ST3A1, was associated in the presence of linkage with parasitaemia. This supports the hypothesis that rs28470223 alters both the expression of HS3ST3A1 and parasitaemia. Functional studies will be required to evaluate whether rs28470223 affects the binding of a transcription factor, and the level of gene expression. However, it cannot be excluded that rs28470223 is in linkage disequilibrium with the causal polymorphism.

Third, given that HS3ST3A1 and HS3ST3B1 encode enzymes with a nearly identical activity, a gene-by-gene interaction analysis was conducted based on the PGMDR approach 28 . Two-, three-, four, and five- locus interactions were systematically evaluated. Seventy-three significant multi-locus models, which included SNPs found in both HS3ST3A1 and HS3ST3B1, were identified. This supports the hypothesis of epistatic interaction between HS3ST3A1 and HS3ST3B1. In addition, 37 out of the 73 significant multi-locus models included rs28470223 located in the promoter of HS3ST3A1, further supporting the hypothesis of a particular role for rs28470223. Other SNPs located in the promoter of either HS3ST3A1 or HS3ST3B1 and synonymous mutations, which may alter gene expression levels, were also included in several significant multi-locus models. Moreover, 38 multi-locus models contained at least one mis-sense mutation. This suggests a possible functional role of rs62636623, rs62056073, rs61729712, and rs9906590, which alter the sequence of amino acids. rs62056073, rs61729712, and rs9906590 are of major interest because they affect the sulphotransferase domain, suggesting that they may alter the enzymatic activity. Interestingly, site-directed mutagenesis experiments have demonstrated that several amino acid changes in the sulphotransferase domain dramatically reduce the enzymatic activity 40 . In addition, rs62056073, rs61729712, and rs9906590 are close to known mutations that result in the loss of the enzymatic activity 40 .

In all, the results suggest that several SNPs within HS3ST3A1 and HS3ST3B1, the genes encoding 3-OST-3A1 and 3-OST-3B1, may cause variations in either gene expression levels or the enzymatic activity, and that this may result in variations in parasitaemia. Two mechanisms may explain how genetic variations in HS3ST3A1 and HS3ST3B1 may result in variations in parasitaemia. Genetic variations in HS3ST3A1 and HS3ST3B1 that alter the 3-O sulphation of HS may affect i) the binding of P. falciparum antigen on host cells and/or ii) the pro-inflammatory response.

Together with previous reports 2 3 4 41 , the results suggest that variations in the 3-O sulphation catalyzed by 3-OST-3A1 and 3-OST-3B1 may affect both the binding of P. falciparum antigen on host cells and the parasite invasion rate. Highly sulphated HS has been shown to promote a productive invasion of cells by P. berghei sporozoites, whereas sporozoites migrate through cells harbouring low sulphated HS 2 . Since the 3-O sulphation is the last step of HS synthesis and occurs after the 2-O and 6-0 sulphation steps, one might assume that highly sulphated HS involved in the sporozoite invasion is 3-O sulphated. This hypothesis is consistent with the data showing that 2-O, 3-O, and N sulphate moieties participate in sporozoite CSP binding 41 . Although the N-sulphation has been shown to be involved in rosette disruption 4 5 6 , and although the O-sulphation has not been reported to influence either the binding of P. falciparum antigen on the human erythrocyte surface or the merozoite invasion rate 3 , the results showing the inhibition of merozoite invasion by heparin and other highly sulphated glycoconjugates 42 43 , and those showing the influence of the O-sulphation on hepatocyte invasion by sporozoites makes this hypothesis relevant.

Genetic variations in HS3ST3A1 and HS3ST3B1 might also alter the immune response, and more specifically, the inflammatory response. Indeed, pro-inflammatory cytokine and chemokine binding to HS that depends on the sulphation profile of HS controls both the tissue targeting and the local accumulation of cytokines and chemokines 44 45 . Interestingly, CCR5 (chemokine (C-C motif) receptor 5) up-regulation was associated with the up-regulation of HS3ST3A1 and HS3ST3B1 in humans infected by HIV; the authors suggested either that interaction between HS and CCR5 causes up-regulation, or that the promoters of CCR5, HS3ST3A1 and HS3ST3B1 share a cis-regulatory motif binding the same transcription factor 46 . Since CCR5 and HS3ST3A1 have been associated with HIV infection 12 47 , this suggests an interaction at the genetic level affecting resistance to HIV infection. By extension, this suggests that the interaction between immune genes and genes involved in HS biosynthesis may contribute to resistance against other infectious diseases, such as malaria. Additional investigations are needed to evaluate the role of such genetic interactions, and to elucidate how the sulphation profile of HS determines cytokine and chemokine binding.

This study shows that HS3ST3A1 and HS3ST3B1 are linked to P. falciparum parasitaemia, that rs28470223 within the promoter of HS3ST3A1 is associated with P. falciparum parasitaemia, and that interactions between HS3ST3A1 and HS3ST3B1 polymorphisms alter P. falciparum parasitaemia. The results also indicate that rs28470223 and four mis-sense mutations within HS3ST3B1 strongly contribute to genetic interactions. This study also suggests that other genes involved in HS biosynthesis may affect malaria resistance. In this way, NDST1 that plays a major role in HS biosynthesis is located within the chromosome 5q31-q33, which is linked to P. falciparum parasitaemia 15 .

CCR5: Chemokine (C-C motif) receptor 5; CSP: Circumsporozoite protein; GRCh37: Genome Reference Consortium Human genome build 37; HIV: Human immunodeficiency virus; HS: Heparan sulphate; HSV-1: Herpes simplex virus 1; HSPG: Heparan sulphate proteoglycan; LD: Linkage Disequilibrium; MAF: Minor Allele Frequency; iRBC: infected red blood cells; PGMDR: Pedigree-based generalized multifactor dimensionality reduction

The authors declare that they have no competing interests.

AA carried out the molecular genetic studies, participated in the sequence alignment and performed the statistical analysis. SG participated in the molecular genetic studies and sequence alignment. SA participated in the statistical analysis. FF participated in the design of the study, and revised the results and the manuscript. PR performed the design of the study, supervised the experiments and the statistical analyses, and wrote the manuscript. All authors read and approved the final manuscript.