Abstract
Allelic heterogeneity is a common phenomenon where a gene exhibit different phenotype depending on the nature of genetic mutations. In the context of genes affecting malaria susceptibility, it allowed us to explore and understand the intricate host-parasite interactions during malaria infections. In this study, we described a gene encoding erythrocytic ankyrin-1 (Ank-1) which exhibit allelic heterogeneity during malaria infections. We employed ENU mutagenesis screen on mice and identified two Ank-1 mutations, one resulted in an amino acid substitution (MRI95845), and the other a truncated Ank-1 protein (MRI96570). Both mutations caused hereditary spherocytosis-like phenotypes and confer protection against Plasmodium chabaudi infections. Upon further examination, Ank-1(MRI96570) mutation was found to inhibit intra-erythrocytic parasite maturation, whereas Ank-1(MRI95845) caused increased bystander erythrocyte clearance during infection. This is the first description of allelic heterogeneity in ankyrin-1 from the direct comparison between two Ank-1 mutations. Despite the lack of direct evidence from population studies, this observation further supported the protective roles of ankyrin-1 mutations in conferring malaria protection. This study also emphasised the importance of such phenomenon to achieve a better understanding of host-parasite interactions, which could be the basis of future studies.
Authors' summary In malaria endemic regions, many individual developed natural resistance against the disease by having certain genetic mutations that affect the ability of malarial parasites to survive within the human body, notably the red blood cells. However, it is often observed that different mutations within the same gene could give rise to different degree of malaria protection. Through studying this phenomenon, we are able to better understand the underlying cause of their protective effects. In this report, we study two mutations of ankyrin-1 gene, MRI96570 and MRI95845, both of which protect mice from malaria infections. However, both of them exhibit stark differences in the way they mediate protection. MRI96570 affects the ability of malarial parasites to develop inside the red blood cells, whereas MRI95845 enhances the destruction of red blood cells during malaria infection. This is the first direct observation of two distinct methods of achieving malaria protection from ankyrin-1 gene. This report also highlights the complex relationship between the human and malarial parasites, and that such phenomenon might be more common than we initially expected.
Introduction
Historically, malarial parasites have been co-evolving with humans for thousands of years and have played a major role in shaping human genome in malaria endemic regions (1, 2). Indeed, many genetic polymorphisms were selected for as they provide significant survival advantages during malaria infections (1, 3), resulting in high frequencies of protective genetic mutations in malaria endemic regions. The majority of them affect the red blood cells (RBCs), and hence the blood stage of malaria infections (3-5).
Interestingly, these genetic mutations or alleles often exhibit varying degrees of malaria protection even if they affect the same gene, which is influenced by the location and the severity of mutations (6, 7). This phenomenon, known as “allelic heterogeneity”, is characterised by multiple different phenotypes arising from mutations in a single gene. It has been described for certain genes affecting malaria susceptibility, which is reflected by their geographical distribution within endemic regions (8). One of the most prominent examples of this is the G6PD deficiency disorder, which could arise from multiple mutations in G6PD gene (7, 9). Many studies have explored the effectiveness of each mutation in protecting individuals from malaria, which corresponds to the distribution of each allele across the globe (10-12). Another example is the β-globin gene, which is well known for its two malaria protective alleles – the HbS and HbC in African populations (13, 14). HbC is restricted to West Africa, whereas HbS is widespread throughout Africa, which is thought to be linked to the effectiveness of each allele to confer malaria resistance, and their associated morbidity (15, 16). Studies on these alleles would not only allow a better understanding of host-parasite interactions, but also give us insights into the dynamics of population genetics in malaria endemic regions (8).
However, allelic heterogeneity could also complicate the characterisation of the malaria protective roles of certain genes, often resulting in conflicting evidence from various studies. One example of such polymorphisms is CD36 deficiency, which was originally thought to be protective against malaria, as evidenced by the positive selection in East Asian and African populations (17-19). While some studies reported increased malaria protection (20), others reported no significant associations (21) or even increased susceptibility (17, 22). It is possible that these contradictive findings are due to confounding factors associated with allelic heterogeneity in CD36 deficiency (6). This also further emphasises the importance of taking allelic heterogeneity into consideration to enable a better design in future studies involving host genetics in malaria, as well as various infectious diseases.
In contrast, the allelic heterogeneity of genes affecting RBC cytoskeleton in terms of malaria susceptibility is poorly understood. Many of the resulting genetic disorders are heterogeneous, such as hereditary spherocytosis (HS), which is characterised by the formation of “spherocytes”, RBCs that exhibit reduced volume due to disruptions in erythrocyte cytoskeletons. HS is caused by mutations in ankyrin, spectrins, band 3 and protein 4.2, with ankyrin mutations contributing to more than 50% of HS cases (23-27). HS also exhibits clinical heterogeneity, where the severity depends greatly on the location and the nature of mutations (28). However, the prevalence of HS in malaria endemic regions is not well studied, where only specific cases were reported (29-32). Nevertheless, in vivo and in vitro studies have repeatedly suggested an association of HS with increased malaria resistance, and several mechanisms have been proposed, although not all of them were consistent (33-36). Based on these observations, we hypothesised that the inconsistencies in resistance mechanisms might be due to the allelic heterogeneity of genes associated with HS.
To explore this hypothesis, we examined mouse models carrying two novel N-ethyl-N-nitrosourea (ENU)-induced ankyrin mutations. These two mouse lines, Ank-1(MRI96570/+) and Ank-1(MRI95845/MRI95845), displayed haematological and clinical features consistent with HS, and a marked resistance to infection by the murine malarial parasite, P. chabaudi. Analysis of the underlying mechanism of resistance to infection revealed both common and distinct features between the strains. RBCs from both lines were similarly resistant to merozoite invasion. However, the Ank-1(MRI95845/MRI95845) erythrocytes were also more rapidly cleared from circulation during an infection, whereas an impairment in intra-erythrocytic parasite maturation was observed in the infected Ank-1(MRI96570/+) erythrocytes. This study highlights the first direct examination of allelic heterogeneity of Ank-1 gene in the context of malaria resistance in mouse models.
Results
MRI96570 and MRI95845 carry mutations in Ank-1 gene
ENU-treated SJL/J male mice were crossed with wild-type female to produce G1 progeny. The G1 progeny carrying heterozygous point mutations in their genome were then subjected to haematological screening to identify genes affecting RBC properties, as potential candidates that might confer malaria protection. G1 mice MRI96570 and MRI95845 were identified from the ENU-dominant screen with mean cellular volume (MCV) three standard deviations below the normal level of the respective parental line - 48.5fl for MRI96570, and 50.6fl for MRI95845, compared to the background of 55.1±1.2fl in SJL/J inbred strain. These mice were crossed with wild-type to produce G2 progeny, where approximately half of them exhibit low erythrocyte MCV values. Two affected MRI96570 and MRI95845 G2 progeny, which also showed a reduction of MCV, were sent for exome sequencing to identify the causative genetic mutations. Unique variants shared between the affected mice were filtered and selected for mice carrying MRI96570 mutation or MRI95845 mutation (S1 Table). A mutation in ankyrin-1 (Ank-1) gene was present in all the affected mice, and co-segregated completely with the reduced MCV phenotype for over three generations of crosses. Sanger sequencing revealed a T to A transversion in exon 34 of Ank-1 gene for MRI96570 strain, and a T to A transversion in exon 5 of Ank-1 gene for MRI95845 strain (S1 Figure). They were predicted to cause a nonsense mutation at amino acid position 1398, located in the spectrin-binding domain for MRI96570 mice, and a substitution of tyrosine for asparagine at amino acid residue 149 in the 4th ankyrin repeat for MRI95845 mice (Figure 1a). MRI96570 and MRI95845 will be referred as Ank-1(MRI96570) and Ank-1(MRI95845) respectively, for the rest of the report.
Both Ank-1(MRI96570) and Ank-1(MRI95845) exhibit HS-like phenotypes
Since ankyrin mutations are usually associated with HS, we examined both Ank-1(MRI96570) and Ank-1(MRI95845) mice in terms of their HS-like phenotypes. When two Ank-1(MRI96570/+) G2 progeny were intercrossed, Ank-1(MRI96570/MRI96570) mice were born with severe jaundice and died within several days from birth (Figure 1b), suggesting homozygosity for Ank-1(MRI96570) mutation caused lethal anaemia. On the other hand, Ank-1(MRI95845/MRI95845) mice appeared healthy with normal lifespan. Haematological analysis of these mice revealed a significant reduction in MCV and mean corpuscular haemoglobin (MCH) and increased red cell distribution width (RDW) (S2 table), indicating microcytosis and anisocytosis, which are the hallmarks for HS. When the RBCs were subjected to osmotic stress, RBCs from Ank-1(MRI96570/+), Ank-1(MRI95845/+) and Ank-1(MRI95845/MRI95845) mice exhibit significantly increased RBC osmotic fragility compared to wild-type RBCs (Figure 1c). 50% haemolysis was observed at approximately 5.6 and 5.4 g/L (equivalent to 104mM and 100mM) sodium chloride, respectively, compared to approximately 4.6g/L (84mM) sodium chloride of wild-type. The Ank-1(MRI95845/MRI95845) RBCS showed further susceptibility towards osmotic stress, with 50% haemolysis at approximately 6.5 g/L (121mM) sodium chloride concentration.
We predicted that the mutant RBCs have shorter half-life, which is also one of the symptoms of HS. Therefore, RBC half-life was determined by biotinylating mouse RBCs and tracking the remaining biotinylated RBCs over time. As shown in Figure 1d, erythrocytes from Ank-1(MRI95845/MRI95845) RBCs have significantly shorter half-life of approximately 9.5 days as opposed to 16 days of wild-type erythrocytes, but no significant difference was observed for erythrocytes from heterozygous mice. The morphology of these RBCs were examined under light and scanning electron microscopy (S2 Figure). Ank-1(MRI96570/+) and Ank-1(MRI95845/+) mice exhibited slight reduction in RBC size, while Ank-1(MRI95845/MRI95845) mice had smaller RBCs and displayed anisocytosis and acanthocytic. On the other hand, blood smears obtained from jaundiced Ank-1(MRI96570/MRI96570) pups showed reticulocytosis, fragmented RBCs and severe anisocytosis.
Another feature of HS is the reduced RBC deformability, which was examined using two different analytical techniques: ektacytometry and an in vitro spleen retention assay. Ektacytometry measures the flexibility of RBCs when subjected to shear pressure, and expresses as an elongation index, which indicates the deformability of RBCs. The Ank-1(MRI96570/+) RBCs showed reduced elongation index compared to wild-type, whereas Ank-1(MRI95845/MRI95845) RBCs showed further reduction in elongation index, indicating significant reduction in RBC deformability (Figure 1e). In addition, the in vitro spleen retention assay was performed by passing the erythrocytes through layer of microbeads of varying sizes, modelling in vivo splenic filtration. RBC deformability was assessed by the ability of RBCs to pass through the bead layer. Figure 1f showed three independent measurements of RBC deformability via splenic retention assay, comparing between wild-type, Ank-1(MRI96570/+), Ank-1(MRI95845/+) and Ank-1(MRI95845/MRI95845) RBCs. An approximately 70% increased retention for Ank-1(MRI96570/+) RBCS was observed compared to wild-type, whereas erythrocytes of Ank-1(MRI95845/+) and Ank-1(MRI95845/MRI95845) mice showed 86% and 90% increased RBC retention compared to wild-type, respectively. However, no significant difference was observed between Ank-1(MRI96570/+) and Ank-1(MRI95845/MRI95845) erythrocytes.
The expression levels of ankyrin and other RBC membrane proteins were examined (S3 Figure). Significant reduction of Ank-1 mRNA levels was observed in Ank-1(MRI96570/+), Ank-1(MRI95845/+), Ank-1(MRI96570/MRI96570) and Ank-1(MRI95845/MRI95845) embryonic livers (S3a Figure). However, coomassie staining and Western blotting of the RBC membrane fractions did not show a significant difference in ANK-1 levels between wild-type, Ank-1(MRI96570/+) and Ank-1(MRI95845/MRI95845) erythrocytes (S3b-d Figure). The predicted truncated ANK-1(MRI96570/+) form (160kDa) was also not evidenced. The levels of other cytoskeletal proteins were also examined to account for possible disruptions to interactions with other binding partners of ankyrin-1. However, no difference was observed for Band 3, α-and β-spectrin, whereas significantly lower protein 4.2 level was observed in Ank-1(MRI95845/MRI95845) erythrocytes (S3d Figure).
Ank-1(MRI96570) and Ank-1(MRI95845)confer protection against P. chabaudi infection
We proposed that mice carrying these mutations have reduced susceptibility to malaria infection, which we examined by injecting with a lethal dose of P. chabaudi, and the percentage of parasitised RBCs (parasitemia) was recorded. As shown in Figure 2a, Ank-1(MRI96570/+) and Ank-1(MRI95845/+) mice showed significant reduction in peak parasitemia of approximately 15-20%, while Ank-1 (MRI95845/MRI95845) mice showed approximately 30% reduction in peak parasitemia compared to wild-type. Ank-1(MRI95845/MRI95845) mice also showed a two-day delay in parasitemia, peaking on day 12 post-infection rather than day 10 as with wild-type. Ank-1(MRI95845/MRI95845) mice also exhibited significantly higher survival rate compared to wild-type during P. chabaudi infection, but no significant difference was observed for Ank-1(MRI96570/+) and Ank-1(MRI95845/+) mice compared to wild-type (Figure 2b). Overall, these results suggested that both Ank-1(MRI96570/+) and Ank-1(MRI95845/+) mice showed moderate resistance, whereas Ank-1(MRI95845/MRI95845) mice exhibited significant resistance towards P. chabaudi infection in relative to the wild-type mice.
From these results, we further investigated and compared the possible mechanisms of resistance mediated by Ank-1(MRI96570) and Ank-1(MRI95845) mutations. We examined three important determinants of parasite growth and survival within the host. Firstly, we studied the ability of parasite to survive within these erythrocytes, since ankyrin-1 mutations have previously been implicated to impair parasite intra-erythrocytic maturation (36). Secondly, the erythrocyte invasion was assessed as the mutations disrupt erythrocyte cytoskeletal structure, which is important for facilitating efficient erythrocyte invasion (37). Thirdly, the mutations might result in an improved detection of parasitised RBCs, thus enhancing their removal from circulation during malaria infection. Since Ank-1(MRI96570/+) and Ank-1(MRI95845/MRI95845) mice exhibited differences in malaria resistance, we hypothesised that they mediate malaria resistance through different pathways.
Ank-1(MRI96570/+) and Ank-1(MRI95845/MRI95845) erythrocytes are resistant to merozoite invasion
First, the ability of parasite to invade erythrocytes was assessed via an in vivo erythrocyte tracking (IVET) assay. Labelled RBCs from either wild-type, Ank-1(MRI96570/+) or Ank-1(MRI95845/MRI95845) mice were injected into infected wild-type mice of 1-10% parasitemia during late schizogony stage and the parasitemia of each genotype was monitored over 36-40 hours to indicate relative invasion rates. The initial invasion period was expected at 30 minutes to 3 hour timepoints, and the results were expressed as a ratio of parasitised RBCs of either, Ank-1(MRI96570/+) to wild-type (Figure 3a), Ank-1(MRI95845/MRI95845) to wild-type (Figure 3b), or Ank-1(MRI96570/+) to Ank-1(MRI95845/MRI95845) (Figure 3c). From Figure 3a and 3b, Ank-1(MRI96570/+) and Ank-1(MRI95845/MRI95845) erythrocytes were less parasitised compared to wild-type (0.6-0.7 for Ank-1(MRI96570/+) and 0.55-0.8 for Ank-1(MRI95845/MRI95845)) from 3 hours up to 36 hours post-injection, indicating both Ank-1(MRI96570/+) and Ank-1(MRI95845/MRI95845) erythrocytes were more resistant to parasite invasion than wild-type. However, no significant differences in parasitemia ratio were observed at 30 minute timepoint. Furthermore, when the invasion rate of both Ank-1(MRI96570/+) and Ank-1(MRI95845/MRI95845) erythrocytes were compared in infected wild-type mice (Figure 3c), no significant difference in parasitemia ratio was observed, suggesting a similar invasion rate between the two mutant erythrocytes.
Ank-1(MRI96570/+) erythrocytes impair parasite maturation
Second, the parasite intra-erythrocytic maturation was determined through a TUNEL assay, which allows the detection of fragmented DNA in RBCs, as an indication of dying parasites (Figure 3d) (38). Samples were collected from infected mice at 1-10% parasitemia at both young ring stage and late trophozoite stage, and the proportion of TUNEL-positive infected RBCs were calculated. As seen from Figure 3e, more TUNEL-positive parasites were observed within Ank-1(MRI96570/+) erythrocytes, in both ring (30.1±3.4% compared to 15.2±3.1% of wild-type) and trophozoite stage (30.8±3.8% compared to 11.7±1.0% of wild-type), whereas no differences were observed for Ank-1(MRI95845/+) and Ank-1(MRI95845/MRI95845) erythrocytes. This result suggested that the growth of parasites within Ank-1(MRI96570/+) erythrocytes was impaired, but was normal in Ank-1(MRI95845/+) and Ank-1(MRI95845/MRI95845) erythrocytes. This also indicate that Ank-1(MRI96570) disrupts parasite maturation, whereas Ank-1(MRI95845) seems to support normal parasite growth, although growth inhibition at other stages were not studied.
Ank-1(MRI95845/MRI95845) erythrocytes are more likely to get cleared during malaria infections, partially via splenic filtration
The proportions of labelled erythrocytes were also monitored during the IVET assays to compare the relative loss of the two labelled RBC populations as the indicator of RBC clearance during malaria infection. No significant reduction in Ank-1(MRI96570/+) erythrocyte numbers was observed during IVET assay compared to wild-type (Figure 4a). On the other hand, the number of labelled Ank-1(MRI95845/MRI95845) erythrocytes decreased significantly compared to wild-type and Ank-1(MRI96570/+) erythrocytes (Figure 4b and c), with approximately 20% and 50% reduction, respectively. However, the parasitemia measurements during the IVET assays were approximately 2% to 16-30% (S4a-b Figure), which did not correlate with the reduction of labelled Ank-1(MRI95845/MRI95845) erythrocytes. This suggested an increased bystander clearance rather than clearance of infected Ank-1(MRI95845/MRI95845) RBCs. To further verify this observation, the RBCs of infected mice from each genotype were biotinylated and the RBC half-life was examined without blood transfusion. As shown in Figure 4d, the Ank-1(MRI96570/+) mice exhibited no significant reduction in RBC numbers, whereas Ank-1(MRI95845/MRI95845) mice were found to have significantly shorter half-life of approximately 6 days, which did not correlate with the parasitemia curve (S4c Figure). This observation of shorter RBC halflife in infected Ank-1(MRI95845/MRI95845) mice is consistent with the increased Ank-1(MRI95845/MRI95845) erythrocyte clearance as shown in IVET assays.
We proposed that the spleen played a major role in mediating this bystander clearance. Therefore, we infected mice which had been splenectomised with P. chabaudi and infused with labelled wild-type and Ank-1(MRI95845/MRI95845) erythrocytes, the proportions of which were monitored over time. As shown in Figure 4e, Ank-1(MRI95845/MRI95845) erythrocyte numbers are approximately two-fold higher (P<0.01) in splenectomised mice compared to non-splenectomised mice. This suggests that the spleen is a major contributor towards Ank-1(MRI95845/MRI95845) erythrocyte clearance, although the clearance was not completely abrogated in the absence of the spleen.
Increased Band 3 mobility in Ank-1(MRI95845/MRI95845) erythrocytes as a likely mechanism for increased clearance
Nevertheless, we hypothesised that this phenomenon is likely due to changes to the cytoskeletal structure of Ank-1(MRI95845/MRI95845) RBCs. In order to support our hypothesis, we examined the band 3 mobility across the RBC membrane as an indicator of disrupted RBC cytoskeleton (39, 40). We fluorescently labelled erythrocytic band 3 with eosin-5'-maleimide and performed Fluorescence Recovery after Photobleaching (FRAP) on erythrocytes, which involved photobleaching with high-powered laser followed by a recovery period where the fluorescence intensity was recorded. Ank-1(MRI95845/MRI95845) RBCS were found to have significantly higher fluorescence recovery compared to wild-type and Ank-1(MRI96570/+) RBCs (Figure 4f), which suggests a higher band 3 mobility in Ank-1(MRI95845/MRI95845) erythrocytes, possibly due to an increased amount of band 3 that was not associated with the RBC cytoskeleton.
Discussion
Ank-1 gene displayed allelic heterogeneity displayed allelic heterogeneity on host mice phenotypes and during malaria infections
Similar to HS in human populations, ankyrin mutations in mice also exhibit differences in clinical symptoms depending on the mutations. As shown in this study, homozygosity for MRI96570 mutation is lethal, while MRI95845 homozygotes appeared healthy, whereas both Ank-1(MRI96570/+) mice and Ank-1(MRI95845/+) mice exhibited HS-phenotypes with similar severity. While both mutations also conferred malaria protection and appeared to impair parasite invasion, they also showed some remarkable differences in mediating this resistance. Parasites in Ank-1(MRI96570/+) erythrocytes were more likely to be TUNEL-positive, indicating impaired intra-erythrocytic maturation, whereas Ank-1(MRI95845/MRI95845) erythrocytes were more likely to be removed from circulation and possibly increased turnover rate.
These findings were not exclusive to these two Ank-1 mice described in this study. In fact, previous studies on other Ank-1 mice also exhibit similar mechanisms of resistance. Notably, Ank-1(MRI23420/+) and Ank-1(nb/nb) mice were both reported to affect the parasite survival within the defective RBCs (33, 36). On the other hand, Ank-1(MRI61689/+1) mice were also found to exhibit increased RBC bystander clearance (41), similar to Ank-1(MRI95845/MRI95845) mice. Nevertheless, this is the first direct report of such allelic heterogeneity described in in vivo mouse models, which highlighted the complexity behind the genetic resistance to malaria, especially in human populations.
Allelic heterogeneity of Ank-1 and its association with malaria
However, due to lack of large scale studies on the HS prevalence in malaria endemic regions, ankyrin-1 has not been associated with malaria protection. Although HS prevalence is more well-characterised in Northern European and Japanese populations, with a prevalence of about 1 in 2000 (42-44), one study proposed an increased HS incidence in Algeria of about 1 in 1000 (45), raising the possibility of positive selection of HS by malarial parasites. However, as the result of extreme allelic heterogeneity of HS-causing genes, many alleles do not reach sufficient frequencies (46) or achieve consistent symptoms (47) to be easily associated with malaria protection. In addition, technical difficulties (29), confounding factors from large genetic variation in African populations (48), as well as poor diagnostics and health systems (48), pose significant challenges for dissecting the connection between HS and malaria. With development of more advanced technologies and better characterisation of the genetic structure of African populations, further studies into the association of HS and malaria could potentially yield beneficial insights into the co-evolutionary relationships between humans and Plasmodium.
Nonetheless, previous in vivo studies have indicated that Ank-1 mutations affect merozoite invasion and maturation (33, 36), both of which were also demonstrated in this study. However, this study also describes the first direct in vivo observation of different mutations in the Ank-1 gene mediating two distinct, independent mechanisms of malaria resistance, where one impairs parasite maturation and the other increases RBC clearance. Ankyrin is one of the key proteins involved in RBC remodelling by parasites (49-51), and maintaining the structure of RBC cytoskeleton (28, 52). It is possible that this allelic heterogeneity is due to the effect each mutation has on the integrity of RBC cytoskeletal structure, as evidenced by the significantly increased band 3 mobility caused by Ank-1(MRI95845), but not Ank-1(MRI96570) mutation (Figure 4f). This suggests that mutations at different locations of the ankyrin-1 protein might have different effects on the RBCs, consequently impacts the ability of parasites to invade and grow. This hypothesis also agrees with the findings from other Ank-1 mice from previous studies (35, 36, 41), where each mutation exhibited differences in terms of mechanisms of malaria resistance, which could be the basis for further studies.
Similarities of allelic heterogeneity in Ank-1 and other malaria susceptibility genes
As evidenced from this study, the protective effect of the Ank-1 gene against malaria is dependent on the nature and the location of mutations within the gene. Similarly, this allelic heterogeneity is also observed in other malaria susceptibility genes in human populations. For instance, although many G6PD deficiency-causing alleles have been implicated with malaria protection (53, 54), the protective effects are often debated, with many studies reporting increased, or no protection, for individuals with certain alleles of G6PD deficiency (55-60). This is thought to be due to the phenotypic complexity of G6PD deficiency associated with malaria susceptibility (7). Indeed, various G6PD alleles have been shown to cause a reduction of G6PD activity with differing severity, and was proposed to correlate with the malaria protection they conferred (55). More recently, Clarke and colleagues proposed reduced G6PD activity is associated with lower risk of cerebral malaria, but in exchange for higher risk of malaria anaemia (12), suggesting a delicate balance underlying the high frequency of G6PD polymorphism in populations of malaria endemic region. Similarly, Ank-1 mutations described in this study, as well as other previous mouse studies (33, 35, 36), exhibit variability in malaria resistance, most likely as the result of allelic heterogeneity.
The heterogeneity in malaria resistance mechanisms of the Ank-1 gene as observed in this study is comparable to the two prevalent alleles of the β-globin gene – the HbS and HbC, which result from amino acid substitution at position 6 from glutamate to valine, or lysine, respectively. They exhibit some similarities in mediating malaria resistance, including impaired parasite growth (61, 62), reduced cytoadherence (63-65) and increased erythrocyte clearance (66). However, HbS erythrocytes were found to be more resistant to all forms of malaria, whereas HbC erythrocytes appeared to be protective against cerebral malaria (15). This difference in malaria protection was proposed to correlate with distribution of HbS and HbC in Africa (16), further emphasising the importance of allelic heterogeneity in understanding host-parasite interactions.
In conclusion, we reported a novel observation where the Ank-1 gene exhibits phenotypic heterogeneity in mediating malaria resistance mechanisms either via impairing intra-erythrocytic parasite growth, or promoting RBC clearance. This study also highlighted that the allelic heterogeneity in relation to malaria resistance is not exclusive to G6PD deficiency, and it could also be much more common than we expected. Further studies should extend the understanding of the effects of various Ank-1 mutations on the development of intra-erythrocytic parasites, as well as the association of HS with malaria in human populations. This could provide new insights into the intricate relationships between RBC cytoskeletal structure and parasite interactions.
Materials and Methods
Mice and ethics statement
All mice used in this study were housed with 12 hour light-dark cycle under constant temperature at 21 °C. All procedures were performed according to the National Health and Medical Research Council (NHMRC) Australian code of practice. Experiments were carried out under ethics agreement AEEC A2014/54, which was approved by the animal ethics committees of the Australian National University.
ENU mutagenesis and dominant phenotype screening
SJL/J male mice were injected intraperitoneally with two dose of 100 mg/kg ENU (Sigma-Aldrich, St Louis, MO) at one week interval. The treated males (G0) were crossed to females from the isogenic background to produce the first generation progeny (G1). The seven-week-old G1 progeny were bled and analysed on an Advia 120 Automated Haematology Analyser (Siemens, Berlin, Germany) to identify abnormal red blood cell count. Mouse carrying MRI96570 or MRI95845 mutation was identified with a “mean corpuscular volume” (MCV) value three standard deviations lower from other G1 progeny. It was crossed with SJL/J mice to produce G2 progeny to test the heritability of the mutations and the dominance mode of inheritance. Mice that exhibited low MCV (<48fL) were selected for whole exome sequencing and genotyping.
Whole exome sequencing
DNA from two G2 mice carrying the abnormal red blood cell parameters (MCV <48fL) were extracted with Qiagen DNeasy blood and tissue kit (Qiagen, Venlo, Netherlands) for exome sequencing as previous described (67). Briefly, 10μg of DNA was prepared for exome enrichment with Agilent Sure select kit paired-end genomic library from Illumina (San Diego, CA), followed by high throughput sequencing using a HiSeq 2000 platform. The bioinformatics analysis was conducted according to the variant filtering method previously described by Bauer, McMorran (68). Private variants that were shared between the two mutants but not with other SJL/J, C57BL/6 mice or previously described ENU mutants were annotated using ANNOVAR (69). Private non-synonymous exonic and intronic variants within 20 bp from the exon spicing sites were retained as potential candidate ENU mutations.
PCR and Sanger sequencing
DNA from mutant mice were amplified through PCR with 35 cycles of 30 seconds of 95°C denaturation, 30 seconds of 56-58°C annealing and 72°C elongation for 40 seconds. The primers used in the PCR are described as below. The PCR products were examined with agarose gel electrophoresis before being sent to the Australian Genome Research Facility (AGRF) in Melbourne, Australia, for Sanger sequencing. Logarithm of odds (LOD) score was calculated based on the number of mice that segregated with the candidate mutations.
Primers for MRI95845 mutation:
Primers for MRI96570 mutation:
RBC osmotic fragility analysis
To assess the susceptibility of RBC membrane to osmotic stress, 5μl of mouse whole blood was diluted 100-fold with phosphate buffer (pH 7.4) containing 0 to 10g/L of sodium chloride, and incubated for at least 10 minutes at room temperature. The cells were centrifuged at 800g for 3 minutes, and the supernatant, which contains free haemoglobin, was measured at 540nm to assess the degree of haemolysis. The absorbance values were expressed as percentage of haemolysis, with haemolysis at 0g/L sodium considered as 100% lysis.
Erythrocyte lifetime assay
Each uninfected mouse was injected with 1mg of EZ-link® Sulfo-NHS-LC Biotin (Biotin) (Thermo Scientific, Waltham, MA) in MT-PBS intravenously. 2ul of blood was collected on day 1, 7, 14, 21 and 28 from the day of injection. Samples were stained and analysed using a flow cytometer (details described in method 2.4.5). The proportion of Biotin-labelled mature RBCs on day 1 was considered as the “starting point” of 100% of labelled cells. For subsequent timepoints, the remaining number of biotin-labelled RBCs were expressed as a percentage of the starting number as the indication of RBC turnover rate.
For infected mice, 1mg of Biotin was injected intravenously as soon as parasitemia was detectable on flow cytometry (approximately 0.05-0.3%). Samples were collected daily and analysed as above.
Ektacytometry
10-15ul of uninfected RBCs were first resuspended in 500ul of pre-warmed polyvinylpyrrolidone (PVP) solution at a viscosity of 30 mPa/second at 37 °C until needed. Samples were analyzed according to the manufacturer's instructions with a RheoScan Ektacytometer (Rheo Meditech, Seoul, South Korea) and the elongation index measured across a range of pressures from 0-20 Pa. Each sample was measured three times to account for technical variabilities. The values were normalized against the wild-type samples.
In vitro spleen retention assay
The RBC deformability were examined according to the protocol described previous by Deplaine, Safeukui (70) with modifications. Briefly, RBCs from each genotype of mice were stained with 10μg/ml of either hydroxysulfosuccinimide Atto 633 (Atto 633) or hydroxysulfosuccinimide Atto 565 (Atto 565) (Sigma-Aldrich, St Louis, MO), followed by three washes with MTRC (154mM NaCl, 5.6mM KCl, 1mM MgCl2, 2.2mM CaCl2, 20mM HEPES, 10mM glucose, 4mM EDTA, 0.5% BSA, pH 7.4, filter sterilized). The stained RBCs were mixed in equal proportion and diluted with unstained wild-type RBCs to give approximately 10-20% of the total RBCs being labelled RBCs. The samples were further diluted to 1-2% haematocrit with MTRC, before passing through the filter bed. The pre-filtered and post-filtered samples were analysed on BD LSRFortessa (BD Biosciences, Franklin Lakes, NJ) flow cytometer to determine the proportion being retained in the filter bed.
Scanning electron microscopy (SEM)
SEM was performed as described previously (41). Mouse blood was fixed overnight in 3% EM-grade glutaraldehyde (Sigma-Aldrich, St Louis, MO) at 4°C immediately upon collection. The samples were washed with MT-PBS 3 times, 10 minute each time. The cells were then adhered to the coverslips with 0.1% polyethylenimine (PEI) for 10 minutes, before washing with MT-PBS. The cells were then dried serially using 30%, 50%, 70%, 80%, 90%, 100%, 100% ethanol, 10 minutes each time. The cells were then soaked in 1:1 ethanol: hexamethyldisilazane solution for 10 minutes, followed by 2 washes with 100% hexamethyldisilazane (Sigma-Aldrich, St Louis, MO), each 10 minutes. The coverslips were then air-dried overnight and coated with gold and examined under JEOL JSM-6480LV scanning electron microscope.
Quantitative PCR and cDNA sequencing
RNA was purified from embryonic livers of E14 embryos using Qiagen RNeasy kit (Qiagen, Venlo, Netherlands), followed by cDNA synthesis using Transcriptor High Fidelity cDNA Synthesis Kit (Roche, Basel, Switzerland), as described previously (41). Quantitative PCR was performed on ViiA™ 7 Real-Time PCR System (Thermo Scientific, Waltham, MA). The ΔΔCT method (71) was used to determine the cDNA levels of Ank-1 and the housekeeping gene β-actin and expressed as a fold-change of the mutants to the wild-type. The primers used for Ank-1 gene spanned exon 2 to 4: Ank-1-F: 5'- TAACCAGAACGGGTTGAACG-3'; Ank-1-R: 5'-TGTTCCCCTTCTTGGTTGTC-3'; β-Actin-F: 5'-TTCTTTGCAGCTCCTTCGTTGCCG-3'; β-Actin-R: 5'- TGGATGCGTACGTACATGGCTGGG-3'.
SDS-PAGE, Coomassie staining and Western blot
The analysis of erythrocytic proteins was carried out as described previously (41). Briefly, RBC ghosts were prepared by lysing mouse RBCs with ice-cold 5mM phosphate buffer (ph7.4) and centrifuged at 20,000g for 20 minutes followed by removal of the supernatant, and repeat until the supernatant became clear. The RBC ghosts or whole blood lysates were denatured in SDS-PAGE loading buffer (0.0625M Tris pH 6.8, 2% SDS, 10% glycerol, 0.1M DTT, 0.01% bromophenol blue) at 95°C for 5 minutes before loading onto a Mini-PROTEAN® TGX™ Precast Gels (Bio-Rad, Hercules, CA). The gels were then either stained with Coomassie blue solution (45% v/v methanol, 7% v/v acetic acid, 0.25% w/v Brilliant Blue G) overnight or transferred to a nitrocellulose membrane. The western blot was carried out using these primary antibodies: anti-alpha 1 spectrin (clone 17C7), anti-beta 1 spectrin (clone 4C3) (Abcam, Cambridge, UK), anti-GAPDH (clone 6C5) (Merck Millipore, Darmstadt, Germany), anti-N-terminal Ank-1 “p89”, anti-Band 3 and anti-protein 4.2 (kind gifts from Connie Birkenmeier, Jackson Laboratory, US). Each primary antibody was detected with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibody at 1:5000 dilution from 1mg/ml stocks. The blots were visualised using ImageQuant LAS 4000 (GE Healthcare Life Sciences, Arlington Heights, IL), and quantified using ImageJ software (72).
Malaria infection
Malaria infections on mice were performed as described previously (41). 200μl of thawed P. chabaudi adami infected blood was injected into the intraperitoneal cavity of a C57BL/6 donor mouse. When the donor mouse reached 1-10% parasite load (parasitemia), blood was collected through cardiac puncture. The blood was diluted with Krebs' buffered saline with 0.2% glucose as described previously (73). Each experimental mouse was infected with 1×104 parasites intraperitoneally. The parasitemia of these mice were monitored either using light microscopy or flow cytometry.
Terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) staining
The TUNEL assay was carried out as described previously (41) with slight modification. 3μl of infected blood containing 1-10% parasitemia were collected during trophozoite stage and fixed in 1 in 4 diluted BD Cytofix™ Fixation Buffer (BD Biosciences, Franklin Lakes, NJ) for at least day until they were needed. Each sample was then washed twice with MT-PBS, and adhered to a glass slide pre-coated with 0.1% polyethylenimine (PEI) for 10 minutes at room temperature. The excess cells were washed and incubated overnight at room temperature with TUNEL labelling solution (1mM Cobalt Chloride, 25mM Tris-HCl pH 6.6, 200mM sodium cacodylate, 0.25mg/ml BSA, 60uM BrdUTP, 15U Terminal transferase). The slides were washed three times, followed by staining with 50μg/ml of anti-BrdU-FITC antibody (Novus Biologicals, Littleton, CO) in MT-PBT (MT-PBS, 0.5% BSA, 0.05% Triton X-100) for 1 hour. The slides were then washed three times with MT-PBS and mounted with SlowFade® Gold antifade reagent with DAPI (Thermo Scientific, Waltham, MA) and sealed. When the slides were dried, they were examined using Axioplan 2 fluorescence light microscope (Carl Zeiss, Oberkochen, Germany) between 600x to 1000x magnification. At least 100 DAPI-positive cells were counted, and each was graded as either positive or negative for TUNEL staining, as an indication of DNA fragmentation.
In vivo erythrocyte tracking (IVET) assays
The IVET assay was carried out as previously described by Lelliott, Lampkin (74), (75). Briefly, at least 1.5ml whole blood was collected from mice of each genotype via cardiac puncture, followed by staining with either 10μg/ml of Atto 633 or 125μg/ml of EZ-LinkTM Sulfo-NHS-LC-Biotin (Biotin) (Thermo Scientific, Waltham, MA) for 45 minutes at room temperature, followed by washing two times with MT-PBS. The blood was mixed in two different dye combinations to correct for any dye effects. Approximately 1×109 erythrocytes were injected intravenously into infected wild-type mice at 1-5% parasitemia during schizogony stage. Blood samples were collected at various timepoints, from 30 minutes up to 36 hours after injection, and analysed using flow cytometry. The ratio of infected labelled erythrocytes was determined, as an indication of the relative susceptibility of RBCs to P. chabaudi infections. The proportion of labelled blood populations was also tracked over time to determine the clearance of these RBCs from the circulation.
Flow cytometry analysis of blood samples
For erythrocyte lifetime assays, 2μl of whole blood samples were stained with 2μg/ml streptavidin-PE-Cy7, 1μg/ml anti-CD71-allophycocyanin (APC) (clone R17217), 1μg/ml anti-CD45-APC eFluor 780 (clone 30-F11) (eBioscience, San Diego, CA), 4μΜ Hoechst 33342 (Sigma-Aldrich, St Louis, MO) and 12μΜ JC-1 (Thermo Scientific, Waltham, MA) in MTRC. The samples were washed once with MTRC and further stained with 2μg/ml streptavidin-PE-Cy7 to capture all biotin-labelled cells. Immediately prior to analysing on flow cytometer, 5μl of 123count eBeads (eBioscience, San Diego, CA) was added to determine the relative anaemic levels.
For both malaria infections and IVET assay, 2μl of whole blood samples were stained with 2μg/ml streptavidin-PE-Cy7 (only for experiments with biotinylated erythrocytes), 1μg/ml anti-CD45– allophycocyanin (APC)–eFluor 780 (clone 30-F11), 1μg/ml anti-CD71 (TFRl)–PerCP–eFluor 710 (clone R17217) (eBioscience, San Diego, CA), 4μΜ Hoechst 33342 (Sigma-Aldrich, St Louis, MO) and 12μΜ JC-1 (Thermo Scientific, Waltham, MA) in MTRC. All samples analysed through flow cytometry were performed on BD LSRFortessa (BD Biosciences, Franklin Lakes, NJ), where 200,000 to 1,000,000 events were collected and visualized on FACSDiva™ and FlowJo software. RBCs were identified by gating on CD71 negative and CD45 negative population, followed by gating on Atto-labelled and Biotin-labelled erythrocytes on appropriate channels (APC for Atto-633, PE for Atto-565 and PE-Cy7 for Biotin). The parasitemia of each labelled erythrocyte population was determined by gating on Hoechst 33342 positive and JC-1 positive population.
Statistical analysis
The statistical analysis was carried out as described in previous study (41). The LOD score method coupled with Bonferroni correction was used to determine the causative mutation for MRI96570 and MRI95845. The statistical significance of the malaria survival was tested using the Log-Rank test. The statistical significance of parasite infection was determined via the statmod software package for R (http://bioinf.wehi.edu.au/software/compareCurves) using the 'compareGrowthCurves' function with 10,000 permutation, followed by adjustments for multiple testing. The statistical significance for the ratios of IVET assays was determined using the one sample t-test with hypothetical mean of 1. For the rest of the results, statistical significance was determined using two-tailed Students t-tests.
Authors' Contributions
H.M.H., D.C.B., P.M.L., M.W.A.D, L.T., B.J.M, S.J.F. and G.B. designed and planned the experimental work. H.M.H., D.C.B. and G.B. performed the research. H.M.H., D.C.B., P.M.L., M.W.A.D, L.T., B.J.M, S.J.F. and G.B. interpreted and analysed the data. H.M.H., D.C.B. and G.B. performed statistical analysis. H.M.H, P.M.L., G.B., B.J.M. and S.J.F. wrote the manuscript. All authors reviewed the manuscript.
Acknowledgement
We would like to acknowledge Shelley Lampkin and Australian Phenomics Facility (APF) for the maintenance of the mouse colonies. We are also grateful for the assistance of the Microscopy Unit of the Macquarie University in the sample preparation and operation of the scanning electron microscope. This study was funded from the National Health and Medical Research Council (NHMRC) of Australia (Program Grant 490037, and Project Grants 605524 and APP1047090), the National Collaborative Research Infrastructure Strategy (NCRIS), the Education Investment Fund from the Department of Education and Training, the Australian Phenomics Network,