Organism-like formation of Schistosoma hemozoin and its function suggest a mechanism for anti-malarial action of artemisinin

The current theories of antimalarial mechanism of artemisinin are inadequate to fully explain the observed effects. In our study, “organism-like” formation of Schistosoma hemozoin granules by attaching to and utilizing erythrocytes to form new ones was observed. This indicates that heme iron is transferred from erythrocytes to hemozoin granules during their formation. However, as a disposal product of heme detoxification, these granules are not completely expelled from the Schistosoma gut, but decomposed again between microvilli in the posterior portion of the gut to transfer iron to eggs. Based on the function of iron transport supported by our observation of the unique process of Schistosoma hemozoin formation, here we propose a new viewpoint of antimalarial mechanism of artemisinin, which emphasizes the final outcome, i.e., interference of iron utilization in parasites by artemisinin, instead of focusing on the mode of interaction between artemisinin and heme or hemozoin. This suggests that artemisinin and its endoperoxides derivatives likely hit the Achilles’ heel of hemozoin-producing and iron-dependent organisms.

make this problem very controversial. Although hypotheses such as the reactive species hypothesis involving oxygen and C-centred radicals or carbocations, and the heme (Fe) activation hypothesis (or heme model) are related, they are also different from each other. At present, there is no decisive evidence pointing towards the absolute validity of any of the proposed theories relating to a direct effect on the parasite. The heme model, which is one of the major theories, has received wide acceptance to explain the mechanism of action of artemisinin. In this model, heme from haemoglobin degradation interacts with artemisinin. Iron in the heme directly reduces the peroxide bond in artemisinin, thereby generating reactive oxygen radicals that damage the parasite, thereby leading to its death 16 . However, evidences from some studies do not support this model. For example, C-centered radicals are too short-lived to favor intermolecular interactions 17 . Furthermore, the parasiticidal effects of artemisinin on early ring-stage malaria parasites with little heme or hemozoin, also argue against heme-dependent activation theory 18 . These observations weaken support to the heme model. On the other hand, they suggest that previous models may be missing a key part of the antimalarial mechanism of artemisinin.
Although there is no consensus about the mechanism of action through which artemisinin kills malaria parasites, several lines of evidence show that artemisinin exerts antimalarial action by radical formation which depends on the endoperoxide bridge. During the process, the interaction of iron in heme with the endoperoxide bridge has been confirmed to generate reactive oxygen radicals. Most researchers have paid much more attention to the radicals and their damages in mechanism of action of artemisinin, but not on the fact that heme is consumed by artemisinin and hemozoin formation is interfered. Since hemozoin is a disposal product formed during detoxification of free heme, it should be removed from malarial parasites or Schistosoma guts in a timely manner. However, hemozoin has always been found to accumulate in malarial parasites or worm guts. Furthermore, hemozoin nearly accompanies their overall mature stages, like a cell "organ" in the malarial parasite cell. Why is it not excluded completely from the cell? It is not easy to investigate the characteristics and rules of hemozoin formation within a malarial parasite. Moreover, the frequently-studied synthetic analogs of malarial hemozoin, such as β -hematin, do not easily reveal hemozoin-related biological processes either. However, Schistosoma hemozoin, which is very similar to malarial hemozoin, is formed in the gut and can be observed in situ or conveniently obtained for in-depth analyse of the characteristics and potential function of hemozoin. In fact, unique process of hemozoin formation suggests its unexpected role in vivo, which can improve our understanding of the formation and potential role of malarial hemozoin.

Results and Discussion
Morphological characteristics of Schistosoma hemozoin granules. In this study, the morphological characteristics of Schistosoma hemozoin granules (SHGs) were observed in different ways to analyze hemozoin formation. SHGs from the gut contents of females showed a characteristic dark brown colour under light microscopy (LM) (Fig. 1a,b), which served as a natural and effective identification mark. Under transmission electron microscopy (TEM), these electron-dense granules appeared to be globular or comma-shaped, with tail lengths varying from 0.2 μ m to 1 μ m ( Fig. 1c and Supplementary Fig. 1). Field emission scanning electron microscopy (FESEM) and TEM showed that electron density in the SHG tail was different from that in its head ( Fig. 1d-f). Moreover, in the presence of SDS, the tail of hemozoin granule is easily dissolved ( Supplementary Fig. 2), indicating that SHG is a complex of hemozoin and lipids. In addition, SHGs shared the same absorption peak at 400 nm with malarial hemozoin (Fig. 1g), suggesting that they have similar chemical compositions. Since SHGs are crystals of heme, they are undoubtedly iron-containing granules, which was confirmed in our previous study 19 .
To investigate the interaction between erythrocytes and hemozoin granules, we observed erythrocytes and hemozoin granules in different portions of the digestive tract of S. japonicum with various methods (Fig. 2). Spit from the mouth of female worms showed that the recently-ingested erythrocytes retained their typical biconcave disc shapes (Fig. 2b). FESEM images revealed that some erythrocytes were "punctuated" with SHGs ( Fig. 2e), revealing the original interaction between erythrocytes and SHGs. In the gut section between the mouth and the ovary, some erythrocytes lost their typical morphology and were 'wrapped' by dark brown SHGs (Fig. 2c). FESEM showed that SHGs with various sizes 'embedded' themselves in the damaged erythrocytes ( Fig. 2f-k). Moreover, some 'wrapped erythrocytes' disappeared and left a group of SHGs with the outline of an erythrocyte (Fig. 2h,j,k). By contrast, the gut section after the ovary and adjacent to the vitelline glands only contained individual dark brown granules or clusters of granules (Fig. 2d,l). Based on the abovementioned observations, the issue is whether the SHGs destroy erythrocytes only by attaching to them or that erythrocytes are attached to and utilized by SGHs to form new SHGs.

Interaction between SHGs and erythrocytes.
To reveal the interaction between SHGs and erythrocytes, we incubated a mixture of erythrocytes and individual SHGs in vitro (Fig. 3a). After 12 h of incubation, we found that SHGs attached to nearly all erythrocytes (Fig. 3b), confirming that SHGs are able to attach to the cells. After 24 h of incubation, tiny SHGs gradually appeared at the surface of erythrocytes or at the bottom of culture plates, and their colour turned browner after 36 h of incubation (Fig. 3c,d). If the mixture did not contain sufficient erythrocytes or heme, SHGs formed slowly or stopped forming. However, after more erythrocytes or supernatant of broken erythrocytes (SBE; containing heme), rather than SHGs, were added, we observed more brown SHGs forming around the degrading erythrocytes (Fig. 3e,f), as in the Schistosoma gut. This process can be clearly observed in Supplementary video file 1. It showed that in the culture, erythrocytes were degraded after SHGs attached to the erythrocytes (Fig. 3g-i), and then new SHG granules were observed to form gradually along the cell membrane ( Fig. 3h-j). Simultaneously, we analysed the quantity of SHG in the mixture. We found that the quantity of SHGs significantly increased from 9.3 nmol to 10.7 nmol and from 9.9 nmol to 12.5 nmol after 2 and 4 days of incubation in chloroquine-sensitive reactions, respectively (Fig. 3q). We could even observe change in erythrocyte debris from a group of colourless debris with a few dark brown SHGs to a group with increasing number of dark brown SHGs, which marked the gradual process of utilization of erythrocyte debris to form more new SHGs ( Supplementary Fig. 3). The abovementioned observation strongly confirmed that the increase of SHGs depended on de novo assembly of free heme released from erythrocytes, rather than aggregation of pre-existing SHGs. Notably, in the culture, only the erythrocyte attached by SHGs can be utilized to form new ones by self-assembly. Malarial hemozoin has also been reported to self-assemble, depending on availability of pre-existing hemozoin 20 . This suggests that SHGs and malarial hemozoin form via similar mechanisms.
We then asked how SHGs become globe-or comma-shaped granules. To reveal this, we observed a series of changes of SHGs. We found that erythrocytes were degraded to debris, based on which globe-shaped SHGs assembled first. Then, the components of debris aggregated to the tail, which resulted in a comma-shaped granule. Supplementary Fig. 4 shows a process of formation from heme microcrystalline in erythrocyte debris to a comma-shaped granule. Thus, the formation of a globe-or comma-shaped SHG granule depends on the electron-dense part of erythrocyte debris. Morphological characteristics of malarial hemozoin. In malarial parasites, we found that, as in SHGs, hemozoin crystal also form from the electron-dense part of cell debris (Fig. 4a), indicating that hemozoin formation depends on the electron-dense part of cell debris. Notably, the electron-dense part of cell debris formed after the cell was attached and degraded by SHGs. It was derived from the cell, and then transformed to a specific form; only then did the hemozoin microcrystal begin to aggregate to form new hemozoin. We observed a clear difference between malarial hemozoin and SHGs in shape (Fig. 4a,b). Such difference likely arises from the different micro-environment in which hemozoin forms. The former is limited to the food vacuole in a cell and maintain a relatively stable state, like a store, whereas the latter forms in Schistosoma guts and enables attachment to erythrocytes. Supplementary Fig. 5 shows that the comma-shaped granules easily attach to various cells, suggesting a specific significance of a comma-shaped granule. In addition, aside from hemozoin crystal-like shape, other different shapes of malarial hemozoin are also observed ( Fig. 4a and Supplementary Fig. 6). Thus, it is possible that at different developmental stages of malarial parasites, these hemozoin show different morphologies. Although In our in vitro assay, SHGs did not occur spontaneously in erythrocytes or its debris, even if they contained lipids and heme. However, in the mixture of SGHs and erythrocytes, they can be utilized to form new SHGs. Obviously, erythrocytes plus SHGs meet the recognized requirements of hemozoin formation, including the following: (1) lipids 21 , which are likely from the cell membrane 22 ; (2) heme, which come from haemoglobin in erythrocytes; and (3) pre-existing SHGs, which provide seed to initiate the formation of new SHGs 20 . Notably, no enzymes or proteins were added to the culture, suggesting that no proteins or enzymes were necessary during the process, as in malarial hemozoin 20 . Interaction between SHGs and other cells aside from erythrocytes. Given that the formation of SHG only requires lipids, heme and pre-existing SHG as seed, we wondered whether cells, aside from erythrocytes, together with heme and SHGs, would likewise promote SHG formation. We found that SHGs attached to and degraded a number of non-erythrocyte cells in the Schistosoma gut, similar to the fate of erythrocytes attached by SHGs (Fig. 3k,l). To confirm this finding, we mixed SHGs with erythrocytes (or supernatant of broken erythrocytes by repeated freezing and thawing) and Saccharomyces cells in vitro. When the mixtures of Saccharomyces cells, SHG and erythrocyte supernatant were incubated, dark brown SHG also attached to Saccharomyces, thereby leading to the degradation of Saccharomyces cells and formation of more SHG (Fig. 3m-p and Supplementary Fig. 7a-d). Upon analysing the amount of SHG in the culture of Saccharomyces cells, SHG and erythrocytes, we found that the amount of SHG increased significantly from 1.3 nmol to 11.9 nmol and 37.5 nmol after 4 and 8 days of incubation, respectively (Fig. 3r). This further revealed that other cells, likely acting as another lipid resource, could also promote SHG formation in the presence of heme and SHGs. Furthermore, it clearly showed that different cells (lipids) could lead to the formation of different SHGs with different sizes and morphologies ( Supplementary Fig. 7). It also suggests that the morphology of hemozoin is determined by lipid resources or cell environments/conditions. Undoubtedly, in the food vacuole of malarial parasites, the hemozoin and the ingested erythrocyte debris constitute an appropriate system to produce hemozoin. Moreover, its special cell environments/conditions likely result in the special morphology of hemozoin.
The abovementioned results clearly showed the relationship or interaction between erythrocytes (or its debris) and hemozoin (crystal or granules), which further elucidated the model of hemozoin formation. Moreover, SHGs and malarial hemozoin require the same conditions for formation, such as heme and lipids and pre-existing hemozoin. Their similarity suggests that they have the same function. Notably, with the formation of hemozoin, iron (or heme) is undoubtedly transferred from erythrocytes to hemozoin. Moreover, the transfer is performed by a series of "organism-like" behaviors of SHGs, such as attachment to cells, cell lysis and self-assembly.
Unexpectedly, we found that hemozoin granules, which are formed in the anterior portion of Schistosoma gut (Fig. 4c3), are degraded again between microvilli in the posterior portion of the gut (Fig. 4c7 and Supplementary Fig. 8), as has been reported in our previous research 19 . Iron was detected in the vitelline gland, vitelline duct, ootype and eggs of female worms (Fig. 4c4-c6). In our previous study, iron was also found in the Schistosoma gut wall of male and female worms 19 . This observation strongly suggested that the worms need to obtain iron through the formation and degradation of SHG for nutrition and egg production. That is to say, iron is so essential for schistosomes that they have to adopt the unique way of collecting, transporting and utilizing it. It is not clear why the parasites transfer and utilize iron using hemozoin, but it is likely that when erythrocytes are degraded and utilized to form iron-containing heme crystals in situ, few free heme are released to the outside of the cells, as suggested by the colorless supernatant of the Schistosoma gut content after centrifugation. It is clear that iron is required during artemisinin activation and artemisinin is capable of killing schistosomes. Thus, the mechanism of action of artemisinin could involve iron, particularly its utilization in the worm.
The high similarity of malarial hemozoin and SHGs implies that malarial parasites require and utilize iron in hemozoin more than previously expected. It is known that many metabolic processes of the malaria parasite are dependent on iron, such as DNA synthesis, carbohydrate metabolism, proteolysis of host hemoglobin, mitochondrial function and electron transport. However, the significance of iron has not been actually recognized, despite the fact that iron chelators have been found to possess antimalarial activity 23,24 . This is because the relationship and interaction between iron utilization and hemozoin formation have not been recognized. Hemozoin has been considered a disposal product of free heme, so its potential function of hemozoin has been neglected. However, based on our observation of formation and degradation of hemozoin in schistosomes, we found that hemozoin is actually a vector or a pool of iron. Formation and degradation of hemozoin serves as a means to store, transport and utilize iron. The growth and development and reproduction of malarial parasites depend so much on iron that iron chelators can kill them. Thus, our findings offer an explanation for why these iron-dependent organisms need to form hemozoin. This is because that hemozoin acts as a safe intermediate to store and transport iron. It is obviously impossible to reach this conclusion simply by observing changes of hemozoin in malarial parasites. However, SHGs in the Schistosoma gut offer us a unique chance to understand the process or mechanism.
When artemisinin "consumes" heme 25,26 (or iron 27 ) and produces free radicals or heme-artemisinin adducts to inhibit heme crystallization 28 , even accelerating the breakdown of hemozoin 5 , it simultaneously inevitably interferes the normal formation and degradation of hemozoin, uses up iron by forming iron-artemisinin within the parasites, interferes or interrupts the transport and utilization of iron, thereby finally leading to death of the malaria parasites. More importantly, the antimalarial effect of iron chelators highlight the greater need and significance of iron in the survival of malaria parasites, compared with other organisms which are not sensitive to iron chelators. In particular, in hematophagous parasites only malarial parasites and schistosomes are capable of producing hemozoin, suggesting that they have a special dependence on iron. Their special demand for iron likely makes them more susceptible to iron deficiency. Artemisinin and its derivatives can deplete cell iron and kill iron-dependent cells 29,30 , and they can also kill ring forms and schistosomula, irrespective of hemozoin, in turn suggesting that malarial parasites and schistosomes are iron-dependent organisms. The parasitized erythrocytes were found to contain more iron than uninfected ones 31 . Moreover, the amount of iron increases as the malaria parasite develops from early ring to late schizont stage 31 . In addition, malarial parasites were reported to require serum iron sources in hosts, even iron deficiency in the host could protest from malaria infection 32 . These works strongly confirmed that malaria parasites actually store, need and depend on iron. For these hemozoin-producing organisms, if hemozoin formation is considered only as the waste of heme detoxification, long-term accumulation of large amounts of hemozoin in cells is difficult to explain. Instead, it is a rational explanation that hemozoin act as an iron reservoir or an iron vector, which continuously provide iron to meet the parasite's requirement. Indeed, it has been proposed that iron or heme-iron play a pivotal role in the mechanisms of action and toxicity of artemisinin 17,25 . Although artemisinin leads to various tissue damages or molecular effects, causing more widespread changes than iron chelators, the ultimate effect of artemisinin or its derivatives is likely on the interference in iron utilization. This hypothesis offers a better explanation for why artemisinin and its derivatives only act on very few types of organisms, such as schistosomes, malarial parasites and other iron-dependent cells, and why a ring form of malaria trophozoite without hemozoin is sensitive to artemisinin. Our hypothesis can reasonably explain most controversial results, whereas other views often cause much controversy. For example, recent proteomics analyses showed that many of the identified artemisinin target proteins were involved in various biological processes, including carboxylic acid metabolic process, cellular biogenic amine metabolic process, nucleoside metabolic process, DNA synthesis, protein synthesis, lipid synthesis, glycolysis and hemoglobin digestion, and so on 25,33 . Thus, some researchers even proposed that activated artemisinin kill the malaria parasite through a promiscuous targeting mechanism 25 . If so, it is more difficult to understand the mechanism of action of artemisinin. Undoubtedly, this view does not offer a reasonable explanation for why very limited kinds of organisms can be killed by artemisinin and its derivatives, and for what role the target proteins, such as PfEXP1, PfATP6, PfTCTP, Pfmdr1, K13-propeller protein and PfPI3K, could play independently in the mechanism of action of artemisinin.
Overall, these phenomena, such as the formation and degradation of SHGs in Schistosoma gut, transport of iron from guts to eggs, high similarity between SHGs and malarial hemozoin, accumulation of large amounts of malarial hemozoin in food vacuole, the antimalarial effect of iron chelators, and so on, support the rationality of our hypothesis. In addition, it is reasonable that in malarial parasites the alkylation of proteins or damage of cell organs, which are caused by artemisinin and its derivatives, also contribute to parasite killing.
Thus, research on the formation and degradation of SHGs in schistosomes improve our understanding on the significance of malarial hemozoin and the antimalarial mechanism of artemisinin. In particular, the conclusions in this study cannot be drawn simply by researching malarial parasites or the interaction of artemisinin and some proteins or tissue structures in malarial parasites, regardless of the function of hemozoin as iron vector. In addition, our opinion does not exclude previously recognized theories [5][6][7][8][9][10][11][12][13] , but instead reveals a hidden mechanism. Although our view still needs the support of more evidences, it undoubtedly has provided unprecedented insight into the mechanism of action of artemisinin. Furthermore, our theory also introduces a novel idea for understanding or revealing the mechanism of resistance of artemisinin. SHGs were collected by cutting females into small sections in sterile normal saline solution in eppendorf tubes and collecting the dark suspensions, which were centrifuged at 1,500 × g for 30 s and the worm tissues pellets were discarded. Then, the supernatant was re-centrifuged at 12,000 × g for 10 min. The resulting dark-brown pellets were washed thrice with sterile normal saline and observed or stored at − 20 °C before use.

Observation of SHGs from different sections of the gut in female adult worms. An intact female
adult worm was placed on a glass slide, and the spit from the mouth of female worms were collected. The worm was then cut into three sections, and the gut contents from each section were collected for observation using a light microscope (LM, Nikon 50i) and a field-emission gun environmental scanning electron microscope (FEG-ESEM, Quanta 200 F). Some were treated with 1-2.5% SDS for 30-60 min, then rinsed with double distilled water three times, finally dried at room temperature for observation using a transmission electron microscopy (TEM).

Quantification of SHGs.
SHGs obtained from the worm gut or from in vitro cultures was washed with 25 mmol/L Tris (pH 7.8) containing 2.5% SDS, followed by centrifugation at 12,000 × g for 10 min. This process was repeated until the supernatant was colorless and without spectrophotometric absorbance at 400 nm. The pellet was dissolved in 250 μ L of 2.5% SDS buffer with 20 μ L of 2.5 mol/L NaOH. Then, the amount of SHGs was quantified by measuring its absorbance at 400 nm using a Unico spectrophotometer.
In vitro incubation assay. SHGs granules were passed through a filter membrane with 3-or 5-μ m pores (Millipore). Then, about 10 nmol or 3 nmol of SHGs was incubated with different cells in 500 μ L RPMI-1640 culture system in 1.5 mL Eppendorf tubes for 2 d to 8 d at 37 °C as follows: (1)  2 × 10 3 to 10 × 10 5 per mL); (3) 100 μ L SHGs and 200 μ L erythrocytes (or supernatant of broken erythrocytes by repeated freezing and thawing) and 200 μ L Saccharomyces cells (final concentration of 0.5 to 5 × 10 3 per mL); (4) 300 μ L RPMI-1640 culture medium and 200 μ L erythrocytes (final concentration of 2 × 10 3 to 10 × 10 5 per mL). All conditions were kept sterile throughout the experiment. The quantity of SHGs was determined by spectrophotometry. Erythrocytes are stained with Giemsa stain (Sigma-Aldrich) according to Giemsa's Staining Protocol.

Observation of SHGs granules and malarial hemozoin by transmission electron microscopy (TEM).
Samples of SHGs granules collected from the gut of adult female worms or from the aforementioned cell incubations. Samples of malarial hemozoin collected from the cell culture dish or from liver tissues of infected mice. Every sample was first fixed with 2.5% glutaraldehyde-PBS (phosphate-buffered saline) buffer (pH 7.4) at 4 °C. The samples were then rinsed and stained with or without 1% uranyl acetate for 30 s to 60 s, and dried at room temperature for 5 min. The samples were examined by a JEOL EW-1230 transmission electron microscope at an accelerating voltage of 80 kV, and images were acquired using a digital photo-documentation system (Gatan Bioscan Camera, model 792).

Observation of SHGs granules by FEG-ESEM.
The SHGs specimens collected from the gut of adult female worms or from the aforementioned cell incubations were fixed with 2.5% glutaraldehyde-PBS buffer (pH 7.4) at 4 °C for 1 h to 2 h, then rinsed and fixed with 1% to 4% osmium tetroxide at 4 °C for 1 h to 2 h, and sequentially dehydrated in graded ethanol series. These steps were followed by critical-point drying in CO 2 and sputter-coating with gold-platinum using standard techniques. The incubated Saccharomyces cells and SHGs were directly observed with ESEM (Quanta 200 F) to observe the interaction between SHGs granules and living Saccharomyces cells.
FESEM can be used for high-resolution imaging. It can be switched between three vacuum modes, thus enabling the investigation of conductive, non-conductive, and high-vacuum incompatible materials. The high vacuum mode (typically at 10 −5 mbar), called the FEG mode in this study, is used for the imaging and microanalysis of conductive and/or conventionally prepared specimens. The low vacuum mode (at < 1.3 mbar) is for non-conductive specimens without preparation, whereas the ESEM mode (at < 26 mbar) for high-vacuum incompatible specimens. Our conventionally prepared specimens were investigated using the FEG mode, while the Saccharomyces cells incubated with SHGs were analyzed using the ESEM mode.

Prussian blue Iron Stain assay. The Iron Stain Kit is used to identify ferric iron deposits in a female
Schistosoma japonicum. Prussian blue iron stain assay is performed according to the previous paper 19 .

Statistical analysis.
Results are presented as mean ± standard deviation from at least three independent experiments. Statistical analyses were performed using one-way ANOVA and Student's t-test. A value of P < 0.05 was considered statistically significant.