Assessment of the in vitro activity and selectivity of Artemisia afra and Artemisia annua aqueous extracts against artemisinin-resistant Plasmodium falciparum

Abstract

Background

See More

The recent emergence of artemisinin resistance in Africa is drawing scrutiny toward the use of alternative anti-malarial therapy based on Artemisia annua and Artemisia afra phytotherapies. This study aimed to determine if either A. annua and A. afra extracts are active against artemisinin-resistant Plasmodium falciparum isolates and determine the selectivity of inhibitory phytotherapies.

Methods

Artemisia extracts were tested in vitro to mimic parasites exposure to extracts in population drinking Artemisia sp. teas. Artemisia extracts were tested in Ring Stage Survival Assays (RSA^{0−3 h}) against Cambodian clinical isolates previously genetically and phenotypically characterized as artemisinin resistant or sensitive. Primary human hepatocytes and a human hepatoblastoma cell line (HepG2 cells) were used to assess the cytotoxicity of Artemisia extracts.

Results

The study revealed a substantially decreased in vitro activity of A. annua extracts when tested on artemisinin-resistant parasites mutated in the Pfkelch13 gene (RSA₅₀ 0.137–2.56 g.L⁻¹) compared to artemisinin-sensitive parasites (RSA₅₀ 0.080 g.L⁻¹). Conversely, the A. afra extracts have a similar activity on the isolates tested whether they are sensitive or resistant to artemisinin (RSA₅₀ 0.537–0.758 g.L⁻¹) However, the selectivity index for A. afra extracts was much lower than for A. annua extracts (A. afra: 4.628, 4.305 and 6.076 vs A. annua: 387.625, 226.350 and 12.099, respectively for WT, C580Y and R539T).

Conclusions

Artemisia annua activity is driven by artemisinin, implicating the same resistance profiles and concerns associated with semisynthetic artemisinin derivatives. Artemisia afra showed artemisinin-independent antiplasmodial activity. However, the molecular basis of this activity is unknown and may not present a sufficient selectivity, thus further characterization of A. afra is essential.

Background

The emergence of Kelch13 (K13) mediated artemisinin resistance in Plasmodium falciparum is now largely documented in Africa [1,2,3,4,5,6,7]. Although this resistance is only partial and does not compromise artemisinin-based combined therapy (ACT) endpoint efficacy, it could be a first step in the acquisition of multiresistance. The genetic analysis of the artemisinin-resistant strains collected so far in Africa has revealed a local emergence as opposed to an importation of Asian-originated parasites [8]. Although the basis of this recent evolution is still debated, the off-label uses of artemisinin(s) monotherapies are suspected to trigger this emergence [9]. In parallel to the inadequate use of commercial treatments, several malaria control stakeholders have raised concerns about the relation between broad use of Artemisia sp. infusions in sub-Saharan Africa and the selection of resistant parasites [9]. This problematic is notably reinforced by the inherent unrealistic standardization and observance of a homemade therapy and by the lack of concrete in vivo efficacy data in the literature [9,10,11]. The two main Artemisia species used in Africa are Artemisia annua and Artemisia afra. Both species contain artemisinin, at relatively high concentration for A. annua but substantially lower for A. afra [12]. In the absence of clear characterization, it could then be assumed that both might participate to the selection of resistant parasites. Previous data that showed artemisinin can readily select K13 mutant parasites reinforce this possibility [13, 14]. These hypotheses are discussed by the proponents of these phytotherapies, who notably argue that Artemisia sp. infusions consist in a poly-chemotherapy that would limit the selection of drug resistant parasites [15]. This is notably justified by data from a rodent malaria model showing that oral consumption of A. annua dried leaves is more active than pure artemisinin. In addition, these studies have showed slower acquisition of artemisinin resistance [16], greater reduction of parasitaemia [16,17,18,19] and higher bioavailability of artemisinin compared to mice receiving the corresponding dose of artemisinin [20,21,22]. However, the mechanisms of resistance in rodent malaria parasites could present discrepancies with P. falciparum artemisinin resistance, such as K13-independent mechanisms [23].

The overall aim of this study is to determine to which extend artemisinin-resistant K13 mutant parasites present a better survivability upon A. afra and A. annua exposure, to evaluate the selection risk through in-population usage of these phytotherapies.

Methods

Plasmodium falciparum clinical isolate collection and culture

Clinical isolates were collected between 2014 and 2020 in Cambodia and adapted to in vitro continuous culture at 2% haematocrit (O + human blood, Centre de Transfusion Sanguine, Phnom Penh, Cambodia) in RPMI-1640 medium supplemented with 0.5% (w/v) albumax II, 2.5% (v/v) decomplemented human plasma (mixed serogroups) under an atmosphere of 5% CO₂ and 5% O₂ and kept at 37 °C. These isolates were previously characterized for their genotype (k13 propeller domain sequence) [14] and their in vitro susceptibility to dihydroartemisinin (DHA) using the ring stage survival assay (RSA^{0−3 h}) [24]. The selected sample set included 12 strains presenting a wildtype k13 haplotype (WT group), 6 strains with the C580Y mutation (C580Y group), and 6 strains with the R539T mutation (R539T group) (Table 1).

Table 1 Characterization of Plasmodium falciparum strains used

Preparation of Artemisia extracts

Artemisia afra (voucher: LG0019528 Université de Liège) and A. annua (voucher: MNHNL 17733 Herbarium Luxembourg) were collected as leaves and twigs and preserved in their original packaging at room temperature. Both products were protected from sunlight until Artemisia extract preparation. The stock extract was prepared as follows: 5 g of Artemisia dried leaves and twigs were poured in 100 mL of pre-boiling molecular grade water and boiled for 5 min at 100 °C. The extract was allowed to cool for another 10 min at room temperature and then centrifuged at 3000 rpm for 10 min to pellet down the plant debris and fine solids. The supernatant was first filtered through a 40 µm cell strainer (Falcon, Corning Brand) and then through a 0.20 µm membrane filter (CA-Membrane), generating an extract stock (50 g.L⁻¹). The stocks of extracts were stored at −20 °C. The quantification method of artemisinin in extracts is described in Ashraf et al. [25]. The artemisinin concentration in 50 g.L⁻¹ extract stocks was 10.75 µM for A. annua and 251 nM for A. afra.

Evaluating anti-malarial activity of Artemisia extracts

Parasites susceptibility to Artemisia extracts was determined using RSA^{0−3 h} according to Witkowski et al. [24]. Briefly, tightly synchronized parasites (rings 0–3 h post-invasion) were obtained by 75% Percoll centrifugation followed by sorbitol treatment and adjusted between 0.5 and 1% parasitaemia. Parasites were exposed to a range of increasing concentration of either A. annua or A. afra for 6 h, before washout. Twofold dilution series from A. annua and A. afra stock solutions (50 g.L⁻¹) were prepared in molecular grade water. Each concentration tested was diluted ten times when added to the culture media, making a 10% final extract concentration in the media. A carrier-control consisted of 10% molecular grade water in media. After 72 h, parasitaemia was measured by counting 1 × 10⁵ red blood cells per sample condition via microscopy. The survival rate was determined as a ratio of parasitaemia in the drug-exposed condition versus the drug-free condition.

Cytotoxicity assessment of Artemisia extracts using primary human hepatocytes

Primary human hepatocytes (BioIVT, lot BGW) were thawed into InVitroGro™ CP Medium (BioIVT, Cat# Z99029) containing a 1 × antibiotic mixture (PSN, Gibco Cat# 15,640,055 and Gentamicin, Gibco, Cat# 15,710,072). Cell viability was recorded using trypan blue exclusion on a haemocytometer, and 18,000 live cells were added to each well of a collagen-coated 384-well plate (Grenier, Cat# 781,956) and maintained at 37 °C and 5% CO₂. Primary human hepatocytes were treated with Artemisia extracts for 6 or 24 h at 7 days post-seeding. Twofold dilution starting from 5 g.L⁻¹ were used to assess cytotoxic effect. Control consisted of 10% molecular grade water in media. After 66 (6 h treatment) or 48 (24 h treatment) additional culture hours following drug removal, hepatocytes were fixed for 1 h with 4% paraformaldehyde (PFA) (Thermo Fisher Scientific) in PBS (Gibco), washed twice with PBS, and stained with 1 μg mL⁻¹ Hoechst 33,342 (Thermo Fisher Scientific, Cat# H3570) to detect hepatic nuclear DNA. Hepatic nuclei were imaged and quantified using a 4 × objective on a Lionheart FX automated microscope (Biotek^®).

Cytotoxicity assessment of Artemisia extracts using HepG2

The HepG2 human hepatocyte line from hepatocellular carcinoma (ATCC HC-8065) was cultured in rat collagen I-coated (5 µg.cm⁻²) flasks (Corning) at 20–90% confluence in DMEM without sugar (Lonza catalog 11,966–025) supplemented with 4.5 g.L⁻¹ glucose (Lonza), 1 mM Sodium Pyruvate (Lonza), 2 mM L-glutamine (Gibco) and 10% fetal bovine serum (Hyclone) in a cell culture incubator at 37 °C and 5% CO₂. Cells were passed by treating with Trypsin LE (Gibco) for 7 min at 37 °C. Toxicity assays were performed in DMEM with glucose replaced with galactose to avoid false negative results due to potential Crabtree effect [26]. Cells were harvested from a flask, counted on a haemocytometer using trypan blue, diluted in DMEM prepared as above, but with 10 mM galactose instead of glucose, and seeded at a density of 2000 live cells in 40 µL per well into rat-tail collagen I-coated 384-well plates (Greiner Bio-one) using a Biomek NX (Beckman Coulter).

Twofold dilution series from A. annua and A. afra stock solutions (50 g.L⁻¹) were prepared in molecular grade water. Each concentration tested was diluted ten times when added to the parasite media, making a 10% final extract in the media. A carrier-control consisted of 10% molecular grade water in media. Assays were performed in two 72 h formats: one in which cells were treated for 24 h before a drug washout with media and 48 h of additional culture, and the other with treatment for the entire 72 h. At 72 h, media was removed and cells were fixed with 4% paraformaldehyde (Thermo Fisher Scientific) in PBS and then stained with 10 µg.mL⁻¹ Hoechst 33,342 for 1 h. The entire culture area of each well of assay plates were imaged with a Lionheart FX automated microscope (Biotek^®) with a 4 × objective, and net hepatic nuclei per well quantified.

Selectivity index determination

The selectivity index (SI) is the ratio that measures the cytotoxicity of the extracts and their anti-malarial activity. It was calculated for the 6 h treatment of P. falciparum clinical isolates with this formula SI = (PI₅₀_hepatocytes_6 h/RSA₅₀_parasites_6 h)) for A. afra and A. annua for the WT, C580Y and R539T groups.

Statistical analysis

All analyses were performed using GraphPad Prism (v7.00). A p value < 0.05 was considered significant. IC_{50 s} were calculated using Quest Graph™ IC₅₀ Calculator (https://www.aatbio.com/tools/ic50-calculator) or GraphPad Prism (v7.00).

Results

The in vitro activity of A. annua (Fig. 1A) and A. afra (Fig. 1B) extracts was measured against three groups of P. falciparum clinical isolates, designed out of their K13 profile (wild-type, C580Y and R539T), using RSA^{0−3 h}. From the RSA^{0−3 h} values obtained for each concentration tested, an inhibitory concentration of 50% of the survival rate (RSA₅₀) after exposure was determined. The median RSA₅₀ values (± standard deviation) of A. annua extracts were 0.080 g.L⁻¹ [± 0.064], 0.137 g.L⁻¹ [± 0.268] and 2.563 g.L⁻¹ [± 2.679] for parasites WT, C580Y and R539T, respectively. The RSA₅₀ values from artemisinin-resistant parasites harbouring C580Y and R539T profile were significantly higher compared to wildtype artemisinin-sensitive parasites (WT) (p = 0.0289 and p = 0.0071 respectively for C580Y and R539T, Mann Whitney test, Fig. 1A). The concentration range of A. annua extracts used was insufficient to decrease parasitaemia to an undetectable level making impossible to determine a minimal inhibitory concentration (MIC) value.

Fig. 1

Ring-stage survival assay (RSA^{0−3 h}) following treatment with Artemisia annua and Artemisia afra. The susceptibility of clinical isolates (WT group, C580Y group and R539T group) to A A. annua and B A. afra measured with a RSA^{0−3 h}. Each dot represents the mean survival of isolates at a given concentration. Error bars represent the SEM. Tables below graphs summarize the median RSA_{50 s} and the statistical analysis performed (Mann Whitney test)

The RSA₅₀ values of A. afra extracts were 0.705 g.L⁻¹ [± 0.185], 0.758 g.L⁻¹ [± 0.259] and 0.537 g.L⁻¹ [± 0.155] for WT, C580Y and R539T parasite groups respectively (Fig. 1B). No significant differences were observed between wildtype versus K13-mutant parasites (p = 0.7392 and p = 0.9860 respectively for C580Y and R539T, Mann Whitney test; Fig. 1B). The concentration range for A. afra extracts resulted in a drop in parasitaemia to an undetectable level for all the strains tested and, therefore, allowed for a MIC determination ranging between 1.25 to 2.5 g.L⁻¹.

Individual RSA₅₀ values were then used to evaluate the cross resistance of Artemisia extracts with DHA. The results showed a positive correlation between A. annua extract RSA₅₀ and RSA^{0−3 h} values (r = 0.4112, p = 0.0459, Pearson correlation test; Fig. 2A) while no correlation was observed for A. afra extracts (r = − 0.1975, p = 0.3549, Spearman correlation test; Fig. 2B).

Fig. 2

In vitro activities of Artemisia annua and Artemisia afra extracts correlate with survival to dihydroartemisinin treatment. RSA_{50 s} of clinical isolates (WT group, C580Y group and R539T group) extrapolated from their survival during RSA^{0−3 h} with A. annua and A. afra correlate with the survival percentage in an RSA^{0−3 h} with a DHA treatment of 700 nM. A A. annua RSA_{50 s} show a positive moderate correlation with RSA^{0−3 h} _DHA values (r = 0.4112, p = 0.0459, Pearson correlation test), B A. afra RSA_{50 s} do not correlate with RSA^{0−3 h} _DHA values (r = −0.1975, p = 0.3549, Spearman correlation test). Each point was labelled according to Table 1 on both panel A and panel B

The hepatotoxicity of the Artemisia extracts was evaluated by treating HepG2 cells, gold standard for hepatotoxicity assessment, in both a 24 h washout and 72 h assay (Fig. 3). Regardless of assay mode, the results showed a similar level of cytotoxicity between A. annua and A. afra. The concentration causing 50% of proliferation inhibition (PI), PI₅₀’s for A. afra extracts from 24 and 72 h assays were 2.025 g.L⁻¹ and 1.855 g.L⁻¹, respectively, while the PI₅₀’s for A. annua were 2.406 g.L⁻¹ and 2.260 g.L⁻¹ for 24 and 72 h assays, respectively. In addition, no significant difference between the PI₅₀ of A. annua and A. afra for the 24 or 72 h treatment was observed (24 h, p = 0.4395 and 72 h, p = 0.4890, Unpaired-t test).

Fig. 3

Cytotoxicity studies of Artemisia afra and Artemisia annua tea extracts on HepG2 cells. Percentage of proliferation inhibition (PI) compared to non-treated condition after 72 h of culture with 24 or 72 h of drug exposure on HepG2 cells after 24 h seeding. N = 2 independent biological replicates and N = 2 technical replicates for each condition and biological replicate. PI_{50 s} for A. afra extracts were 2.025 g.L⁻¹ and 1.855 g.L⁻¹ for 24 and 72 h exposure, respectively. PI_{50 s} for A. annua extracts were 2.406 g.L⁻¹ and 2.260 g.L⁻¹ for 24 and 72 h exposure, respectively. No significant difference between the PI₅₀ of A. annua and A. afra for the 24 or 72 h treatment was observed (24 h, p = 0.4395 and 72 h, p = 0.4890, Unpaired-t test). Similar results after 24 and 72 h of treatment show that the compounds are cidal. Each dot represents the mean proliferation inhibition percentage of isolates at a given concentration. Error bars represent the SEM

The selectivity of the Artemisia extracts against P. falciparum was then evaluated by treating primary human hepatocytes cultures with extracts for 6 h and 24 h. The results showed that exposure to A. annua extracts presented low cytotoxicity at the highest concentrations for the exposure duration tested. The PI₅₀ was determined to be greater than 5 g.L⁻¹, with an estimated to be 31.010 g.L⁻¹ and 26.900 g.L⁻¹ for 6- and 24-h exposure, respectively (Fig. 4A). In contrast, A. afra extracts showed cytotoxicity at higher test concentrations, resulting in PI₅₀’s of 3.263 g.L⁻¹ and 2.124 g.L⁻¹ for 6 and 24 h exposure, respectively (Fig. 4B). The selectivity index (SI) for the 6 h treatment of P. falciparum clinical isolates (using the formula SI = (PI₅₀_hepatocytes_6 h/RSA₅₀_parasites_6 h)) for A. afra was 4.628, 4.305 and 6.076 for WT, C580Y and R539T groups, respectively, whereas for A. annua it was higher, with 387.625, 226.350 and 12.099 for WT, C580Y and R539T groups, respectively (Table 2).

Fig. 4

Cytotoxicity studies of Artemisia annua and Artemisia afra tea extracts on primary human hepatocytes. Percentage of proliferation inhibition (PI) compared to non-treated condition after 72 h of culture with 6 or 24 h of drug exposure on hepatocytes 7 days after seeding. N = 4 independent biological replicates for A. annua and N = 5 independent biological replicates for A. afra. For each biological replicate, 3 to 6 technical replicates were made. Means of proliferation inhibition percentage of hepatocytes with SEM are represented on the graphs. PI_{50 s} for A. afra extracts were 3.263 g.L⁻¹ and 2.124 g.L⁻¹ for 6 and 24 h exposure, respectively. PI_{50 s} for A. annua were undetermined (PI₅₀ > 5 g.L⁻¹), but were estimated by Graphpad prism7 to 31.010 g.L⁻¹ and 26.900 g L⁻¹ for 6 and 24 h exposure, respectively

Table 2 Selectivity index (SI) for 6 h Artemisia annua or Artemisia afra exposure of Plasmodium falciparum clinical isolates

Discussion

The results obtained here showed that the A. annua extract is largely less effective against the artemisinin-resistant P. falciparum isolates tested here compared to the artemisinin-sensitive isolates. Additionally, a cross resistance with DHA was observed. These data suggest the in vitro activity of A. annua extracts is mainly driven by its content in artemisinin and, therefore, might trigger the selection of resistant parasites.

This finding is consistent with previous studies [27,28,29] and notably Czechowski et al. [27] that showed that A. annua with a knocked-out artemisinin synthesis pathway lost its anti-malarial activity. Consequently, there is a non-negligeable risk in the use of A. annua as anti-malarial treatment surrogate leading to artemisinin resistance selection. Furthermore, these in vitro data do not suggest pharmacological benefits of A. annua phytotherapies over ACT. Several studies on animal models suggest that the therapeutic interest of A. annua lies in increased artemisinin plasmatic concentration due to co-present molecules, such as chrysosplenetin [15, 30]. However, this potential advantage should be put in perspective with human clinical data showing that neither susceptible nor resistant parasites clearances were achieved by higher artemisinin derivative concentrations [31].

The data obtained with A. afra are contrasting and do not support a potential role for artemisinin resistance selection. A. afra extracts presented a notable in vitro activity regardless of the K13 genotypes tested and showed no cross resistance with dihydroartemisinin. These findings are in accordance with the very low artemisinin content in A. afra and rather suggest the presence of potent, non-endoperoxide anti-malarial pharmacophores absent in A. annua. However, these results only reflect an activity based on an aqueous extraction and might not represent the whole extent of the anti-malarial activity. These likely differences were evidenced by Kane and colleagues showing the level of phytochemical compounds of A. afra is modulated by the origin of the plant but also to the extraction solvent chosen and subsequent the intrinsic anti-malarial activity as well [10]. Several studies have assumed a possible anti-malarial activity of Artemisia sp. related to flavonoids [15], but the nature of the anti-malarial activity of A. afra aqueous extracts remains an open question. Moreover, the in vitro anti-malarial activity showed here occurred at high doses unlikely to be reached out of reported tea consumption [9, 11, 32].

In addition, the absence of published data on the therapeutic efficacy of A. afra does not enable discussing this point further. This important in vitro anti-malarial activity of A. afra could be explained by pharmacophore(s), but that lack specificity. HepG2 cells, considered as the gold-standard in vitro model for hepatotoxicity, were used to verify the toxicity level of Artemisia extracts. A cytotoxic effect of both A. annua and A. afra extracts was observed. However, it is relevant to note that at least the toxicity of A. annua was expected since HepG2 cells have already been described as artemisinin sensitive [33]. Therefore, a second model was chosen to assess the selectivity of Artemisia extracts. Thus, cytotoxicity data on primary human hepatocytes showed a much more marked cytotoxicity of A. afra aqueous extracts than those prepared with A. annua, regardless of the duration of treatment. Discrepancies with A. annua on primary human hepatocytes might be due to the anti-tumoral activity of artemisinin as previously observed [34, 35].

The cytotoxicity of these A. annua and A. afra extracts was previously tested on primary human and monkey hepatocytes [25]. In this report, these extracts presented low toxicity, contrasting with the results in this study. These differences might be explained by inter-donor variability within both studies [36]. Either way, these data highlight a toxicity concern, that might be different when observed in vivo after per os consumption of Artemisia than was observed in vitro on direct hepatocytes exposure. However, toxicity has already been noted in some case reports describing adverse events related to Artemisia tea consumption [37, 38]. This toxicity could be due to the same compounds found in other Asteraceae such as Artemisia absinthium or Artemisia dracunculus that are notoriously hepatotoxic at high doses [39]. At this stage it is still unclear if the molecule(s) responsible for A. afra anti-malarial activity are the same that exert cytotoxicity, but the narrow in vitro selectivity indices (4.048 to 6.423) question the safety of its use. These observations advocate for conducting proper pharmacovigilance studies within population using A. afra.

Conclusion

In summary, this study raised several concerns regarding Artemisia phytotherapies. The use of A. annua as anti-malarial surrogate could trigger the selection and facilitate the emergence of artemisinin resistant parasites. Further, based on these in vitro results, A. afra exhibits toxicity issues, despite a possible content in anti-malarial pharmacophores that are unlikely to trigger selection of artemisinin resistance. All these aspects associated with the absence of factual efficacy data emphasize the need for additional investigations, but also lead to renewed reservation about anti-malarial phytotherapies.