Coordination of apicoplast transcription in a malaria parasite by internal and host cues

No genes are annotated as an apicoplast σ subunit in the Plasmodium genome databases. Using Escherichia coli σ⁷⁰ as the basis for a sequence similarity search, we predicted that PF3D7_0621000 encodes a nuclear-encoded apicoplast σ subunit in P. falciparum, which we named apSig (gene) and ApSigma (protein; SI Appendix, Fig. S2). σ⁷⁰ family σ subunits have conserved regions (regions 1.1 to 4.2) (8, 17), which include functional domains that interact with specific DNA promoter elements and the RNA polymerase core enzyme. All these conserved regions, except for region 1.1, occur in ApSigma (SI Appendix, Figs. S2 and S3). We also predicted the 3D structures of the conserved regions of ApSigma (SI Appendix, Fig. S4). Regions 2 and 4 were highly conserved in both primary amino acid sequence and predicted 3D structure (Fig. 1A and SI Appendix, Figs. S2 and S4). These regions are important domains of the sigma subunit that recognize the -10 and -35 regions of the promoter (18, 19). The conservation of these structural similarities in ApSigma suggests that it functions as a σ subunit. ApSigma has a long N-terminal extension, which was predicted to be a transit sequence for localization to the apicoplast. To test this prediction, we generated ApSigma antibodies and performed localization analysis, which confirmed that ApSigma localizes to the apicoplast in the trophozoite, schizont, and ring stages (Fig. 1B and SI Appendix, Fig. S5). Therefore, the long N-terminal extension may function as the apicoplast-targeting sequence. We wished to test whether ApSigma has activity as an RNA polymerase sigma factor. We previously attempted to knock out the P. falciparum apSig gene, but could not isolate the mutant. We interpret this as indicating that the apSig gene is likely to be essential for viability. Constitutive overexpression of the apSig gene was also unsuccessful, which might be due to toxicity of ApSigma protein overproduction. Furthermore, bacterial-overproduced ApSigma protein was insoluble, so we could not obtain a biochemically testable ApSigma protein. Since biochemical purification of the apicoplast RNA polymerase from Plasmodium sp. is not practical, we tested ApSigma activity through heterologous expression of a chimeric protein formed from ApSigma and the E. coli alternative sigma factor σ^S (SI Appendix, Fig. S6A). In E. coli, σ^S recognizes the katE promoter, which shares a common -10 promoter element with other consensus-type promoters, but is not recognized by other sigma factors (20). Thus, in σ^S mutant E. coli where katE is not expressed, katE promoter activity provides a tool to test for sigma factor activity. In our chimeric protein, parts of the -10 promoter element-recognition helix [regions 2.3-2.4, (21)] of σ^S was substituted with the corresponding amino acid sequence of ApSigma (SI Appendix, Fig. S6A). Using this, we found that the ApSigma peptide sequence responsible for promoter -10 region recognition can partially complement the activity of the corresponding sequence of E. coli σ^S (SI Appendix, Supplementary Experiment and Fig. S6B). This suggests that ApSigma has sigma factor activity, and can function as the apicoplast σ subunit.

Other species included in the apicomplexa, such as haemosporidians and coccidians, also harbor one gene that is homologous to ApSigma. Furthermore, a homolog of ApSigma is present in Vitrella brassicaformis, a chromerid species related to the Apicomplexa. Our phylogenetic analyses revealed that apicomplexan σ subunit proteins constitute a monophyletic clade after divergence from the chromerid protein, and each of the haemosporidian and the coccidian proteins forms independent clades within the apicomplexan clade (SI Appendix, Fig. S7). This implies that the apicomplexan species inherited their σ subunits from a common ancestor, without horizontal gene transfer events.

We hypothesized that if ApSigma functions as an apicoplast RNA polymerase σ subunit, it will interact specifically with DNA within apicoplast promoter regions. To examine this, five sites from putative promoter regions estimated from the apicoplast genome structure (R1-4 and R6) and seven sites from other regions (R5 and R7-12) were selected. The interaction of these regions with ApSigma was analyzed by chromatin immunoprecipitation (ChIP) (Fig. 2A and SI Appendix, Fig. S8). This found that ApSigma preferentially interacted with R1–4, 8 and 11 compared with R9, the protein-coding region with the lowest binding value, which we used as a negative control (Fig. 2B and SI Appendix, Fig. S9). Weak interaction of ApSigma to the apicoplast genome was observed for the negative control, R9 (Fig. 2B). A recent study indicates that the σ⁷⁰ subunit binds to bacterial RNA polymerase core and elongation complex not only during transcription initiation, but also during the elongation reaction (22). Therefore, the weak binding occurring within R9 might be this ApSigma interaction with the RNA polymerase elongation complex. We suggest that the regions R1–4, 6, and 11, which showed significantly stronger ApSigma binding than R9, are promoter regions. It was previously suggested that two long polycistronic transcripts (>15 kb) are produced, initiating from tRNA gene clusters between the large and small subunit ribosomal RNA genes in the apicoplast genome (14). The preferential interaction of ApSigma with R1 to R4 (Fig. 2B and SI Appendix, Fig. S9) is consistent with this prediction. A significant ApSigma interaction was also identified from other sites (R6 and R11), suggesting the existence of two additional promoters or RNA polymerase pausing regions (Fig. 2A; orange arrows, SI Appendix, Fig. S8). This experimental evidence reveals the promoter locations within the P. falciparum apicoplast genome.

It was previously shown that P. falciparum retains its own circadian rhythm to facilitate intraerythrocytic development (23). Given that circadian rhythms of chloroplast gene expression in plants are regulated by nuclear-encoded PEP subunits (13), we hypothesized that a similar mechanism could be present in P. falciparum. To investigate this, we extracted the transcriptional profiles of P. falciparum in erythrocytes from published microarray analyses (9, 10) (Fig. 2C and SI Appendix, Fig. S10A). Using this, we identified with periodicity-detecting algorithms (JTK_CYCLE, Lomb-Scargle) (24) that all apicoplast genes are expressed periodically (Bonferroni–Hochberg adjusted P < 0.01 for all genes, in both datasets) (BH.Q in SI Appendix, Tables S1 and S2). In one dataset (9), accumulation of about 70% and 30% of the gene transcripts had estimated periods of 48 h and 24 h, respectively (Fig. 2C and SI Appendix, Fig. S10 B and C and Table S1). In another dataset (10), only transcripts with a 48-h period were detected (SI Appendix, Table S2). The 48-h period transcript set peaked in abundance at the early schizont stage in both datasets. We found that the nuclear apSig transcript also had a 48-h period, but its phase of oscillation preceded the apicoplast genes by 7 to 8 h, and peaked at the early to mid-trophozoite stage. Interestingly, transcripts for the apicoplast RNA polymerase α subunit genes, rpoA1 and rpoA2, encoded in the nucleus, had high correlation coefficients with apicoplast transcripts having a 48-h period, which may indicate a mechanism to synchronize the apicoplast and nuclear gene expression (SI Appendix, Table S3).

The mammalian hosts of malaria parasites have their own circadian rhythm, and the rhythm is communicated across their body through mechanisms including the blood hormone melatonin. Given the periodicity of apSig and apicoplast transcript accumulation (Fig. 2C), we were interested in the coordination of parasite gene expression programs with melatonin levels. We examined this, focusing on the nuclear-encoded apicoplast σ subunit ApSigma. For this, accumulation of protein-coding apicoplast transcripts in trophozoite cells was examined after 30 or 90 min of melatonin treatment. Melatonin is present in human blood up to about 200 pM (25). We performed a preliminary experiment with 200 pM and detected increased apSig transcripts (SI Appendix, Fig. S11). However, the stability of reproduction under experimental conditions was poor, so we used 10 nM melatonin, as reported by Furuyama et al. (26), because this gave consistent and reproducible results. The apSig transcripts and the apicoplast gene transcripts sufB, rpoC2, and tufA accumulated in response to melatonin, suggesting that melatonin positively regulates apicoplast gene expression through ApSigma function (Fig. 3A). The increase in apSig transcripts was specific, because nuclear-encoded act1 and rpoA1/2 genes encoding actin and apicoplast RNA polymerase α subunits, respectively, were unchanged. SufB is a protein involved in the biosynthesis of iron–sulfur clusters that drive the methylerythritol phosphate (MEP) pathway for isoprenoid synthesis, which is essential for malaria parasite survival (27). In addition to the apicoplast sufB gene, nuclear-encoded sufC and sufD gene transcripts for the FeS cluster biosynthetic enzyme SufBCD (6), and those of ispG and ispH, encoding key enzymes of the MEP pathway (27), were increased by melatonin (SI Appendix, Fig. S12). This suggests that MEP pathway activity in the apicoplast was affected by the host circadian rhythm. In contrast, expression of the apicoplast transcript rpl4 did not respond to melatonin. Therefore, multiple transcriptional regulatory mechanisms might exist for apicoplast gene transcription.

To further understand the interaction and coordination between the parasite and host rhythmicity, we examined whether there is circadian or developmental gating of the response of apSig expression and apicoplast transcription to melatonin. Melatonin was administered at various times during synchronous culture, with cells sampled to investigate the melatonin response of apSig, sufB, and act1 transcripts (Fig. 3B). Melatonin had a time-restricted effect, inducing apSig and sufB prominently at 22 h to 24 h after synchronization, after which the response decreased (Fig. 3B). This change in melatonin sensitivity over time might be due to circadian gating of the response, or developmental stage-specific sensitivity. Together, this suggests that extrinsic circadian cues, intrinsic circadian cues and developmental timing cues are integrated to coordinate the transcription of the apicoplast genome (Fig. 4).

Finally, we investigated the mechanism of activation of apSig expression in response to melatonin. Two distinct pathways are considered to mediate melatonin signaling in P. falciparum (Fig. 3C) (26, 28). One is the inositol trisphosphate (IP₃) pathway via phospholipase C (PLC), and another is the cyclic AMP (cAMP) pathway via adenylyl cyclase (AC). We analyzed the melatonin response of apSig and apicoplast sufB genes in the presence of the melatonin receptor inhibitor luzindole, the PLC inhibitor U73122, or the AC inhibitor MDL12330. Since melatonin-induced transcript elevation occurred also in nonsynchronized cultures, this experiment was performed under nonsynchronized conditions. In this experiment with nonsynchronized cells, melatonin caused approximately 1.5-fold increase of apSig and sufB transcripts (Fig. 3D). This response was inhibited by luzindole or MDL12330, whereas U73122 had no effect (Fig. 3D). Furthermore, cAMP addition mimicked the effect of melatonin (Fig. 3D). Together, these results suggest that melatonin induces apSig gene expression through the second messenger cAMP, which results in the upregulation of apicoplast gene expression.

P. falciparum strain 3D7 was cultured at 3% hematocrit with type A+ human erythrocytes in RPMI 1640 medium (Thermo Fisher Scientific), supplemented with 25 mM NaHCO₃, 10 μg/mL gentamicin sulfate, 10 μg/mL hypoxanthine, 25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 0.8 mg/mL L-glutamine, and 0.5% (w/v) Albumax II (Thermo Fisher Scientific) (38). Cultures were maintained under 5% O₂, 5% CO₂, and 90% N₂ at 37°C. Parasitemia was determined with thin blood smears stained with Giemsa. The experiments using human erythrocytes were performed under the guidelines of the ethics committee of the University of Tokyo (#10050-(2)). Human erythrocytes were obtained from Japanese Red Cross Society (No. 25J0207). Synchronization of the parasite cultures was performed by 5% (w/v) sorbitol treatment for 10 min (39).

For analysis of effect of melatonin on the parasite, erythrocytes infected by late trophozoites and schizonts were collected using 63% (v/v) Percoll density centrifugation (40) from parasite cultures pretreated with 5% (w/v) sorbitol (41). After an incubation for 4 h, the Percoll-purified erythrocytes were treated with 5% (w/v) sorbitol and removed trophozite and schizont stages. After an incubation for 22 h, the erythrocytes with parasites highly synchronized to early trophozoite stage were incubated with or without 10 nM melatonin (Sigma-Aldrich) for 30 or 90 min. After the treatment, erythrocytes were collected by centrifugation, washed with phosphate-buffered saline (PBS), and stored at −80 °C until use. For analysis of signal transduction in the parasite, erythrocytes infected by early trophozoites were treated with luzindole (melatonin receptor inhibitor; Sigma-Aldrich) (26), 1 μM MDL12330A (adenylate cyclase inhibitor; Sigma-Aldrich) (26), 10 μM U73122 (PLC inhibitor; Tocris Bioscience, Bristol, UK) (28), or no inhibitor, for 90 min at 37°C. Subsequently, parasite cultures were incubated with 10 nM melatonin for 90 min at 37°C. After incubation, parasite cultures were then pelleted by centrifugation, washed by PBS, and stored at −80 °C until use.

Apicomplexan σ subunits in the amino acid sequence database were searched for by BLAST analysis at PlasmoDB (https://plasmodb.org/plasmo/app), using conserved domain sequence of σ⁷⁰ of E. coli (BAB37373) as the query. One σ subunit-like sequence was identified in each of P. falciparum 3D7 (XP_966194.1 encoded by PF3D7_0621000), Eimeria brunetti (CDJ52515), Eimeria necatrix (XP_013434793), Eimeria maxima (XP_013337375), Cyclospora cayetanensis (XP_026192347), Toxoplasma gondii VEG (ESS33313), Toxoplasma gondii CAST (RQX72928), Neospora caninum (XP_003881399), P. berghei (XP_034422407), Plasmodium vivax (SCO73651), Plasmodium relictum (XP_028533929), and V. brassicaformis (CEL93836). Amino acid sequences (regions 2 to 4) were aligned using ClustalW, and the alignment was refined by eye. Phylogeny of the σ subunits was analyzed by the maximum likelihood method using MEGA program, and bootstrap values were calculated replicating 1,000 analyses with LG+G model.

In the gregarines and piroplasmids, no σ subunit homologs are not identified. In gregarines, the apicoplast is lost (42, 43). The structure of the apicoplast genome of piroplasmids differs from the general plastid genome because all genes, including those of rRNAs and tRNAs, occupy one strand of the genome (44). Probably, a different mechanism that does not involve σ subunit regulates apicoplast gene expression in piroplasmids.

For ApSigma structural predictions, we used a version of AlphaFold (version 2) available at https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb. that allows single predictions (45). The PDB file output from AlphaFold was visualized, edited, and colored with UCSF Chimera (46).

A 654-nt-long synthetic DNA fragment encoding the 2.4 region of ApSigma in a modified codon usage that matches the one of highly expressed proteins of E. coli (SI Appendix, Table S4), and another fragment with the complementary sequence, were obtained from Azenta. The double-strand DNA fragment composed from the two fragments was inserted into SmaI-digested pGEX-4T-1 (GE Healthcare). Expression of the recombinant proteins in E. coli and their purification were performed as described previously (47). A guinea pig was immunized with the recombinant protein, and polyclonal antibodies against ApSigma were purified by Tanpaku Seisei Kougyou.

Human erythrocytes infected by P. falciparum were fixed in 2% (w/v) paraformaldehyde containing 0.075% (w/v) glutaraldehyde in PBS for 10 min. The reaction was quenched using 0.1 M glycine in PBS for 15 min. After blocking in 3% (w/v) bovine serum albumin (BSA), 0.2 % (w/v) Tween 20 in PBS for 1 h, the erythrocytes were incubated with anti-ApSigma antibody (diluted 1:200) and rabbit anti-HU antibody (diluted 1:200) or rabbit anti-ATG8 antibody (a kind gift from Noboru Mizushima, The University Tokyo, diluted 1:200) in 1% BSA, 0.2% (v/v) Tween 20 in PBS for 1 h (48, 49). Then, erythrocytes were washed in 0.2% (v/v) Tween 20 in PBS three times and incubated with Alexa Flour 561-conjugated goat anti-guinea pig IgG (1:1,000) and Alexa Flour 488-conjugated goat anti-rabbit IgG (1:1,000) for 1 h. The fluorescence images were captured by a confocal microscopy system with Zeiss LSM780 and LSM980 (Carl Zeiss).

Erythrocytes infected by P. falciparum 3D7 at high parasitemia (above 10%) were collected from 20 mL cultures by centrifugation at 800 × g for 5 min at room temperature. The collected cells with parasites were fixed with 1% (v/v) formaldehyde at 37 °C for 10 min, followed by 0.125 M glycine for 5 min. The fixed erythrocytes were incubated in PBS containing saponin [0.075% (w/v)] and released parasites were collected by centrifugation. After washes with PBS buffer, the parasites were stored at −80 °C until use. The parasites were resuspended in 0.5 mL ChIP lysis buffer [50 mM Tris-HCl, 140 mM NaCl, 1 mM ethylenediaminetetraacetic acid (EDTA), 0.1% (w/v) sodium dodecyl sulfate (SDS), 1 % (v/v) Triton X100, 0.1 % (w/v) Sodium Deoxycholate, Complete Mini (Sigma-Aldrich), EDTA-free, protease inhibitor, pH 8.0], and cell disruption and shearing of genomic DNA were achieved by sonication [Branson Sonifier 250 (Emerson Electric), Duty Cycle 50, Output Control 2.0, 20 s, 10 times]. After sonication, it was experimentally confirmed that the size of genomic DNA fragments was within the range of 100 and 300 bp. The sonicated parasite suspension was centrifuged and the supernatant containing the fragmentated genomic DNA was collected. To a 0.4 mL of the collected supernatant, a 19-fold volume of ChIP lysis buffer was added, and pretreated with nProtein A Sepharose 4 Fast Flow (Sigma-Aldrich) for 4 h at 4 °C. After removal of nProtein A Sepharose beads by column filtration, the collected solution was divided into two. Each of these was subjected to immunoprecipitation. Immunoprecipitations were performed with 5.0 μL crude serum containing anti-ApSigma antibody or the preimmune serum at 4 °C overnight, then a 30 μL of 50 % (v/v) slurry of magnetic Dynabeads (Thermo Fisher Scientific) was added and further incubated for 5 h. The beads were washed twice with each of: RIPA150 buffer [50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 1% (v/v) Triton X-100, 0.1% (v/v) SDS, 0.1% (v/v) sodium deoxycholate, pH 8.0], RIPA500 buffer [50 mM Tris-HCl, 500 mM NaCl, 1 mM EDTA, 1% (v/v) Triton X-100, 0.1% (v/v) SDS, 0.1% (v/v) sodium deoxycholate, pH 8.0], LiCl wash solution [10 mM Tris-HCl, 250 mM LiCl, 1 mM EDTA, 0.5% (v/v) Nonidet P40, 0.5% (w/v) sodium deoxycholate, pH 8.0], and TE buffer: in this order. For reversions of the cross-link, the beads were resuspended in ChIP direct elution buffer [10 mM Tris-HCl, 300 mM NaCl, 5 mM EDTA, 0.5% (v/v) SDS, pH 8.0] and incubated at 65 °C overnight. After a treatment with RNase A and proteinase K, DNA was extracted from the beads using phenol:chloroform:isoamylalcohol (25:24:1) and precipitated with ethanol using Ethachinmate (Nippon Gene) as a carrier. The resultant pellets were dissolved in 100 μL water and analyzed by qPCR using relevant sets of primers (SI Appendix, Table S5), as reported previously (50) with modifications. Briefly, amplifications were done by incubating reaction mixtures at 95 °C for 2 min prior to 40 cycles of 10 s at 95 °C followed by 15 s at 40 °C and 30 s at 60 °C. Standard curves were constructed with several serial dilutions (1 to 1 × 10^–4) of input DNA to estimate percent of input of each DNA fragment relative to the input DNA.

We used MetaCycle in R package (https://cran.r-project.org/web/packages/MetaCycle/index.html) with the JTK_CYCLE and Lomb-Scargle algorithms (24, 51, 52) to detect periodicity of a length between 20 and 48 h in the expression profiles of the transcripts from the gene.

Total RNA was isolated from the parasites infecting the erythrocytes with Trizol LS reagent (Thermo Fisher Scientific), according to the manufacturer’s instructions. First-strand synthesis of cDNA was performed using 1 µg RNA and ReverTra Ace qPCR RT Master Mix with gDNA remover (Toyobo) according to the manufacturer’s instructions. The abundance of each transcript was quantified by qPCR. qPCR was performed as described (47), using primers shown in SI Appendix, Table S6. Expression of each gene was normalized with the value of 18S rRNA. A Shapiro–Wilk test was performed on all data groups after analysis. All data groups had P > 0.05, so Student’s t test was performed to determine significant differences.

Y.K., A.N.D., K. Kita, and K.T. designed research; Y.K., K. Komatsuya, S.I., T.N., Y.-i.W., S.S., and K.T. performed research; Y.K., K. Komatsuya, S.I., and S.S. analyzed data; and Y.K., K. Komatsuya, S.S., A.N.D., K. Kita, and K.T. wrote the paper.

The authors declare no competing interest.