Multifunctional Involvement of a C2H2 Zinc Finger Protein (PbZfp) in Malaria Transmission, Histone Modification, and Susceptibility to DNA Damage Response

ABSTRACT In sexually reproducing organisms, meiosis is an essential step responsible for generation of haploid gametes from diploid somatic cells. The quest for understanding regulatory mechanisms of meiotic recombination in Plasmodium led to identification of a gene encoding a protein that contains 11 copies of C2H2 zinc fingers (ZnF). Reverse genetic approaches were used to create Plasmodium berghei parasites either lacking expression of full-length Plasmodium berghei zinc finger protein (PbZfp) (knockout [KO]) or expressing PbZfp lacking C-terminal zinc finger region (truncated [Trunc]). Mice infected with KO parasites survived two times longer (P < 0.0001) than mice infected with wild-type (WT) parasites. In mosquito transmission experiments, the infectivity of KO and Trunc parasites was severely compromised (>95% oocyst reduction). KO parasites revealed a total lack of trimethylation of histone 3 at several lysine residues (K4, K27, and K36) without any effect on acetylation patterns (H3K9, H3K14, and H4K16). Reduced DNA damage and reduced expression of topoisomerase-like Spo11 in the KO parasites with normal Rad51 expression further suggest a functional role for PbZfp during genetic recombination that involves DNA double-strand break (DSB) formation followed by DNA repair. These finding raise the possibility of some convergent similarities of PbZfp functions to functions of mammalian PRDM9, also a C2H2 ZnF protein with histone 3 lysine 4 (H3K4) methyltransferase activity. These functions include the major role played by the latter in binding recombination hotspots in the genome during meiosis and trimethylation of the associated histones and subsequent chromatin recruitment of topoisomerase-like Spo11 to catalyze DNA DSB formation and DMC1/Rad51-mediated DNA repair and homologous recombination.

M alaria, a disease caused by blood infection with protozoan parasites of the genus Plasmodium, remains a major public health challenge, with an estimated 214 million cases and~438,000 deaths recorded worldwide in 2015 (1). An effective vaccine to control malaria has not been developed; antigenic variation, genetic diversity, and antimalarial drug resistance continue to impede control efforts. The lifecycle of the parasite alternates between a vertebrate host and an invertebrate mosquito vector. For continuation of the life cycle from one host to another, it is essential for the parasites to differentiate and develop into sexual stages. Plasmodium organisms are haploid throughout their life cycle except for the brief stage postfertilization within the mosquito midgut when haploid gametes fuse to form diploid zygotes. Within hours, these undergo meiosis and develop into ookinetes.
Studies in the malaria parasite Plasmodium falciparum have provided evidence of a process that begins with frequent random mating involving haploid gametes of genetically distinct parasite clones. The diploid zygote then undergoes meiotic division, and genetic recombination during meiosis contributes to the genetic diversity of the parasite, as with other sexually reproducing eukaryotes (2)(3)(4)(5)(6). Homologous recombination (HR) is central to meiosis in almost all organisms, and reciprocal recombination events, also known as crossing-over events, lead to the orderly segregation of homologous chromosomes. Recombination during meiosis is also responsible for the creation of genetic diversity among offspring of diverse parental backgrounds (7). In mammalian cells, crossing-over events during meiotic recombination are not distributed randomly but are clustered in short-interval regions along chromosomes called "hotspot" (8).
Studies on identification and genome-wide mapping of recombination (9)(10)(11) have revealed a correlation between hotspot activity and enrichment of trimethylated histone 3 at lysine 4 (H3K4me3) (12,13). Further studies in mammals have identified PRDM9, a meiosis-specific chromatin-modifying protein, as the key determinant of recombination hotspots and a master transcription regulator of entry into meiosis (14)(15)(16)(17). The PRDM9 protein contains a C-terminal module comprised of multiple C 2 H 2 -type zinc fingers (ZnF) which binds the degenerate 13-bp motif that was previously found to be associated with~40% of human hotspots, formation of doublestrand breaks (DSB), and recruitment of the recombination machinery leading to exchange of chromosomal material and DNA repair. Additionally, the ZnF domain (Zfd) in PRDM9 is fused to a SET domain which catalyzes the trimethylation of histone 3 lysine 4 (H3K4) and histone 3 lysine 36 (H3K36) associated with the hotspots (18,19). The SET domain initially characterized in Drosophila is an evolutionarily well-conserved domain associated with many chromosomal proteins and with proteins necessary for histone lysine trimethylation in mammalian cells (20).
The C2H2-type Zfd, represented by two cysteine and two histidine residues separated by~12 amino acids acting as ligands to a zinc atom, are widespread in most DNA-binding proteins (21,22). In animals, proteins with multiple C 2 H 2 ZnFs are transcription factors regulating gene expression and acting during cell differentiation and development. Binding of C 2 H 2 ZnF proteins to DNA often helps recruit other proteins which mediate epigenetic changes, including posttranslational modifications of histones. These include histone methylases as well as enzymes catalyzing demethylation of H3K4me3 and other histone methyl transferases transferring methyl groups to H3K9 and H3K27. Depending upon the sites methylated, histone methylation is involved in both repression and activation of transcription (23)(24)(25). For example, trimethylation of K4 in H3 (HeK4me3) is associated with activation, whereas trimethylation of K9 (H3K9me3) and trimethylation of K27 (H3K27me3) are generally associated with transcription repression (20,26). In contrast, acetylation of histone lysine residues relaxes chromatin structures, leading to increased transcriptional and gene expression activities in the cell.
Meiotic recombination involves a series of orchestrated and programed biochemical events beginning with DNA DSB and DNA repair (27)(28)(29). Genetic mapping and hotspot identification have also been performed in Plasmodium spp. (30)(31)(32)(33). Previously, our laboratory identified and investigated the role of molecules involved in HR. These include P. falciparum Rad51 (PfRad51), P. falciparum Rad54 (PfRad54), and two types of P. falciparum replication protein A1 (PfRPA1L and PfRPA1S) involved in DNA strand exchange and DNA-DSB repair processes in erythrocytic stages of P. falciparum (34)(35)(36) and Dmc1 (disrupted meiotic cDNA homolog, a meiotic recombinase) involved in the sporogonic life cycle of P. berghei (a murine malaria parasite) ensuing after meiotic nuclear division in developing zygotes in the mosquito midgut (37). However, it is not known how recombination machinery is recruited to the site of DNA damage and whether Plasmodium employs molecular mechanisms similar to those deployed in mammals for recognition of putative recombination hotspots. In this study, we investigated if Plasmodium utilizes a molecule analogous to mammalian PRDM9, particularly with respect to hotspot choice, hotspot-associated H3K4 trimethylation, and DNA-DSB repair mechanisms.

RESULTS
Characteristic features of PbZfp. While proteins with runs of multiple (Ͼ3) instances of C 2 H 2 ZnF are common in animals, the P. berghei Zfp (PbZfp) protein is the only such example conserved across Plasmodium species: PbANKA_06077 in P. berghei, Pf3D7_1209300 in P. falciparum, PKH_13092 in P. knowlesi, Py17X_0610200 in P. yoelii, and PVX_084495 in P. vivax. The protein is comprised of a poorly conserved N-terminal low-complexity region and a C-terminal region with 11 C 2 H 2 ZnFs, which is where the sequence similarity is most pronounced. The C 2 H 2 ZnFs are arranged in an initial run of 3 ZnFs followed by a disordered poorly conserved linker connecting them to the remaining 8 ZnFs. Unlike the animal PRDM9 protein, PbZfp lacks the methyltransferase SET domain and does not have KRI domains typical of the members of the KRAB family of animal C 2 H 2 ZnF proteins. Thus, while not directly related to these animal proteins, PbZfp shows an unusual resemblance to animal C 2 H 2 ZnF proteins in having a multiplicity of C 2 H 2 Zfd (22). Notably, orthologs of PbZfp exhibiting greater sequence divergence are also seen in other Apicomplexa species, such as piroplasms (e.g., Theileria spp.), coccidians (e.g., Toxoplasma spp.), and Cryptosporidium spp. Additionally, bioinformatics analysis predicted three potential phosphorylation sites in the P. falciparum homolog of PbZfp, a result which was experimentally validated by phosphoproteome analysis of erythrocytic asexual life cycle stages, demonstrating phosphorylation at serine residues 512, 596, and 1230 (38). PbZfp conservation in a single copy across Apicomplexa is unlike the more lineage-specific conservation patterns of transcription factors (39). Further, the presence of a run of multiple C 2 H 2 ZnFs indicates that the protein might contact an extended binding site as is typical of recombination hotspots. These observations pointed to potential roles in addition to or beyond that of a conventional transcription factor and prompted us to investigate if it might play a role in recombination.
Targeted disruption of the PbZfp gene. To study the biological role of PbZfp, we employed two PbZfp gene disruption strategies. In the first, we generated loss-offunction parasite clones by targeting the full-length endogenous gene (Fig. 1B, panel III) using an insertion plasmid (Fig. 1B, panel I). As shown in panel IV of Fig. 1B, HR within the 5=utr and 3=utr regions of the genomic locus (Fig. 1B, panel III) and the targeting plasmid was expected to result in the disruption of genomic PbZfp. We also sought to generate parasites expressing truncated (Trunc) PbZfp, lacking the C-terminal C 2 H 2 ZnFs (Fig. 1A, panel II). A truncation plasmid (Fig. 1B, panel II) was designed to introduce a stop codon within the coding sequence at the start of the Zfd. Panel V of Fig. 1B shows the expected locus in the truncated parasites after HR within the coding sequence and the 3=utr regions of the genomic locus and the targeting plasmid. Parasites obtained from successful transfection were used for further cloning, and two independent clones of each type (knockout [KO] and Trunc) were employed in various studies. Insertion-specific diagnostic PCRs were done using different primer combinations (see Fig. S1 in the supplemental material; locations of primers are indicated in Fig. 1, and sequences are listed in Table S1 in the supplemental material). Integration at the 5= end was confirmed using primer pair 815/921 (panel A of Fig. S1 shows an Loss of expression of PbZfp in disrupted parasites revealed by RT-PCR. Next, we demonstrated the absence of expression of full-length PbZfp in the KO parasites. Reverse transcriptase PCR (RT-PCR) performed using primers 811/812 revealed an 850-bp transcript band in WT parasites, whereas no band was seen in the KO parasites ( Fig. 2A). In parallel, we tested RNA from WT and PfZfp KO parasites for expression of Rad51 and detected the presence of an~500-bp transcript in both, ruling out any nonspecific gene-targeting event. Results of reactions performed in the absence of a reverse transcriptase step were negative. We also tested gene expression in the Trunc parasites. Similarly to KO parasites, the results obtained using primers 807/808 with Trunc parasites were negative, whereas WT parasites showed an expected 650-bp fragment (Fig. 2B). Interestingly, when primers 807/810 were employed, the Trunc and WT parasites revealed~500-bp and~1,000-bp fragments, respectively (Fig. 2C), establishing the expression pattern for PbZfp truncated by removing the Zfd coding region. Expression of truncated PbZfp (upstream sequence encoding the N-terminal region) was further confirmed by positive RT-PCR results obtained using primer pair 811/812 and a smaller RT-PCR band in Trunc parasites obtained using primers 811/810 (data not shown).

Asexual-and sexual-stage growth and transmission competence of PbZfp KO and Trunc parasites.
To evaluate whether genetic disruption or truncation of PbZfp would affect various life cycle parameters, we followed the asexual growth kinetics of the WT, KO, and Trunc parasites. Swiss Webster mice (n ϭ 3 per group) were infected intravenously (i.v.) with 10 5 parasites in four independent experiments, and parasitemia was monitored daily in the infected mice. The asexual growth kinetics of the WT, KO, and Trunc parasites were comparable, with the exception that mice infected with KO parasites survived significantly longer (Fig. 3A). Figure 3B shows Kaplan-Meier survival plots for mice infected with WT, KO, and Trunc parasites. Mice infected with KO parasites lived nearly twice as long as those infected with WT parasites. Even mice infected with Trunc parasites showed a survival advantage over WT-infected mice. These data implicate PbZfp as a potential determinant of parasite virulence.
Next, to examine the transmission competence of WT, KO, and Trunc parasites, Swiss-Webster mice (3 mice/group) were infected (i.v.) with 10 5 WT, KO, or Trunc parasites. A drop of blood from infected mice was used to examine exflagellation of male gametocytes, and mice were used to infect Anopheles stephensi mosquitoes on day 4 postinfection. At 9 to 11 days postfeeding, mosquitoes were dissected, individual midguts were stained with 0.1% mercurochrome, and oocysts were enumerated microscopically at ϫ100 magnification. As shown in Fig. 4, both KO and Trunc parasites exhibited significantly reduced levels of transmission competence compared to WT parasites. Oocyst numbers in mosquitoes infected with KO and Trunc parasites were 98% to 99% lower and 75 to 80% lower, respectively, than those seen with mosquitoes infected with WT parasites. Panels A and B of Fig. 4 show transmission results with two independent clones of KO and Trunc parasites. The differences in the oocyst burdens and in the percentages of infected mosquitoes between the WT and KO groups and the WT and Trunc groups were statistically significant (P Ͻ 0.05).
To ensure that the differences seen in the oocyst numbers were indeed due to gene disruption (KO or truncation) and not due to loss of infectiousness of parasites due to experimental manipulations performed during the transfection and cloning steps, we tested the transmission competence of parasite clones that went through the entire transfection process but that had a wild-type PbZfp genotype. As shown in Fig. 4C, these transfection control (Tc) parasites had numbers of oocysts comparable to those seen with the WT, indicating that the differences in the numbers of oocysts indicated in Fig. 4A and B were indeed due to disruption of PbZfp gene.
Additionally, we also compared gametocyte sex ratios in the blood of mice infected with WT, KO, and Trunc parasites on the day of mosquito infection (day 4). As shown in Fig. 4D, there was no significant association between the sex ratios and subsequent  Table S1. All the samples were also tested using primers for an unrelated gene (pbRad51) to establish the specificity and quality of RNA samples. mosquito midgut stage development. Finally, we also compared ookinete formation differences between the WT, KO, and Trunc parasites using an in vitro ookinete development assay. On the day of mosquito feed, blood from infected mice was cultured for 24 to 30 h at 19°C and then examined microscopically. Figure 4D shows average ookinete counts for the three parasites and comparable morphologies of Giemsa-stained representative ookinetes (Fig. 4E).
Histone lysine methylation differences in WT, KO, and Trunc parasites. We next considered the possibility that PbZfp with 11 C 2 H 2 ZnFs might affect histone methylation in the parasites. Purified parasites from mice infected with WT, KO, or Trunc parasites were analyzed by Western blotting. Lysates of parasites (10 6 per lane) were fractionated by 15% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and probed with a panel of histone lysine methylation-and histone lysine acetylation-specific antibodies. Anti-H3K4me3, anti-H3K36me3, and anti-H3K27me3 antibodies showed strong reactivity with WT parasites (band just below the 17-kDa standard). However, all three antibodies failed to react with KO parasites (Fig. 5A). In sharp contrast to the histone methylation results, the WT and KO parasites showed comparable levels of reactivity, with antibodies recognizing H3K9ac, H3K14ac, and H4K16ac (Fig. 5B). Surprisingly, the levels of antibody reactivity with Trunc parasites lacking C 2 H 2 ZnFs were comparable to those seen with WT parasites (Fig. 5A), suggesting that there is no apparent functional role of Zfd in histone modification.
To test the specificity of these antibodies and to revalidate these findings, we tested purified histones from WT, KO, and Trunc parasites. Figure 5C shows the purity of various histones (H2A, 18 kDa; H2B, 14 kDa; H3, 16 kDa; H4, 14 kDa) from WT, KO, and Trunc parasites. We then probed purified histones with various antibodies recognizing histone trimethylation and acetylation at specific lysine residues. Similarly to the results obtained with crude parasite lysates, all three methylation-specific antibodies recog- nized histones from WT and Trunc parasites and, once again, no reactivity was seen with purified histones from KO parasites, demonstrating the inability of KO parasites to trimethylate H3K4, H3K36, or H3K27 (Fig. 5D). Purified histones from all three parasite lines (WT, KO, and Trunc) were strongly recognized by antibodies recognizing H3K14ac and H4K16ac (Fig. 5E).

Methyl methanesulfonate (MMS)-induced DNA damage response in WT, KO, and Trunc parasites.
Previous studies have shown that chromatin modifier proteins such as PRDM9 (14,15) target sites in the genome where DSB in DNA occur. Our observations confirming that PbZfp KO parasites completely lacked H3K4, H3K27, and H3K36 trimethylation led us to hypothesize that the KO parasites would exhibit reduced susceptibility to DNA-damaging agents. To test our hypothesis, we exposed WT, KO, and Trunc parasites for 6 h to various concentrations (0.05%, 0.005%, and 0.0005%) of the DNA-damaging agent MMS using conditions similar to those used in our previous studies with P. falciparum (35) followed by Comet assay analysis (18). There was an MMS dose-dependent increase in Olive tail moment (OTM) values for WT, KO, and Trunc parasites (Fig. 6, top). However, the OTM values in KO parasites exposed to 0.005% and 0.0005% concentrations of MMS were significantly lower than those seen with the WT and Trunc parasites, supporting our hypothesis that PbZfp disruption leads to reduced susceptibility to DNA damage responses. Even at the highest concentration (0.05%) of MMS, KO parasites still showed reduced DNA damage (P value ϭ 0.054, approaching significance).
To quantify the percentage of damage in the parasites exposed to MMS, we counted the total number of cells, the total number of intact nuclei, and the total number of

C2H2 Zinc Finger Protein in Malaria Transmission
® Comets for each sample and then calculated the percentage of damaged cells (18). As seen in the bottom panel in Fig. 6, there was a dose-dependent increase in the percentage of damaged cells, with Ͼ95% damage at an MMS concentration of 0.05% in the WT and Trunc parasites. In contrast, KO parasites had only about 61% damage, once again demonstrating reduced susceptibility of KO parasites to the DNA-damaging agent. These results also suggest a direct role for the region upstream of the C 2 H 2 ZnFs in mediating susceptibility to DNA-damaging stimuli.

Induction of representative genes in the presence of DNA-damaging agent.
Previously, we showed that exposure to MMS induces expression of Rad51, Rad54, RPA1L, and RPA1S in P. falciparum (35,36), indicating their role in DNA damage repair. In view of the fact that we observed reduced susceptibility of KO parasites to MMS, we studied the MMS-induced changes in the expression of select genes likely to be involved in DNA damage and repair mechanisms. We used quantitative real-time PCR to compare the relative amounts of and average threshold cycle (⌬⌬C T ) values for select genes in WT, KO, and Trunc parasites by the Pfaffl method (40). As shown in the left panel in Fig. 7, there were significant increases in the levels of expression of PbZfp, Spo11, and Rad51 (a mitotic recombinase gene) in WT parasites exposed to MMS. As expected, Dmc1 (a meiotic recombinase gene and homolog of Rad51) was not induced by MMS in mitotically active blood stages. In KO and Trunc parasites, PbZfp and Spo11 were not induced, while they maintained normal inducibility of Rad51 (Fig. 7, middle and right panels). This finding also suggests that Spo11 probably acts in concert with PbZfp in the parasites.

DISCUSSION
In organisms reproducing sexually, meiosis is an essential step responsible for generation of haploid gametes from diploid somatic cells. The combination of alignment of homologous chromosomes and the reciprocal exchange between two homologs creates new allelic combinations via HR. Plasmodium organisms in the vertebrate host and invertebrate mosquito vector are haploid throughout their life cycle except for a brief diploid zygote stage occurring in the mosquito midgut. Haploid male and female gametes fertilize in the mosquito midgut, and within a few hours the resulting zygote undergoes meiotic nuclear division, resulting in the development of haploid stages within developing oocysts. In mammals, DNA-binding C 2 H 2 ZnFs and the H3K4 methyltransferase activity of the SET domain of PRDM9 play a major role in the identification of recombination hotspots in the genome during meiosis. H3K4 trimethylation leads to the recruitment of the evolutionarily conserved Spo11 (a topoisomerase-like enzyme) (41) to chromatin, which initiates DNA DSBs. Single-strand DNA tails created by DSBs allow further recruitment of Dmc1 and Rad51 proteins to facilitate genetic exchange via HR mechanisms. The processes of proper segregation of chromosomes during meiosis and formation of novel allelic genotypes depend upon genetic recombination occurring at unique sites termed "hotspots." We have previously described several molecular components involved in HR in Plasmodium, thus unequivocally demonstrating the presence of active functional HR  (Fig. 6) was used for transcriptional analysis by real-time RT-PCR. RNA was purified from parasites that were left untreated or were exposed to the indicated concentrations of MMS in duplicate wells, and purified RNA (1.0 g) was reverse transcribed as described in Materials and Methods. Three different amounts of cDNA (undiluted, diluted 1:10, and diluted 1:100) from each were analyzed by PCR using specific primers for PbZfp, PbSpo11, PbRad51, and PbDmc1 (Table S1) and a BioRad IQ5 real-time PCR detection system. Pb18rRNA gene served as an internal control. Relative C T values from real-time PCR in the control and treated parasites were calculated by the Pfaff method, and fold change between untreated and treated samples was determined as previously described (35).

C2H2 Zinc Finger Protein in Malaria Transmission
® machinery in the parasite. These included homologs of Rad51, Rad54, RPA1 (in two forms, L and S), and meiotic recombinase Dmc1 in human and murine malaria parasites (35)(36)(37). However, the markers for chromosomal sites undergoing DNA DSBs during meiosis in Plasmodium are not well characterized. We investigated if PbZfp containing 11C 2 H 2 ZnFs might mediate functions analogous to PRDM9 in mammals and might be important during meiosis and also an important determinant for transmission through the mosquitoes. We employed a reverse genetics approach to understand the functions of various protein domains in PbZfp. We developed P. berghei lacking full-length PbZfp (KO) and parasites expressing PbZfp lacking C-terminal Zfd (Trunc). We were able to successfully knock out the gene for PbZfp in the parasites, showing that it is not required for maintenance of erythrocytic asexual growth of parasites. While our studies did not find significant differences in proliferation rates among WT, KO, and Trunc parasites, a P. falciparum homolog of PbZfp was previously revealed by quantitative trait locus mapping analysis to be one of the several putative genes involved in regulating the duration of the cell cycle (42). It is noteworthy that the KO parasites seemed to be less virulent in mice despite showing peak parasitemia levels comparable to those seen with the mice infected with WT and Trunc parasites. The most striking phenotypic difference revealed was the loss of transmission competence of KO and Trunc parasites, thus pointing to a functional role of PbZfp during mosquito stage development.
Epigenetic mechanisms have been shown to play important roles in malaria parasite development, antigenic variation, Var gene silencing, and pathogenesis (43)(44)(45)(46)(47)(48)(49)(50)(51)(52). We found that trimethylation of H3K4, H3K27, and H3K36 was completely blocked in PbZfp KO parasites. On the other hand, parasites expressing PbZfp lacking the C-terminal Zfd behaved like WT parasites, thus implicating the functionality of the N-terminal region of PbZfp, upstream of the Zfd, in mediating H3 lysine trimethylation. Interestingly, this was unique to histone 3 lysine methyl transferase activity since acetylation of histones (H3K9ac, H3K14ac, and H4K16ac) was not affected in either the KO parasites ( Fig. 4B and E) or the truncated parasites. The simplest interpretation of the results would be that the N-terminal region of PbZfp recruits multiple histone methylases to chromatin. Paradoxically, genomic analysis has failed to identify any E(z) family H3K27 methylase in P. falciparum, and if the same holds true for P. berghei, it is puzzling that we detected H3K27 trimethylation in P. berghei. This observation suggests a possibility that 1 of the other SET domain proteins among the 9 to 10 found in Plasmodium species might play a role in mediating H3K27 trimethylation.
The KO parasites also exhibited trimethylation of H3K36. Not much is known about the functional role of HeK36me3 in Plasmodium. In humans, H3K36me3 has been shown to regulate DNA mismatch repair (53). Other recent studies have provided evidence of important mechanistic involvement of H3K36me3 in the control of DNA damage signaling and DSB (54,55). Because of the possible involvement of PbZfp in mosquito stage development, including meiosis, we compared the levels of susceptibility of DNA DSB formation in WT, KO, and Trunc parasites using MMS-induced DNA damage analysis and the Comet assay technique. PbZfp KO parasites were significantly less sensitive to MMS-induced DNA damage than WT and truncated parasites as revealed by lower OTM values and percentages of accumulation of damaged nuclei upon exposure to MMS. Reduced susceptibility to MMS-induced DNA damage was further substantiated by quantitative RT-PCR (qRT-PCR) analysis. MMS caused induction of PbZfp, Spo11, and PbRad51 in WT parasites. Not too surprising, PbDmc1 encoding a meiotic recombinase was not induced by MMS in mitotically active blood-stage parasites. In contrast, Spo11 was not induced in the KO and Trunc parasites, providing additional support for findings of reduced susceptibility to MMS-induced DNA damage in Comet assays. The fact that the induction of PbRad51, a key HR repair component, was not affected in the KO and Trunc parasites suggests that these parasites maintain efficient HR repair machinery independently of PbZfp, resulting in efficient repair of damaged DNA. The results presented here indicate a key biological role for a ZnF protein during Plasmodium transmission via mosquitoes on one hand and regulation of epigenetic modifications and susceptibility of the parasite to DNA-damaging conditions on the other.

Identification of PbZfp gene in P. berghei.
Plasmodium parasites, like other apicomplexans, have relatively few transcription factors and, in contrast to animals, almost none with C 2 H 2 ZnFs (56). Nevertheless, they contain a well-conserved C 2 H 2 ZnF protein present across Plasmodium species, including P. falciparum, identified using a PSI-BLAST sequence profile for C 2 H 2 ZnFs (39). In P. berghei, this protein is encoded by a 3,747-bp open reading frame on chromosome 6 (PBANKA_060770). The product of this gene (termed PbZfp here) has been annotated Љputative Krox1-like transcription factorЉ in the PlasmoDB although it has no specific relationship to the mammalian Krox1 protein beyond the presence of multiple C 2 H 2 ZnFs (Fig. 1A). PbZfp contains 11 C 2 H 2 ZnFs in the C-terminal region extending from approximately amino acid 800 to amino acid 1248.
Plasmids for gene targeting in P. berghei. A gene knockout plasmid (pB3D backbone) was constructed by sequential insertion of following sequences: (i) PCR-amplified (574 bp; primers 778/779; pDEF plasmid template) P. berghei EF1␣ promoter (PbEF1␣) between BamHI and EcoRI sites; (ii) PCRamplified (711 bp; primers 776/777; pB3D-mCherry template) mCherry between EcoR1 and EcoRV sites; (iii) PCR-amplified (450 bp; primers 780/781; pDEF plasmid) PbDHFR 3=utr in the EcoRV site; (iv) PCRamplified (755 bp; primers 809/810; P. berghei genomic DNA template) PbZfp 3=utr between HindIII and KpnI sites; and (v) PCR-amplified (773 bp; primers 803/804; P. berghei genomic DNA template) PbZfp 5=utr in the Sac2 and Spe1 sites. The PbZfp gene encodes a protein containing a Zfd consisting of 11 C 2 H 2 zinc fingers toward the C terminus (Fig. 1A). To analyze the functional role of this region, we also developed a second gene-targeting plasmid to generate parasites expressing truncated PbZfp lacking Zfd. First, a PbZfp coding sequence corresponding to amino acids 568 to 862 was obtained by PCR amplification (882 bp; primers 805/806; P. berghei genomic DNA template) and cloned between Sac2 and Spe1 sites of the plasmid obtained after step iv (see above). Next, we amplified a PfZfp genomic sequence (610 bp in length; primers 823/824, P. berghei genomic DNA template) comprised of the last two codons (including the stop codon) and the 3=utr of PbZfp immediately downstream for insertion in the Spe1 and BamHI sites. The plasmid pB3D encodes Toxoplasma gondii DHFR-TS, which confers resistance to pyrimethamine and allows selection of transfected parasites. Primer details are given in Table S1 in the supplemental material. At each cloning step, plasmids were verified by restriction digestion and sequence analysis.
Transfection and diagnostic PCR analysis of PbZfp knockout parasites and truncated parasites. For transfection, plasmid DNA was purified using a plasmid Maxi kit (Qiagen) and linearized by digestion with SacII and Kpn1. Linearized DNA (10 micrograms) and an Amaxa human T cell nucleofector kit (U033 settings) were used for each transfection. Transfected parasites were injected i.v. into naive Swiss-Webster mice, and mice were put on drinking water containing pyrimethamine (70 g/ml) 24 h posttransfection to select for drug-resistant transformants. Genotypes and various recombination events in the drug-resistant parasites were characterized by PCR analysis prior to and after cloning by limiting dilution. Two clones of parasites with PfZfp knockout (KO) and PfZfp truncation (Trunc) were selected from two independent transfections for these studies.
Reverse transcriptase PCR to examine expression of PbZfp. Total RNA was extracted from WT, KO, and Trunc parasites obtained from infected Swiss-Webster mice using Trizol reagent (Invitrogen), and concentrations were determined using a NanoDrop instrument (model 2000/2000c). RNA (1 microgram) was reverse transcribed using a Qiagen One-Step RT-PCR kit, and the resulting cDNAs were subjected to PCR amplification for 35 cycles using gene-specific primers (Table S1). Control reactions were carried out without a reverse transcriptase step to confirm that no genomic DNA contamination was present in the RNA samples. The products were run on a 1% agarose gel and stained with ethidium bromide.
Characterization of asexual growth kinetics, ookinete formation, and transmission to mosquitoes. Swiss-Webster mice (female, 6 to 8 weeks old) were infected i.v. with 10 5 infected red blood cells (RBCs) of WT, KO, and Trunc parasites (3 mice per group). Beginning on day 3, parasitemia was monitored daily by examining Giemsa-stained thin blood smears. Mice were monitored twice daily to record any mortality. Blood smears were also examined on day 4 or day 5 postinfection for determination of the relative abundances of male and female gametocytes. To compare ookinete stage development, blood was drawn from mice infected with WT, KO, or Trunc parasites with approximately equivalent levels of gametocytemia on day 4 postinfection. The blood was then diluted 1:5 in RPMI 1640 medium supplemented with 10% fetal bovine serum and incubated at 19°C for 24 to 30 h. Thin smears were prepared and subjected to Giemsa staining to examine the morphology of ookinetes. Multiple replicates of cultured diluted blood were used to enumerate ookinetes by the use of a hemocytometer at ϫ400 magnification, and counts were expressed as numbers of ookinetes per milliliter. Female A. stephensi mosquitoes (4-to-5-day-old adults; n ϭ 25 to 30 per mouse) starved for 4 to 5 h were allowed to engorge on infected mice (typically on day 4 postinfection) for 30 min. Blood-fed mosquitoes were maintained on 10% dextrose in an incubator at 70% to 80% relative humidity and 19°C (optimal for P. berghei development in the mosquitoes). Mosquitoes were dissected on days 9 to 11 to determine the oocyst load in the midgut.
Purification of histones. Histones were purified from parasites using an acid extraction method (57, 58) with minor modifications. Briefly, 1 to 3 ϫ 10 9 parasites were harvested from mice infected with P. berghei (WT, KO, or Trunc). The parasite pellets were washed two times with cold phosphate-buffered saline (PBS) and two times with a buffer (25 mM Tris-HCl [pH 7.8], 1 mM EDTA, 0.2% Nonidet P-40) to remove hemoglobin. After washing three times with 0.8 M NaCl, the pellets were treated two times with C2H2 Zinc Finger Protein in Malaria Transmission ® 0.25 M HCl to extract histones. Acid-soluble fractions were pooled and precipitated with equal volumes of 25% trichloroacetic acid on ice for 1 h. Precipitated histones were collected by centrifugation at 12,000 ϫ g for 15 min at 4°C, washed with 500 l of acetone, and air dried. Samples were resuspended in 100 l of PBS (pH 7.0) and analyzed by 17.5% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting using antibodies against histone 3 and histone 4.
Evaluation of DNA damage by Comet assay. Swiss-Webster mice were infected (i.v.) with 10 5 WT, KO, or Trunc parasites (5 mice/group). Mice were bled by cardiac puncture, and the blood was passed through cellulose columns to remove white blood cells (WBCs) and resuspended at a hematocrit level of 5% in RPMI 1640 medium containing 20% fetal bovine serum and gentamicin. Infected mouse blood was plated in duplicate wells in 6-well tissue culture plates with various concentrations of methyl methanesulfonate (MMS) (0%, 0.05%, 0.005%, and 0.0005%). Parasites were incubated for 6 h (36), lysed with 0.05% saponin, washed with PBS (pH 7.4), and finally resuspended in 1.0 ml PBS for Comet assay analysis (35). Diluted cells (30 l) were mixed with 300 l Comet LMAgarose (Trevigen, Gaithersburg, MD) (1% low-melting-temperature agarose), and the mixture was added to wells of Trevigen slides, incubated at 4°C for 30 min, and immersed in chilled lysis solution (Trevigen) at 4°C for 60 min. Slides were then transferred to a clean tray and immersed in alkaline lysis solution (200 mM NaOH and 1 mM EDTA). Lysed cells were electrophoresed for 30 min at 1 V/cm in 1ϫ Tris-borate-EDTA (TBE) buffer, followed by fixation in 70% ethanol and drying overnight. Wells were stained with 50 l freshly prepared SYBR green I and photographed at ϫ200 magnification using a Nikon Eclipse 80i (Nikon) high-performance camera with Sensicam QE (Cooke Corporation). Results corresponding to Comet head and tail length, head and tail percent intensity, tail migration, and Olive tail moment (OTM) (the product of the tail length and the fraction of total DNA in the tail) were scored using Comet assay IV software. For each sample (control or MMS treatment), three replicates were performed and averages of their OTM values were used for data analysis. Comets were also analyzed visually to score round undamaged (intact) and Comet (damaged) cells to calculate percent damage.
Real-time qPCR analysis. RNA from infected red blood cells from mice infected with parasites (WT, KO, Trunc) was purified using Trizol reagent (Invitrogen) and used for cDNA synthesis as described above. Samples for quantitative PCR (qPCR) analysis were prepared in IQ SYBR green Supermix, containing SYBR green dye, iTaq DNA polymerase, deoxynucleoside triphosphates (dNTPs) and buffers (Bio-Rad), 0.4 M gene-specific oligonucleotide primers (Table S1), and a 1:100 dilution of the cDNA reaction products. Amplification was performed on a Bio-Rad IQ5 real-time PCR detection system using 95°C for 2 min for the initial denaturation and enzyme activation followed by further incubation at 95°C for 3 min and 45 cycles of 95°C for 30 s, 50°C for 30 s, and 72°C for 30 s. Primers for the P. berghei 18S rRNA gene (PlasmoDB.org gene identifier [ID]: PBANKA_0521221) were used as an internal control. Each sample was analyzed as three independent replicates, and the average values were used for data analysis.
Data analysis and statistical evaluation. GraphPad Prism Software (GraphPad Software, Inc., CA) was used for statistical analyses of the data. Regression analysis was used to determine the differences in asexual growth kinetics of WT, KO, and Trunc parasites. The Mann-Whitney U test was used to compare the oocyst counts from the WT, KO, and Trunc parasites.
For analysis of the OTM values from the WT, KO, and Trunc parasites, a Shapiro-Wilk test was performed first to establish the normalcy of the data. Further, a Mann-Whitney U test was performed to determine two-tailed P values. Relative threshold cycle (C T ) values from RT-PCR analysis of the induced and uninduced samples were calculated by using the Pfaffl method (40), and ΔΔC T values were used to estimate fold change as previously described (35). P values of less than 0.05 were considered statistically significant.

ACKNOWLEDGMENTS
We thank members of the Kumar laboratory for critical discussions during various phases of the work and Melody C. Baddoo for the Comet assays performed in the core