In this study, we characterized the Puf family gene member Puf3 in the malaria parasites Plasmodium falciparum and Plasmodium yoelii. Secondary structure prediction suggested that the RNA-binding domains of the Puf3 proteins consisted of 11 pumilio repeats that were similar to those in the human Puf-A (also known as PUM3) and Saccharomyces cerevisiae Puf6 proteins, which are involved in ribosome biogenesis. Neither P. falciparum (Pf)Puf3 nor P. yoelii (Py)Puf3 could be genetically disrupted, suggesting they may be essential for the intraerythrocytic developmental cycle. Cellular fractionation of PfPuf3 in the asexual stages revealed preferential partitioning to the nuclear fraction, consistent with nuclear localization of PfPuf3::GFP and PyPuf3::GFP as detected by immunofluorescence. Furthermore, PfPuf3 colocalized with the nucleolar marker PfNop1, demonstrating that PfPuf3 is a nucleolar protein in the asexual stages. We found, however, that PyPuf3 changed its localization from being nucleolar to being present in cytosolic puncta in the mosquito and liver stages, which may reflect alternative functions in these stages. Affinity purification of molecules that associated with a PTP-tagged variant of PfPuf3 revealed 31 proteins associated with the 60S ribosome, and an enrichment of 28S rRNA and internal transcribed spacer 2 sequences. Taken together, these results suggest an essential function for PfPuf3 in ribosomal biogenesis.
Translational control of gene expression plays an important role during the development of eukaryotes. This regulation is often mediated by cis-acting elements in the 3′ untranslated region (UTR) of the mRNA. Binding of specific regulators to these cis elements activates or represses translation of the mRNAs (Gray and Wickens, 1998; Waters et al., 1989). One of these well-characterized regulators is the Puf family of RNA-binding proteins (RBPs), which were named after the two founding members, Pumilio (Pum) in Drosophila melanogaster and fem-3 binding factor (FBF) in Caenorhabditis elegans (Zamore et al., 1997; Zhang et al., 1997). Members from this family represent a highly conserved group of RBPs present in many eukaryotic organisms including animals, plants, fungi and protists. In different model organisms, the number of Puf-encoding genes varies. Drosophila has only one Puf gene, whereas 26 and 19 Puf-like genes have been found in the Arabidopsis and rice genomes, respectively (Tam et al., 2010). Puf members have diverse functions during development of eukaryotic organisms. Puf members share a highly conserved RNA-binding domain (RBD), which mostly consists of eight imperfect tandem repeats of ∼36 amino acids, termed the Pum repeats (Zamore et al., 1997; Zhang et al., 1997).
Recent structural and functional studies have revealed three distinct subfamilies within the Puf family proteins. Classical Puf members contain eight Pum repeats organized in a crescent shape, with the concave surface recognizing specific single-stranded RNA sequences. These classical Puf proteins are predominantly localized within the cytoplasm of the cell, which is consistent with their roles in translational repression or activation of mRNAs (Archer et al., 2009; Gu et al., 2004; Lublin and Evans, 2007; Macdonald, 1992; Moore et al., 2003; Zhang et al., 1997). Classical Puf members perform a wide range of functions. In D. melanogaster, binding of Pum to the Nanos response element (NRE) in the maternal hunchback mRNA determines abdominal formation by repressing its translation (Murata and Wharton, 1995; Wharton et al., 1998; Wharton and Struhl, 1991). In C. elegans, the Puf proteins FBF-1 and FBF-2 were originally identified by their ability to bind regulatory elements in the 3′ UTR of fem-3 mRNA and repress its translation (Zhang et al., 1997), which is required for the switch from spermatogenesis to oogenesis in hermaphrodites (Ahringer and Kimble, 1991; Hodgkin, 1986). In addition, the FBF proteins were also reported to bind to and activate the translation of the target mRNA egl-4, which is important for adaptation to odors (Kaye et al., 2009). In the budding yeast Saccharomyces cerevisiae, the Puf protein Puf3p represses expression of the cox17 mRNA, promoting its deadenylation and subsequent decay (García-Rodríguez et al., 2007; Jackson et al., 2004; Olivas and Parker, 2000), while Puf5p and Puf6p have been shown to control mating type switching by regulating the expression of the ho gene (Goldstrohm et al., 2006; Gu et al., 2004; Tadauchi et al., 2001).
Another subfamily of Puf proteins is represented by the human Puf-A (also known as PUM3) and S. cerevisiae Puf6 proteins (Li et al., 2009; Qiu et al., 2014). Crystal structures of Puf-A and Puf6 revealed 11 Pum repeats that are arranged in an ‘L-like’ shape, with the concave surfaces of the N- and C-terminal subdomains binding single- or double-stranded nucleic acids without apparent sequence specificity (Qiu et al., 2014). Both proteins primarily adopt a nucleolar localization, and are involved in biogenesis of the 60S ribosome (Chang et al., 2011; Li et al., 2009; Qiu et al., 2014). A closer look at Puf6 revealed that this protein is required for pre-rRNA processing and 60S ribosome export, and it also facilitates loading of the ribosome protein Rpl43 (Yang et al., 2016). Moreover, S. cerevisiae Puf6 is also found in the cytoplasm, where it acts as a translational repressor of the ash1 mRNA and is required for the asymmetric localization of ash1 to the bud tip of the daughter cell (Gu et al., 2004). Whereas human Puf-A appears to be important for tumorigenesis (Fan et al., 2013), knockdown of Puf-A in zebrafish results in abnormal eye development and primordial germ cell migration (Kuo et al., 2009).
The third Puf subfamily is represented by S. cerevisiae Nop9, a nucleolar protein that is essential for 40S ribosome biogenesis (Thomson et al., 2007). Recently, structural and functional analyses showed that Nop9 contains 11 Pum repeats and forms a ‘C-shaped’ structure (Zhang et al., 2016). Consistent with its role in processing the 20S pre-rRNA to produce mature 18S rRNA, Nop9 recognizes the base of the internal transcribed spacer (ITS) 1 RNA stem loop, which prevents premature cleavage of the 20S pre-rRNA by Nob1 in the nucleolus (Zhang et al., 2016). In Arabidopsis, APUM23, a Nop9 subfamily member, is involved in pre-rRNA processing, and disruption of apum23 leads to the accumulation of 35S pre-rRNA and unprocessed 18S and 5.8S rRNA precursors without affecting the steady-state levels of the mature rRNAs (Abbasi et al., 2010, 2011). Consistent with its role in ribosomal biogenesis, apum23 disruption results in an abnormal growth phenotype reminiscent of that in ribosomal protein gene mutants. In the protozoan parasite Trypanosoma brucei, two nucleolar Puf proteins, T. brucei (Tb)PUF7 and TbPUF10, may also be Nop9 subfamily members, and their knockdown (through RNAi) affects rRNA processing, leading to slow-growth phenotypes (Droll et al., 2010; Schumann Burkard et al., 2013). These examples have illustrated somewhat convergent functions for the Puf-A/Puf6 and Nop9 subfamilies of the Puf proteins in pre-rRNA processing and ribosomal biogenesis.
It has been increasingly recognized that translational regulation plays essential roles in the development of malaria parasites, especially during developmental stage transitions (Cui et al., 2015). A large number of RNA-binding proteins have been predicted and identified from the Plasmodium falciparum genome by in silico and genome-wide RNA-pulldown analyses (Bunnik et al., 2016; Reddy et al., 2015). Among them, the two classical Puf proteins, P. falciparum (Pf)Puf1 and PfPuf2, are differentially expressed and utilized during development (Cui et al., 2002). Functional studies have shown that PfPuf2 disruption leads to an increase in gametocytogenesis, especially for male gametocytes, suggesting a role for PfPuf2 in repressing gametocytogenesis and differentiation in the human parasite P. falciparum (Miao et al., 2010). In the rodent malaria parasites Plasmodium berghei (Pb)Puf2 and Plasmodium yoelii (Py)Puf2 proteins play important roles in maintenance of the sporozoite infectivity in the salivary glands of the vectors (Gomes-Santos et al., 2011; Lindner et al., 2013a; Müller et al., 2011). Whereas PbPuf1 appears to be dispensable throughout the P. berghei life cycle, disruption of PfPuf1 results in a significant reduction in the number of mature female gametocytes, indicating that PfPuf1 is required for the maintenance of female gametocytes (Shrestha et al., 2016).
In this study, we identified and functionally characterized two orthologs of a putative Puf protein, PfPuf3 (PF3D7_0621300) in P. falciparum and its syntenic ortholog PyPuf3 (PY17X_1121600) in P. yoelii. Our results showed that both PfPuf3 and PyPuf3 were essential for the asexual parasite erythrocytic cycle. Subcellular localization studies showed that both proteins were limited to the nucleus and nucleolus in the asexual stages. However, we also observed changes of PyPuf3 localization to cytosolic granules in oocysts, oocyst sporozoites and early liver stage parasites, as well as its absence in salivary gland sporozoites and in late liver stage parasites. Analyses of molecules associated with PfPuf3 identified the majority of associated proteins as being from the large ribosome subunit and the associated RNAs as originating from the internal transcribed spacer (ITS) 2 RNA sequence. This suggests that PfPuf3 is a member of the Puf-A/Puf6 subfamily and may participate in ribosomal biogenesis in Plasmodium parasites.
PfPuf3 shows higher similarity to the Puf-A subfamily than the classical Puf subfamily
A search of the PlasmoDB database using the Pumilio Puf domain identified another putative Puf gene (PF3D7_0621300) in the P. falciparum genome, which we here call PfPuf3. The PfPuf3 open reading frame is 2382 bp, located on chromosome 6 and contains no introns. PfPuf3 encodes a protein of 793 amino acids (aa) with a predicted molecular mass of ∼94.3 kDa. The predicted Puf RBD of PfPuf3 is located from aa 166 to 526. Whereas the amino acid sequences of the Puf RBD of PfPuf1 and PfPuf2 are 41 and 58% identical to the respective sequences of the Pum RBDs from Drosophila melanogaster (Cui et al., 2002), PfPuf3 and Pum RBD only have 15% amino acid identity overall. Like the Puf1 and Puf2 genes, Puf3 is highly conserved in Plasmodium spp. and a Puf3 ortholog is present in the genome of each sequenced Plasmodium species (Fig. 1A). The full-length PfPuf3 sequence shares 45% amino acid identity with the orthologous PyPuf3 (PY17X_1121600) (Fig. S1A). A search for homologous structures based on the PfPuf3 primary sequence by using NCBI Blastp identified Homo sapiens (Hs)Puf-A as the most similar structure in the Protein Data Bank (Qiu et al., 2014). However, sequence alignment showed that PfPuf3 shares only 15.1% amino acid identity in the 511-residue (aa 156–774) fragment that aligned with HsPuf-A (Fig. S1B). Within the aligned region, PfPuf3 has a long insertion spanning aa 573–687 (Fig. 1B) that was predicted to be intrinsically disordered and therefore was left as an unmodeled region. Similar to PfPuf3, PyPuf3 shares only 16.3% amino acid identity in the fragment (aa 129–665) that aligned with HsPuf-A (Fig. S1C). Further phylogenetic analysis using the 11 Puf proteins from Plasmodium spp. and ten representatives of the three Puf subfamilies from model organisms showed that Plasmodium Puf3 proteins form a subgroup together with the HsPuf-A and S. cerevisiae Puf6 proteins (Fig. 1A).
To further predict the structural features of PfPuf3, homology-based structural modeling was carried out using the HsPuf-A crystal structure as the template, based on our preliminary search results. Compared with the crescent-shape structure of the classical Puf protein Pumilio and the C-shaped structure of Nop9, the homologous model of PfPuf3 more readily adopts an L-like shape, similar to that of HsPuf-A (Fig. 1B,C) (Qiu et al., 2014). By using a similar modeling approach, we found that the PyPuf3 model also superimposed well with the HsPuf-A (Fig. S1D,E). Taken together, these sequence comparisons and predicted structures suggest that PfPuf3 and PyPuf3 are likely members of the Puf-A subfamily. Comparison of the predicted RNA base-interacting residues in the Puf-A group showed relatively higher levels of conservation in the first eight repeats, whereas the last three repeats showed little sequence conservation (Fig. 1D).
Puf3 is essential for asexual development
To study the function of PfPuf3, plasmid pHD22Y/Puf3KO-GFP was transfected into 3D7 parasites to disrupt PfPuf3. Southern blot analysis of resistant parasites that emerged after drug selection showed that none of the six independent transfection experiments led to the disruption of PfPuf3 (Fig. S2A). In P. yoelii, we attempted to genetically delete PyPuf3 by double homologous recombination in two experiments. Similarly, the genotyping PCR results demonstrated that PyPuf3 could not be deleted in either attempt (Fig. S2B). These results from two parasite species collectively indicate that Puf3 is likely essential for the intraerythrocytic developmental cycle (IDC).
PfPuf3 is differentially expressed in the IDC
Microarray analysis shows that PfPuf3 mRNA is present throughout the IDC, with its mRNA abundance being the lowest in the late trophozoite stage (∼35 h) (PlasmoDB). To detect PfPuf3 protein expression, the endogenous PfPuf3 was tagged with a PTP tag (comprising a protein C epitope, a TEV protease site and a protein A epitope) at the C-terminus. Correct integration of the transfected plasmid at the PfPuf3 locus was verified by integration-specific PCR, which produces the tagged protein (detected by western blotting) when proper insertion occurs. Analysis of three clones from each parasite strain (3D7, NF54) by genotyping PCR revealed bands that were consistent with the expected size of the transgenic locus (1233 bp) (Fig. S2C). Moreover, western blotting of a wild-type 3D7 strain and PfPuf3-PTP clone 1 showed that a protein band of ∼114 kDa was detected with the anti-protein C antibodies only in the PfPuf3–PTP clone, which is consistent with the fusion of the 20 kDa PTP tag to the 94.3 kDa PfPuf3 (Fig. S2C). Further examination of PfPuf3 expression dynamics by western blotting with anti-protein C antibodies revealed PfPuf3 expression during the entire IDC, with the highest and lowest expression in the early and late trophozoites, respectively (Fig. 2A,B), which is similar to the PfPuf3 mRNA expression profile.
PfPuf3 is a nucleolar protein
In order to analyze the subcellular distribution and localization of PfPuf3, asexual blood stage parasites were separated into nuclear and cytoplasmic fractions. SDS-PAGE and immunoblotting of these fractions showed that PfPuf3–PTP is predominantly present in the nuclear fraction (Fig. 2C), as determined by using histone H3 and PfAldolase as markers of the nuclear and cytoplasmic fractions, respectively (Thavayogarajah et al., 2015).
To study the subcellular localization of PfPuf3, we tagged the endogenous PfPuf3 with GFP at its C-terminus. Correct integration of the transfected plasmid at the PfPuf3 locus was verified by genotyping PCR, which generated a 1249 bp PCR product with primers gF and g-GR (see Table S1) in three clones of each of the two transfection attempts using 3D7 and NF54 (Fig. S2C). Fluorescence microscopy analysis of the PfPuf3::GFP parasites detected expression in all asexual stages, with foci that were preferentially located towards the nuclear periphery in a Hoechst 33342-negative area in ring, early trophozoite and early schizont stages (Fig. 3A). This pattern of localization is reminiscent of the nucleolar localization reported for two nucleolar proteins NuProC2 and NuProC3 (Oehring et al., 2012). To determine whether PfPuf3 is indeed localized in the nucleolus, we performed colocalization analysis using the known nucleolar marker PfNop1 (Figueiredo et al., 2005). Using the PfPuf3::GFP line with PfNop1-tdTomato expressed from a nuclear plasmid, we found excellent overlap of these two tagged proteins in ring, trophozoite and schizont stages (Fig. 3A), indicating that PfPuf3 is a nucleolar protein.
PyPuf3 displays differential expression and localization patterns during mosquito and liver stage development
Similarly, we interrogated the expression and localization of Puf3 in the rodent-infectious P. yoelii parasite, as analyses throughout the entire parasite life cycle are practical in this species. PyPuf3 was successfully tagged with GFP at its C-terminus as confirmed by genotyping PCR (Fig. S2D). Similar to P. falciparum, immunofluorescence assay (IFA) micrographs showed that PyPuf3::GFP expression in the asexual blood stage was confined to the nucleus with a pattern that was also suggestive of nucleolar localization (Fig. 3B). After transmission of the transgenic parasite line to mosquitoes, PyPuf3::GFP was observed to be concentrated into banded foci in the developing oocysts (Fig. 3C). When sporozoites were dissected from the developing oocysts on day 10, PyPuf3::GFP was found in punctate foci of the oocyst sporozoites (Fig. 3C). Some puncta are located adjacent to the DAPI-stained nuclear DNA, in agreement with its observed localization in asexual blood stages. However, additional puncta were found in the cytoplasm in positions similar to that of PyPuf2 (Fig. 3C; Fig. S3A) (Lindner et al., 2013a). Intriguingly, the PyPuf3::GFP foci were not observed in mature salivary gland sporozoites (Fig. 3C, Fig. S3A). During mid-liver stage development (24 h), large foci containing PyPuf3::GFP were visible, and they were adjacent to DAPI-stained DNA but had minimal overlap (Fig. 3D). However, in late (48 h) liver stage schizonts, PyPuf3::GFP expression was not detectable (Fig. 3D, Fig. S3B). These data indicate that the parasite can rapidly modulate the abundance of PyPuf3 during parasite maturation, and in addition to its putative role in the nucleolus, that it may also play additional roles in the cytoplasm in the mosquito stages of parasite development.
PfPuf3 is associated with ribosomal proteins
In order to identify proteins that associate with PfPuf3 that might indicate its role in the nucleolus, we used a tandem affinity purification (TAP) procedure combined with mass spectrometry (MS) to analyze PfPuf3 in asexual blood-stage parasites. The presence of PfPuf3–PTP in the transgenic parasite lysates and TAP fractions was verified by western blotting using rabbit anti-protein C antibodies, and was not detected in wild-type parasites (Fig. S4). Liquid chromatography coupled to tandem MS (LC/MS/MS) analysis of the proteins after TAP detected a total of 124 proteins, including PfPuf3, in three biological replicates. A total of 31 proteins were identified after establishing a highly stringent false discovery rate (FDR) at 1% by using the Significance Analysis of INTeractome (SAINT) algorithm (Table 1). Among them, 26 proteins are putative 60S ribosomal proteins. The remaining five include the nuclear import receptor karyopherin β, the putative nucleolar GTP-binding protein 1 (Nog1), the putative ribosome biogenesis protein MRT4, and histones H3 and H4. All of these are associated with ribosome biogenesis in model organisms (Chook and Süel, 2011; Harnpicharnchai et al., 2001; Jakel and Gorlich, 1998; Jensen et al., 2003; Kallstrom et al., 2003; Rout et al., 1997). These data suggest that PfPuf3 might be specifically involved in the biogenesis of the 60S ribosome.
PfPuf3 is associated with ITS2 sequences of rRNA
To identify potential target RNAs of PfPuf3, an RNA immunoprecipitation (RIP) with anti-protein C antibodies followed by RNA-seq analysis was conducted for the transgenic PfPuf3–PTP and control 3D7 parasites in the asexual blood stage. Illumina sequencing showed that in both immunoprecipitated and input RNAs from the PfPuf3–PTP parasite, the overwhelming majority of the sequence reads were from rRNAs, mostly 28S, 18S and 5.8S rRNAs from the A-type rDNA clusters located in chromosomes 5 and 7. Specifically, 28S rRNAs from chromosomes 5 and 7 had 1.47- and 1.49-fold enrichment, respectively, whereas reads from the 18S and 5.8S rRNAs were 0.47- and 0.25-fold lower in the immunoprecipitated RNA than the input RNA (Fig. 4A). In comparison, ITS1 and ITS2 sequences from these two rDNA clusters were at least 100 times less abundant than the mature rRNA sequences (Fig. 4B). Despite this, ITS2 sequences from chromosomes 5 and 7 had 14.7- and 18.4-fold enrichment, respectively, in the immunoprecipitated RNA compared to the input RNA, whereas there was essentially no enrichment for the ITS1 sequences (Fig. 4B). A side-by-side comparison of sequence enrichment in the RIP versus input between transgenic PfPuf3–PTP and wild-type 3D7 control parasites clearly demonstrated significant enrichment of the ITS2 sequences from both rDNA clusters (Fig. 4C). This is further illustrated when sequence reads from the two A-type rDNA clusters were combined (Fig. 4D). Taken together, this analysis suggests that PfPuf3 is primarily associated with the ITS2 sequences.
More than 100 Puf protein members have been identified in various eukaryotes from unicellular species to human. They fall into three distinct subfamilies. Here, we have identified a novel Puf member, Puf3, that is found in each of the sequenced Plasmodium genomes and shares limited sequence identity with the classical Puf protein Pum. A phylogenetic analysis showed that Puf3 diverges from Puf1 and Puf2, and is grouped with the HsPuf-A/ScPuf6 subfamily (Fig. 1A). We provide data that indicates that Puf3 is likely involved in large ribosomal subunit biogenesis, and perhaps in the maturation of the 28S rRNA.
A previous study showed that HsPuf-A/ScPuf6 subfamily proteins are involved in large ribosomal subunit biogenesis (Li et al., 2009; Qiu et al., 2014; Yang et al., 2016). Consistent with this, we also found that PfPuf3 is a nucleolar protein and colocalized with a nucleolar marker PfNop1 in the asexual stages (Figueiredo et al., 2005; Mancio-Silva et al., 2008, 2010). In addition, PfPuf3 was co-purified with 26 putative 60S ribosomal proteins and five assembly factors including the nuclear import receptor karyopherin β, the putative nucleolar GTP-binding protein 1 (Nog1), a putative ribosome biogenesis protein MRT4, and H3 and H4. These proteins all have important functions in the biogenesis of the large ribosomal subunit in yeast (Chook and Süel, 2011; Jensen et al., 2003; Keener et al., 1997; Rodríguez-Mateos et al., 2009a,b; Woolford and Baserga, 2013). Although none of these factors have been well studied in P. falciparum, the association of PfPuf3 and other 60S ribosomal proteins suggests that they may have conserved functions in ribosome assembly. It is noteworthy that whereas these 32 proteins associated with the 60S ribosomes were identified under stringent conditions, additional 40S ribosomal proteins were also detected when the stringency was lowered. These results suggest that PfPuf3 might be involved in 60S ribosomal subunit biogenesis. Further analysis of PfPuf3 and putative ribosome assembly factors in the Puf3 complex may yield mechanistic insights into how these factors cooperate in the 28S rRNA maturation and large ribosome subunit assembly.
It is noteworthy that PfPuf3 expression peaks in early trophozoites, which is compatible with active protein synthesis during trophozoite growth (Fig. 2A). In late trophozoite stage when nuclear division begins, PfPuf3 showed minimal levels of expression. While the reason for the PfPuf3 decline is not known, we speculate that at the earlier stage of nuclear division, the nucleoli might be less well organized in the newly formed nuclei, and presumably rRNA transcription and ribosome assembly decrease to a minimum due to a lesser requirement for the activities and functions of PfPuf3 and ribosome assembly factors. In schizonts, rRNA synthesis and ribosome assembly occur in individual merozoites, coincidentally with increased PfPuf3 expression and localization of PfPuf3 in individual nucleoli. Furthermore, the three putative ribosome assembly factors (karyopherin β, Nog1 and MRT4) all show a parallel mRNA expression profile to that of PfPuf3, suggesting that they might be integral components of the ribosome assembly complex during different stages of the IDC. Expression and localization studies of PyPuf3 in asexual blood stages show a similar pattern.
Interestingly, PyPuf3 expression in mosquito and liver stage parasites is different. Detectable expression is limited to banded foci in oocysts, and foci in oocyst sporozoites, and large foci in early liver stage parasites (Fig. 3C,D). In addition to DAPI-adjacent puncta, cytosolic puncta are also observed for PyPuf3 in oocyst sporozoites, which is reminiscent of what is seen for S. cerevisiae Puf6, which undertakes the function of classical Puf proteins in the translational repression of mRNA (Gu et al., 2004). This shifting pattern of PyPuf3 expression and localization during mosquito stage development, as compared to putative nucleolar localization in asexual blood stage, oocyst sporozoite and mid-liver stage parasites suggests that PyPuf3 may play additional roles in sporozoites. Further investigations should address what these potential functions of Puf3 are. Classical Puf proteins adopt a crescent shape and recognize single-stranded RNA in a sequence-specific manner. Similarly, the S. cerevisiae Nop9 family proteins bind specifically to the ITS2 D-A2 fragment of 20S pre-rRNA to protect ITS1 from premature cleavage in the nucleolus (Zhang et al., 2016). Distinct from these two subfamilies, the HsPuf-A/ScPuf6 family proteins can interact with single- or double-stranded RNA through non-specific sequence contacts with the phosphate backbone (Qiu et al., 2014). Homology modeling of Puf3 shows that excellent superimposition of Puf3 and HsPuf-A is possible, suggesting that Puf3 may also nonspecifically interact with nucleic acids. Our RIP and subsequent RNA-seq analysis with PTP-tagged PfPuf3 clearly showed association of PfPuf3 with the 28S rRNA and the ITS2, strongly suggesting that PfPuf3 is involved in the processing of the large ribosome subunit. While significant enrichment of the ITS2 RNA over ITS1 RNA suggests a sequence-specific interaction, it is unknown whether the interaction between PfPuf3 and the ITS2 region of the pre-rRNA is direct, or whether it indirectly associates via a specific ITS2-binding protein.
Whereas Puf-A appears to play critical roles in both human and zebrafish development (Fan et al., 2013; Kuo et al., 2009), the S. cerevisiae Puf6 is nonessential under standard culture conditions. Deletion of S. cerevisiae Puf6 results in a slow-growth phenotype at 20°C and it is linked to 60S ribosome biogenesis defects (Li et al., 2009). Our evidence strongly indicates a similar involvement of Plasmodium Puf3 in the biogenesis of the large ribosome subunit, and this function is likely essential in Plasmodium, as the Puf3 gene could not be deleted in either P. falciparum or P. yoelii. Future studies could use conditional knockdown systems to investigate which steps of the 60S ribosome assembly are affected by Puf3, whether this interaction is direct or indirect, and whether the binding involves a specific sequence motif in the ITS2.
MATERIALS AND METHODS
Identification of PfPuf3, and phylogeny and homology-based modeling
PfPuf3 (PF3D7_0621300, PFF1030w) was identified by searching the PlasmoDB database (http://www.plasmodb.org) with the Pum domain using a hidden Markov model. A total of 19 GenBank entries with complete Puf domains, representing three Puf subfamilies, were retrieved for phylogenetic analysis. The Puf domains of individual Puf proteins were trimmed and used to generate the data matrix to infer phylogenetic relationships among these Puf family members. Multiple alignment was performed using the CLUSTALW program and phylogenetic analysis was performed using MEGA 7.0. Preliminary prediction of the protein model was carried out by submitting the full-length PfPuf3 sequence to the online (PS)2-v2 server (http://184.108.40.206/~ps2v2/index.php), which is good for homologous modeling when sequence similarities between the targets and templates are 15–25% (Chen et al., 2009). This program employs a S2A2 matrix and a position-specific sequence profile (PSSM) generated by PSI-BLAST, which together ‘blend’ the amino acid and structural propensities. For homology-based modeling, instead of using the default ‘Automatic’ template selection, we specified the 2.15 Å resolution crystal structure of HsPuf-A as the modeling template based on the preliminary model search result.
Generation of transgenic P. falciparum lines
To disrupt PfPuf3, a PfPuf3 fragment [nucleotides (nts) 90–1005] was amplified from P. falciparum genomic DNA using primers F1 and R1 (all primers used in this study are listed in Table S1). Then the PCR product was cloned into pBluescript, to fuse with GFP, and pDT 3′ UTR. This cassette was subcloned into pHD22Y at the BamHI and NotI sites to produce pHD22Y/Puf3KO-GFP (Fidock and Wellems, 1997). To tag PfPuf3 with GFP and PTP at the C-terminus (Schimanski et al., 2005; Takebe et al., 2007; Xu et al., 2010), a PfPuf3 nt 1450–2361 was amplified using primers F2 and R2, and cloned into pBluescript SK, to fuse with GFP and PTP, and pDT 3′ UTR. This cassette was digested with BamHI and NotI and subcloned into pHD22Y to generate pHD22Y/Puf3-GFP/PTP. Colocalization studies were performed with the PfPuf3::GFP parasites containing an additional pCC4 plasmid for expression of the nucleolar marker PfNop1 with a C-terminal tdTomato tag. To create this plasmid, the full-length PfNop1 and tdtomato coding sequences were amplified with two pairs of primers (Nop1F and Nop1R, tdTomF and tdTomR). The Nop1 PCR product was cloned into pBluescript to fuse with tdTomato and pDT 3′ UTR. The PfNop1 expression cassette was cloned into the modified vector at SpeI and NotI sites to obtain the construct PfNop1-tdTom-pCC4. For transfection, P. falciparum 3D7 and NF54 parasites were maintained and synchronized through sorbitol treatment (Lambros and Vanderberg, 1979). Parasite transfection was performed by using the RBC loading method (Rug and Maier, 2013). To enrich parasites that have integrated the constructs into the genome, parasites were subjected to two rounds of on–off drug selection (Crabb and Cowman, 1996). Drug-resistant parasites were screened by integration-specific PCR and Southern blotting to detect plasmid integration, while western blotting was used to confirm the generation of the variant protein. Single clones of parasites with stable integration of the constructs were obtained by limiting dilution (Rosario, 1981).
Generation of transgenic P. yoelii with PyPuf3 tagging and knockout
To study the function of the PfPuf3 ortholog PyPuf3 in P. yoelii, two clonal transgenic parasite lines were created by double homologous recombination. First, the PyPuf3 coding sequence was replaced with a drug resistance cassette (HsDHFR) and a GFPmut2 expression cassette by using targeting sequences that were PCR amplified from genomic DNA with the Phusion polymerase (NEB) supplemented with 5 mM MgCl2 via primers yKO5F, yKO5R, yKO3F and yKO3R (Table S1). The PCR products were gel purified, fused by means of sequence overlap extension PCR, and cloned into the pDEF-GFP vector (pSL0444). Second, the PyPuf3 protein was tagged with GFP at its C-terminus using the same strategy as above, but using primers yKO3F, yKO3R, yCtOF and yCtOR for the amplification of targeting sequences (Table S1). PCR products were fused and inserted into a pDEF-GFP-Ct plasmid (pSL0442). Linearized plasmids were introduced into P. yoelii (17XNL) parasites using an Amaxa Nucleofector 2b. Transfected parasites were intravenously injected into Swiss Webster mice (Envigo) and selected with pyrimethamine supplied in the drinking water as previously described (Munoz et al., 2017). The presence of transgenic parasites (PyPuf3−, PyPuf3::gfp) was verified by genotyping PCR and by the presence of GFP. For the P. yoelii work, a total of 22 female Swiss Webster mice were used. The animal use protocol was approved by the Pennsylvania State University Institutional Animal Care and Use Committee (#42628).
PfPuf3 expression during IDC
To study PfPuf3 protein expression during the IDC, synchronization was performed by two rounds of sorbitol treatment at the ring state. Synchronized parasites were lysed by sonication (three pulses of 10 s each). Protein concentration was measured by using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific), and equal amounts of parasite lysate (50 μg) at each developmental stage were separated by 10% SDS-PAGE and transferred to nitrocellulose membranes. Western blotting was performed by using a standard procedure with rabbit anti-Protein C (1:1000, GenScript, Cat# A00637) and rabbit anti-PfAldolase (1:3000, kindly provided by Dr Tobias Spielmann, Parasitology Section, Bernhard Nocht Institute for Tropical Medicine, Germany) as the primary antibodies and horseradish peroxidase-conjugated goat anti-rabbit-IgG (1:3000) as the secondary antibodies. The results were visualized with the ECL detection system using X-ray film (Invitrogen, Cat# WP20005).
To estimate the distribution of PfPuf3 in the cytoplasmic and nuclear compartments of the parasite, ∼100 μl of parasite pellet of the PfPuf3–PTP line was resuspended in 500 μl of a hypotonic buffer A [10 mM Hepes, pH 7.9 at room temperature, 1.5 mM MgCl2, 10 mM KCl, 0.5 mM DTT, 0.5 mM EDTA and 1% (v/v) protease-inhibitor cocktail (Roche)] and incubated on ice for 10–15 min. The parasites were mechanically lysed by at least 40 strokes in a Dounce homogenizer with a loose pestle and then centrifuged at 700 g for 20 min at 4°C. The supernatant was centrifuged at 10,000 g for 10 min to obtain the cytoplasmic extract. The pellet was resuspended in five volumes of buffer B [20 mM HEPES, pH 7.9 at room temperature, 20% (v/v) glycerol, 200 mM KCl, 0.5 mM DTT, 0.5 mM EDTA, 0.5% (v/v) NP40 and protease inhibitor cocktail] and dounced for 40 times with a tight pestle. The homogenate was centrifuged at 10,000 g for 10 min to obtain the nuclear extract. Protein extracts were resolved by 10% SDS-PAGE and detected by immunoblotting using the anti-Protein C antibodies (1:1000, GenScript, Cat# A00637) and anti-histone H3 antibodies (1:1000, Sigma, Cat# H0164) as a control for nuclear proteins. For a cytoplasmic protein control, a rabbit antiserum against aldolase was used, which specifically reacts with PfAldolase in the parasite (Knapp et al., 1990).
A P. falciparum 3D7 line with PfPuf3::GFP and tdTomato-tagged PfNop1 were used for protein localization studies. Culture medium was removed gently and infected red blood cells were resuspended in 1× PBS with Hoechst 33342 (1:8000) (Chazotte, 2011) and incubated at room temperature for 5 min. Cells were centrifuged at 500 g for 3 min, and washed with 1× PBS. Then cells were placed onto a slide and covered with a coverslip. Images were captured by using a Nikon Eclipse E600 epiﬂuorescence microscope.
PyPuf3 expression in P. yoelii was examined by using the transgenic PyPuf3::GFP parasite line. PyPuf3::GFP expression was observed in blood stages, oocyst and salivary gland sporozoites, and liver stages by performing an indirect immunofluorescence assay (IFA), while its expression in day 7 and day 10 oocysts was observed by live-cell fluorescence microscopy. All samples for IFA were prepared as previously described (Lindner et al., 2014). Parasites were stained with the following primary antibodies: rabbit anti-GFP (1:1000, Invitrogen, Cat# A11122), rabbit anti-PyACP (1:1000), mouse anti-GFP (1:1000, DSHB, Clone 4C9), rabbit anti-merozoite surface protein (MSP) (1:1000, BEI Resources, MRA23) and mouse anti-P. yoelii circumsporozoite protein (PyCSP) (1:1000, Clone 2F6) (Lindner et al., 2013b) antibodies. Secondary antibodies used for all stages were Alexa Fluor-conjugated (Alexa Fluor 488 or Alexa Fluor 594) and specific to rabbit or mouse IgG (1:1000, Invitrogen, Cat# A11001, A11005, A11008, A11012). 4′,6-Diamidino-2-phenylindole (DAPI) was used to stain nuclei and samples were mounted with the VectaShield antifade reagent (VWR), covered with a coverslip, and sealed with nail polish prior to visualization. Fluorescence and DIC images were taken by using a Zeiss fluorescence and phase contrast microscope (Zeiss Axioscope A1 with 8-bit AxioCam ICc1 camera) with a 40× or 100× oil objective and processed with Zen imaging software.
Isolation of a PfPuf3 protein complex and MS analysis
TAP of PfPuf3 was performed by using the PfPuf3-PTP parasite line or wild-type (3D7) parasites as control (Schimanski et al., 2005; Takebe et al., 2007; Xu et al., 2010). Parasites were isolated as described above. The parasite pellet was lysed with five times the volume of the parasite pellet of pre-cooled PA150 buffer [150 mM KCl, 20 mM Tris-HCl, pH 7.7 at room temperature, 3 mM MgCl2, 0.5 mM DTT, 0.1% (v/v) Tween 20] containing a protease inhibitor cocktail (one pill in 10 ml PA150 buffer, Roche). Then the parasites were manually lysed by 60–100 strokes in a Dounce homogenizer with a tight pestle and centrifuged at 16,000 g for 20 min at 4°C. The supernatant was incubated with 100 μl (settled volume) of IgG–agarose beads (GE Healthcare) at 4°C for 2 h. The beads were washed three times with pre-cooled PA150 and equilibrated twice with the TEV buffer [150 mM KCl, 20 mM Tris-HCl, pH 7.7 at room temperature, 3 mM MgCl2, 0.5 mM EDTA, 1 mM DTT, 0.1% (v/v) Tween 20]. Then the beads were incubated overnight with 2 ml of TEV buffer containing 100–150 U of TEV protease with gentle rotation at 4°C. The supernatant was obtained by centrifugation at 500 g for 5 min at 4°C. Subsequently, 200 μl of anti-protein C affinity matrix (Roche) equilibrated with 1 ml of PC150 [150 mM KCl, 20 mM Tris-HCl, pH 7.7 at room temperature, 3 mM MgCl2, 1 mM CaCl2, 0.1% (v/v) Tween 20] was added into the supernatant with 7.5 μl 1 M CaCl2. After incubation for 2 h at 4°C, the beads were washed four times with PC150 and eluted with a buffer containing 10 mM EGTA and 5 mM EDTA. Western blotting was used to check for the presence of PfPuf3 in the purified complex; PfAldolase served as a control. The eluted protein complex was analyzed by nano-LC/MS/MS with a Waters NanoAcquity HPLC system interfaced to a Q Exactive™ Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo Scientific). Peptides were loaded on a trapping column and eluted over a 75 μm analytical column at 350 nl/min. MS and MS/MS were performed at 70,000 full width and half maximum (FWHM) and 17,500 FWHM resolution, respectively. The 15 most abundant ions were selected for MS/MS and Mascot Generic Files (MGF) from RAW files were extracted by use of Proteome Discoverer v1.4 (Thermo Scientific). Parasite proteins were identified by searching the Uniprot P. falciparum protein database (v01/2014, 5369 entries) concatenated with the SWISS-PROT human database (20,160 entries). The combined database (25,529 entries) was reversed and appended back to the forward database (51,058 entries total). All peak list files (MGFs) were analyzed by using Mascot (Matrix Science; London, UK; version 2.5.1). Search parameters included trypsin digestion (C-terminal K and R cleavage) full cleavage with 2 missed sites, and a fixed modification of +57 on C (carbamidomethyl), and variable modifications −17 on n (Gln->pyro-Glu), +1 on NQ (deamidated), +16 on M (oxidation), +42 on n (acetyl), fragment ion mass tolerance of 0.02 Da, peptide mass tolerance of 10 ppm. Mascot DAT files were parsed into the Scaffold software for validation, filtering and to create a nonredundant list per sample. Data were filtered with a 1% FDR for both protein and peptide analyses, and proteins reported had at least two unique peptides detected. Unique proteins were identified after controlling the FDR at 1% using the SAINT algorithm (Table S2) (Choi et al., 2012). Data from these proteomic experiments have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE database (dataset identifier PXD007978).
RIP and RNA-Seq
The transgenic parasite line PfPuf3-PTP was employed for RIP and RNA-seq analysis. RNA was isolated from the eluate after the TEV protease cleavage step. RNA libraries were made using the KAPA stranded RNA-seq library preparation kit instruction (Roche). Briefly, RNA was fragmented by heating at 94°C for 8 min. Then cDNA was generated from the RNA template with random primers by means of a reverse transcriptase reaction. The first-strand cDNA was converted to double-stranded DNA, which was subsequently end-repaired to create blunt ends. Addition of a single nucleotide to the 3′ end of the double-stranded DNA fragments was performed. Afterwards, 3′ dTMP adapters were ligated to the 3′ dAMP library fragments. PCR amplification was performed to increase the amount of the library, and the RNA library was analyzed for size distribution and quality. Finally, the libraries were sequenced on an Illumina HiSeq 2500 to produce 150 base pair single end reads. Illumina adapter sequence removal and quality trimming of reads were performed with Trimmomatic (Bolger et al., 2014). A sliding window approach was used to trim the reads once the average quality within a 4 base pair window dropped below the threshold of 18. Only reads that had a minimum length of 50 base pairs were retained. Reads were then mapped to the P. falciparum reference genome (Pf3D7_01_v3) with Tophat2 (Kim et al., 2013). Reads that mapped to each gene were counted by using featureCounts (Liao et al., 2014). Read counts were normalized to their library size by using a DESeq normalization method (Anders and Huber, 2010). Normalized read counts were used for further analysis. Final RNA-seq datasets for both wild-type and PfPuf3-PTP have been deposited in the GEO database under the accession number GSE105126, and are also provided as Table S3.
Conceptualization: L.C.; Methodology: X. Liang; Software: G.D., I.A.; Formal analysis: X. Liang, K.H., G.D., A.S., J.M., S.L., L.C.; Investigation: X. Liang, K.H., F.S., A.S., X. Li, I.A., J.M.; Data curation: X. Liang, A.S.; Writing - original draft: X. Liang, K.H., J.M.; Writing - review & editing: S.L., L.C.; Supervision: S.L., L.C.; Project administration: L.C.; Funding acquisition: S.L., L.C.
This research was supported by the National Institutes of Health (grants R01AI104946 to L.C., R01AI123341 and K22AI101039 to S.L.). Deposited in PMC for release after 12 months.
Final RNA-seq datasets for both wild-type and PfPuf3-PTP have been deposited in the GEO database under the accession number GSE105126 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE105126). Proteomic data ProteomeXchange Consortium via the PRIDE database under dataset identifier PXD007978 (http://proteomecentral.proteomexchange.org/cgi/GetDataset?ID=PXD007978).
The authors declare no competing or financial interests.