Analysis of the interactome of the Ser/Thr Protein Phosphatase type 1 in Plasmodium falciparum

Background Protein Phosphatase 1 (PP1) is an enzyme essential to cell viability in the malaria parasite Plasmodium falciparum (Pf). The activity of PP1 is regulated by the binding of regulatory subunits, of which there are up to 200 in humans, but only 3 have been so far reported for the parasite. To better understand the P. falciparum PP1 (PfPP1) regulatory network, we here report the use of three strategies to characterize the PfPP1 interactome: co-affinity purified proteins identified by mass spectrometry, yeast two-hybrid (Y2H) screening and in silico analysis of the P. falciparum predicted proteome. Results Co-affinity purification followed by MS analysis identified 6 PfPP1 interacting proteins (Pips) of which 3 contained the RVxF consensus binding, 2 with a Fxx[RK]x[RK] motif, also shown to be a PP1 binding motif and one with both binding motifs. The Y2H screens identified 134 proteins of which 30 present the RVxF binding motif and 20 have the Fxx[RK]x[RK] binding motif. The in silico screen of the Pf predicted proteome using a consensus RVxF motif as template revealed the presence of 55 potential Pips. As further demonstration, 35 candidate proteins were validated as PfPP1 interacting proteins in an ELISA-based assay. Conclusions To the best of our knowledge, this is the first study on PfPP1 interactome. The data reports several conserved PP1 interacting proteins as well as a high number of specific interactors to PfPP1. Their analysis indicates a high diversity of biological functions for PP1 in Plasmodium. Based on the present data and on an earlier study of the Pf interactome, a potential implication of Pips in protein folding/proteolysis, transcription and pathogenicity networks is proposed. The present work provides a starting point for further studies on the structural basis of these interactions and their functions in P. falciparum. Electronic supplementary material The online version of this article (doi:10.1186/s12864-016-2571-z) contains supplementary material, which is available to authorized users.


Background
Malaria continues to be a major health problem and a leading cause of child mortality of the inter-tropical regions despite extensive research efforts. The present situation is also increasingly complicated by the emergence of parasite resistance to multiple drugs much more rapidly than the development of novel anti-malarials. A major obstacle in devising new control tools is our limited knowledge of basic parasite biology and the paucity of identified potential intervention targets. Plasmodium species are obligate intracellular protozoan parasites that undergo a number of developmental stages in the vertebrate host and the invertebrate vector. During the last decade, several studies strongly indicated that protein phosphorylation and dephosphorylation processes, governed by kinases and phosphatases respectively, play a central and essential role in Plasmodium cell cycle and developmental regulation [1,2].
In Plasmodium falciparum (Pf), the most deadly apicomplexan parasite, efforts have begun for an examination of the biological roles of protein kinases and phosphatases and their potential as drug targets. In this context, biochemical and cloning studies have reported enzymatic activities and the genes of catalytic subunits related to plasmodial phosphatases. These include PfPP1c, PfPP2A, PfPP2B and PfPP2C, which have been shown to be highly conserved during evolution, suggesting an essential role for these enzymes [3][4][5]. With respect to the Pf Protein Phosphatase type 1 catalytic subunit or PfPP1c, it cannot be considered per se as a suitable drug target (> 80 % identity with human and yeast PP1 or GLC7). Indeed, complementation studies showed that PfPP1 was able to rescue yeast mutated in PP1 [6]. In P. falciparum functional studies suggested that its activity predominates over the other phosphatases in blood parasites [7]. Finally, phenotypic PfPP1 gene knockdown and the use of phosphatase inhibitors indicated that it seems to be essential for parasite asexual development in erythrocytes [2,4]. At this stage, it is important to note that free PP1c seems to be toxic for the cell [8] but a considerable number of proteins interacting with PP1c have been reported that direct the localization, specificity and the level of its activity in yeast and higher eukaryotic cells [9]. Indeed, about 200 proteins have been shown to physiologically interact with PP1c, constituting a 'PP1c platform' that provides a framework offering diverse functions to this enzyme in cell division, transcription and apoptosis for instance [10]. Structural and functional studies revealed that the interaction of regulators with PP1 involved several binding sites, including mainly the so-called RVxF and SILK motifs [8,11]. The most studied binding motif is the RVxF for which it has been established a first short consensus sequence as [RK]x 0-1 [VI]{P} [FW] where x could be any amino acid and {P} any amino acid except P [12].
The importance of these proteins in the regulation of PP1 activity in vitro and in vivo prompted us to identify the regulators of PfPP1c and to address their functions. Our initial focus based on the completion of the Pf genome and the delineation of known PP1 regulators has revealed the existence of four conserved regulators. So far, three gene products, Pf Leucine Rich Repeat 1 (PfLRR1), Pf Inhibitor 2 (PfI2) and Pf Inhibitor 3 (PfI3) have been thoroughly explored to define their functions in this parasite. Biochemical and structure-activity relationship studies demonstrated that these regulators indeed bind to PfPP1. In these studies, we showed that P. falciparum PP1 is submitted to a control of its activity by PfLRR1 (a homolog of yeast SDS22), Inhibitor-2 (PfI2) and -3 (PfI3) with substantial differences compared to the human orthologs for Inhibitor 2 and 3 [13][14][15][16][17][18]. PfI2 exhibits an inhibitory role on PfPP1 activity, the short RVxF binding motif, not present in human I2, a secondary Fxx[RK]x[RK] binding motif and a peptide sequence 30 % shorter than its human homolog. PfI3, although it contains the RVxF consensus motif, does not seem to be an inhibitor but rather an activator of PfPP1 in vitro and is unable to complement I3 deficient yeast, suggesting a specific role for PfI3 on PfPP1. In P. falciparum, our reverse genetic analyses suggest that PfI2 as well as PfI3 are essential for parasite growth. Further, NMR analysis showed that the main domain of PfI3 binding to PfPP1 is within residues P 39 -W 45 . The mutation of the RVxF motifs in both proteins significantly affects the binding to PP1. Finally, peptides derived from these binding motifs have been shown to inhibit P. falciparum growth in vitro [17,18].
These observations underscore the importance of PP1 regulators and promote an increased interest to identify P. falciparum PP1 interacting proteins (Pips). In the present work, biochemical approaches, yeast two-hybrid screens and in silico analysis were used to define the Pips that are the starting point to elucidate PP1 interaction networks in P. falciparum and to provide new pathways for further studies to reveal potential targets for specific pharmacological intervention.

Identification of Pips by affinity/Mass Spectrometry
In order to identify Pips expressed by P. falciparum, affinity purification on PfPP1 followed by mass spectrometry was carried out. In this approach, soluble extracts from blood stage parasites cleared as described in Materials and Methods were incubated with PfPP1 covalently bound to sepharose beads. The eluate was separated by SDS-PAGE and the proteins detected by coomassie blue staining were identified by MS-MS. As shown in Table 1, only 6 P. falciparum proteins were clearly identified based on the number of peptides analysed and the corresponding mascot scores for each protein (Additional file 1: Table S1). The identified proteins, Ornithine aminotransferase, GAPDH and Phosphoethanolamine N-methyltransferase contained the motifs KTVKF, KLVSW [19,20] which are in agreement with the above data. Further, amino acid sequence analysis of PfHSP70 showed 70 % identity with mouse and human HSP70 and 2 potential binding motifs, however only the binding motif Fxx [RK]x[RK] is present in the latter (Additional file 2: Table S2). Overall, the above approach led to the detection of a low number of direct or indirect Pips when compared to the high number of reported regulators/substrates of PP1c in other species. The low recovery by the affinity/mass spectrometry analysis could be attributed to several factors including the presence of very few free Pips in parasite extracts, their low solubility in buffers suitable for affinity column and/or the competition between Pips.

Yeast two hybrid screening of a Pf cDNA library
To further detect additional Pips in P. falciparum, a cDNA library was screened in a yeast two hybrid (Y2H) system using full-length PfPP1c as bait and clones growing on high stringency selection media to reduce the number of false positive clones. Five independent screens yielded 189 clones from a total of 3.3 × 10 6 clones screened. After sequencing, all clones showed open reading frames encoding Pf genes with 53 clones in frame with GAL4AD and 136 clones out of frame. Amino acid sequence comparison revealed that 8 proteins in frame with GAL4AD were also found among the out-of-frame clones. These data are in agreement with an earlier observation indicating that out-of-frame proteins could be considered as valid interactors in Y2H screening [21] and with recent results suggesting that translational frameshift could be means for yeast to reduce the level of expression of exogenous proteins in order to grow on selective media. Indeed, this was observed with DHFR where most of isolated clones were not in frame with GAL4AD. In addition, this work showed that a reduction of growth rate of yeast was observed when the screening was carried out with DHFR fused in frame with GAL4AD [22]. Consequently, the GAL4AD in frame and out-of-frame clones were included in our analysis. In total, sequences identified from 189 clones represented 134 different proteins (Table 2) of which 27 were obtained more than once with 10 clones for PF3D7_1202600 and 9 clones for PF3D7_0919900 (Additional file 3: Table S3). Analysis of the sequences revealed that HSP70 (PF3D7_0818900) was isolated by Y2H screens, reinforcing the idea of its interaction with PfPP1 that we observed using parasite extracts and a PfPP1 affinity column. This supports the interaction of HSP70 with PP1 and suggests that besides its chaperone function it may control PP1 activity and/or regulate its binding with other partners. The Y2H screens also identified the gene PF3D7_1020900 as a Pip. Its gene product shares 72 % identity with a yeast ADP Ribosylation Factor (ARF1) which has been shown to interact with yeast PP1 [23] (Additional file 2: Table S2). Analysis of the Pf isolated fragment did not show the presence of known binding motifs to PP1. However the analysis of the full length sequence revealed the presence of 1 RVxF motif, suggesting the presence of at least 2 PfPP1 potential binding motifs in this protein.
When the identified sequences were analyzed for the presence of known binding motifs to PP1, 30 proteins share the RVxF binding motif and 20 contain the Fxx[RK]x[RK] motif, supporting the idea of their role as potential regulators of PP1. Indeed, it was reported that 70 % of PP1 interacting proteins containing the RVxF motif are regulators of PP1 activity [26].
Further examination of Y2H screening revealed that some identified clones correspond to different portions of the same protein ( Table S3). For these proteins it seems that their binding to PfPP1 could take place through at least 2 to 3 regions, potentially involving novel binding motifs.
Among the isolated clones, 8 cDNAs representing 6 known kinases were identified as potential interacting    proteins for PP1 but none of the fragments contains any known binding motifs. This supports the idea that these kinases encompass new PP1-binding region/motifs. Examination of the full length proteins of these 6 kinases showed the presence of at least one binding motif (Additional file 3: Table S3). It would be important to check whether these motifs are also involved in interaction with PP1. Taken together, it could be speculated that these kinases could be direct substrates and/ or regulators of PfPP1 and vice versa. In addition to the interaction of PfPP1 and kinases, PfH2A, PfH2B and Pf Histone deacetylase 2 (PfHDA2) have been identified as Pips. All of these proteins are associated with nucleosome modifications and could direct the control of gene expression. Previous studies showed that the phosphorylation marks of H2A and H2B are detected along with lysine acetylation [27,28]. The interaction of H2A and PP1 has also been reported in yeast [24,29] (Additional file 2: Table S2). Taken together, these observations suggest a potential role and/or a control of the PP1 network in the marks of histones to govern diverse nuclear actions in P. falciparum.
Identification of Pips by in silico screening of the Pf genome PP1 partners have been identified by biochemical and Y2H approaches, both of which can have their limitations: the quality and quantity of protein preparations, or the incapacity of yeast to produce Pf proteins (or proteins toxic for the growth of yeast). In order to further examine the presence of Pips genes in Pf, we used an extended RVxF consensus sequence mentioned in Materials and Methods. The choice of this extended consensus sequence, referred thereafter as RVxF ext , was based on previous studies indicating its high specificity [30,31]. The first consensus sequence defined by Wakula et al. [12] was not used in this study because it is less restrictive and it occurs randomly in~25 % of all proteins [30]. The bioinformatics analysis was performed with ORFs present in PlasmoDB (v24). This screening showed the presence of 55 RVxF ext motif containing proteins in P. falciparum of which 5 proteins contain at least 2 motifs (Table 3). Two RVxF ext containing proteins were excluded as one corresponds to a pseudogene and the second is annotated twice in PlasmoDB. Interestingly, among the 55 in silico Pips, 42 exhibit at least one additional binding motif to PP1 (SILK and/or Fxx[RK]x[RK]) (Additional file 4: Table S4). In addition, 8 proteins identified in silico were also isolated by Y2H screening while 1 protein was common between the affinity/MS-MS approach and Y2H screening (Fig. 1). Among the 8 shared proteins, 6 clones identified by Y2H screening followed the RVxF ext consensus sequence. Further, homologs to 2 potential in silico PfPips (PF3D7_1107700, PF3D7_1220100) have been shown to interact with PP1 in S. cerevisiae (Additional file 2: Table S2) [23,32,33]. The in silico screening applied in this work cannot include PfLRR1 and Pf Inhibitors 2 and 3, known to bind and to regulate PfPP1 activity. Indeed, PfLRR1 binds PP1 through LRR motifs, however PfI2 and PfI3 encompass the short RVxF motif.

Validation of interaction of Pips identified
All interactors identified by the above approaches could be capable of binding directly to PfPP1. To explore this further, we decided to investigate the physical interaction of PfPP1 with some of PfPips produced as recombinant proteins. To this end, a binding assay with purified Histagged fusion Pips in E. coli and biotin labeled PfPP1 was used. This approach has already been used and validated using recombinant proteins of known partners of PfPP1 or derived peptides from binding motifs [17]. PfI2 and PfI3, previously described to interact with PfPP1, were used as positive controls. A total of 37 PfPips were successfully produced as recombinant proteins in E. coli and purified out of 68 candidates tested and corresponding to 66 different Pips.
In the case of the Pips identified by PfPP1 affinity column, 5 proteins were produced out of 6. Binding data presented in Fig. 2a showed that biotin-PfPP1 binds to the 5 proteins tested. These data likely suggests that the identified Pips are direct interactors of PfPP1.
Next, we searched to check the direct interactions of Pips identified by Y2H screening and PfPP1. Fifteen recombinant PfPips were produced out of 30 cloned cDNAs tested (Additional file 5: Figure S1, Additional file 6: Table S5). The failure for the production of some proteins could be due to the lack of expression and/or to the toxicity for bacteria. For the binding with H2A and H2B, commercially available human recombinant proteins were used as they showed 67 and 63 % identity with PfH2A and PfH2B respectively (Additional file 7: Figure S2). As depicted in Fig. 2b, biotin-PfPP1 was able to bind to 14 recombinant proteins out of 17 tested. These data confirmed a direct interaction between PfPP1 and these Pips. Moreover, this binding assay showed that Pips with or without the RVxF motif can bind PfPP1 and confirmed that proteins which are out of frame of GAL4AD could be true Pips. However, under the same conditions, 3 proteins did not show significant binding when compared to BSA. The lack of binding is unlikely to be due to an absence of coating of purified protein to wells as the quality of coating was checked using an anti-His antibody (data not shown). It could be argued that these proteins are not direct partners of PfPP1 or their interaction may require additional partner(s) expressed by the yeast.
Finally, in order to examine the binding capacity of several in silico Pips identified by genome analysis, 32 fragments which correspond to 30 potential candidates were tested for protein production. The fragments were about 190 residues long with the RVxF ext motif in the middle of the recombinant proteins (Additional file 5: Figure S1, Additional file 6: Table S5). The expression and purification were successful for only 17 proteins. Their ability to bind PP1 was then examined as described above. Significant interactions for 16 out of 17 proteins tested were observed (Fig. 2c). The lack of interaction of PfPP1 with PF3D7_0305500 could be attributed to the fact that its RVxF is not a valid binding motif. However, it is important to point out that 6 Pips identified by Y2H and in silico screens (PF3D7_0220000, PF3D7_0610100, PF3D7_0919900, PF3D7_1008100, PF3D7_1031600 and PF3D7_1202600) were confirmed to interact with PfPP1 by ELISA (Fig. 2b and c), highlighting the accuracy of the RVxF consensus sequence used throughout this study.
To further explore the implication of the RVxF motifs in PfPP1 binding, 2 Pips identified by Y2H screening and in silico analysis were mutated in their RVxF motifs and the corresponding recombinant proteins were produced. Results presented in Fig. 2d showed a significant decrease of the binding with PfPP1 (about 86 % for PF3D7_0220000 and 70 % for PF3D7_0919900). This confirms that the RVxF motifs of these 2 Pips are the main contributors for binding to PfPP1. However, it could not be excluded that these interactions could involve secondary structures of the protein to stabilize and to lead to a functional complex. This could be attributed to the fact that the RVxF binding motif, in most regulators, is often present in intrinsically disordered regions and the protein could adopt a structure upon binding [16,17].
The newly RVxF-containing PfPips were then used to reexamine the features of the RVxF motif and the nature  Figure S3). Interestingly, the amino acids between V and F/W are enriched by S and H (~50 %). At the N-terminal positions of the RVxF (−1, and −2), the most conserved amino acid is the K residue. We have also found a high frequency of L residue at −2. Based on these data, the RVxF consensus sequence for PfPips could be extended and refined as

PfPP1 interaction networks
The identification in this study of 186 new Pips provides new insights on the functional diversity of PfPP1 functions. Among the 186 identified proteins, 108 (58 %) have unknown function. For the 78 remaining proteins, the most abundant proteins correspond to metabolism, DNA maintenance, translation and phosphorylation processes (Fig. 3). Based on an earlier published P. falciparum interactome [34] and on the identified PP1 interacting proteins reported in this work, we were able to establish three networks where several Pips could be involved. The transcription and DNA maintenance network illustrated in Figure S4A (Additional file 9) contained 9 direct Pips (white circles) detected by the different approaches described above and 4 connecting proteins (gray circles). The folding/proteolysis network involved 8 direct Pips and 2 connecting proteins (Additional file 9: Figure S4B). The pathogenicity network comprised 18 direct Pips and 16 connecting proteins (Fig. 4). In the latter network, 2 homologs of direct Pips in P. berghei (PBANKA_0408500 homolog to PF3D7_0310400 and PBANKA_0926700 homolog to PF3D7_1121600) have been shown to be essential for parasite survival as no viable KO parasites can be obtained [35,36]. It is worthy of note that PF3D7_1023900, a connecting protein, contains a region of about 700 amino acids with 40 % identity with a yeast protein (YER164W) which has been shown to interact with yeast PP1 [23]. Its homolog in P. berghei (PBANKA_050810) has been shown to be essential for parasite survival [35]. The network analysis raises the possibility that new Pips  However, it cannot be excluded that these proteins could be also substrates of PfPP1.

Conclusion
The diversity of proteins identified by three different and complementary approaches suggests that PP1c has a wide interaction network which controls several aspects of parasite physiology. A total of 186 proteins were identified by these approaches. One protein was common to both Y2H screening and PfPP1-affinity column approach and 8 proteins to Y2H and in silico screenings. This provides experimental support to determine the contribution of the RVxF motif to PfPP1 binding when present in Pips, to identify new binding regions/motifs specific of Plasmodium and to explore the role of Pips as regulators, substrates or both. Further experiments need to be done to determine PfPP1 substrates, for instance by identifying the phosphoproteomes of Pf in vitro in the absence and presence of recombinant active PfPP1, and to evaluate whether the new Pips could be able to regulate PfPP1 activity. These future studies will help to better elucidate PP1 signaling in Plasmodium and to increase and improve the choice of parasite targets for drug design.

Affinity purification of Pips
For affinity purification, 10 mg of PfPP1 recombinant protein (produced as previously described [17]) were covalently coupled to 0.5 ml of CNBr-activated Sepharose 4B according to the supplier's instructions (Sigma). The coupling efficiency was > 90 %. Before use, PP1 beads were submitted to 3 wash cycles with 3 M NaSCN and PBS. Ten mg of total soluble parasite extract (prepared as previously described [17]) were precleared on 2 ml of activated sepharose beads blocked with ethanolamine. The precleared extract recovered by centrifugation was filtered, divided in 2 equal volumes and incubated with Fig. 2 Direct interaction of Pips to PfPP1. The binding capacity of recombinant Pips (25pmol/well) to biotin-PfPP1 (4pmol/well) was measured using an ELISA-based assay. a represents the binding of recombinant Pips identified by PP1-affinity column (gray bars). PfI2 (PF3D7_0320000) and PfI3 (PF3D7_1031700) were used as positive controls (white bars) and BSA was used as negative control (black bars). b represents the binding of Pips identified by Y2H screening (gray bars). AAN59960 and AAN59961 are human H2A and H2B proteins respectively which present high identity with PfH2A (PF3D7_0617800) and PfH2B (PF3D7_1105100). c represents the Pips identified by genome in silico analysis (gray bars). d contribution of the RVxF motifs to the binding of PF3D7_0220000 and PF3D7_0919900 to PfPP1 (gray bars). The interaction of PfPP1 to wild proteins was considered at 100 % binding. Results are presented as mean ± SD of 2 independent experiments preformed in duplicate. *p < 0.05, **p < 0.01 and ***p < 0.001 (Mann-Whitney U test), compared to the negative control Fig. 4 PfPP1 Pathogenicity network. The graph shows interactions (lines) between Pips identified by our different approaches (white circles) and PfPP1 (square box). The gray circles represent connecting proteins based on the results of Lacount et al. [34]. When available, gene names given in PlasmoDB are shown in brackets BSA-sepharose beads or PP1-beads overnight at 4°C. After 5 washings with 50 mM Tris, 0.3 M NaCl, proteins were eluted with 2 × 0.5 ml of 3 M NaSCN. The proteins were then precipitated with acetone and separated on SDS-PAGE. No stained bands were observed for eluates of BSA-sepharose beads. Stained bands observed for eluates of PP1 column affinity were cut for MS/MS identification as previously described [37].

Plasmid and yeast strain
The cDNA library of Plasmodium falciparum, cloned in pGAD-HA vector, was purchased from Dualsystems Biotech. In this plasmid, the expressed parasite proteins are fused to the GAL4 activation domain. Yeast Y2HGold strain (Clontech) was transformed with the pGAD-cDNA library according to the method for highefficiency library scale transformation of Dualsystems Biotech, and allowed to form colonies on 150 mm plates with SD agar medium lacking leucine (SD/-Leu) at 30°C (3-4 days). Y2HGold contains four integrated reporter genes (AUR1-C, HIS3, ADE2 and MEL1) under the control of three distinct GAL4 responsive promoters used to detect two-hybrid interaction and in order to reduce false positives. All clones were scraped, harvested with a freezing medium (YPDA medium with 25 % glycerol) and stored in 1 ml aliquots at −80°C for further experiments. This library-transformed yeast strain is referred below as the prey strain. The cDNA encoding for P. falciparum protein phosphatase type-1 (PfPP1, PlasmoDB accession number: PF3D7_1414400) was cloned in the pGBKT7 vector (Clontech) to generate a PfPP1-pGBKT7 plasmid allowing the production of PfPP1 fused with the GAL4 DNA binding domain. Y187 yeast competent cells (Clontech) were then transformed with this construct according to the manual of Yeastmaker™ Yeast Transformation System 2 (Clontech). The expression of PfPP1 was checked by immunoblot using anti-GAL4BD antibody. This strain was used as a bait to screen protein interactions with the P. falciparum cDNA library by mating.

Yeast two-hybrid assay
All reagents and methods for yeast-two hybrid assays were from the Matchmaker™ Gold Yeast Two-Hybrid System (Clontech).
One fresh and large colony of bait yeast strain (pGBKT7-PfPP1 in Y187) was inoculated into 50 ml of SD/-Trp/Kan (Kanamycin) liquid medium agitated at 250 rpm, 30°C until the OD600 reached 0.8. The supernatant was discarded after centrifugation for 5 min at 1,000 g. The pellet was suspended in 5 ml of SD/-Trp/Kan liquid medium ([cells] >1 × 10 8 /ml). 1 ml of library prey strain (after rechecking the titer) and 5 ml of bait strain Y187 were combined with 45 ml of 2xYPDA liquid medium and incubated for 20-24 h at 30°C at 40 rpm. The culture was centrifuged for 10 min at 1,000 g, then 50 ml of 0.5xYPDA liquid medium (with 50 μg/ml Kan) was added to resuspend the pellet. After a centrifugation for 10 min, cells were suspended in 10 ml of 0.5xYPDA liquid medium (with 50 μg/ml Kan) and the total volume was measured. To calculate the number of screened clones and the mating efficiency, 100 μl of the fusion culture (diluted by 10 1 gradient) were plated on the selective medium SD/-Trp, SD/-Leu, and SD/-Trp/-Leu (Double dropout medium, DDO) and incubated for 3-5 days at 30°C. More than 1 million diploids were screened to maximize chances of detecting genuine interactions on Aureobasidin A (AbA) plates. The remaining fusion culture was spread onto selective medium SD/-Trp/-Leu/Kan supplemented with 125 ng/ml AbA (DDO/A), 200 μl per 150 mm plates and incubated for 3-7 days at 30°C. Colonies obtained were re-streaked onto higher stringency SD/-Trp/-Leu/-His/AbA/Kan (TDO/A, Triple dropout supplemented with AbA) and SD/-Trp/-Leu/-His/-Ade/ AbA/Kan (QDO/A, Quadruple dropout supplemented with AbA) agar plates. This screening was carried out five times independently.

Identification of potential Pips
The prey plasmids from positive yeast clones were isolated using the Plasmid DNA Purification kit (Macherey-Nagel). To facilitate yeast lysis, the equivalent of 100 μl of glass beads (Sigma) was added in the resuspension buffer. To amplify the plasmid obtained, it was transformed to E. coli DH5α cells (Life Technologies) followed by selection on LB/Ampicillin and LB/Kanamycin plates. Plasmids were isolated once again using the Plasmid DNA Purification kit and then to estimate the sizes of the specific inserts on positive prey plasmids, plasmids were digested using the SfiI restriction enzyme.
The positive prey plasmids were characterized by sequencing and BLAST searches to identify the corresponding P. falciparum genes in the PlasmoDB database (v24).

In silico screening
The consensus sequence used to identify potential interactors of PfPP1, was [KR] [KR][ACHKMNQRSTV]V [CHKNQRST] [FW]. It was based on the crystal of PP1 with a RRVSFA peptide [38] and on consensus sequences defined by different publications [12,30,31]. After identification of the putative partners, the positions −10 to +10 of RVxF motifs were aligned by BioEdit software (v7.2.5) to establish a new consensus sequence.

Recombinant proteins expression and purification
In order to verify the interaction between PfPP1 and its new interactors, His-tagged recombinant proteins were