The Pf Inhibitor2 gene (PfI2) encodes a protein of 144 amino acids related to the I2 proteins of different organisms, which are known to inhibit PP1c activity in vitro. Of the three central regions identified in the I2 protein as binding motifs to PP1, the KGILK, RVxF, and HYNE motifs, PfI2 contained only a consensus RVxF (12KTISW16) and the 102HYNE105 sequences. The lack of KGILK in PfI2 was supported by bioinformatics analysis indicating the absence of this sequence in all potential open reading frames upstream of the PfI2 gene and was further confirmed by a 5′cDNA walking approach. The KGILK motif present in vertebrate I2 was found to be involved in the interaction with PP1 through the region of amino acids 50–59 in PP1c . In addition, deletion of the N-terminal side of I2 containing this site and mutation of the first Lys or the Ile dramatically reduced the inhibition capacity of I2 (>500 fold decrease) [33–35]. These observations emphasize the importance of this site in the binding and activity of vertebrate I2, which represents a major difference compared with PfI2, which lacks this motif. Concerning the RVxF site, vertebrate I2 does not contain the canonical motif falling within the consensus sequence [R/K]X0-1[V/I]X0-1[F/W]. However, studies on the crystal structure of PP1c-I2 revealed that the sequence KSQKW, where the consensus Val/Ile residue is replaced by a Gln is docked in the PP1 groove, which usually binds the RVxF motif . Structure-activity studies on the implication of KSQKW site showed that the mutation of Trp in mammalian I2 drastically reduced the inhibitory activity of I2 . It is worth noting that almost all I2 proteins contain Gln at the position of Val/Ile. However, in P. falciparum, the I2 protein does contain an Ile in the RVxF motif, a second important dissimilarity between PfI2 and other I2 proteins. The comparison of PfI2 with putative I2 of Toxoplasma gondii or Neospora canium (TGGT1_114760, NCLIV_032710), revealed the presence of the consensus RVxF sequence where V/I is replaced by Leu, maintaining the hydrophobicity of the residue and suggesting its conservation within other Apicomplexa parasites. Studies on the third site of interaction, HYNE, have shown that the His and Tyr residues are important in the interaction with PP1c and it has been proposed that this motif functions as a ‘degenerate’ RVxF motif [33, 53]. More recent studies clearly showed that the region containing the HYNE motif interacts directly with the active site of PP1c (involving 272Tyr and 96Arg residues of PP1c) with a major contribution of His and Tyr residues [33, 37]. This excludes completely the possibility of a competition of binding to PP1c between the RVxF and HYNE motifs and suggests that the His and Tyr residues of I2 promote the displacement of the catalytic metal ion. In the PfI2 protein, these two residues are conserved.
Among the three binding sites of I2, the best-identified and most widely found in PP1 partners is the [R/K]X0-1[V/I]X0-1[F/W] consensus motif, which corresponds to KTISW in PfI2. The presence of RVxF in about 25-30% of eukaryotic proteins is not a sufficient indicator in itself to classify a protein as a PP1c regulator . These observations, together with the fact that PfI2 is the shortest I2 protein identified so far (144 amino acids for PfI2 versus 164–205 for other I2), the absence of one binding site (KGILK) and the fundamental difference in the RVxF motif (KTISW) raised the question of the capacity of PfI2 to bind and to regulate PfPP1. Using wild-type recombinant proteins, we showed that labeled PfPP1 was able to bind to PfI2 and vice versa. This was further validated by the use of a yeast two-hybrid system that confirmed the interaction of wild-type PfI2 with PfPP1c and suggested that it was strong since the mated PfI2 and PfPP1 yeast strains were able to grow under stringent conditions (SD-LWHA medium). In order to explore the contribution of PfI2 RVxF and HYNE motifs for the interaction with PfPP1, two types of constructions were used, one deleted for the Nt moiety of PfI2 and the other with a single mutation in the RVxF motif. Binding was unaffected on SD-LWH medium, whatever the construction tested and only one strain, carrying the PfI2 Y103A, mutant was unable to grow under the most stringent conditions (SD-LWHA medium). These observations show that there is no one, major site of interaction in PfI2 unlike Pf Inhibitor-3 (PfI3), for which we showed that the mutation of 16 W (localized within the RVxF domain of PfI3) completely abolished its binding/function . PfI3 exhibits a totally disorganized structure and seems to bind first to PfPP1 via the RVxF groove and folds afterwards to accomplish its function . Regarding I2, previous studies suggested a major role for the RVxF motif along with secondary binding sites which should be intrinsically structured for efficient binding to PP1c [33–35]. PfI2 secondary structure analysis predicted that the RVxF motif is a part of an unstructured region, while the HYNE is within an α-helix. The role of this structure in PfI2-PfPP1c interaction was substantiated by the lack of binding of PfI2 deleted for the region containing the α-helix (PfI2 (1–94)). In the case of mutated PfI2, the yeast two-hybrid method supported a role for 103Tyr (localized within the HYNE domain of PfI2) in the stabilization of PfI2-PfPP1 binding under stringent culture conditions.
It has been shown that most I2 proteins are able to drastically decrease PP1c activity towards different non-specific substrates such as Phosphorylase A and pNPP [34, 35, 38, 50]. As expected, the addition of PfI2 in the nanomolar range significantly decreased PfPP1 activity up to 80%. To investigate the impact of KTISW (RVxF) and HYNE motifs on PfI2 regulatory activity we used deleted or mutated recombinant proteins. The contribution of the RVxF motif (KTISW) is key to the function of PfI2 as both Nt deleted PfI2 (PfI2(19–144)) and mutated PfI2 (PfI2W16A) were unable to inhibit PfPP1 activity, whereas the involvement of the HYNE domain seems to be less important. Thus, although the PfI2W16A mutant is still able to bind to PfPP1, 12KTISW16 is a vital and a primary site for the inhibitory activity of PfI2. To further evaluate the inhibitory activity of PfI2 and the role of the two motifs, we took advantage of the Xenopus model where oocytes are physiologically arrested in G2/M prophase I [55, 56]. The injection of Xenopus I2 (spanning 188 residues and containing the KGILK, KSQKW and HYNE motifs) or anti PP1 antibodies into oocytes induced germinal vesicle breakdown or GVBD [51, 57]. Plasmodium I2 is able to substitute for the Xenopus orthologue in this system since the microinjection of PfI2WT into oocytes promoted the progression to M phase, inducing GVBD and co-immunoprecipitation experiments confirmed the interaction of PfI2 with Xenopus PP1c. This confirmed that PfI2 can function in cells without the need for the KGILK site and are in accordance with previous studies that showed the involvement of Xenopus I2 in the G2/M transition in acellular extracts  or the implication of Glc8 (yeast inhibitor 2) in the cell cycle [38–40]. Deletion, mutation or RNA interference studies carried out on inhibitor 2 have demonstrated its implication in the cell cycle, chromosome segregation and embryogenic development [38, 39, 57]. In the case of PfI2, when deleted PfI2 (PfI2(19–144)) lacking 12KTISW16 or mutated PfI2 (PfI2W16A or PfI2Y103A) were microinjected, no GVBD was observed, demonstrating the importance of both PfPP1 binding sites in the functional capacity of PfI2. Since the PfI2 mutated proteins are able to bind PP1 but unable to inhibit its function we sought to determine whether the pre-injection of deleted or mutated PfI2 proteins may block the role of wild PfI2. The pre-injection of either PfI2(19–144) or PfI2W16A were able to block the induction of GVBD while PfI2Y103A did not. One explanation for these observations is that the HYNE-dependent binding is critical as the injection of PfI2WT is able to displace this mutated protein and to induce GVBD. When the HYNE site is not mutated the binding of PfI2 is sufficiently stable to prevent its displacement.
Closer examination of the PfI2 peptide sequence revealed the presence of a consensus PXTP motif (37PNTP40), also present in other I2, in which the phosphorylation of the T within this site abrogated the function of I2 [32, 57]. In PfI2, the replacement of T by D (mimicking phosphorylation) did not impact either the binding or the function of PfI2 (not shown), tending to exclude the phospho-regulation of I2 at this site. These data are in agreement with the recent P. falciparum phosphoproteome characterization showing the phosphorylation of PfI2 at positions T13, S48, S50, S115, T117 and S142, but not at T39 within the PXTP motif. The assessment of the impact of PfI2 phosphorylation will await further investigations on these phosphorylated residues as well as the “T” within the PXTP motif. At this stage, it is important to mention that, beside the capacity to interact with PP1c, human I2 has been shown to participate in a direct kinase-dependent signaling network. It was found that I2 was able to bind and to activate Nek2 and Aurora-A kinases [58, 59]. For these functions, I2 seems to operate through its C-terminal domain as the protein deleted in this domain (I2(1–118)) failed to interact with these kinases, excluding a role for the KGILK and RVxF motifs. Although the PfI2 sequence is 61 amino acids shorter than its human homologue, the capacity of PfI2 to bind P. falciparum kinases of the NIMA and Aurora families (for which active recombinant enzymes are available [60–64]) should be evaluated.
In P. falciparum, microarray analysis detected PfI2 mRNA in all blood parasite stages and gametocytes (data available in PlasmoDB, ). In this work, co-immunoprecipitation experiments with anti-PfI2 antibodies followed by Western blotting and the use of a PfPP1 affinity column clearly revealed the expression of PfI2 protein by P. falciparum and of its capacity to bind PfPP1. Transfection of live parasites with the tagged PfI2-GFP protein showed that its distribution is nucleocytoplasmic, like PfPP1 , with a strong accumulation in the nucleus, is in agreement with the localization of other I2 proteins . Indeed mammalian I2 fused to GFP was localized in both the cytoplasm and the nucleus, with an active import to the latter compartment, supported by the presence of two putative nuclear localization signals [49, 65, 66]. In the case of PfI2, bioinformatics analysis also revealed a putative nuclear localization signal, supporting its nuclear localization. We previously reported that PfLRR1 and Pf inhibitor-3, the first identified regulatory subunits of PfPP1c, localized to the nucleus, evoking a specific role in this compartment [28, 29]. The present study suggests an additional role for the PfI2 regulatory subunit of PP1c, present in the nucleus but also in the cytoplasm. Our reverse genetic studies strongly suggest a critical role for PfI2 in the erythrocytic asexual cycle in vitro as no parasites with a disrupted PfI2 gene were detectable. Definition of the PfI2 role(s) during the life cycle necessitates further work, requiring the development of a powerful inducible expression system for P. falciparum.
The ability of PfI2 to bind and to inhibit PP1c both in vitro and in cellular conditions (Xenopus oocytes) through the two main motifs: the RVxF motif (KTISW) and the HYNE motif, together with the fact that a tight and appropriate regulation of PP1c is crucial for cellular functions, led us to explore whether derived ‘competing’ peptides from PfI2 could bind to PP1c and inhibit downstream signaling pathways. Only peptides containing the KTISW or HYNE motifs were able to bind to PfPP1c. However, the incubation of these peptides with PfPP1 or their injection into oocytes failed either to inhibit phosphatase activity or to promote GVBD respectively. However, the pre-injection of the KTISW and HYNE peptides did block the PfI2-dependent GVBD. Moreover, there was no interaction between Xenopus PP1 and PfI2 in extracts of oocytes pre-injected with the KTISW or HYNE peptides. This encouraged us to investigate the ability of these peptides to inhibit the growth of P. falciparum. To do this, the capacity of the peptides to cross membranes was first improved by including a short basic peptide, which has been shown to be highly efficient in increasing the permeability of peptides and to promote accumulation within infected red blood cells . Peptides encompassing the RVxF degenerate motif R/KX0-1 V/I X0-1 F/W (KTISW or KVVRW) inhibited the growth of 3D7 P. falciparum strain at low micro-molar concentrations. The substitution of amino acids essential for binding with PfPP1 validated that the growth inhibition was RVxF-dependent. The difference in the observed IC50 values of KTISW and KVVRW containing peptides could be related to a higher affinity of the latter for PfPP1 and the fact that it proved able to accumulate not only in merozoites but also in parasites within infected red blood cells. Unexpectedly, the second PP1 binding peptide containing the HYNE motif, although it was found functional in oocyte model, was not active as an antiplasmodial suggesting that native PfI2 expressed by P. falciparum could displace the HYNE peptide. One possible explanation for the anti-parasitic activity of RVxF containing peptides is that an increase in PP1 activity due to its reduced interaction with regulators could result in uncontrolled protein dephosphorylation, leading in turn to an inhibition of parasite differentiation/growth. This implies that each competing active peptide can block its respective protein but that cross-inhibition of other partners using the same docking site cannot be excluded. These peptides might prove very useful as fundamental research tools to dissect pathways and processes controlled by PP1 in Plasmodium falciparum.