|Home | About | Journals | Submit | Contact Us | Français|
New drugs are needed to treat toxoplasmosis. Toxoplasma gondii calcium-dependent protein kinases (TgCDPKs) are attractive targets because they are absent in mammals. We show that TgCDPK1 is inhibited by low nanomolar levels of bumped kinase inhibitors (BKIs), compounds designed to be inactive against mammalian kinases. Cocrystal structures of TgCDPK1 with BKIs confirm that the structural basis for selectivity is due to the unique glycine gatekeeper residue in the ATP-binding site at residue 128. We show that BKIs interfere with an early step in T. gondii infection of human cells in culture. Furthermore, we show that TgCDPK1 is the in vivo target of BKIs because T. gondii cells expressing a glycine to methionine gatekeeper mutant enzyme show significantly decreased sensitivity to this class of selective kinase inhibitors. Thus, design of selective TgCDPK1 inhibitors with low host toxicity may be achievable.
The food-borne apicomplexan protozoan Toxoplasma gondii is the causative agent of toxoplasmosis and may be the most common infectious eukaryotic parasite of humans, based on sero-surveys1. Most infections are probably transmitted by undercooked meats, but accidental ingestion of cat feces can also transmit toxoplasmosis to humans2. T. gondii is an obligate intracellular pathogen that can invade and replicate in most nucleated mammalian cells. Toxoplasmosis remains a major health concern in pregnancy, where it causes severe birth defects or miscarriage, and in immunocompromised hosts, where it causes toxoplasmic encephalitis3–5. Infection is often asymptomatic in immunocompetent hosts, but prolonged disease does occasionally occur6. Attack rates in immunocompetent individuals can be high, as evidenced by laboratory-acquired infections7, and acute symptoms can be debilitating8, warranting T. gondii’s classification as a class B agent of concern for biodefense9. Toxoplasmosis is extremely difficult to treat, and only limited therapeutics are currently available in the United States (sulfadiazine and pyrimethamine or clindamycin)10. These drugs are problematic in that they can cause rash, leucopenia and nephrotoxicity11, and sulfadiazine and pyrimethamine can result in complications during pregnancy. New therapeutics against T. gondii are needed.
Calcium levels have long been associated with T. gondii’s interrelated processes of invasion, gliding motility and secretion12. The intracellular Ca2+ level oscillates during gliding motility and is promptly dampened upon cell invasion, preventing T. gondii from immediately gliding out of cells13. Calcium oscillations control many targets in the cell, and the mediation of invasion, micronemal secretion and gliding motility is thought to be largely due to T. gondii calcium-dependent protein kinases (TgCDPKs)12,14.
Protein kinases are generally attractive targets for new drugs against eukaryotic pathogens15 and TgCDPKs are no exception, in part because they are more closely related to CDPKs found in plants and algae than any protein kinases found in mammalian cells16,17. TgCDPK1 is especially attractive for treatment of toxoplasmosis because, in addition to its key role in parasite invasion14, it contains a unique sequence variation in the ATP binding pocket: namely, a glycine at the so-called ‘gatekeeper position’. This variation distinguishes TgCDPK1from all known mammalian kinases. Thus, design of selective inhibitors may be possible. To this end, we show here that TgCDPK1 is inhibited by low nanomolar levels of bumped kinase inhibitors (BKIs), a class of compounds designed to be inactive against mammalian kinases. Crystal structures of TgCDPK1 in complex with BKIs confirm the structural basis for selectivity and provide a structural look at the Ca2+-binding regulatory domain in this class of kinases. Furthermore, we find that BKIs interfere with host-cell invasion in culture, and reverse genetic experiments with live parasites expressing a gatekeeper-residue mutant confirm that TgCDPK1 is indeed the specific target of BKIs in vivo. These studies pave the way for the design of selective drugs against toxoplasmosis.
Two types of enzyme assays were developed to follow TgCDPK1 activity: a radiometric scintillation proximity assay measured the labeled γ-phosphate of ATP added to a biotinylated peptide substrate, and an ATP consumption assay monitored ATP consumption by luciferase and light production (KinaseGlo). Both assays gave similar results for the Km of the substrates (less than two-fold differences in Km values, Supplementary Table 1 and Supplementary Fig. 1a–d) and inhibitor concentrations for 50% enzyme inhibition (IC50 values) (Supplementary Tables 1 and 2). As expected, activity of TgCDPK1 on the peptide substrates required addition of exogenous calcium (Supplementary Fig. 2). Indeed, calcium titration revealed that a concentration of 4.7 µM gave 50% of TgCDPK1 kinase activity.
We have determined X-ray crystal structures of Ca2+-free TgCDPK1 in the apo form and in complex with two potent inhibitors (Fig. 1 and Table 1). The structure of the catalytic domain is typical of serine/threonine-type protein kinases. The two Ca2+-binding EF hand lobes and the connecting extended helical stem of the calmodulin-like regulatory domain lie along one face of the kinase domain, adjacent to the active site (Fig. 1a). The kinase and regulatory domains are connected by an intervening helical junction domain characteristic of this class of kinases18. The conformation represented by these structures is likely an inactive form of the enzyme, as the calmodulin-like domain occludes the surface required for recognition of target proteins and peptides. Notably, the ATP binding site remains accessible to small-molecule substrates and inhibitors. Comparison of the calcium-free TgCDPK1 structure (Fig. 1a) with the calcium-activated structure19 (PDB 3HX4) shows that the regulatory domain undergoes major structural rearrangement and is repositioned to lie against the opposite surface of the kinase domain upon calcium binding. This dramatic structural change allows access of the protein substrates to the active site.
Most known kinase inhibitors bind in the ATP binding pocket of the active site20,21. These inhibitors exploit many of the same hydrophobic contacts as the purine ring of ATP and make at least one conserved hydrogen bond to the hinge region. Potent inhibitors also occupy at least one hydrophobic pocket adjacent to the ATP binding site. These additional hydrophobic interactions increase both binding affinity and target selectivity of the inhibitor because sequence heterogeneity is greater among different kinases in these regions. Examination of the TgCDPK1 sequence in the vicinity of the ATP binding pocket (Fig. 1b) shows that it contains a glycine residue at a position that has been termed the gatekeeper residue because it constrains access to the ATP binding site22–24. The glycine at this position in TgCDPK1 (Gly128) is expected to create a much larger pocket off the ATP binding site than is typically observed in protein kinases. Comparison of the TgCDPK1 structure with other kinases shows that this is indeed the case. This difference in the active-site architectures may be exploited for design of selective inhibitors against TgCDPK1.
Previous work25 has shown that mutation of bulky gatekeeper residues to glycine renders mutant kinases uniquely susceptible to inhibition by BKIs. BKIs are analogs of 4-amino-1-tert-butyl-3-phenylpyrazolo[3,4-d]pyrimidine (Table 2) that are derivatized at the C3 position with bulky aromatic groups26. The large side chain of the gatekeeper residue in most kinases prevents bulky aromatic substituents at the C3 position from accessing a hydrophobic pocket at the back of the catalytic cleft, rendering them insensitive to BKI inhibition22,23. Large gatekeeper residues, like methionine or phenylalanine, severely restrict access by the BKIs to a large hydrophobic pocket, whereas small gatekeeper residues, such as the glycine present in TgCDPK1, allow access of BKIs to this hydrophobic pocket23,24,27,28. Studies with genetically engineered mice that express mutant kinases with small gatekeeper residues (glycine or alanine), have shown that BKIs preferentially target the mutant kinases29–31. The well-documented lack of BKI inhibition of mammalian kinases suggests that this class of compounds may be very selective for TgCDPK1 during T. gondii infection29–31.
Based on structural and sequence analysis of the differences in ATP binding sites of typical mammalian kinases and TgCDPK1 with a small gatekeeper (Fig. 1a,b), we synthesized three BKIs (NA-PP1, NA-PP2 and NM-PP1) and showed that two of the three potently inhibit TgCDPK1, with IC50 values in the low double-digit nanomolar range (Table 2). To show that the glycine gatekeeper is the primary determinant of BKI inhibition, we engineered a glycine-to-methionine gatekeeper mutant of TgCDPK1 (G128M). Indeed, the wild-type enzyme was inhibited much more effectively by BKIs compared to the G128M mutant enzyme (Table 2). For NA-PP1, there was a 175-fold difference between IC50 values for wild type versus the G128M mutant; for NA-PP2, a 1,200-fold difference; and for NM-PP1, a 550-fold difference (Table 2).
We determined the crystal structures of TgCDPK1 in complex with two potent BKIs (NA-PP2, Fig. 1c and Supplementary Fig. 3a; NM-PP1, Fig. 1d and Supplementary Fig. 3b) and, as predicted, these inhibitors bound in the ATP binding site, with the C3 bulky aromatic substituent occupying the pocket adjacent to the glycine gatekeeper. Superposition of an orthologous CDPK containing a methionine gatekeeper onto our BKI complexes showed impairment of the favorable binding mode observed for the BKIs in TgCDPK1 (Fig. 1e). The bulky gatekeeper side chain clashes with the C3 bulky aromatic substituent, providing a structural basis for the insensitivity of typical kinases toward BKI inhibition, as verified experimentally with the G128M mutant of TgCDPK1 (Table 2). These results implicate TgCDPK1 with its unique glycine gatekeeper as a promising drug target for selective treatment of toxoplasmosis.
The mode of action of TgCDPK1 is unknown, but its localization in the cell might provide clues to its cellular function. To this end, we fused TgCDPK1 with green fluorescent protein (GFP) to facilitate its localization within live intracellular T. gondii. We found GFP-labeled TgCDPK1 in both the cytosol and the nucleus (Fig. 2a). We obtained similar findings when the GFP tag was replaced with a C-terminal hemagglutinin (HA) tag (data not shown). Thus, TgCDPK1 may phosphorylate target proteins in both the cytosol and the nucleus.
We tested three BKIs for effects on host-cell invasion and parasite growth, as earlier work on TgCDPK1 (ref. 14) suggested a key role for this enzyme in T. gondii invasion. BKIs profoundly reduced T. gondii proliferation if added simultaneously with cellular infection (Fig. 2b) and modestly reduced proliferation when added 4 h after the start of invasion (Fig. 2c). Thus, inhibition of TgCDPK1 activity by BKIs has a stronger effect on invasion than on intracellular growth. This was confirmed by microscopic examination of invasion (data not shown) and is consistent with previous observations on the role of TgCDPK1 (refs. 12,14). We expect that longer exposure to BKIs would increase the impact of the drug on T. gondii growth as egressed parasites are prevented from invading new cells. We tested each BKI for its effects on human fibroblast replication; the EC50 values were found to be ~1,000-fold higher (not shown) than those seen in the T. gondii ‘invasion’ assay.
We transfected the T. gondii cell line with expression plasmids encoding an HA tag fused to the C terminus of either wild-type TgCDPK1 or the G128M gatekeeper mutant. Immunoblot analysis showed that the wild-type and G128M mutant TgCDPK1s were expressed to similar levels (Fig. 3a). Compared to the parental cell line, parasites expressing the G128M mutant were relatively resistant to BKIs NA-PP2 and NM-PP1 when they were added before invasion (Fig. 3b,c). In contrast, parasites overexpressing the wild-type TgCDPK1 protein showed only a small shift in resistance. Although NA-PP2 was somewhat less effective against wild-type parasites in this experiment, the results were identical in the relative potency of the inhibitors, in that NA-PP2 was more potent than NM-PP1 in blocking T. gondii invasion, following the potency observed against the TgCDPK1 enzyme. Furthermore, in both a microscopic assay of invasion32 (Supplementary Fig. 4) and a mixed cellular infection using wild-type and G128M mutant TgCDPK1 (Supplementary Fig. 5), parasites expressing the gatekeeper mutant were markedly resistant to NA-PP2, whereas cells expressing either wild-type TgCDPK1 or GFP controls were not. These findings show that BKIs work primarily through the TgCDPK1 target to prevent T. gondii mammalian-cell entry.
We have shown that TgCDPK1 is a promising drug target for the therapy of toxoplasmosis. Its kinase activity is uniquely sensitive to inhibition by BKIs, and BKI treatment in turn blocks entry of the parasite into mammalian cells. Blocking cell entry is important because T. gondii is an obligate intracellular parasite and cannot replicate without invasion. Clinical toxoplasmosis is caused by the actively dividing tachyzoite form of the parasite, which exits its host cell and invades a new cell every few days. TgCDPK1 has a unique ATP binding site with a small gatekeeper residue, as opposed to the larger gatekeeper residues present in mammalian protein kinases. This key difference in the ATP binding pockets of mammalian kinases and TgCDPK1 allowed us to use BKIs to selectively inhibit TgCDPK1 without untoward effects on the mammalian host cell. Crystal structures revealing the binding mode of BKIs in TgCDPK1 (Fig. 1c,d and Supplementary Fig. 3a,b) support the selectivity we observed. Consistent with the proposal that TgCDPK1 is the target of BKIs, the order of increasing potency (NA-PP2 > NM-PP1 > NA-PP1) was identical in both the enzymatic and cellular assays. TgCDPK1 sensitivity to BKIs was altered by deliberate mutation of the gatekeeper glycine residue to a methionine (G128M) that is more typical of mammalian kinases. Expression of this G128M mutant led to resistance in three different types of assays when assessed using model compound NA-PP2 (Fig. 3b and Supplementary Figs. 4 and 5). These data support the contention that TgCDPK1 is the major target for BKIs in T. gondii cells. These experimental findings, therefore, address any concerns about quantitative differences between enzyme inhibition and cellular effects, thereby validating the utility of TgCDPK1 structures in complex with BKIs to drive drug development for toxoplasmosis therapy. The concept of using BKIs for the therapy of toxoplasmosis is bolstered by the fact that BKIs have been used in mouse studies with no demonstration of toxicity or troublesome effects to the animals29–31. Therefore, BKIs have promise as a selective drug for toxoplasmosis therapy because they prevent cell entry, and thus the replication, of T. gondii and are likely to be nontoxic to the mammalian host.
Because we determined the structure for the inactive, Ca2+-free form of TgCDPK1, there may be some concern that this form would not be optimal to guide the design of small-molecule inhibitors with improved potency and selectivity. It is apparent from structural and biochemical studies presented here, however, that small-molecule inhibitors can still access the ATP binding site of the Ca2+-free conformation of TgCDPK1. Notably, the structure of the active site near the ATP binding pocket, particularly in the vicinity of the gatekeeper residue, the hinge region and the activation loop, is not substantially[AU: Correct?] altered between the structures shown here and that of the Ca2+-bound enzyme in complex with adenylyl-imidodiphosphate (AMP-PNP)19 (PDB 3HX4). Thus, structure-guided optimization of small-molecule inhibitors that target this region is possible using the inactive, Ca2+-free form of TgCDPK1.
TgCDPK1 was localized in the cytoplasm but also found in the nucleus. Some plant CDPKs are also partially localized to the nucleus. In those CDPKs, nuclear localization is mediated by a signal in the junction domain18, but the T. gondii protein is not homologous in this region, and no nuclear localization signal is predicted by standard programs. Because the size of the protein is above the threshold for free diffusion through the nuclear pore, we propose that the protein could bear a noncanonical nuclear-localization sequence or could piggyback into the nucleus on another protein. These results raise the possibility that TgCDPK1 phosphorylates specific nuclear proteins in addition to its presumably cytosolic targets involved in gliding motility. Furthermore, as previously noted, CDPK1 undergoes a large Ca2+-dependent structural rearrangement that repositions the regulatory domain to the opposite side of the catalytic domain.19 This raises the intriguing, but speculative, possibility that Ca2+ could modulate CDPK1 localization by revealing or occluding the region of the kinase required to mediate nuclear import.
Drug-resistant mutations of the TgCDPK1 gatekeeper to a bulky residue could eventually emerge under selective pressure of BKI therapy. A primary strategy to suppress the emergence of resistance is the coadministration of two drugs targeting different proteins. As TgCDPK1 is not the target of any existing drug, the development of an anti-TgCDPK1 compound could provide a partner drug for coadministration with an existing drug. Most transmission of T. gondii is not from person to person (although this can occur in pregnancy, transplantation or transfusion) but rather through zoonotic cycles where drug pressure is not exerted1. This suggests that, if drug resistance emerges, it will largely be confined to the individual, posing little threat to the utility of the drug in other infected persons.
As other apicomplexan pathogens use CDPK enzymes with a small gatekeeper residue, this work may have broader applicability. For instance, the TgCDPK1 ortholog in Cryptosporidium parvum, also an apicomplexan parasite, likewise has a glycine residue at the gatekeeper position (Fig. 1b), suggesting that BKIs targeting T. gondii could also be effective for the therapy of cryptosporidiosis, another potentially life-threatening infection with poor therapeutics. CDPKs from other apicomplexan parasites have smaller gatekeeper residues, such as serine and threonine (Fig. 1b). An extended search of all reported human kinase ATP binding pockets found none with a glycine or alanine gatekeeper residue, though ~20% do contain threonine22. Thus, the results presented here may have implications in the rational design of anti-apicomplexan CDPK agents devoid of toxic side effects to the host cells.
Methods and any associated references are available in the online version of the paper at http://www.nature.com/nsmb/.
The authors would like to acknowledge the generous assistance of F. Ghomashchi and M. Gelb in delineating the calcium dependence and enzyme kinetics of TgCDPK1. This work was funded by US National Institute of Allergy and Infectious Diseases grants R01AI080625 (W.C.V.V.), R01AI50506 (M.P.) and AI067921 (C.L.M.J.V., F.S.B., W.G.J.H., E.A.M. and W.C.V.V.) and financial support from Gerry and Kristine Pigotti. K.K.I. was supported by a US National Institutes of Health grant from the Fogarty International Center 2D43 TW000924. Portions of this research were carried out at the Stanford Synchrotron Radiation Lightsource, a national user facility operated by Stanford University on behalf of the US Department of Energy, Office of Basic Energy Sciences.
Accession codes. Protein Data Bank: Atomic coordinates and structure factors have been deposited with accession numbers 3I79 (apo), 3I7c (NA-PP2 complex) and 3I7b (NM-PP1 complex).
Note: Supplementary information is available on the Nature Structural & Molecular Biology website.
AUTHOR CONTRIBUTIONSK.K.O., K.R.K., K.K.I. and W.C.V.V. were involved in the biochemical characterization and testing of inhibitors of TgCDPK1; L.J.C., K.K.O., K.R.K., A.J.N., C.L.M.J.V., F.S.B. and W.C.V.V. selected, cloned and purified the recombinant wild-type and mutant TgCDPK1 protein; E.T.L., J.E.K., T.L.A., L.Z., W.G.J.H. and E.A.M. crystallized and solved the structure of TgCDPK1; R.M. and D.J.M. synthesized the inhibitors; A.E.D. and M.P. performed the cellular T. gondii experiments; K.K.O., E.T.L., A.E.D., D.J.M., M.P., E.A.M. and W.C.V.V. wrote the paper; all authors reviewed and edited the paper.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Reprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions/.