Gene transfer in the evolution of parasite nucleotide biosynthesis

Abstract

Nucleotide metabolic pathways provide numerous successful targets for antiparasitic chemotherapy, but the human pathogen Cryptosporidium parvum thus far has proved extraordinarily refractory to classical treatments. Given the importance of this protist as an opportunistic pathogen afflicting immunosuppressed individuals, effective treatments are urgently needed. The genome sequence of C. parvum is approaching completion, and we have used this resource to critically assess nucleotide biosynthesis as a target in C. parvum. Genomic analysis indicates that this parasite is entirely dependent on salvage from the host for its purines and pyrimidines. Metabolic pathway reconstruction and experimental validation in the laboratory further suggest that the loss of pyrimidine de novo synthesis is compensated for by possession of three salvage enzymes. Two of these, uridine kinase-uracil phosphoribosyltransferase and thymidine kinase, are unique to C. parvum within the phylum Apicomplexa. Phylogenetic analysis suggests horizontal gene transfer of thymidine kinase from a proteobacterium. We further show that the purine metabolism in C. parvum follows a highly streamlined pathway. Salvage of adenosine provides C. parvum's sole source of purines. This renders the parasite susceptible to inhibition of inosine monophosphate dehydrogenase, the rate-limiting enzyme in the multistep conversion of AMP to GMP. The inosine 5' monophosphate dehydrogenase inhibitors ribavirin and mycophenolic acid, which are already in clinical use, show pronounced anticryptosporidial activity. Taken together, these data help to explain why widely used drugs fail in the treatment of cryptosporidiosis and suggest more promising targets.

Fig. 2.

C. parvum harbors a UK-UPRT with similarity to plant and algal UK-UPRT. The tree shown is the single most parsimonious tree. Numbers above the branches indicate the bootstrap values for parsimony and neighbor-joining analyses followed by maximum likelihood puzzle frequencies. Only values >50% are shown. If a given method did not provide significant support for the relationship indicated, an “X” has been inserted. The scale is as indicated. GenBank and TIGR EGO accession numbers are as listed. See Materials and Methods and supporting information for details on alignment and analysis.

Fig. 1.

The C. parvum nucleotide biosynthetic pathway is a phylogenetic mosaic. Enzymes labeled in red show strong phylogenetic association with eubacteria, those in green show association with plants and algae. Metabolic reconstruction and phylogenetic analyses presented here are based on the C. parvum type 2 IOWA genome [analysis of the type 1 data set (36) yielded identical results]. Supporting comparative genomic analyses are provided in Table 1. Two arrows indicate two or more enzymatic steps. Most nucleoside mono- and diphosphate kinase and phosphorylase steps have been omitted for simplicity. A complete set of these genes is present in the genome. The membrane topology of the feeder organelle (37) has been schematized. ^*, Transporter: the localization of nucleoside transporters is hypothetical, a localization outside the feeder organelle is equally possible. 1, adenosine transporter; 2, AK; 3, adenosine monophosphate deaminase; 4, IMPDH; 5, guanosine monophosphate synthase; 6, UK-UPRT; 7, uracil phosphoribosylltransferase; 8, TK; 9 ribonucleotide diphosphate reductase; 10, cytosine triphosphate synthetase; 11, deoxycytosine monophosphate deaminase; 12, dihydrofolate reductase-thymidylate synthase.

Fig. 3.

C. parvum expresses a TK of eubacterial origin. (A) Phylogenetic analysis of TK. The tree shown is one of four most parsimonious trees obtained. Solid bullets indicate branches not found on the other most parsimonious trees. Numbers above the branches (space permitting) indicate the bootstrap values for parsimony and neighbor-joining analyses. Only percentages >50% are shown. Asterisks below branches indicate nodes with ≥50% frequency support from maximum likelihood puzzle analysis. GenBank accession numbers, multiple sequence alignment, and analysis details are available in Materials and Methods and supporting information. Many organisms lack a TK gene including δ- and ε-proteobacteria. (B) Genomic Southern blot of C. parvum type 2 DNA probed with TK coding sequence. Lane 1, EcoRI; lane 2, BamHI. Sizes are as indicated (expected sizes based on genome sequence data are 2.4 and >11.5 kb, respectively). (C) C. parvum type 1 meronts were identified in infected tissue cultures based on their reactivity with the C. parvum-specific mAb c3c3 (D) and the presence of six to eight nuclei (E, I, and M). These nuclei could be detected after BrdUrd labeling with an antibody specific for incorporated BrdUrd (H). No antibody labeling was observed without prior BrdUrd labeling (L) or in T. gondii (P), which lacks a TK gene. DAPI, 6′-diamidino-2-phenylinidole; PI, propidium iodide; Cp, C. parvum; Tg, T. gondii; nu, host cell nucleus. [Scale bar represents 1 (C-N) or 10 μm (O-R), respectively.]

Fig. 4.

C. parvum encodes an active AK. Transgenic expression of the putative C. parvum AK gene complements enzymatic activity in the T. gondii AK null mutant (measured by conversion of [^2,8,3H]adenosine to AMP, DE81 filter retained cpm versus incubation time) (A) and also partially restores the sensitivity of this mutant to the AK activated prodrug adenosine-arabinoside (AraA) (B). Parasite growth under drug was measured by the monolayer disruption assay (9) in 24-well tissue cultures (white wells represent complete host cell lysis due to uninhibited parasite growth, and dark wells indicate parasite inhibition).

Fig. 5.

C. parvum lacks HXGPRT, making it susceptible to IMPDH inhibition. (A) C. parvum development in tissue culture is resistant to the HXGPRT-activated prodrugs 6-thioxanthine and 8-azaguanine (6-TX, 8-AG, number of type 1 meronts per 25 microscopic fields, data points represent three independent coverslip cultures with bars indicating the standard deviation) but susceptible to the IMPDH inhibitors ribavirin and mycophenolic acid (D). This pattern is equivalent to the T. gondii HXGPRT null mutant (C and G) growth for T. gondii was measured daily by fluorescence in YFP-YFP expressing lines (8); rfu, relative fluorescence units; four independent wells per data point. Note susceptibility of wild-type RH T. gondii to 6-TX (B). Supplementation of the medium with 100 μM guanine completely rescues growth under mycophenolic acid (MPA) in host cells (E) and wild type T. gondii (F), which can salvage guanine through H(X)GPRT but not in C. parvum (E) or the T. gondii HXGPRT null mutant (G).