XVII. Lipid metabolism of Pneumocystis : toward the de¢nition of new molecular targets

Drugs targeted against lipids or their biosyntheses have proved highly eiective against fungi. Polyene antibiotics such as amphotericin B bind to sterols (particularly with ergosterol), the complexes aggregate, and then form pores in the organism's membranes, thus abolishing ion gradients. Fluconazole and other related drugs inhibit speci¢c steps in fungal sterol biosynthesis, and are eiective against systemic mycoses. Parasites typically scavenge nutrients from their hosts, and metabolize them or utilize them directly for elaborating cellular structures. Thus, membranes can be formed by inserting some hostderived molecules into the bulk phase of lipid bilayers. However, it is believed that some lipids, intimately associated in domains serving important physiological functions (e.g., pumps, enzymes, signal transduction) require parasite lipids having speci¢c three-dimensional con¢gurations. Thus, if host lipids cannot provide the appropriate molecules, the parasite must synthesize at least a low amount of these lipids for those domains. Vital pathogen-speci¢c lipids that are not available from the host have been described as `metabolic' lipids [1,2] because they enable the organism to function properly and proliferate. These make attractive targets for elimination of the pathogen.


Introduction
Drugs targeted against lipids or their biosyntheses have proved highly e¡ective against fungi. Polyene antibiotics such as amphotericin B bind to sterols (particularly with ergosterol), the complexes aggregate, and then form pores in the organism's membranes, thus abolishing ion gradients. Fluconazole and other related drugs inhibit speci¢c steps in fungal sterol biosynthesis, and are e¡ective against systemic mycoses. Parasites typically scavenge nutrients from their hosts, and metabolize them or utilize them directly for elaborating cellular structures. Thus, membranes can be formed by inserting some hostderived molecules into the bulk phase of lipid bilayers. However, it is believed that some lipids, intimately associated in domains serving important physiological functions (e.g., pumps, enzymes, signal transduction) require parasite lipids having speci¢c three-dimensional con¢gurations. Thus, if host lipids cannot provide the appropriate molecules, the parasite must synthesize at least a low amount of these lipids for those domains. Vital pathogen-speci¢c lipids that are not available from the host have been described as`metabolic' lipids [1,2] because they en-able the organism to function properly and proliferate. These make attractive targets for elimination of the pathogen.

Puri¢ed organisms
Studies on Pneumocystis carinii lipids have been performed on organism preparations for which purity was not rigorously de¢ned by both qualitative and quantitative criteria [3^8]. It became apparent that extensive ground work was required before credible, reproducible and interpretable lipid biochemical data were to be obtained from organisms isolated from infected animal models. The alveolar lining in which P. carinii proliferates is bathed in lung surfactant, which is composed primarily of lipids and lesser amounts of proteins and carbohydrates. The lipids of lung surfactant are characterized by the predominance (about half) of dipalmitoylphosphatidylcholine (disaturated PC). Lung surfactant lipids [3] and proteins [9] bind avidly to P. carinii surfaces which are not removed by washing with ordinary bu¡ered salt solutions. Treatment with reagents such as the divalent cation chelator EDTA is required to remove surfactant components (e.g., surfactant protein A, SP-A) [9]. Thus the levels of host tissue and molecular contaminants in organism preparations had to be measured. Light microscopic monitoring of intact host cells is not su¤cient for detecting contaminating host cell fragments or molecules.
A puri¢cation protocol was developed ( Fig. 1) which prioritized purity over yield. The corticosteroid-immunosuppressed model [10] was adopted in which P. carinii pneumonia (PcP) was induced by intratracheal inoculation of cryopreserved bacterialand fungal-negative organisms into viral antigennegative P. carinii-free rats. Included in this isolation procedure was a mucolytic agent (sulfhydryl reagent) which caused the detachment of organisms from each other and from type I pneumocytes of the lung epithelium presumably by breaking disul¢de linkages. Hence, alveolar type I cells (to which organisms attach) and delicate structures, such as tubular extensions of trophozoite cell surfaces, remained intact (Fig. 2). Also, these preparations readily passed through micro¢lters in the ¢nal step of the puri¢cation protocol [11]. The^SH agent also aided in the removal of surfactant proteins and other host contaminants from organisms surfaces; the resultant preparations lacked SP-A and other exogenous substances (see below). Glutathione was selected as theŜ H agent of choice since it is a normal component of the alveolar lining £uid, and it is milder than the other chemicals tested.
The critical aspect of developing a contaminantfree preparation was documentation that it is free of host cell fragments and molecules. Thus, the purity of the preparations was demonstrated by multiple criteria including light and electron microscopy, biochemical, and immunochemical analyses (Table 1).
(1) By light microscopy, intact host cells were not detected in stained preparations observed under bright ¢eld optics or in unstained preparations observed under phase-contrast and di¡erential interfer-  Pneumocystis was identi¢ed by its distinctive thick glycocalyx; membrane fragments with thick glycocalyces were identi¢ed as P. carinii; those without were scored as contaminant. ELISA, enzyme-linked immunosorbent assay. GLC, gas-liquid chromatography. d Resuspending the pellet obtained after the ¢rst low spin, and pooling the supernatant from the ¢rst low spin resulted in higher recoveries. e Lungs from individual immunosuppressed, uninfected rats processed through the same protocol. f Typical P. carinii ¢nal preparations from individual rats contained approx. 5 mg protein.
ence optics. Electron microscopy veri¢ed the absence of intact host cells.
(2) Transmission electron microscopic analysis was performed on 28 separate preparations, and 2^4 blocks were prepared from each. Each block was sectioned at v2 levels, and 80^100 micrographs of each preparation were randomly taken at low magni¢cation (maximal area). Quantitative analysis of the micrographs was performed by identifying organisms by their distinctive thick glycoca-lyx; all isolated membrane fragments that lacked a thick glycocalyx (even if they could be P. carinii cytoplasmic membranes) were scored as contaminant. Purity was expressed in Table 1 as the lowest (not average) estimate. (3) The lung surfactant marker SP-A was quanti¢ed using monospeci¢c polyclonal antibodies directed against rat SP-A in an enzyme-linked immunosorbent assay (ELISA). Again, organism purity in Table 1 is expressed as the lowest (not average), based on these analyses. (4) To determine whether exogenous host lipids were adhered to organism surfaces, stigmasterol (not normally detectable in P. carinii) was added during the early homogenization step in the isolation procedure. Aliquots were analyzed at each step of the isolation and puri¢cation procedures; stigmasterol was not detected during the series of high-and low-speed centrifugations, midway in the protocol. Thus potential exogenous lipid contaminants were shown to be eliminated in this protocol. These quantitative analyses showed that the preparations were s 95^100% free of host contamination. Several P. carinii lipid analyses have since been performed using this protocol [12^17]. Microscopic analysis of preparations stained by the dual (live/dead) staining procedure, using calcein acetoxymethyl ester plus propidium iodide [18,19], indicated 80^95% of the organisms were viable. In these analyses, moribund cells exhibiting dual staining, and empty cysts containing residual doublestranded nucleic acids, were scored as dead. The high levels of ATP measured in these preparations con¢rmed the high percentage of viable organisms [11]. Incorporation of radiolabeled precursors into P. carinii compounds also showed that these preparations were of metabolically active organisms [212 3]. The yield from a single heavily infected rat was 10 V^1 0 W organisms, typically containing approximately 5 mg protein. Process controls, in which the lungs from individual P. carinii-free immunosuppressed rats were subjected to the same protocol, resulted in a maximum (not average) of 6 300 mg protein.

Sterols
An early signi¢cant ¢nding on the nature of P. carinii lipids was the inability to detect ergosterol, which is the major sterol of most fungi that have been analyzed [2]. This observation explained the ine¤cacy of the antimycotic drug amphotericin B. The C PU sterol cholesterol (Fig. 3A), which is apparently scavenged from the mammalian host, was present in the highest amount in P. carinii. However, P. carinii was shown to have at least 24 sterol components, most of which were not detected in the lungs of normal rats and immunosuppressed, uninfected rats [6,8,12,13]. De novo isoprenoid biosynthesis in P. carinii was further demonstrated by the detection of lovastatin-inhibitable HMG-CoA reductase activity [20], the key regulatory enzyme in isoprenoid metabolism. The product of this enzyme is mevalonic acid, from which sterols and other isoprenoid compounds are synthesized. Only trace to low levels of HMG-CoA reductase activity were detected in lungs from normal rats and from immunosuppressed, uninfected rats.
The P. carinii-speci¢c sterols were characterized by an alkyl group at C-24 of the sterol side chain (Fig.  3B^D), which was identi¢ed as a drug target since mammals are unable to alkylate sterol C-24 [13,22]. Another unique feature of the P. carinii sterols is that they had a double bond at C-7 of the sterol nucleus (v U ). These putative`metabolic' sterols are not found in mammals or in most infectious agents (e.g. Candida, Aspergillus) and are therefore also considered`signature' lipids of the pathogen. Detection of these sterols in a sample would highly indicate the organism's presence. Examples of major P. carinii-speci¢c`signature' and putative`metabolic' sterols are fungisterol (ergost-7-en-3L-ol; 24L-methylcholest-7-en-3L-ol; Fig. 3B), which constituted 21% of the total, and 29% of the free non-cholesterol Fig. 3. Structures of sterols found in P. carinii. A: Cholesterol, which is probably scavenged from the mammalian host. B: Fungisterol, a C PV sterol with a v U unsaturation, is a major P. carinii-speci¢c sterol. C : Stigmast-7-3L-ol (24 methyl cholest-7-3L-ol), a C PW sterol. D: Pneumocysterol, a C QP sterol which is a C-24-alkylated lanosterol molecule. sterol components) and stigmast-7-3L-ol (24 ethylcholest-7-3L-ol; Fig. 3C).
Alkylation of sterol C-24 utilizes transferase enzymes and S-adenosylmethionine (SAM) as the methyl donor. Unlike cholesterol, C PV sterols have an additional methyl or methylene group at C-24, and C PW sterols have an ethyl or ethylene group with two carbons resulting from sequential additions of two methyl groups transferred from SAM. Two inhibitors of C-24 alkylation, 22,26-azasterol and 24,25-epiiminolanosterol, were shown to inhibit ratderived P. carinii proliferation in a monoxenic culture system and to alter its sterol composition [8]. The drugs caused a relative increase in the total percentage of C PU sterols with a concomitant decrease in the percentage of C PV and C PW sterols. The C PU sterols that accumulated in drug-treated organisms were not detectable, or present in only trace amounts, in untreated controls. The e¡ects of sterol analogues with side groups linked by a direct P^C bond were also tested on P. carinii and were found deleterious to the organism [23].
Since the P. carinii-speci¢c signature sterols include those with v U unsaturation, the formation of this double bond may be vital for the organism [6,8,13]. Thus, structural characterizations of P. carinii-speci¢c sterols have also identi¢ed v V to v Uisomerase of the pathogen as a potential chemotherapeutic drug target.
Recently, very unusual and rare sterols were found in lung and bronchoalveolar lavage samples contain-ing P. carinii hominis. They were identi¢ed as a C QI (euphorbol) and a C QP sterol, which was given the trivial name pneumocysterol [25] (Fig. 3D). Pneumocysterol may be present in trace amounts in P. carinii carinii isolated from rat lungs, but it can accumulate in P. carinii hominis in very high concentrations. The C QI and C QP sterols were identi¢ed as C-24-alkylated lanosterol molecules, previously reported in some plants. In many cells, lanosterol (C QH sterol with a double bond at C-8 and three methyl groups attached to the sterol nucleus) is an intermediate compound. Three methyl groups on the sterol nucleus are usually removed during processing of the compound to the ¢nal accumulated products (e.g. in the biosynthesis of C PU cholesterol). Identi¢cation of the C QI and C QP sterols further indicate that C-24 alkylation is important and highly active in P. carinii, and that the v PRPS and v PRPV sterol methyltransferases responsible for C-24 alkylation are excellent drug targets.

Ubiquinone (coenzyme Q, CoQ)
In addition to sterols, isoprenoid biosynthesis also involves branch pathways leading to the formation of a number of products, including the polyprenyl chain of ubiquinone. Polyprenyl chains are formed from isopententyl groups (C S ), and the number of these units in the ubiquinone polyprenyl chain designates CoQ homologs. For example, the CoQ W po- lyprenyl chain contains nine C S isopentenyl units (45 carbons total). A completed chain, the length of which depends on the organism and tissue, is added to p-hydroxybenzoate (benzoquinone ring of CoQ) by chain length-speci¢c polyprenyl transferases (Fig. 4). The hydrophobic polyprenyl chain allows insertion into the inner mitochondrial membrane bilayer where CoQ plays a central role in respiration, mediating the transfer of electrons from a number of dehydrogenases to the cytochrome electron transport chain or the alternative oxidase system (Fig. 5) [26]. The cytochrome bc I complex contains ubiquinone (CoQ) which is reduced in reactions at the matrix side of the inner mitochondrial membrane (Q i site), whereas ubiquinol (CoQH P ) oxidation occurs at the side facing the cytoplasm (Q o site). High levels of CoQ are also found in other cytomembranes such as those of the Golgi apparatus and the cell surface where CoQ apparently participates in electron transfer [27].
Since CoQ has a pivotal metabolic function, it is an attractive chemotherapeutic target in parasites. Di¡erences in the cyt b gene nucleotide sequence of humans and a number of parasitic protists suggest Q i or Q o sites as attractive drug targets [26]. Several analogs of ubiquinone (e.g., the hydroxynaphthoquinone atovaquone) and 8-aminoquinolones (e.g., primaquine) are e¡ective against malaria and also exhibit anti-P. carinii activity [25,27,28]. 8-Aminoquinolone metabolites include compounds similar to hydroxynaphthoquinones, hence the mechanism of action of these two groups of drugs may be similar [26]. The antimalarial activity of atovaquone results from blocking electron transport to CoQ from dihydroorotate dehydrogenase (DHOD), an important enzyme in pyrimidine biosynthesis. Since Plasmodium spp. cannot scavenge host pyrimidines, inhibition of de novo pyrimidine synthesis has cidal consequences for these organisms. Unlike Plasmodium, Pneumocystis DHOD activity is relatively insensitive to atovaquone [29]. The anti-P. carinii activity of atovaquone and other hydroxynaphthoquinones may occur by inhibition of electron transport and cellular respiration [30] resulting from blocking CoQ synthesis and/or displacing CoQ from the membrane. The CoQ molecule is comprised of two parts. In most microbes and plants, the benzoquinone moiety is synthesized de novo from chorismate, a key product of the shikimic acid pathway (Fig. 5). From chorismate, several pathways diverge and lead to the syntheses of vital compounds, including the aromatic amino acids, folic acid, and vitamin K (Fig. 6). Since mammals, including humans, lack the shikimic acid pathway, the products formed from chorismate are essential, or dietary requirements for mammals. Mammals can produce ubiquinone by incorporating dietary aromatic amino acids (tyrosine or phenylalanine) into the molecule. Since humans lack the shikimic acid and most post-chorismate pathways, reactions in the pathogen for synthesizing these products are excellent drug targets. For example, folic acid synthesis in P. carinii is an attractive drug target, and a number of anti-folates are active against P. carinii viability and proliferation [31]. One pathway from chorismate leads to the formation of p-hydroxybenzoic acid, the direct precursor of the CoQ benzoquinone moiety.
The major CoQ homolog in P. carinii carinii was found to be CoQ IH ; CoQ W was also present [15,20]. In contrast, rat lung controls contained CoQ W as the major homolog, and CoQ IH was not detected. These observations suggested that P. carinii had the enzymatic capacity for synthesizing the polyprenyl chain of at least CoQ IH , and to attach it to the ring structure. Radiolabeled mevalonic acid was incorporated in vitro into a band on thin-layer chromatography (TLC) plates containing ubiquinones [15], indicating that the organism was capable of synthesizing the polyprenyl chain of CoQ. Recently, radiolabeled mevalonate and several precursors of the benzoquinone ring were shown to be incorporated in vitro into P. carinii CoQ W and CoQ IH ( Table 2) [24]. These studies demonstrated that the organism has functional pathways for synthesizing both moieties of CoQ W and CoQ IH molecules.

cis-9,10-Epoxy stearic acid
Thus far, the only fatty acid in P. carinii that was not detected in lung controls is cis-9,10-epoxy octadecanoic acid [16] (Fig. 7). This epoxy fatty acid is synthesized from oleic acid (E.S. Kaneshiro, unpublished) which has a single double bond in the middle of the chain. Most known lipoxygenases use fatty acid substrates with two or more conjugated cis methylene-interrupted double bonds (e.g., arachidonic acid). It is likely that this rare fatty acid is essential for P. carinii viability, and that the lipoxygenase enzyme responsible for the formation of the epoxide ring in 18:1 is unique to the pathogen. The structure of the parasite enzyme catalyzing epoxide fatty acid formation from oleate would be expected to di¡er from the host enzymes that use substrates with conjugated cis methylene-interrupted double bonds. Thus, the enzyme in P. carinii that catalyzes the biosynthesis of the signature lipid, cis-9,10-epoxy stearic acid, has been identi¢ed as a potential molecular target for the elimination of this pathogen. Fig. 7. The structure of the unique P. carinii fatty acid, cis-9,10epoxy octadecanoic acid (cis-9,10-stearic acid). Fig. 6. Metabolism of chorismate. A number of pathways lead to compounds that mammals, including humans, cannot synthesize. Mammals obtain most from their diets. The benzoquinone moiety of CoQ is a product of one of those pathways. Total protein of P. carinii preparations.

Conclusions
Characterizations of P. carinii lipids and those of lungs from normal rats and corticosteroid-immunosuppressed rats have proved productive in identifying several organism lipids that are not present or synthesized by the host. Reliable data using organism preparations shown to be of high purity by several independent quantitative analyses enabled the distinction between molecules of P. carinii and those of the host's lung. Thus far, several unique v U C-24alkylated sterols, ubiquinone, and cis-9,10-epoxy stearic acid (and the reactions in their biosyntheses) have been identi¢ed as molecular targets for PcP. Other targets may be identi¢ed after other lipid classes and lipid species present in low concentrations have been characterized in greater detail.