Characterization of Metabolically Quiescent Leishmania Parasites in Murine Lesions Using Heavy Water Labeling

Information on the growth rate and metabolism of microbial pathogens that cause long-term chronic infections is limited, reflecting the absence of suitable tools for measuring these parameters in vivo. Here, we have measured the replication and physiological state of Leishmania mexicana parasites in murine inflammatory lesions using 2H2O labeling. Infected BALB/c mice were labeled with 2H2O for up to 4 months, and the turnover of parasite DNA, RNA, protein and membrane lipids estimated from the rate of deuterium enrichment in constituent pentose sugars, amino acids, and fatty acids, respectively. We show that the replication rate of parasite stages in these tissues is very slow (doubling time of ~12 days), but remarkably constant throughout lesion development. Lesion parasites also exhibit markedly lower rates of RNA synthesis, protein turnover and membrane lipid synthesis than parasite stages isolated from ex vivo infected macrophages or cultured in vitro, suggesting that formation of lesions induces parasites to enter a semi-quiescent physiological state. Significantly, the determined parasite growth rate accounts for the overall increase in parasite burden indicating that parasite death and turnover of infected host cells in these lesions is minimal. We propose that the Leishmania response to lesion formation is an important adaptive strategy that minimizes macrophage activation, providing a permissive environment that supports progressive expansion of parasite burden. This labeling approach can be used to measure the dynamics of other host-microbe interactions in situ.


Introduction
A number of medically important bacterial, fungal and protozoan pathogens are associated with persistent chronic infections that can reactivate to cause acute disease long after initial infection [1][2][3][4][5]. With few exceptions [6], very little is known about the growth rate or physiological state of these pathogens during chronic stages of infection, reflecting limitations in current methods for measuring microbial growth in situ. This information is crucial for modeling host-pathogen dynamics and developing therapies that target these stages.
Leishmania spp are protozoan parasites that are associated with long-term chronic infections, as well as acute disease, ranging from self-healing cutaneous lesions to fatal visceral infections, in millions of people worldwide [7]. Infection is initiated by flagellated promastigote stages that are injected into the skin by a sandfly vector. Following their uptake by macrophages and other phagocytic cells, promastigotes differentiate to aflagellate amastigotes that proliferate in the phagolysosome compartment of these host cells [8,9]. A hallmark of all Leishmania infections is the formation of localized tissue lesions or granulomas composed primarily of infected and uninfected macrophages, at the site of the sandfly bite or in distal tissues such as the liver and spleen [10][11][12]. Depending on the Leishmania species involved and host genetics, lesion formation can be associated with immune control (but usually not eradication of the parasite) or parasite expansion and systemic infection. In murine models of infection, host resistance is associated with the development of a T-helper type 1 response, while lesion development occurs in susceptible animals that mount a T-helper type 2 response [13,14]. In contrast to our understanding of the host immune responses that underlie these different outcomes, very little is known about the growth rate or physiological state of Leishmania in these tissues. Transgenic parasite lines expressing luciferase or different fluorescent proteins have been developed and used to visualize parasite dynamics in vivo [15][16][17][18]. However, these approaches only provide a measure of net changes in parasite burden, which are determined by rates of parasite death and migration from infected tissues, as well as rates of replication. Furthermore, attempts to infer the physiological status of lesion amastigotes from transcriptomic and proteomic analyses have been hampered by the absence of conventional gene-specific transcriptional control in these parasites and the paucity of coordinated changes in the abundances of individual mRNA and proteins in different insect and mammalian-infective stages [19,20].
In this study we introduce the use of 2 H 2 O labeling to measure Leishmania growth rate and metabolic activity in murine inflammatory lesions. In the presence of 2 H 2 O, cells stably incorporate deuterium into a wide range of metabolites, which are subsequently incorporated into cellular macromolecules, providing a quantitative read-out of global rates of DNA replication, transcription, protein turnover and membrane lipid biosynthesis (S1 Fig., Fig. 1A) [21,22]. Stage-specific changes in Leishmania growth rates. A. Schematic overview of 2 H 2 O labeling protocol. Parasite stages were cultivated axenically in the presence 5% 2 H 2 O, or isolated from infected macrophages or BALB/c lesion incubated or infused with 2 H 2 O (final concentration 5%). Parasite stages were harvested at multiple time points and extracts containing total DNA/RNA, or total proteins and lipids generated from purified parasite fraction. Levels of deuterium enrichment in constituent dRib/Rib, amino acids and fatty acids were subsequently quantitated by GC-MS. B. Kinetics of 2 H-labeling of DNA dRib in cultured promastigotes (Pro log , Pro stat ) and amastigotes (Ama axenic ), and in amastigotes isolated from macrophages (Ama Mø ) and murine lesions 2 H 2 O is easily and safely administered to animals for periods of weeks to months and rapidly equilibrates across all tissues [23], making it suitable for measurement of slowly growing microbial populations in infected tissues. Using this approach, we show that amastigotes exhibit a constant, but very slow growth rate, in non-healing lesions and appear to enter into a distinct semi-quiescent metabolic state characterized by low rates of transcription and protein turnover. This quiescent state is distinct from that measured in non-dividing (insect) promastigote stages and may represent an adaptive response to a growth-restrictive intracellular microenvironment in granulomas. Using this approach we also identified parasite-specific metabolic pathways, such as polyunsaturated fatty acid biosynthesis that are up-regulated in situ. This approach has provided the first global analysis of the physiological state of the major mammalian-infective stage of Leishmania and is generally applicable to studying the in vivo growth and physiology of other microbial pathogens.

Measurement of L. mexicana amastigote replication rates in vivo
As previously reported [24], cultivation of L. mexicana promastigotes stages in standard medium containing 5% 2 H 2 O, leads to the deuterium-labeling of the deoxyribose (dRib) moiety of DNA (Fig. 1B). The labeling of dRib occurs as a result of gluconeogenesis, various sugar phosphate isomerization/epimerization reactions, the pentose phosphate pathway and ribonucleotide reductase (S1 Fig.), and the rate of incorporation of deuterium into promastigote DNA was growth-dependent. Specifically, the rate of labeling of exponentially growing promastigotes (Pro log ) indicated a doubling time of 9 hours, concordant with cell counts, while no incorporation was observed in the DNA of non-dividing promastigotes (Pro stat ) (Fig. 1B). Significantly, labeling of the deoxyribose moiety of DNA was not affected by supplementation of the culture medium with ribose or a range of nucleosides and nucleotides to the culture medium (S2 Fig.), indicating that de novo synthesis of ribose/dRib is not affected by salvage pathways. Pro stat were induced to differentiate to axenic amastigotes (Ama axenic ) by acidification of the medium and cultivation at elevated temperature. Following differentiation, Ama axenic exhibited a doubling time of 4.2 days, substantially slower than Pro log (Fig. 1B, C). These data support the notion that amastigote differentiation is associated with activation of the parasite stress responses and a reduction in maximum growth rate, independent of exogenous nutrient levels [24].
Having shown that 2 H 2 O labeling can be used to quantitate rates of parasite replication; we extended this approach to directly measure amastigote proliferation in murine tissue lesions. BALB/c mice were infected with L. mexicana parasites and subsequently labeled with 2 H 2 O following the appearance of cutaneous lesions. A constant level of 5% 2 H 2 O in the body water was established by providing mice with a bolus of 100% 2 H 2 O and subsequent inclusion of 9% 2 H 2 O in the drinking water for up to several months (S3 Fig.). Mice were culled at various time points and lesion amastigotes (Ama lesion ) isolated from infected tissues. Histological examination showed that these lesions primarily comprised heavily infected host cells (with parasites in large communal vacuoles) and no detectable necrosis (S4 Fig.). Ama lesion purified from these tissues were free of intact host cells or nuclei, as determined by DAPI staining, and were further treated with DNAse to remove any extracellular host DNA released during tissue disruption (S5 Fig.). Contamination of parasite DNA with host DNA was estimated to be less than 20% as determined by direct quantitation of DNA (S5 Fig.). Deuterium enrichment in parasite DNA-dRib, increased with time, reaching a maximum 15% enrichment (EM 1 ; excess molar fraction of M 1 ) after about 40 days of labeling (Fig. 1B). Based on the rate of enrichment, Ama lesion were found to have a remarkably constant doubling time of~12 days, irrespective of the age of the lesion when the 2 H 2 O labeling was initiated (ranging from 4 weeks to 4 months postinfection) (Fig. 1B,C). This growth rate is 32-fold slower than the maximum growth rate of promastigotes stages and approximately 4-fold slower than that measured for Ama axenic (Fig. 1C), indicating that parasite growth in the granuloma microenvironment is highly constrained.
To investigate whether the slow growth of lesion amastigotes reflected growth-limiting conditions in the phagolysosome of infected macrophages, we measured amastigote replication in J774 macrophages. J774 macrophages were infected with L. mexicana promastigotes and internalized parasites allowed to differentiate to amastigotes, before cultures were incubated in the presence of 5% 2 H 2 O. Infected macrophages were labeled for up to 20 days and amastigotes (Ama Mø ) isolated at various time points. Intracellular Ama Mø were found to have a doubling time of 4 days (Fig. 1B), comparable to that of Ama axenic and substantially faster than observed in Ama lesion (Fig. 1B). These results suggest that the phagolysosomal compartment of nonactivated macrophages is not growth limiting for amastigotes per se, and that additional factors in the lesion environment constrain parasite growth and/or induce a slow growth phenotype.
Quantitative analysis of hematoxylin and eosin (H&E) stained sections of BALB/c lesions harvested at day 85 post-infection showed that the majority of granuloma macrophages harbored between 100-200 amastigotes/phagolysosome (Fig. 1D). These values are in close agreement with parasite vacuole densities calculated assuming; (i) that each vacuole was established by a single parasite invasion event, (ii) a constant amastigote doubling time of 12 days and (iii) zero parasite death (120 parasites/vacuole). The presence of macrophages with fewer than 100 amastigotes/phagolysosome (28% of all infected macrophages) may reflect influx of uninfected macrophages that have been infected for a shorter period of time, or a subpopulation of slower growing parasites. On the other hand, the presence of small number of hyper-infected macrophages (up to 400 amastigotes/phagolysosome) may reflect a subpopulation of amastigotes with a faster growth rate and/or the uptake of multiple promastigotes/amastigotes during the initial infection. Overall, these analyses indicate that parasite and host cell turnover in lesions is minimal and that a significant majority of the granuloma host cells were infected very early after injection of parasites.

Measurement of RNA and protein turnover in lesion amastigotes
To further define the physiological state of Ama lesion , we assessed the rate of incorporation of deuterium into the ribosyl moiety of RNA and proteinogenic amino acids in different parasite stages. Measurement of deuterium enrichment in the RNA ribosyl moiety primarily reflects ribosome biosynthesis, one of the most energy intensive processes in the cell [25] and is thus a measure of the metabolic state of a cell. As expected, Pro log exhibited the highest rate of RNA turnover, comparable to the rate observed for DNA synthesis ( Fig. 2A). Appreciable levels of RNA turnover were also observed in Pro stat , confirming that this non-dividing stage remains transcriptionally and metabolically active ( Fig. 2A). Strikingly, all three amastigote stages (Ama axenic , Ama Mø and Ama lesion ) exhibited lower rates of RNA turnover than Pro stat , providing direct evidence that amastigote differentiation is linked to a general shut-down of energy intensive processes ( Fig. 2A-C).
The rate of protein synthesis/turnover provides another proxy for the metabolic state of a cell. As expected, deuterium was incorporated into a range of Leishmania proteinogenic amino acids via different transamination reactions and pathways of de novo biosynthesis [8,26] (S6A Fig.). While maximum levels of deuterium enrichment in different amino acids varied, with highest levels of enrichment in alanine (11% EM 1 ) and glutamate (9% EM 1 ), similar rates of protein turnover were calculated after normalization to maximum labeling, regardless of the amino acid used (S6B Fig.). Both dividing and non-dividing promastigote stages exhibited higher rates of protein synthesis than Ama axenic and Ama lesion ( Fig. 3A-C). Interestingly, while the turnover of protein in Pro log and Ama axenic stages occurred at approximately the same rate as DNA replication, protein turnover in Ama lesion occurred at nearly twice the rate of DNA synthesis. Thus although Ama lesion stages have the lowest absolute rates of protein turnover, the cellular proteome is turned over more times per cell division cycle than in other stages. Leishmania amastigotes also exhibit reduced membrane turnover in vivo, but are dependent on de novo synthesis of key fatty acids The turnover of membrane phospholipids is intimately linked to cell division, organelle biogenesis and dynamic cellular functions, such as secretion and endocytosis. To investigate whether membrane turnover is reduced in Leishmania amastigotes we measured global rates of fatty acid turnover in different cultured and lesion-derived stages. Fatty acids are primarily incorporated into phospholipids, with little incorporation into other lipids (such as triacylglycerol), providing a direct measure of membrane biogenesis. As with proteinogenic amino acids, the extent to which 2 H-is incorporated into different Leishmania fatty acids varies, depending on the extent to which they are generated via de novo synthesis or scavenged from the media or host (S7 Fig., S8 Fig.). However, when rates of labeling were normalized to maximum labeling, similar rates of turnover were obtained regardless of the fatty acid measured. As expected, Pro log exhibited fastest rates of fatty acid turnover, while both Pro stat and Ama axenic exhibited turnover rates that were comparable to, or slightly faster, than determined rates of protein turnover, respectively (Fig. 4). Strikingly, Ama lesion exhibited very low rates of fatty acid turnover (t 1/2~7 .8 days, compared to 1.4 days in Ama axenic ), indicating a dramatic slowdown in global rates of membrane biogenesis in lesion stages.
To further investigate whether the 2 H-labeling of Ama lesion fatty acids reflects de novo synthesis or uptake of 2 H-labeled fatty acids from the host [8,27], we measured the maximum levels of 2 H-enrichment in both parasite and host fatty acids derived from plasma or lymph nodes after a prolonged period of labeling (Fig. 5, S9 Fig.). The labeling of C14:0, C16:0 and C16:1 fatty acids was comparable in both Ama lesion and serum samples, and somewhat lower than maximum levels of enrichment in Pro log , indicating that lesion amastigotes are largely dependent on salvage pathways for these fatty acids. In contrast, maximum 2 H-enrichment in Ama lesion C18:0, C18:1 and C18:2 was higher than in equivalent host fatty acids. This was particularly pronounced for parasite C18:1 (oleic acid) and C18:2 (γ-linoleic acid) which were 2-fold or >50-fold more highly labeled than corresponding host fatty acids. The absence of 2 H-enrichment in plasma C18:2 is consistent with the absence of Δ12 oleic acid desaturase in animals, while the elevated levels of labeling of amastigote C18:1 and C18:2 indicate that these fatty acid are predominantly synthesized by the parasite. Significantly, the rate of turnover of γ-linoleic acid was very similar to that Ama lesion DNA providing additional support for the notion that these stages have a slow replication rate of~12 days in vivo. Ama lesion also contained higher levels of very long chain, polyunsaturated fatty acids, than cultured promastigotes (S8, S9 Figs). These included C20:4 n-6 (arachidonic acid) and C22:6 n-3 which were labeled to the same extent as the equivalent host fatty acids (Fig. 5A). Because the additional 2 H-enrichment in parasite γ-linoleic acid is not observed in these downstream fatty acids, they are most likely salvaged directly from the host cell (Fig. 5B). Collectively, these results show that fatty acid/membrane turnover is dramatically reduced in lesion amastigotes. Notwithstanding their reduced requirements, these stages appear to be dependent on both salvage and de novo synthesis for maintaining fatty acid levels. In particular, our data suggest that they are likely to be critically dependent on the de novo synthesis of C18:2, which is not synthesized by the host and appears to be depleted in the macrophage phagolysosome compartment.

Discussion
Information on host-parasite dynamics within Leishmania-induced lesions is limited, reflecting the technical difficulty of measuring the growth rate or physiological state of parasites within these tissues. Estimates of parasite growth based on direct enumeration of parasite numbers or detection of transgenic parasite lines expressing luciferase or fluorescent proteins are limited in sensitivity and do not distinguish between dynamic changes in the rate of pathogen replication, death or migration out of infected tissues. These approaches also require the generation of transgenic parasite lines which may alter virulence phenotypes and in which expression of reporter proteins may vary. In this study we have utilized 2 H 2 O labeling to explicitly measure the growth rate and other key physiological parameters of wild type L. mexicana amastigotes in non-healing cutaneous lesions. We show that lesion amastigotes replicate very slowly  Lesion amastigotes utilize both salvage and de novo biosynthetic pathways to supply their fatty acid needs. A. The maximum level of 2 H-enrichment (EM1, %) in major cellular fatty acids of Pro log and Ama lesion were determined after labeling for 7 days and >6 weeks, respectively. 2 H-enrichment in the total plasma lipid of infected mice was also measured to determine the potential contribution of labeled host fatty acids to the parasite labeling. Note that while saturated and unsaturated C18 fatty acids are predominant fatty acids in both stages, the fatty acid composition of Ama lesion differs from cultured promastigotes in containing elevated levels of C20:4 n-6, and polyunsaturated very long chain fatty acids (C22:4 n-6, C22:6 n-3) (S8 Fig.). The C:D nomenclature refers to overall chain length and number of double bonds in each fatty acid, respectively. n-3 and n-6 refers to the two major biosynthetic pathways involved in unsaturated fatty acid biosynthesis (where-3 and-6 refer to the position of double bond relative to the methyl carbon). B. Stagespecific differences in the levels of 2 H-enrichment in fatty acid pools can be used to infer the contributions of de novo biosynthesis and salvage pathways. In particular, levels of 2 H-enrichment in the major Ama lesion C18 fatty acids, C18:1 (oleic acid) and C18:2 (linoleic acid), were appreciably higher than in host plasma, indicating that these stages are dependent on de novo biosynthesis. Conversely, the elevated levels of 2 Henrichment in C20:4 n-6 and C22:6n-3 compared to C18:1 precursor (comparable to plasma pools) indicate that these very long chain polyunsaturated fatty acids are primarily scavenged from the host cell. (doubling time~12 days) throughout lesion development and appear to enter a distinct semiquiescent state, characterized by low rates of transcription, protein turnover and membrane biogenesis. Significantly, the calculated rates of amastigote growth account for the increase in the total parasite burden in isolated lesions, as well as the mean parasite densities in the phagolysosomes of lesion macrophages, suggesting that parasite death and turnover in these lesions occurs to a minimal extent. These analyses also provide direct evidence that infected macrophages in L. mexicana-induced granulomas have a long life span. Specifically, based on the average parasite burden of infected macrophages (100-200 amastigotes/ phagolysosome) it is likely that a significant majority of infected host cells in the lesion had been infected very early in the infection and have sustained intracellular parasites for >12 weeks. While it is possible that some macrophage death and turnover could be masked by direct transfer of parasite-laden vacuoles from a ruptured macrophage to a naive recipient host cell [28], these findings are broadly consistent with a growing body of evidence suggesting that L. mexicana amastigotes repress a number of signaling pathways in macrophages, including those that activate apoptosis, autophagy or necrosis [29,30]. Overall, these findings suggest that L. mexicana-induced lesions are characterized by slow parasite growth and low macrophage turnover. We propose that the slow replication rate of intracellular amastigotes (triggered in part by intrinsic amastigote differentiation signals), minimizes overgrowth of the phagolysosome and contributes to the long life-time of infected macrophages. Slow growth may therefore be a key factor in generating a stable, permissive tissue niche within which the parasite burden can progressively expand.
A number of other bacterial, fungal and protozoan parasites induce granulomatous structures in their host, of which the most intensively studied are the pulmonary granulomas induced by Mycobacterium tuberculosis [31,32]. Our findings suggest that host-pathogen dynamics in the L. mexicana induced lesions differ substantially from those in M. tuberculosis granulomas in several respects. In particular, the M. tuberculosis-induced granulomas are generally characterized by high rates of macrophage infiltration and host cell death [31,32]. There is also increasing evidence that bacterial replication and turnover in these granulomas may be relatively high as a result of host-mediated bacterial killing and/or immune clearance [6,12,32,33]. This contrasts with the earlier view that M. tuberculosis bacilli have a low replication rate, and corresponding low death rates, leading to the observed plateauing in bacterial numbers during chronic infections [6,12,32,33]. These observations highlight marked differences in the way different pathogens adapt to, and potentially exploit the host's attempt to wall off persistent infection with granulomas. Whether these differences are defined by intrinsic differences in pathogen growth rate remain to be determined.
While a number of microbial pathogens are thought to switch to a quiescent or semiquiescent state during long-term chronic phases of infection, relatively little is known about the physiological/metabolic state of these stages [3,4]. Here we show that slowly replicating lesion amastigotes strongly repress energy-expensive processes such as transcription, protein synthesis and membrane lipid turnover. The down-regulation of these processes in amastigotes was more pronounced than in non-dividing promastigotes, highlighting the fact that metabolic quiescence is not necessarily linked to growth rate and that non-dividing stages can remain metabolically active. It was notable, that repression of RNA synthesis was less pronounced than for either protein or fatty acid biosynthesis. Leishmania lack transcription factors and gene-specific transcriptional control [19] and a higher basal level of RNA turnover may be needed for post-transcriptional regulation of gene expression. The strong repression of protein synthesis in Ama lesion (10-fold and 5-fold lower than in Pro stat or Ama axenic , respectively) is consistent with recent studies demonstrating that promastigote to amastigote differentiation results in activation of the PERK kinase and phosphorylation of eIF2a, both of which regulate and repress protein translation and are required for amastigote virulence [34][35][36]. A general shut down in energy intensive processes, such as protein and fatty acid synthesis is also consistent with recent 13 C-flux studies on isolated Ama lesion , which identified a unique stringent metabolic response in these stages, characterized by reduced carbon utilization and more efficient mitochondrial catabolism of sugars and fatty acids [24]. Thus the physiological responses of amastigotes to the lesion microenvironment are complex and regulated at the level of DNA replication, transcription and protein synthesis, in addition to remodeling of central carbon metabolism.
Entry into this semi-quiescent state is likely to be triggered by a number of factors. Both Ama axenic and Ama Mø exhibited slow maximum rates of growth (doubling time 3.5-4 days) and RNA turnover, suggesting that entry into this state is a hardwired response to elevated temperature and/or low pH used to induce differentiation in vitro. These findings also suggest that the phagolysosome of non-activated J774 macrophages are not restrictive for amastigote growth. On the other hand, there is evidence that the growth of L. major amastigotes in cutaneous lesions is restricted by the chronic production of sub-lethal levels of nitrous oxide [18]. However, L. mexicana amastigotes reside within larger communal phagolysosome compartments and are intrinsically more resistant to macrophage microbicidal processes, including nitric oxide or reactive oxygen species [14,37]. They also appear to have evolved additional mechanisms for inhibiting macrophage activation and nitric oxide production [38]. The slow growth rate of L. mexicana amastigotes in lesions may therefore reflect both intrinsic parasite responses, as well as adaptive responses to other host microbicidal processes, nutrient deprivation and/or physical stresses in this niche [24].
The high parasite inoculum used in our studies is expected to lead to rapid recruitment of macrophages and early induction of lesion development [39]. In contrast, infection of mice with a low dose inoculum is associated with a significant delay in lesion formation that is preceded by an exponential increase in parasite numbers [40]. In the L. major-C57BL murine model, this silent expansive phase is associated with an increase in parasite burden consistent with an apparent parasite doubling time of~2.3 days [40]. This growth rate is similar to the replication rate we observed for L. mexicana amastigotes in non-activated J774 macrophages, raising the possibility that Leishmania amastigotes may switch between different growth states during acute and long-term chronic phases of infection. The possibility that amastigotes exhibit a range of growth rates within lesion was also suggested by the detection of a small number (<20%) of hyper-infected macrophages with more than 200 parasites/macrophage. The existence of distinct amastigote growth/physiological states is analogous to the situation in the sandfly vector. Initial colonization of the mid-gut of this host is mediated by 'procyclic' promastigotes that are rapidly dividing and generally more sensitive to a variety of physiological and nutritional stresses. These stages transition though a number of physiological states before differentiating to non-replicating, metacyclic promastigotes (related to Pro stat ) in the foregut [41]. Metacyclic promastigotes are resistant to a number of stresses (including elevated temperature), suggesting that slow-growth represents a generalized response to elevated stresses in both the insect and mammalian hosts.
Many microbial pathogens acquire essential nutrients or metabolites via a combination of salvage pathways or de novo synthesis, leading to a level of redundancy that complicates efforts to identify and validate drug targets. While administration of 2 H 2 O to infected mice results in metabolic labeling of both parasite and host lipids, analysis of the relative rates of labeling of parasite/host pools can be used to infer the contribution of salvage versus de novo biosynthetic pathway to parasite lipid homeostasis. In particular, we found that 2 H-enrichment in Ama lesion C18 fatty acids was substantially higher than in the equivalent host fatty acids, indicating that they are synthesize de novo by intracellular amastigotes. This was particularly striking for C18:2, which was labeled to a significant extent in Ama lesion , but unlabeled in mouse plasma samples. The absence of labeling of plasma C18:2 was not surprising given that animals lack the Δ12 desaturase needed to synthesize linoleic acid and are dependent on uptake of this fatty acid in the diet. The high maximum labeling of C18:2 in Ama lesion indicates that this fatty acid is limiting for parasite growth in the phagolysosome compartment and therefore that amastigotes may be dependent on their own Δ12 desaturase for intracellular growth [42,43]. On the other hand, Ama lesion contained elevated levels of very long chain, polyunsaturated fatty acids, including C20:4 n-6 (arachidonic acid), C22:4 n-6 (adrenic acid), C22:5 n-6 (osbond acid) and C22:6 n-3 (cervonic acid) compared to promastigotes. These fatty acids were 2 H-enriched to a higher level than parasite C18 precursors fatty acids and were labeled to the same extent as the equivalent host fatty acids, suggesting that they are derived primarily via salvage pathways. These findings add to accumulating evidence that Leishmania amastigotes acquire a range of lipids from the host cell [24,44,45] and suggest that salvage as well as de novo biosynthesis pathways are potential drug targets.
In summary, we show that 2 H 2 O labeling can be used as a universal labeling procedure to measure microbial growth, physiology and metabolism in their animal hosts. This approach is well suited for studying the in vivo growth and metabolism of other microbial pathogens.

Materials and Methods
Cell culture L. mexicana promastigotes (Pro log , Pro stat ) were cultured in RPMI 1640 medium supplemented with 10% (v/v) heat-inactivated fetal calf serum (FCS) at 27°C. Axenic amastigote (Ama axenic ) were generated by adjusting the pH of the medium of stationary phase promastigotes (day 5, Pro Stat ) to pH 5.5 with 1 M HCl and addition of 10% FCS (20% v/v final), followed by incubation at 33°C for four days.

H O labeling
Cultivated parasites stages (Pro log , Pro stat , Ama axenic ) were cultured in medium supplemented directly with phosphate buffered saline-2 H 2 O (99.9%, Cambridge Isotopes) to give a final concentration of 5% (v/v) 2 H 2 O.
J774 macrophages (4 × 10 6 ) were grown overnight in RPMI 1640 medium supplemented with 10% (v/v) FCS, penicillin and streptomycin at 33°C [46], before being infected with L. mexicana Pro stat at a MOI of 3. Non-internalized parasites were removed after 4 hr by washing macrophages twice with fresh RPMI medium. After 3 days, the medium of infected macrophages was replaced with fresh medium containing 5% 2 H 2 O and cells harvested at various time points (ranging from 4 h and 16 days). For long term labeling experiments, the medium containing 5% 2 H 2 O was replaced every 5 days.
BALB/c mice (6 week old) were infected with L. mexicana Pro stat (10 6 in 50 μl PBS) near the base of the tail and lesion size monitored as previously described [46]. Infected mice were injected intra-peritoneally with 2 H 2 O (99%, 35 μl/g body weight) containing 0.9% NaCl after the development of nascent lesions and serum 2 H 2 O concentration subsequently maintained at 5% by supplementation of the drinking water with 9% 2 H 2 O. 2 H 2 O levels in the urine were routinely monitored as previously described [47]. To determine maximum labeling parasite metabolites (EM 1 ), mice were labeled with 2 H 2 O immediately after the infection and parasites harvested one month after the development of a granulomatous lesion.

Harvest of parasites
Cultured parasites stages were harvested with rapid metabolic quenching as previously described [48] and cell pellets (triplicate samples) were stored at -80°C prior to extraction. Infected J774 macrophages were metabolically quenched by chilling plates on ice and replacing the overlying culture medium with ice-chilled PBS. Infected macrophages were scraped from the plastic surface and lysed by repeated passage through a 25G needle (x10). After low speed centrifugation to remove host cell debris (60 g, 5 min, 4°C), amastigotes were recovered by sequential filtration through 5 μm and 3 μm pore filters and centrifugation (1500g, 10 min, 4°C) of the filtrate. The parasite pellet was washed three times with chilled PBS and any residual host cell DNA, removed by treatment of the pellet with 1000 U of DNAase in PBS with 5 mM MgCl 2 for 2 h at 33°C.
Mice were culled humanely post-labeling (24 h to 150 days) and granulomatous lesions excised and immediately chilled in cold PBS. All subsequent procedures were carried out at <4°C. The isolated tissue mass was disrupted by passage through a cell strainer and host cells lyzed by passage through a 27 G syringe needle (x5). Intact host cells and host cell debris were removed by centrifugation (60 g, 10 min, 4°C) and released parasites harvested by centrifugation (1500 g, 10 min, 4°C). The pellet was washed three times with chilled PBS and the purity of the parasite extract confirmed by light microscopy. Samples for DNA analysis were treated with 1000 U of DNAase in PBS with 5 mM MgCl 2 for 2 h at 33°C to remove any host cell DNA.
Histology BALB/c lesions (85 day post-infection) were fixed with 10% paraformaldehyde in PBS, embedded in paraffin and tissue sections stained with hematoxylin and eosin (H&E) reagents.
All derivitized samples were analyzed by GC/MS using a DB5 capillary column (J&W Scientific, 30 m, 250 μm inner diameter, 0.25 μm film thickness), with a 10 m inert duraguard. The injector insert and GC/MS transfer line temperatures were 270 and 250°C, respectively. The oven temperature gradient was set to: 70°C (1 min); 70°C to 295°C at 12.5°C/min;295°C to 320°C at 25°C/min; 320°C for 2 min. All metabolites were identified based on GCretention times and mass spectra of standards. The excess M 1 fraction and the fraction of new cells/molecules were calculated as described [51] following the measurement of the Total lipids were extracted in chloroform/methanol/water (1:2:0.8 v/v) as described previously [48]. The organic phase was dried in a centrifugal evaporator and resuspended in chloroform/methanol (2:1 v/v, 30 μl) and total fatty acids analyzed as their methylesters after addition of Meth-Prep II reagent (5 μl, Grace Davison, Alltech) and direct injection and analysis by GC/MS in electron impact (EI) mode and positive chemical ionization mode. The delipidated protein pellets were hydrolyzed in 6 M HCl (200 μl, 110°C, 18 hr) and insoluble material removed by centrifugation (16,100 g, 5 min, RT). The supernatant was dried under nitrogen and released amino acids converted to their TBDMS derivatives by addition of MTBSTFA-1% TBDMCS (30 μl) and pyridine (30 μl, 60°C, 30 min) prior to GC-MS analysis in positive chemical ionization mode as described above.

Enumeration of parasite numbers in lesion macrophages
BALB/c lesions (85 day post-infection) were fixed with 10% paraformaldehyde in PBS, embedded in paraffin and tissue sections stained with hematoxylin and eosin (H&E) reagents. Slides were imaged using a Zeiss Axioplan microscope and digital computer images were recorded with a Zeiss camera control unit and the corresponding dedicated software. The numbers of amastigotes was counted in 8 sections of 2 lesions evaluating >1200 phagolysosome in total. The number of parasites/vacuole was calculated assuming that (1) the phagolysosome compartment is a sphere, (2) that the average cross sectional diameter of the phagolysosome is equivalent to the largest vacuolar sections seen in the tissue sections and (3) that parasites are primarily arranged around the periphery of the phagolysosome at equal density. The average cross sectional diameter of the parasite occupied PVs was found to be 25.6 μm, while the average diameter of intracellular amastigotes was 2.5 μm.
To determine the maximum number parasites that fit in a vacuole section the ratio of the areas was calculated To confirm that the level of contaminating host DNA is low and largely removed by DNase treatment, we quantitated total DNA and extracellular DNA in the lesion preparations (pre-and post-DNase treatment). DNA levels were determined by spectrophotometric measurement of DAPI fluorescence (fluorescence units) (error bars are for two technical replicates). These analyses show that contaminating host DNA accounts for less than 20% of the total parasite DNA after DNase treatment. C. To further confirm that DNase treatment effectively removes contaminating extracellular DNA, Ama lesion were metabolically labeled with 13 C-glucose (to incorporate uniformly labeled dRib into the DNA) and preparations spiked with excess unlabeled salmon sperm DNA (5-fold over parasite DNA). The level of contamination with salmon DNA was subsequently determined by GC-MS analysis of DNA-dRib and quantitation of the ratio of M +5 /M 0 isotopomers. Pretreatment of live parasites with DNase restored the M +5 /M 0 ratio of the DNA-deoxyribose to that found in parasites which had not be exposed to extracellular DNA, confirming that DNase effectively removes contaminating DNA.