The Inflammatory Effects of Dietary Lipids Regulate Growth of Parasites during Visceral Leishmaniasis

ABSTRACT Visceral leishmaniasis is a potentially fatal disease caused by the protozoon Leishmania donovani or L. infantum (Li). Although previous studies revealed that high lipid intake reduces parasite burdens in Leishmania donovani-infected mice, the specific contributions of dietary lipids to Li-associated pathogenesis are not known. To address this, we evaluated parasite growth, liver pathology, and transcriptomic signatures in Li-infected BALB/c mice fed either a control, high-fat, high-cholesterol, or high-fat–high-cholesterol diet. Using quantitative PCR (qPCR), we observed significantly reduced liver parasite burdens in mice fed the high-fat–high-cholesterol diet compared to mice fed the control diet. In contrast to the liver, parasite expansion occurred earlier in the spleens of mice fed the experimental diets. Histological examination revealed an intense inflammatory cell infiltrate in livers predominantly composed of neutrophils caused by the high-fat–high-cholesterol diet specifically. After 8 weeks of infection (12 weeks of diet), Illumina microarrays revealed significantly increased expression of transcripts belonging to immune- and angiogenesis-related pathways in livers of both uninfected and Li-infected mice fed the high-fat–high-cholesterol diet. These data suggest that increased fat and cholesterol intake prior to Li infection leads to a hepatic inflammatory environment and thus reduces the parasite burden in the liver. Defining inflammatory signatures as well as pathology in the liver may reveal opportunities to modify the therapeutic approach to Li infection. IMPORTANCE Leishmaniasis is a spectrum of diseases caused by Leishmania species protozoa that is most common in warm climates, coinciding with impoverished regions. Visceral leishmaniasis is a potentially fatal disease in which parasites infect reticuloendothelial organs and cause progressive wasting and immunocompromise. The distribution and demographics of visceral leishmaniasis have changed over recent years, coinciding with modernizing societies and the increased availability of Western diets rich in lipid content. We report here that increased dietary fat and cholesterol intake affected disease pathogenesis by increasing inflammation and reducing localized parasite burdens in the liver. These diet-induced changes in disease pathogenesis might explain in part the changing epidemiology of visceral leishmaniasis. A relationship between diet and inflammatory responses may occur in leishmaniasis and other microbial or immune-mediated diseases, possibly revealing opportunities to modify the therapeutic approach to microbial infections.

IMPORTANCE Leishmaniasis is a spectrum of diseases caused by Leishmania species protozoa that is most common in warm climates, coinciding with impoverished regions. Visceral leishmaniasis is a potentially fatal disease in which parasites infect reticuloendothelial organs and cause progressive wasting and immunocompromise. The distribution and demographics of visceral leishmaniasis have changed over recent years, coinciding with modernizing societies and the increased availability of Western diets rich in lipid content. We report here that increased dietary fat and cholesterol intake affected disease pathogenesis by increasing inflammation and reducing localized parasite burdens in the liver. These diet-induced changes in disease pathogenesis might explain in part the changing epidemiology of visceral leishmaniasis. A relationship between diet and inflammatory responses may occur in leishmaniasis and other microbial or immune-mediated diseases, possibly revealing opportunities to modify the therapeutic approach to microbial infections. KEYWORDS dietary lipids, inflammation, leishmaniasis, neglected tropical diseases, transcriptome L eishmaniasis, a vector-borne disease caused by Leishmania species parasites, leads to human morbidity or mortality in more than 100 countries (1,2). Leishmaniasis refers to a spectrum of human diseases ranging from self-healing cutaneous leishmaniasis to visceral leishmaniasis (VL); the latter is the most severe form and causes most leishmaniasis-related fatalities (3). Clinical manifestations with different strains of each species of Leishmania can be highly variable, but the infecting species is a major determinant of the type of disease syndrome (3). VL is usually due to infection by Leishmania infantum (Li) or Leishmania donovani (4). Even after cure of symptomatic disease, parasites can be found in a quiescent state within host tissues of many, and possibly all, infected individuals (5,6). It is not fully understood what host-or parasitespecific factors determine whether some patients remain asymptomatic whereas others develop severe VL, and these factors must be critically important determinants of the outcome of infection.
Sandflies, the insect vector of leishmaniasis, inoculate infectious parasites into the human dermis, after which the parasites are phagocytized by myeloid cells and can manipulate the host's immune response to their own survival advantage (4,(7)(8)(9). In addition to these perturbances of the immune response, individuals infected with Leishmania species parasites have characteristic changes in serum lipid profiles. More specifically, human patients with VL have elevated triglycerides and reduced cholesterol levels (10)(11)(12) as well as decreased levels of some apolipoproteins (13). Additionally, at least two transcriptional profiling studies highlight lipid metabolic pathways among the "most relevant" pathways modulated by Leishmania species infection of mouse bone marrow-derived macrophages (14,15).
Rather than a unique mechanism incited by Leishmania species parasites, the above-described metabolic changes may reflect the changes in lipid metabolism occurring during many infectious diseases (11,16). Although these changes are transient, they almost certainly affect immune responses to the infection. Indeed, one study reported that the parasite metalloprotease gp63 alters host microRNA 122 (miR-122), leading to lower serum cholesterol levels (17). Furthermore, the addition of exogenous host microRNA 122 during L. donovani infection led to the return of cholesterol to normal levels and a reduction of the parasite burden (17).
In regions of Brazil where Leishmania is endemic, increased prevalences of obesity and food insecurity reflect dietary changes (18,19), and the effects of these dietary changes on the outcome of Leishmania infection have not been fully investigated. Ghosh et al. demonstrated that C57BL/6 mice fed a high-fat-high-cholesterol diet (21.1% fat and 1.3% cholesterol) prior to infection with L. donovani had significantly reduced parasite burdens in spleens and livers at 3 and 6 weeks postinfection, compared to mice fed a control diet (20). In contrast, another study showed that C57BL/6 mice fed a high-sugar and -butter diet had increased Li parasite burdens in their spleens and livers at 8 weeks postinfection (21). Parasite species and diet are possibly the explanation for these discrepancies. Nevertheless, we hypothesized that there might be more mechanisms underlying these divergent effects of elevated dietary lipids on the outcome of Leishmania infection.
Here, we investigated the effects of dietary manipulation of fat and cholesterol on the transcriptional profiles of mouse livers and the outcome of Li infection. Our data revealed not only changes in liver transcriptional profiles but also changes in parasite loads, which were divergent between the spleen and the liver. These data raised the possibility that transient manipulation of dietary intake during leishmaniasis might be advised as a synergistic approach with drug therapy in the management of leishmaniasis. duration of the experiment. Macronutrient and cholesterol compositions of the diets can be found in Table 1. Eight weeks after inoculation (12 weeks on the diets), liver weights were significantly increased in both uninfected and infected mice fed the high-fat-high-cholesterol diet compared to mice fed the control diet. The spleen weight was also significantly increased in Li-infected mice after 12 weeks of the highfat and the high-fat-high-cholesterol diets (Fig. 1A).
Murine leishmaniasis usually follows a pattern of the parasite burden peaking in livers after 4 to 6 weeks of infection; this is followed by resolution in the liver and a gradual increase of the parasite burden in the spleen (22,23). Using quantitative PCR (qPCR), we observed that parasite burdens in the livers and spleens of Li-infected mice fed the control diet followed these expected kinetics (Fig. 1B). Parasite burdens in mice fed the experimental diets deviated significantly from the usual kinetics, and the changes were most exaggerated in mice fed the high-fat-high-cholesterol diet (Fig. 1B). Mice fed the high-fat-high-cholesterol diet had lower liver parasite burdens after 4 and 8 weeks of infection (also illustrated by an in vitro imaging system [IVIS] at 4 weeks) (Fig. 1C). In contrast, parasite loads in the spleens of mice fed the high-fathigh-cholesterol diet rose to significantly higher levels after 8 weeks of infection than in animals fed the control diets (Fig. 1B). Thus, a high-fat-high-cholesterol diet changed the organ-specific amplification of parasites during the first 2 months of infection, making the liver refractory and the spleen more susceptible.
The drastic decrease in liver parasite loads in the livers of mice fed the high-cholesterol-high-fat diet led us to investigate diet-induced changes in the liver environment of uninfected mice, the environment in which the parasites were introduced. Histological slides of livers were stained with hematoxylin and eosin (H&E) to determine cellular content and with Oil-Red-O to examine neutral lipids (Fig. 2). Microscopic enumeration of the H&E-stained slides revealed significant leukocytic infiltration, which was composed primarily of polymorphonuclear leukocytes (PMNs). For mice fed the high-fat-high-cholesterol diet, we observed significant increases in both lymphocytes and PMNs ( Fig. 2A).
Oil-Red-O staining indicated that liver parenchymal cells, likely hepatocytes or stellate cells (24), of mice fed the experimental diets were enriched in lipid bodies compared to the livers of mice fed the control diet (Fig. 2B). The lipid bodies were substantially larger in mice fed the experimental diet than in mice fed the control diet, and the largest lipid bodies were observed in mice fed a high-fat diet. This suggested that there might be substantial changes in hepatocytes contributing to the altered liver environment prior to infection with Li.
A high-fat-high-cholesterol diet caused the most extreme transcript changes. Total liver transcriptomes were sequenced to observe potential mechanisms underlying the shift in parasite growth kinetics in animals fed the experimental diets. RNA was extracted from the total liver homogenates of mice after 8 weeks of Li infection (12 weeks of the diets), as outlined in Table 1.  Both macroscopic appearance and histological examinations indicated that the host microenvironment into which parasites were introduced was very different in mice fed the control versus experimental diets. This impression was verified by an overall assessment of transcriptomes of uninfected mice fed experimental diets. The mean F ratio when comparing the high-fat, high-cholesterol, or high-fat-high-cholesterol diet to the control diet was 3.34, 4.44, or 6.65, respectively. The mean F ratios of .1 led us to conclude that the experimental diets significantly contributed to the variance of the samples. Compared to uninfected mice fed the control diet, uninfected mice fed the highfat, high-cholesterol, or high-fat-high-cholesterol diet had 1,812, 2,476, or 3,192 significantly changed (unadjusted P value of ,0.05) transcripts, respectively. Using a false discovery rate (FDR) adjustment, there were 0, 2, or 1,274 significantly (q value of ,0.10) changed transcripts in the livers of mice fed the high-fat, high-cholesterol, or high-fat-high-cholesterol diet, respectively (Fig. 3). Consistent with the magnitude and severity of differences in parasite load and histology, the greatest numbers of transcript changes were observed in the livers of mice fed the high-fat-high-cholesterol diet. The two transcripts that were significantly changed in uninfected mice fed the high-cholesterol diet were 2410002O22Rik (q = 0.0234413, P = 2.36E206, and fold change = 0.402691) and Slc45a3 (q = 0.0234413, P = 3.44E206, and fold change = 0.45553).
Comparisons of transcriptomes of Li-infected livers also revealed F ratios of .1. The F ratio between infected mice fed the control diet and infected mice fed the high-fat, high-cholesterol, or high-fat-high-cholesterol diet was 3.34, 2.98, or 5.79, respectively. This further supports that the experimental diets significantly contributed to the variance in transcript abundance. Compared to infected animals fed a control diet, 1,806, 1,490, and 3,125 transcripts were significantly changed (unadjusted P value of ,0.05) in infected animals fed high-fat, high-cholesterol, and high-fat-high-cholesterol diets, respectively. Again, using the FDR adjustment, 639 transcripts were significantly changed (q value of ,0.10) in infected mice fed the high-fat-high-cholesterol diet compared to infected mice fed the control diet (Fig. 3). Mice fed either the high-fat or high-cholesterol diet had zero significantly changed transcripts (q value of ,0.10). ) were compared to those in uninfected and Li-infected mice fed the control (Con) diet, respectively. The fold change of each transcript was calculated in Partek (n = 3 for each diet). The P and q values were also calculated as a function of one-way ANOVA performed in Partek. All transcripts in the reference genome are depicted via a volcano plot, with significantly increased transcripts in green, significantly decreased transcripts in pink, and transcripts that did not significantly change in gray. A q value of ,0.10 was considered significant (n = 3 mice for each diet).
Pathway analyses further revealed inflammatory changes in the liver. To gain further insight into the mechanisms by which diet might influence the liver parasite burden, pathway analyses were performed using the PANTHER classification system. As pathway analyses are a more robust predictor of functional significance than individual transcripts considered as though they were independent markers, the abovedescribed unadjusted P values were used for further analyses of pathways. Statistical overrepresentation and functional classification tests were performed using the PANTHER Pathways annotation set. The pathway analyses were performed on the 1,812, 2,476, and 3,192 transcripts significantly changed (P value of ,0.05) in uninfected animals fed the high-fat, high-cholesterol, or high-fat-high-cholesterol diets as well as on 1,806, 1,490, and 3,125 transcripts significantly changed (P value of ,0.05) in infected animals fed the high-fat, high-cholesterol, or high-fat-high-cholesterol diet.
Compared to mice fed the control diet, transcripts significantly increased in the livers of both uninfected and Li-infected mice fed the high-fat-high-cholesterol diet were significantly overrepresented in a number of pathways (Tables 2 and 3). Of particular interest are the overrepresented pathways related to immune function, which include T cell activation, B cell activation, the interleukin signaling pathway, and the Toll  a Observed, number of transcripts in the data that represent a given pathway; expected, number of transcripts that are expected to represent a given pathway for a data list of the size of the list entered; PI3, phosphatidylinositol 3. b Pathway also overrepresented in uninfected mice fed the same diet.
receptor signaling pathway (Tables 2 and 3). While many of these pathways did not reach significant overrepresentation in mice fed the high-fat or high-cholesterol diet, the functional classification tests denoted changes in these pathways for these diets (Fig. 4). Changes in the angiogenesis pathway are also of interest (Tables 2 and 3 and Fig. 4), given that angiogenesis has been associated with organomegaly (25) and given that the WHO cites hepatomegaly as a signature symptom of VL (26). While eight of the most significantly overrepresented pathways in the livers of Liinfected mice fed the high-fat-high-cholesterol diet were also significantly overrepresented in uninfected mice fed the same diet, a number of pathways were not shared between infected and uninfected mice fed the high-fat-high-cholesterol diet (Tables 2  to 5). For instance, in Li-infected mice fed the high-fat-high-cholesterol diet, the interferon gamma (IFN-g) signaling pathway was significantly overrepresented (P = 0.00959; FDR = 0.0498) compared to Li-infected mice fed the control diet (Table 3). But this pathway was not overrepresented in uninfected mice fed the same diet. Because the patterns observed in uninfected mice fed the experimental diets were exacerbated in infected mice fed the experimental diets, we can conclude that diet and infection likely synergistically played a role in the phenotypic changes.
Examination of transcripts of interest via qPCR. Several transcripts of interest were examined by qPCR. Uninfected mice fed the control diet served as the mathematical control during the comparative threshold cycle (C T ) method (Fig. 5). Further statistical comparisons were made between infected mice fed the control diet and infected mice fed the experimental diets. We acknowledge that, individually, several of our transcripts of interest may not have significantly changed in the mice. However, in concert with other transcript changes, our transcripts of interest may have impacted functional pathways and the outcome of infection.
The expression of stat1 in infected mice fed either the high-fat or the high-fat-high-cholesterol diet was significantly increased compared to that in infected mice fed the control diet. The expression of Nfkbia was significantly increased in infected mice fed the high-fathigh-cholesterol and showed only a trend toward decreased expression in mice fed the high-cholesterol diet. Consistently, the expression of nfkb1 trended toward increased expression in mice fed the high-cholesterol diet, and there was a lack of change in mice fed the high-fat-high-cholesterol diet. The only other transcript reaching statistical significance among our chosen gene products was il1b. Amplification of this transcript could reflect either the differing cellular compositions of livers induced by the high-fat-high-cholesterol diet, upregulation in endogenous cells of the liver (Kupffer and stellate cells), or both.
Other transcripts of interest did not reach statistical significance, but this does not rule out the possibility that these contributed to the significant upregulation of a functional pathway. These transcripts that did not reach significance, only some of which are shown, included nfkb1, fcgr4, ifi30, myd88, il1a, and ccl7. Fgr3 and ccl3 transcripts did not show a trend toward significance, and the error bars for tlr2 transcript data were so high as to be nonconvincing. A full microarray was not repeated, but a microarray on three independent mice per experimental diet revealed the differential expression of Fcgr4, Stat1, Ifi30, Il1b, and Tlr2 at a statistically significant level, with fold changes 2.98-, 1.01-, 2.54-, 2.53-, and 7.47-fold higher than those of mice fed a control diet.

DISCUSSION
The changing prevalences of obesity and food insecurity in areas where leishmaniasis is endemic (18,19) likely reflect changing dietary habits and access to health care. However, the impacts of diet composition on visceral leishmaniasis are not fully understood. Previous evidence of the impacts of dietary lipids on parasitic disease includes a study of C57BL/6 mice fed a high-lipid diet prior to Plasmodium berghei infection, which led to increased survival and reduced parasitemia (27). Studies of Trypanosoma cruzi have led to mixed results: CD-1 mice fed a high-fat diet prior to or coincident with T. cruzi infection had increased survival rates (28,29). In contrast, CD-1 mice fed that same high-fat diet experienced worsened cardiac disease when infected with a lower dose of the same T. cruzi strain (30), and mice FIG 4 Pathway analyses of increased transcripts. Transcripts found to be significantly (P , 0.05) increased (A) or decreased (B) via one-way ANOVA when mice fed the experimental diets were compared to mice fed a control diet were analyzed using functional classification tests via the PANTHER classification system. The PANTHER pathway identification number (in parentheses) is included with the name of the pathway. For each diet, the percentage of significantly changed transcripts that represent a given pathway was graphed (the key depicts 0% to 1.5%). Pathways for which more than 1.5% of transcripts were representative are in pink. Fewer than 4.5% of transcripts were assigned to any given pathway. Only pathways that the statistical overrepresentation tests found to be significantly overrepresented in at least one of the experimental diets fed to either infected or uninfected mice are included (n = 3 uninfected and infected mice for each experimental diet). HC, high cholesterol; HF, high fat; HF-HC, high fat-high cholesterol.  fed a high-fat diet and infected with a different strain of T. cruzi, VL-10, had increased parasite loads (31). Studies of Leishmania spp. have similarly shown conflicting results. Like with T. cruzi VL-10, C57BL/6 mice fed a high-sugar and -butter diet had increased L. infantum burdens in their livers and spleens after 8 weeks of infection (21), but L. donovani-infected C57BL/6 mice fed a high-fat-high-cholesterol diet had significantly reduced parasite burdens (20). Because the impact of dietary lipids specifically on the outcome of VL remains poorly characterized and contradictory, we examined the hypothesis that dietary lipid intake would impact the severity and outcome of chronic Li infection.
Similar to the above-mentioned studies, the present study showed that increased dietary lipid intake via any of three experimental diets reduced Li liver parasite burdens after 4 weeks of infection compared to infected mice fed a control diet. More specifically, liver parasite burdens were significantly decreased and nearly undetectable in mice fed the high-fathigh-cholesterol diet. Our data also showed significant increases in splenic parasite burdens after 8 weeks of infection in mice fed the high-fat-high-cholesterol diet. These data raised the possibility that tissue-specific responses and kinetics of infection matter in the outcome of VL. Together, these data suggested that there may be an immediate protective effect of diet on Li infection but that chronic infection could be promoted by the high-lipid diet. Histological examination showed an inflammatory response in mice fed the high-fat-highcholesterol diet.
Also of note, other experiments by our group and other investigators show that granulomatous responses are associated with the clearance of parasites in murine leishmaniasis (22). We observed a predominance of polymorphonuclear rather than mononuclear leukocytes in the livers of mice fed the high-fat-high-cholesterol diet, and we suspect that this aberrant cellular response resulting from the high-fat-highcholesterol diet would more likely be associated with parasites failing to establish infection in the liver rather than granuloma-associated clearance of parasites.
To characterize the effect of excess lipid intake specifically during experimental VL, we examined the transcriptomes of liver tissues of mice fed the three different experimental diets. Uninfected mice fed the high-fat-high-cholesterol diet had transcripts that specified the overexpression of genes related to chemokines and their receptors in inflammatory responses. Other pathways or groups of genes whose expression was significantly upregulated in both uninfected and infected mice fed the combined highfat-high-cholesterol diet included Toll receptor signaling pathways, T cell activation, B cell activation, interleukin signaling pathways, and integrin signaling pathways. These results demonstrate that even prior to infection, the livers of mice fed the high-lipid diets had become an inflammatory environment. Also of interest and in need of further investigation, angiogenesis, which often accompanies organomegaly (25), was significantly overrepresented in both uninfected and Liinfected mice fed the high-fat-high-cholesterol diet. Furthermore, our histological sections showed considerable steatoses in all diets, which were likely stellate cell lipid bodies. Consistent with the involvement of stellate cells as well as endothelial and epithelial cell proliferation, platelet-derived growth factor (PDGF) and epidermal growth factor (EGF) receptor signaling pathways were overrepresented by transcripts from the livers of uninfected and Li-infected mice fed the high-fat-high-cholesterol diet. Both pathways are associated with angiogenesis but are not, by any means, the only drivers (32).
We hypothesize that the hepatic inflammatory and proangiogenesis environment might underlie the apparent protected or refractory state of livers observed in Li-infected mice fed the experimental diets, especially in mice fed the high-fat-high-cholesterol diet. A few studies of mice coinfected with microorganism-induced inflammatory stimuli have demonstrated similar phenomena. Infection with the helminthic pathogen Trichinella spiralis or Trichuris muris prior to infection with Leishmania spp. reduced Leishmania parasite burdens in murine livers, possibly due to increased inflammatory mediators such as interleukin 4 (IL-4) and IFNg (33,34). In contrast, prior infection with another helminth, Schistosoma mansoni, led to increased L. donovani burdens in both livers and spleens (35).
As opposed to the above-described coinfections in murine models of VL, infection with Staphylococcus aureus in the skin prior to infection with L. major, a cause of cutaneous leishmaniasis (4), had no significant effect on the parasite burden in the ear but significantly exacerbated the severity of inflammatory skin lesions 4 weeks after coinfection (36). Therefore, we tentatively conclude that the nature of the inflammatory stimulus coinciding with Leishmania infection and the resultant activated immune mediators will likely lead to distinct effects on the course of leishmaniasis, depending on whether the immune responses induced by the costimulus suppresses or synergizes with pathological responses to the Leishmania species.
Utilizing the comparison with diet as the single independent variable, we were able to observe that eight of the most significantly overrepresented pathways in Li-infected mice fed the high-fat-high-cholesterol diet were also significantly overrepresented in uninfected mice fed the same diet. We can therefore conclude that the observed increase in these pathways may have been largely influenced by diet and not solely due to subsequent infection by Li. Very briefly, the sorting of mapk3 into each of the 10 most significantly overrepresented pathways in the infected mice fed the high-fat-high-cholesterol diet is another interesting observation that we were able to make with this comparison.
To summarize, the interplay between excess dietary lipid intake and consequent aberrant inflammatory changes and chronic infections is not well defined. The data presented here show that increased dietary lipid intake leads to inflammatory changes and, thus, changes in the infection kinetics and severity of chronic VL. Prolonged and increased dietary fat and cholesterol in combination provided a localized protective effect against the expansion of parasites causing VL, leading to reduced parasite burdens in the livers. However, parasite loads significantly increased in spleens of mice fed the high-fat-high-cholesterol diet, the site where parasites are maintained long term in the infected host (37,38).
What we cannot discern from the current study is whether the ultimate effects of elevated-lipid diets on chronic infection, in this case, in the spleens of infected mice, might be a prolongation of the chronic carrier state rather than amelioration of disease. The study raises the possibility that temporary changes in dietary lipid intake might enhance the proinflammatory environment acutely during treatment, possibly providing an additive approach to treatment along with antiparasitic agents. Our next challenges are to determine whether dietary changes similarly change parasite loads in human hosts and to discern other conditions under which similar changes caused by dietary intake cause important changes in disease outcomes.

MATERIALS AND METHODS
Experimental groups. Female BALB/c mice were maintained on the following diets: control (standard, 18% protein diet, with 4% fat and ,0.04% cholesterol), high fat (diet TD.88137; 21.2% fat and 0.2% cholesterol), high cholesterol (TD.01383; 5.7% fat and 2% cholesterol), or high fat-high cholesterol (TD.02028; 21.2% fat and 1.3% cholesterol). Mice were started on their respective diets 4 weeks prior to infection. The diets were obtained from Envigo (Indianapolis, IN). The macronutrients of each diet are listed in Table 1.
Histology. Liver tissue was collected after 4 weeks on each diet. The tissues were then sectioned and fixed. Fixed samples were stained with hematoxylin and eosin (H&E) as well as Oil-Red-O. Slides were then examined by light microscopy, and 10 high-power fields were chosen at random in uniform quadrants on each slide. Leukocyte composition was enumerated in all 10 fields and summed for each mouse. The average number of leukocytes was calculated for each diet.
Infection and monitoring of infection dynamics. A strain of Li isolated from a patient with visceral leishmaniasis in Brazil (MHOM/BR/00/1669) was maintained by serial passage in hamsters. Metacyclic promastigotes were isolated by Percoll density gradients as we previously described (39). Mice were infected with 1 Â 10 6 metacyclic, recombinant Li promastigotes expressing firefly luciferase via intravenous injection in the tail 4 weeks after the onset of the experimental diets. An in vivo imaging system (IVIS) was used to monitor the trafficking and expansion of parasite loads. Images were collected with the Xenogen IVIS 200 system. At weekly time points, animals were anesthetized with isoflurane and inoculated with luciferin by intraperitoneal injection 10 min prior to imaging. Parasite localization was assessed using Living Image software (40). The representative images included here were taken at 4 weeks postinfection.
qPCR quantification of parasite burden. At 4 and 8 weeks postinfection, groups of five mice from each diet were euthanized, livers and spleens were homogenized, and DNA was extracted as we previously described (41). Parasite loads in each organ were determined via quantitative PCR (qPCR) using the kDNA7 primers and probe described previously by Weirather et al. (41).
Transcriptomes. After 12 weeks of each diet (8 weeks after infection or PBS injection), mice were euthanized, and RNA was extracted from liver tissues. Each experimental group, uninfected and infected mice fed any of the four diets, included three biological samples, totaling 24 mice. Total RNA was purified using the TRIzol method followed by DNase treatment and Qiagen column purification, and transcriptional profiles were measured via Illumina MouseRef-8 V2.0 microarrays at the Iowa Institute of Human Genetics, Genomics Sequencing Division, University of Iowa.
After quality control, a custom report was made of the raw microarray data using Illumina GenomeStudio V2011.1 (42). The report was then imported into Partek 7.19.1125 (43) using the gene expression workflow. Samples were annotated with attributes, and separate spreadsheets were created for each comparison to be made. Because we were examining more samples than could fit on one Illumina bead chip, principal-component analyses and histograms were assessed to ensure that samples clustered by experimental group rather than by chip. No samples were removed in this assessment.
By individually comparing each experimental diet to the control diet, differentially expressed transcripts were determined using one-way analysis of variance (ANOVA), and the source of variation of transcript expression was examined. q values, FDR-adjusted P values, were then calculated and are depicted as volcano plots. Gene lists were exported and included all transcripts that could be found in the reference library so that further analyses could be completed. For further analyses, transcripts with a P value of 0.05 were considered significantly differentially expressed.
Validation of the differential expression of transcripts of interest was performed using SYBR green qPCR on the samples used for the murine microarray. The comparative C T method was used to calculate a fold change for each transcript (44). More specifically, the fold change between infected mice fed the control diet or the experimental diets and uninfected mice fed the control diet was calculated. gapdh was used as the internal control.
Pathway analyses. Transcripts found to be significantly changed (P , 0.05) by each experimental diet were entered into the PANTHER classification system (45)(46)(47). Both statistical overrepresentation and functional classification tests were performed. For both tests, the PANTHER Pathways annotation set was used. For statistical overrepresentation tests, Mus musculus was selected as the default whole-genome list, and the Fisher test was used with an FDR correction. Data were exported as .txt files to be further analyzed.
Further statistical analyses. GraphPad Prism 8.4.3 (48) was utilized for further statistical analysis of data. When appropriate, one-and two-way ANOVAs were completed, and outliers were identified with ROUT.
Ethical approval. All studies using vertebrate animals for research were first reviewed and approved by institutional review boards at the Iowa City Department of Veterans' Affairs Medical Center (Animal Component of Research Protocol [ACORP] 2090301 and ACORP 1890601) and the University of Iowa (protocol ACURF 0081099).

ACKNOWLEDGMENTS
These studies were funded in part by grants BX001983 and BX000536 from the Department of Veterans' Affairs Research and Development (M.E.W.). The work was conducted while E.T.K. was funded by grant T32 AI007511 from the National Institutes of Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
We are grateful to Noah Butler and Lilliana Radoshevich for their helpful reviews of the manuscript.