Plasmodium Infection Induces Dyslipidemia and a Hepatic Lipogenic State in the Host through the Inhibition of the AMPK-ACC Pathway

Malaria is a major parasitic disease of humans and is a health public problem that affects more than 100 countries. In 2017, it caused nearly half a million deaths out of 219 million infections. Malaria is caused by the protozoan parasites of the genus Plasmodium and is transmitted by female mosquitoes of the genus Anopheles. Once in the bloodstream, Plasmodium merozoites invade erythrocytes and proliferate until the cells lyses and release new parasites that invade other erythrocytes. Remarkably, they can manipulate the vertebrate host’s lipid metabolism pathways, since they cannot synthesize lipid classes that are essential for their development and replication. In this study, we show that mice infected with Plasmodium chabaudi present a completely different plasma profile from control mice, with marked hyperproteinemia, hypertriglyceridemia, hypoglycemia, and hypocholesterolemia. In addition, white adipose and hepatic tissue and analyses from infected animals revealed the accumulation of triacylglycerol in both tissues and free fatty acids and free cholesterol in the liver. Hepatic mRNA and protein expression of key enzymes and transcription factors involved in lipid metabolism were also altered by P. chabaudi infection, leading to a lipogenic state. The enzyme 5′ AMP-activated protein kinase (AMPK), a master regulator of cell energetic metabolism, was also modulated by the parasite, which reduced AMPK phosphorylation levels upon infection. Pretreatment with metformin for 21 days followed by infection with P. chabaudi was effective in preventing infection of mice and also lowered the hepatic accumulation of lipids while activating AMPK. Together, these results provide new and important information on the specific molecular mechanisms induced by the malaria parasite to regulate hepatic lipid metabolism in order to facilitate its development, proliferation, and lifespan in its vertebrate host.


P. chabaudi infection induces plasma dyslipidemia and altered profile of cytokines in swiss mice.
To evaluate the plasma profiles of infected mice, plasma samples were analyzed for total cholesterol, triacylglycerol (TAG), glucose, total protein, tumor necrosis factor alpha (TNFα), interleukin (IL)-1β, and IL-10 concentrations (Fig. 1). Total cholesterol and glucose levels decreased significantly in infected mice compared to controls (p < 0.05 and p < 0.01, respectively; Fig. 1A,C), whereas TAG and total protein levels increased in infected animals (p < 0.001 and p < 0.01, respectively; Fig. 1B,D). TNFα, IL-1β, and IL-10 concentrations increased significantly in infected mice compared to control group (p < 0.0001, p < 0.0001 and p < 0.0001, respectively; Fig. 1E-G). Together, these results revealed marked changes in plasma lipids, protein, and cytokines in infected versus control mice, suggesting dyslipidemia and inflammation response induced by infection. Values are presented as mean ± standard error. * p < 0.05, ** p < 0.01, and *** p < 0.001. P. chabaudi promotes changes in lipid classes in the liver and white adipose tissue (WAt) of infected mice. Analysis of hepatic and WAT lipid profiles was performed to investigate the possible alterations in lipid classes during Plasmodium infection. As shown in Fig. 2A,B, hepatic TAG (p < 0.05), free cholesterol (p < 0.05), free fatty acids (p < 0.05), and sphingomyelin (SM; p < 0.05) levels increased in infected animals relative to controls, whereas the WAT of infected animals exhibited elevated TAG (p < 0.05; Fig. 2C) and decreased phosphatidylethanolamine (PE; p < 0.05; Fig. 2D), with no significant differences found in the other lipid classes. Enzymatic colorimetric assays were used to confirm the hepatic and WAT changes in the TAG concentration and validate the increase of this lipid class in both tissues in the infected versus control groups (inset in Fig. 2A,C). Together, these data indicate that P. chabaudi infection induces an increase in lipid accumulation in tissues that are critical to lipid metabolism in mice.

P. chabaudi infection leads to an altered hepatic expression of genes associated with lipid metabolism.
After demonstrating that Plasmodium infection in mice induces a wide spectrum of lipid changes, including alterations in plasma lipids and hepatic and WAT lipid classes, we next hypothesized that P. chabaudi specifically impairs the gene expression of key enzymes and transcription factors involved in lipid degradation or biosynthesis (Table 1). In this respect, the transcription factors peroxisome proliferator-activated receptor-α (PPAR-α) and PPAR gamma coactivator-1α (PGC-1α) and the β-oxidation enzyme CPT1-A were chosen for investigation as genes involved in lipid degradation pathways; all decreased significantly in infected mice (p < 0.001; Fig. 3A-C). The transcription factor PPAR-γ and the enzymes FAS and ACC were chosen to assess the mediators of lipid biosynthesis pathways; PPAR-γ and FAS levels increased in infected mice (p < 0.05 and p < 0.01; Fig. 3D,E), with no significant difference in the ACC gene expression (Fig. 3F). Together, these findings demonstrate that malaria induces significant changes in hepatic gene expression, whose combination points to an increase in lipogenic and a decrease in lipolytic pathways. Statistical analyses: multiple t-test. Values are presented as mean ± standard error. * p < 0.05 and ** p < 0.01.

P. chabaudi infection modifies the phosphorylation levels and protein expression of hepatic enzymes and transcription factors involved in lipid metabolism.
AMPK is a master regulator of cellular energetic metabolism 20 . When phosphorylated, this kinase is activated and also uses phosphorylation to downregulate anabolic and stimulate catabolic pathways in order to restore cellular energy balance 20 . Thus, we hypothesized that hepatic phosphorylation levels of AMPK and its downstream targets would be altered by P. chabaudi infection in these animals. We then demonstrated that AMPK phosphorylation levels decrease significantly upon infection (p < 0.05; Fig. 4A and Supplementary S1A), which is accompanied by decreased phosphorylation of ACC, the transcription factor sterol regulatory element binding protein-1c (SREBP-1c), and PPARα (p < 0.001; Fig. 4B,C,F, respectively and Supplementary S1B,C,F). Also, there was an increased protein expression of FAS and PPARγ (p < 0.05; Fig. 4D,E; Supplementary S1D,E). pretreatment with metformin reduces P. chabaudi infection in mice and protects the liver against steatosis. In order to validate the critical role of AMPK-mediated lipid metabolism alteration induced  by infection, mice were pretreated with the AMPK activator metformin (1,1-dimethyl biguanide) prior to P. chabaudi inoculation (see Materials and Methods). Metformin pretreatment completely reversed the decrease in hepatic AMPK phosphorylation levels observed following P. chabaudi infection (p < 0.001; Fig. 5A and Supplementary S2A). Similarly, metformin pretreatment abolished the decrease and even increased the hepatic phosphorylation of ACC (a downstream target of AMPK) in the infected metformin group compared to the infected (p < 0.001) and control (p < 0.001; Fig. 5B and Supplementary S2B) groups. Remarkably, while infected mice presented a gradual and marked increase in parasitemia percentage, reaching 80% on the 5th day after infection (Fig. 6), pretreatment with metformin resulted in a dramatic reduction of the parasitemia percentage from the 3rd to the 5th day after infection (p < 0.001; Fig. 6). This significant drop in the infection rate was also accompanied by reduced clinical signs of infection, such as lethargy, piloerection, hypothermia, starvation, and anemia were all abrogated in the pretreated infected animals compared to the untreated infected mice. In addition, pretreatment with metformin induced a decrease in the percentage of all hepatic lipid classes both in infected and in noninfected mice (p < 0.001; Fig. 7), except for total phospholipids, which were higher than in the nontreated groups (p < 0.001; Fig. 7). Pretreatment with metformin caused a significantly decrease on TNFα, IL-1β and IL-10 both in plasma and liver of infected mice compared to control group ( Fig. 8A-F).

Discussion
In this study, we demonstrated that P. chabaudi infection induces severe changes in the vertebrate host's lipid metabolism that manifest as plasma, hepatic, and WAT metabolic modifications. These changes are mediated by AMPK downregulation and the subsequent activation of pathways that lead to increased lipid biosynthesis and accumulation. Pretreatment with metformin reversed hepatic changes induced by infection, suggesting the establishment of a protective metabolic state that inhibited pathogen proliferation and consequently dissemination in the mammal host.
The balance between host pro-and anti-inflammatory immune responses plays a critical role in determining the outcome of Plasmodium infection. A weak pro-inflammatory response may result in unregulated replication of parasites, while an active pro-inflammatory response may cause tissue damage, such as occurs in severe malaria syndromes, including cerebral malaria and multi-organ failure. Here, we showed an increased TNFα, IL-1β, and IL-10 in plasma and liver of infected mice compared to control group ( Fig. 1). High levels of plasma TNFα have been associated with anemia and high density P. falciparum infection 21 , as well as with other severe malaria complications such as renal failure 22 . Data from mouse models of malaria indicates that IL-10 is required to protect host tissue from inflammation, but may also promote growth of parasites and associated disease manifestations.  www.nature.com/scientificreports www.nature.com/scientificreports/ Thus, IL-10 appears to play a critical role in regulating the effects of TNF during malaria, but at the same time, IL-10 may promote high-density infections that result in other complications, including accumulation of infected red blood cells in tissue that can cause hypoxia and direct damage to the vasculature 23,24 .
P. chabaudi promoted hypocholesterolemia, hypertriglyceridemia, hypoglycemia, hyperproteinemia, and altered profile of cytokines, which clearly indicated plasma lipid dysregulation and increased proinflammatory response that is consistent with previous studies [25][26][27][28][29] . However, we also discovered that these plasma changes are accompanied by alterations in the lipid profile of WAT and with increased percentages of TAG and SM; to our knowledge, this is the first time that these effects have been reported in the literature. Considering that infected erythrocytes adhere to capillary endothelial cells in multiple tissues (including WAT 30 ), it is possible that the parasite acts indirectly on the adipocytes from the endothelium and stimulates lipid accumulation to benefit from these lipids for its development and multiplication in the host. Further studies are needed to better understand the molecular pathways involved in this regulation and to address its relevance to host-parasite interactions.
Our study was focused on the liver because this tissue possess the complete set of enzymatic machinery necessary to synthesize lipids, since it is the main organ for lipid exchange in the body 31,32 . Liver from infected mice exhibited increased percentages of TAG, free fatty acids, and free cholesterol, suggestive of hepatic dyslipidemia, and our quantitative polymerase chain reaction (qPCR) results corroborated this by revealing a decreased hepatic mRNA expression of PPAR-α, PGC-1α, and CPT1-A during infection. PPAR-α is the most abundant transcription factor in healthy livers. It is responsible for activating fatty acid uptake and oxidation, and PPAR-α hepatocyte-specific knockout impairs free fatty acid homeostasis and affects the entire body in fed or fasting mice 33 . PGC-1α is a transcriptional coactivator induced by fasting, and it activates fatty acid oxidation and gluconeogenesis via PPAR-α, FoxO1, and HNF-4α stimulation [34][35][36] . CPT1-A is a key enzyme in the β-oxidation www.nature.com/scientificreports www.nature.com/scientificreports/ process and is regulated not only by malonyl-CoA, but also by thyroid hormone and insulin 37,38 . Together, our results all point to a lower fatty acid oxidation state in infected compared to healthy livers.
In contrast, we observed an increased hepatic mRNA and protein expression of PPAR-γ and FAS in infected compared to control mice. FAS is key enzyme in the process of fatty acid synthesis, and its knockout in mice is lethal, indicating a crucial role for de novo lipogenesis in embryonic development 39 . PPAR-γ is a transcriptional regulator of hepatic fatty acid storage, and its liver-specific knockout in obese mice reduces lipid accumulation 40 . Thus, the increase in PPAR-γ and FAS gene expression implies increased lipid biosynthesis during infection.
The enzyme AMPK is a central regulator of the biosynthesis and degradation of lipids, carbohydrates, and proteins 41 . Here we showed that hepatic AMPK is strongly regulated during Plasmodium infection, which decreases AMPK phosphorylation and leads to the activation of an anabolic state (Figs 4A and S1A). ACC, SREBP-1c, and PPARα are also essential in controlling lipid synthesis/degradation pathways, and they are targets for AMPK phosphorylation 42,43 . Plasmodium infection decreased ACC, and SREBP-1c phosphorylation (i.e., induced their activation) (Figs 4B,C and S1B,C) and decreased PPARα phosphorylation (i.e., decrease their activation) (Figs 4E and S1E). In addition, infection increased the protein expression of FAS and PPARγ (Figs 4D,F and S1D,F). FAS is stimulated by SREBP-1c 44 , resulting in favored lipogenesis over lipolysis during infection.
Metformin is a classic AMPK activator and is the most widely prescribed drug for treating hyperglycemia in patients with type 2 diabetes mellitus (T2DM) 45 . It has also been used for prevention in prediabetic populations 46,47 , for avoiding vascular complications associated with T2DM 48 , and for reducing inflammation and hepatic steatosis 49 . Malaria has been associated with T2DM previously 50 . A case-control study conducted in Ghana reported that patients with T2DM had a 46% increased risk of infection with P. falciparum 51 due to impaired defense against liver and blood-stage parasites, decreased T-cell mediated immunity, and increased . Statistical analyses: one-way ANOVA followed by Dunnett's multiple comparisons test. Values are presented as mean ± standard error. ** p < 0.01 and *** p < 0.001. glucose availability for the parasites 52 . Recent work has shown that maintaining glycemia is essential for tolerance to murine malaria, since severe hypoglycemia contributes to increased brain inflammation and increased lethality of infected mice 53,54 . The use of metformin by T2DM patients significantly reduces the incidence of malaria infection compared to those not on metformin treatment 55 . In addition, metformin has been shown to reduce P. falciparum prevalence and to exhibit synergistic effects with antimalarial drugs, including atovaquone 51,56 , and the combination of artesunate/esomeprazole -which was even more effective in reducing glucose levels and attenuating metabolic syndrome symptoms, promoting effective parasite clearance and preventing its recrudescence than the metformin alone 57 .
In this study, we showed that pretreatment with metformin protected P. chabaudi infected mice from developing clinically significant malaria. The role of AMPK in infection severity has been previously studied in viral infections 58 and malaria 59 ; however, the latter study was performed during the malaria hepatic cycle. Our study consists of a distinct and novel experimental design, which was used to address the Plasmodium intraerythrocytic cycle, where the parasite no longer infects hepatocytes and malarial signs and symptoms are more intense. Thus, we have demonstrated here that malarial disease may be preventable by pretreatment with metformin.
The mechanisms by which metformin pretreatment inhibits Plasmodium infection still need to be determined. We suggest, however, that the lipogenic state induced by AMPK downregulation favors Plasmodium multiplication and dissemination within the mammal host, while metformin-induced AMPK activation limits the lipid supply to the parasite, thereby impairing infection. Pathogens extensively utilize their hosts' cellular lipidome pathways, and over millions of years of parasitism they have evolved very sophisticated mechanisms to lengthen their lifespans in their hosts. The acquisition of host lipids is critical in modulating host-pathogen interactions and in improving the pathogens' virulence 60 . Lipids are an energy source for the pathogens, which cannot synthesize fatty acids, and so they rely on the acquisition of these molecules from the host and subsequent induction of β-oxidation 61 . In the case of Plasmodium, acquisition of host lipids is fundamental and particularly important for the intraerythrocytic stage, when this parasite loses the ability to synthesize fatty acids and is only capable of storing lipids within its parasitophorous vacuole 62 . Furthermore, host lipids provide building blocks for pathogen assembly 63-65 and a convenient carbon source, as has been described for Mycobacterium tuberculosis 66,67 , Salmonella 68 , Candida albicans 69 , and Cryptococcus neoformans 70 .
Pretreatment with metformin significantly decreased the concentration of TNFα, IL-1β, and IL-10 in infected compared to noninfected mice (Fig. 8). Metformin has been shown to attenuate proinflammatory response in several diseases [71][72][73] . Therefore, metformin-induced AMPK activation suppresses the uncontrolled inflammation triggered by the presence of the parasite in the blood, protecting the liver from injury and consequent lipid accumulation, which will culminate with inhibition of the development and multiplication of Plasmodium in the blood. Figure 9 shows a working model of the molecular mechanisms involved in Plasmodium infection, indicating a possible route that elicits the accumulation of hepatic lipids during infection in mice. Reduced AMPK activation leads to elevated ACC activation, which stimulates lipid synthesis, and to reduced lipid oxidation due to the inhibition of CPT1-A, a β-oxidation rate-limiting enzyme. Lower AMPK also leads to increased SREBP-1c activation, which stimulates the transcription of genes involved in lipid synthesis, such as FAS. A complementary axis supporting this pathway converges from the increased protein expression of PPARγ, a transcription factor involved on lipid synthesis and decreased phosphorylation levels of PPARα and mRNA expression of PGC-1α, both fatty acid oxidation inducers 74 . Metformin activates hepatic AMPK and inhibits ACC activity, possibly reversing all the signaling pathways that contributed to lipid accumulation in the liver during Plasmodium infection in mice. In summary, the data we have presented suggest that AMPK represents an attractive target for new drugs for the prevention and treatment of malaria.

Materials and Methods
Parasites and mice. Parasites  and control (CTR, n = 14) groups. Infective inocula (1 × 10 6 RBCs containing P. chabaudi) were then prepared and intraperitoneally injected into the INF group. CTR mice received similar inocula but lacking the parasite (1 × 10 6 uninfected RBCs). After injection, parasitemia was monitored using daily blood smears fixed and stained with the Fast Panoptic method (Laborclin ® ). When parasitemia of INF animals reached 80% (five days after infection), all mice were euthanized in a CO 2 chamber and blood was collected into vials containing 3% sodium citrate as an anticoagulant agent after cardiac puncture. Livers and retroperitoneal WAT were collected and stored in a liquid nitrogen tank. Blood samples were centrifuged at 500 g for 15 min for plasma collection and then stored at −20 °C until needed for analysis. Lipid extraction and analysis. Lipid extraction was performed using a methodology modified from Bligh and Dyer (1959) 77 . Liver and WAT samples (10 mg) were added to a 500 µL of lysis buffer (50 mM Tris-HCl, pH 7.4, 1% NP-40, 250 mM NaCl, 5 mM EDTA, and 50 mM NaF), homogenized, and centrifuged at 2,500 g for 5 min. Supernatants were added to glass conical tubes along with a 4 mL of chloroform-methanol-water solution (2: 1: 0.8 mL) and shaken intermittently for 2 h. Extracts were then centrifuged at 730 g for 20 min at 4 °C in a High-Speed Refrigerated Centrifuge (Hitachi, Ltd., TO, Japan). The supernatant was collected, and 2 mL of water and chloroform (1: 1 mL) were added, and the solution was centrifuged at 730 g for 20 min at 4 °C prior to collection of the organic phase (containing lipids). The extracted lipids were analyzed by thin-layer chromatography (TLC) using a hexane/diethyl ether/acetic acid solvent (60: 40: 1 v/v) 78 . To visualize the lipids, plates were immersed in a carbonization solution consisting of 8% CuSO 4 and 10% H 3 PO 4 for 10 s and heated at 110 °C for 20 min 79 . Plates were analyzed by densitometry using ImageMaster TotalLab software (TotalLab, Newcastle, UK).

Experimental design 2. Male
Colorimetric enzyme assays. Plasma TAG, total cholesterol, and glucose concentrations were determined using commercial enzyme kits (Doles ® , Goiânia, Brazil) and by following the manufacturers' instructions.
Protein concentration determination. Plasma protein concentration was determined by the Lowry method 80 , using bovine serum albumin (0.1%) as a standard.
RNA extraction and cDNA synthesis. Liver samples (10 mg) were homogenized in 500 μL TRIzol reagent (Invitrogen Corporation, Carlsbad, CA, USA) and centrifuged at 7,500 g for 15 min at 4 °C. The supernatant was transferred to microtubes containing 200 μL of chloroform (Merck, Darmstadt, Germany). Samples were shaken for 10 s and centrifuged again at 7,500 g for 15 min at 4 °C. Supernatants were collected and transferred to new microtubes containing 200 μL of isopropanol (Merck). Samples were mixed gently by inversion and incubated at room temperature for 10 min and then centrifuged at 7,500 g for 15 min at 4 °C. The supernatants were discarded, and the precipitate was resuspended in 200 μL of absolute ethanol. Samples were then centrifuged at 3,000 g at 4 °C for 15 min, the supernatants were discarded, and the pellets were resuspended in 30 μL of RNase-free water. The RNA concentration was quantified by spectrophotometry using a NanoDrop (Thermo Scientific, Wilmington, NC, USA). Cytokines measurements. TNFα, IL-1β, and IL-10 cytokines analysis was performed on plasma and liver samples. The protocol used followed the manufacturer's specifications (R&D, Systems Corporation, Mineapolis, MN). Briefly, 96-well microplates were sensitized with anti-TNFα, IL-1β, and IL-10 monoclonal antibodies, diluted in PBS and incubated for 24 hours at room temperature. The plates were blocked with PBS/4% bovine albumin serum (SAB-Sigma Chemicals, St. Louis, MO, USA) and incubated for 2 hours at room temperature. After following three wash cycles with 0.05% PBS-Tween-20 buffer, samples and standard curve serial dilutions were added to the plates and incubated for 1 hour at room temperature. Thereafter, biotin-conjugated antibodies were added to the plates and incubated for 1 hour. Following, streptavidin peroxidase was added for 20 minutes. After further 3 wash cycles, the 3′, 3′, 5, 5′ tetramethylbenzidine susbtrate (TmB, Zymed) was added to the plates and the reaction was stopped by addition of sulfuric acid 1 M. The plates were read using an ELISA microplate reader at 450 nm. Statistical analysis. Statistical analyses were performed using GraphPad Prism software version 6.0 (GraphPad Inc., CA, USA) and included unpaired Student's t-test, one-way ANOVA followed by Dunnett's multiple comparisons test, and two-way ANOVA followed by Tukey's multiple comparisons test. Specific statistical analyses are described in the figures' legends. Results are expressed as mean ± standard error, and differences were considered significant when p < 0.05.

Data Avaliability
All data generated or analysed during this study are included in this published article (and its Supplementary  Information Files).