Lipid Body Organelles within the Parasite Trypanosoma cruzi: A Role for Intracellular Arachidonic Acid Metabolism

Most eukaryotic cells contain varying amounts of cytosolic lipidic inclusions termed lipid bodies (LBs) or lipid droplets (LDs). In mammalian cells, such as macrophages, these lipid-rich organelles are formed in response to host-pathogen interaction during infectious diseases and are sites for biosynthesis of arachidonic acid (AA)-derived inflammatory mediators (eicosanoids). Less clear are the functions of LBs in pathogenic lower eukaryotes. In this study, we demonstrated that LBs, visualized by light microscopy with different probes and transmission electron microscopy (TEM), are produced in trypomastigote forms of the parasite Trypanosoma cruzi, the causal agent of Chagas’ disease, after both host interaction and exogenous AA stimulation. Quantitative TEM revealed that LBs from amastigotes, the intracellular forms of the parasite, growing in vivo have increased size and electron-density compared to LBs from amastigotes living in vitro. AA-stimulated trypomastigotes released high amounts of prostaglandin E2 (PGE2) and showed PGE2 synthase expression. Raman spectroscopy demonstrated increased unsaturated lipid content and AA incorporation in stimulated parasites. Moreover, both Raman and MALDI mass spectroscopy revealed increased AA content in LBs purified from AA-stimulated parasites compared to LBs from unstimulated group. By using a specific technique for eicosanoid detection, we immunolocalized PGE2 within LBs from AA-stimulated trypomastigotes. Altogether, our findings demonstrate that LBs from the parasite Trypanosoma cruzi are not just lipid storage inclusions but dynamic organelles, able to respond to host interaction and inflammatory events and involved in the AA metabolism. Acting as sources of PGE2, a potent immunomodulatory lipid mediator that inhibits many aspects of innate and adaptive immunity, newly-formed parasite LBs may be implicated with the pathogen survival in its host.


Introduction
The dynamic nature of lipid bodies (LBs), also known as lipid droplets has lead to their recognition as highly active organelles within most cell types involved in different biological functions and containing not only lipids but also many proteins (reviewed in [1][2][3]). During the last decades, the structure and function of these lipid-rich organelles, which are surrounded by a single layer of phospholipids, have been consistently investigated in mammalian cells and attracted great attention due to their association with human diseases (reviewed in [1][2][3][4]).
LB biogenesis is a well-documented process that happens in vivo within many types of mammalian cells during inflammatory reactions of varied causes, including infectious diseases with different pathogens such as bacteria, parasites and virus. Host-pathogen interaction leads to increased formation of LBs within cells from the immune system mainly macrophages. In these cells, LBs serve as intracellular sites for metabolic transformation of arachidonic acid (AA) into biologically active inflammatory mediators (eicosanoid derivatives) (reviewed in [5][6][7]). Thus, LBs in mammalian cells are remarkably linked to inflammatory responses and are considered structural markers of inflammation [5,6,8].
In recent years, there has been a renaissance of interest and recognition in the structure, composition and function of lipid-rich organelles formed within pathogenic prokaryotes and lower eukaryotes. In the past, pathogen-derived LBs were mostly considered as lipid deposits with slow rates of turnover (reviewed in [1]). However, evidence begins to accumulate that cytosolic LBs of intracellular bacteria and parasitic protozoa have more dynamic roles. These pathogens are able to usurp host lipids or to encode their own lipid biosynthesis machinery, thus allowing formation of LBs independently of their host (reviewed in [9,10]). For example, the infection of erythrocytes with the malaria parasite Plasmodium falciparum induces LBs formation within the parasite [11]. These newly synthesized LBs accumulate in the food parasite vacuole and are involved in the detoxification of heme, enabling parasite persistence [11]. During the infection with Toxoplasma gondii [12] or Mycobacterium tuberculosis [13], host-derived lipids are imported and used for the synthesis of cholesteryl esters or triglycerides that are deposited in pathogen LBs. More recently, it was demonstrated that LBs numbers as well as the expression of prostaglandin F 2 alpha (PGF 2 α) synthase (PGFS) increases during the development of Leishmania infantum chagasi to a virulent metacyclic stage [14].
Here, we studied the formation of LBs within the parasite Trypanosoma cruzi and the ability of parasite-derived LBs to produce eicosanoid in response to exogenous AA. T. cruzi is an obligate intracellular parasite capable of infecting different types of nucleated cells of humans and warm-blooded animals, and responsible for Chagas' disease, which remains a major problem with a great impact on public health in Latin America [15].
We demonstrate, for the first time, that T. cruzi LBs are formed and respond to both host interaction and AA-stimulation, are able to incorporate AA and can be sources of prostaglandin E 2 (PGE 2 ), a potent immunomodulatory lipid mediator known to inhibit many aspects of innate and adaptive immunity [16][17][18]. Our results raise the possibility that pathways of AA metabolism of potential pathophysiologic significance may exist within human-living T. cruzi.

Animals and infection
and free (non-interiorized) parasites in the medium were citospun onto slides. At the end of the exposure period (24h), non-interiorized parasites were removed by repeated washings. For in vivo infection, female Holtzman rats aged 27-30 days (obtained from Universidade Federal de Minas Gerais animal facility) were inoculated intraperitoneally with of 3 X 10 5 of T. cruzi Y strain as described elsewhere [20]. Fresh blood samples taken from the tail showed living trypomastigotes in all animals at 12 days after inoculation. At this time, animals were euthanized in a CO 2 chamber and fragments of the heart (atria) were processed for both histopathological and ultrastructural studies as described below.

Ethics statement
This study was carried out in full accordance with all international and Brazilian accepted guidelines and was approved by the Oswaldo Cruz Foundation Ethics Committee on Animal Use (CEUA-Comissão de Ética no Uso de Animais, under protocol CEUA: P-0069). CEUA follows the Brazilian national guidelines recommended by CONCEA (Conselho Nacional de Controle em Experimentação Animal). Animals were monitored daily for survival and wellbeing status (home cage evaluation, body condition, skin lesions, mobility and general conditions such as diarrhea). No animals died prior to the experimental endpoint (12 days of infection).

Culture of trypomastigotes
T. cruzi Dm 28c strain was obtained and kept in the laboratory as previously described [21]. T. cruzi cultures was grown in Brain Heart Infusion medium (BHI) (BD, Franklin Lakes, NJ, USA) at 27°C in a B. O. D. (Biochemical Oxygen Demand) incubator (Thermo Scientific, Walthan, MA, USA) for 7 days, containing approximately 100% of epimastigotes. At the end of the exponential phase, parasites were centrifuged at 2200 rpm for 15 min at 10°C, resuspended in artificial triatomine urine (TAU) (190 mM NaCI, 8 mM phosphate buffer pH 6.0, 17 mM KCl, 2 mM CaCI 2 , 2 mM MgCI 2 ), and incubated for 2 h at room temperature (RT). The parasites were diluted to a final concentration of 5 x 10 6 parasites ml -1 in TAU supplemented with 2.5% (v/v) sodium bicarbonate 1.4%, 500 units penicillin mL -1 , 10 mM L-proline (TAU-P medium) and incubated at 27°C in tightly closed culture flasks in a B. O. D. incubator for 5 days, obtaining approximately 100% of metacyclic trypomastigotes [22].

Cells and tissue preparation for TEM
Samples from cultured trypomastigotes alone were immediately fixed in a mixture of phosphate buffer 1%, pH 7.3 and freshly prepared aldehydes (1% paraformaldehyde and 1% glutaraldehyde) [23] for 1 h, at room temperature (RT), washed in the same buffer and centrifuged at 1500 g for 1 min. Samples were then re-suspended in molten 2% agar in phosphate buffer 1% and quickly re-centrifuged. To obtain optimal morphology, T. cruzi-infected macrophages directly on the slide surface were fixed as above, after 24 h of infection. Atria fragments collected from infected animals were also fixed in the same fixative for 4 h. After fixation, all samples were kept in phosphate buffer at 4°C for subsequent EM procedures.

Histological analyses
Fragments of the heart (atria) from controls and infected animals were fixed in 4% paraformaldehyde in buffered phosphate, pH 7.3, 0.1 M for 24 h, dehydrated and embedded in plastic resin-glycol metacrilate-(Leica, Heidelberg, Germany) as previous work [24]. Semi-serial 5μm-thick sections were cut on a microtome (RM 2155; Leica) stained by hematoxylin and eosin and examined for evaluation of the inflammatory processes and parasitism.

TEM
Cultured trypomastigotes alone, T. cruzi-infected macrophages and heart fragments from infected animals were fixed in a mixture of freshly prepared aldehydes (1% paraformaldehyde and 1.25% glutaraldehyde) in 0.1 M sodium cacodylate buffer for 1 h (isolated cells) or 4h (tissue) at RT and processed as before [25]. Samples were post-fixed in 1% osmium tetroxide in Sym-Collidine buffer (pH 7.4) for 2 h at RT. After washing with sodium maleate buffer (pH 5.2), pellets were stained en bloc in 2% uranyl acetate in 0.05 M sodium maleate buffer (pH 6.0) for 2 h at RT and washed in the same buffer as above before dehydration in graded ethanol's and infiltration and embedding with a propylene oxide-Epon sequence (Eponate 12 Resin; Ted Pella, Redding, CA, USA). After polymerization at 60°C for 16 h, thin sections were cut using a diamond knife on an ultra-microtome (Leica, Baden-Württemberg, Germany). Sections were mounted on uncoated 200-mesh copper grids (Ted Pella) before staining with lead citrate and viewed with a transmission electron microscope (CM 10; Philips, or Tecnai-G2-20-FEI 2006, Eindhoven, the Netherlands) at 60 kV.

TEM quantitative analyses
In addition to qualitative observations, quantitative study was made in electron micrographs. The area and electron-density of cytoplasmic LBs within parasites were evaluated in thin sections of peritoneal and heart macrophages infected with amastigotes. LBs were classified as Strongly Electron-Dense (SED), Electron-Dense (ED) or Electron-Lucent (EL) according to a grayscale (0-85, 86-170, 171-255, respectively) where 0 is absolute black and 255 absolute white. For LB evaluation, a total of 36 electron micrographs, 50 parasites and 125 LBs were carefully analyzed. Quantitative EM analyses were also performed in electron micrographs randomly taken from metacyclic trypomastigotes to investigate the occurrence of morphological alterations. For this analysis, a total of 50 electron micrographs (25 from control and 25 from AA-stimulated) were used for evaluation of the parasite area and length. All analyses were performed using the ImageJ 1 software (National Institutes of Health, Bethesda, MD, USA).

LB purification
LB purification from control and AA-stimulated (7.5 μM trypomastigotes was performed by modifications of prior methods [27]. The trypomastigotes (1.5 x 10 9 ) in 20 nM/L Tris, 1 mM/L EDTA, 1 mM/L EGTA, 100 mM/L KCl buffer (pH 7.4) containing 10 μg/mL leupeptin, 10 μg/ mL benzamidin, 0.7 μg/mL pepstatin, and 0.1 mM/L phenylmethylsulfonylfluoride were disrupted by nitrogen cavitation at 700 C for 5 min at 4°C. The homogenates were centrifuged at 1800 rpm for 5 min to remove the nuclei. The supernatants were overlaid sequentially with 1.5 mL each of 0.27 M/L sucrose buffer, 0.135 M/L sucrose buffer, TEE solution [25 mM/L Tris-HCl, 1 mM/L EDTA, and 1 mM/L EGTA (pH 7.4)] and centrifuged at 35,000 rpm at 4°C for 70 min. The LBs were collected from the first and second top fractions. After that, a lactate dehydrogenase activity assay was performed to demonstrate that the samples were free of cytoplasmic contaminating contents.

Raman spectroscopy
Trypomastigotes (2 x 10 6 cells per mL) were stimulated for 1h with 7.5 μM AA as above and centrifuged. Cells were then resuspended in 3.7% formaldehyde overnight, pelleted and analyzed by Raman spectroscopy without any labeling. Raman spectra of cells were recorded using a FT-Raman spectrometer model RFS 100S coupled to RamanScopeIII (Bruker Optik GmbH, Ettlingen, Germany), equipped with a ND:YAG laser with an excitation line at 1064 nm. For acquisition of the spectra, the laser power was adjusted to 500 mW (at source) and a good signal/noise ratio was obtained by performing 2048 scans in the region of 3500-50 cm -1 with a spectral resolution of 4 cm -1 . The acquisition of Raman spectra was performed by OPUS 6.0 software (Bruker).
To investigate the presence of AA directly in LBs, purified LBs from untreated and 7.5 μM AA-treated groups were placed over 20 mm CaF 2 windows (cat. number 63207; Edmund Optics, Barrington, NJ, USA) and the data collected with a laser power of 20 mW, 50 s integration time and 5 co-additions, without any labeling. The Raman spectra were obtained in a Senterra Raman spectrometer (Bruker) based in a 180°backscattering configuration and using a 50x objective and the 632.8 nm wavelength of He-Ne laser output as excitation. A spectral resolution of 3-5 cm -1 and slit width of 50x1000 μm were chosen.

MALDI mass spectroscopy
All MALDI spectra were obtained using a time of flight mass spectrometer. Mass spectrometry matrix-assisted laser desorption/ionization-time-of-flight (MALDI-TOF) [28,29] experiments were performed using a pulsed nitrogen laser at 337 nm of a Shimadzu Biotech Axima Performance MALDI-TOF at the Physics Department, Federal University of Juiz de Fora. The matrix was alpha-cyano-4-hydroxycinnamic acid (α-CHCA), dissolved in acetonitrile-HPLC/milli-Q water quality (50:50 v:v) at a concentration of about 5×10 −2 mol/L. Samples of purified LBs were dissolved in phosphate buffer and 10 μL of this final sample solution were added to 10 μL of the matrix solution. This mixture (~0.5 μL) was then deposited in the stainless steel multiprobe and allowed to dry before introduction into the mass spectrometer. A typical starting laser power is 40-50. The instrument was set in the high-resolution in positive reflector ion mode and spectra were taken from 0 to 500 m/z. The experimental setup includes an automatic sample manipulator, where 200 scans were accumulated with 20 repetitions each.

Macrophage culture and BCG in vitro infection
Infected macrophages were used as positive controls for Western Blotting analyses of eicosanoid-forming enzymes as below. For the in vitro infection, peritoneal macrophages from C57BL/6 mice were harvested with sterile RPMI 1640 cell culture medium. Cells (1x10 6 cells/ mL) were allowed to adhere in culture plates (6 wells) for 2 h at 37°C in a 5% CO 2 atmosphere and were vigorously washed twice with PBS to remove nonadherent cells. Macrophages were infected with BCG (Mycobacterium bovis BCG (Moreau strain) vaccine from the Fundação Athaulpho de Paiva, Rio de Janeiro, Brazil), multiplicity of infection (MOI) 1:1 and incubated for 24 h at 37°C in a CO 2 atmosphere with RPMI 1640 cell culture medium containing 2% Fetal Bovine Serum (FBS) as before [30].

Western blotting
Macrophages (1x10 6 cells) and trypomastigotes lysates (3x10 6 cells) were prepared under reducing and denaturing conditions and subjected to SDS-PAGE. Samples were submitted to electrophoresis in 10% acrylamide gradient SDS-PAGE gels. After transferring onto nitrocellulose membranes, nonspecific binding sites were blocked with 5% nonfat milk in TBST (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 0.05% Tween 20). Membranes were probed with the polyclonal antibody (Ab) anti-PGE synthase (Santa Cruz Biotechnology, Dallas, TX, USA, S-16; sc-12268), anti-cyclooxygenase-2 (COX-2) (Santa Cruz Biotechnology, C-20; sc-1745) and antiactin monoclonal Ab (BD Transduction Laboratories, 612657) in TBST with 1% nonfat dry milk. Proteins of interest were then identified by incubating the membrane with HRP-conjugated secondary Abs in TBST, followed by the detection of antigen-Ab complexes by Supersignal Chemiluminescence (GE Healthcare, Fairfield, CT, USA, ECL™ Prime Western Blotting System, RPN2232). A colleague blind to the identity of the sample performed the spotting and the analysis parameters.
Immunodetection of PGE 2 at its sites of production Immunolocalization of PGE 2 at its in vivo sites of production was performed in AA-stimulated trypomastigotes [31]. Briefly, unstimulated (kept in vehicle) and AA-stimulated (7.5 μM) cells were incubated, after 1h of stimulation, with EDAC (N-(3-Dimethylaminopropyl)-N 0 -ethylcarbodiimide hydrochloride) (Sigma-Aldrich), at 37°C. Besides the precise positioned coupling of an immuno-detectable eicosanoid at its sites of formation, EDAC enables: (I) the ending of cell stimulation step; (II) cell fixation; (III) cell permeabilization, allowing the penetration of both anti-eicosanoid and the detecting fluorochrome-conjugated antibodies into cells; and, importantly, (IV) the relative preservation of lipid domains, such as membranes and droplets, which dissipate with air drying or commonly used alcohol fixation [31]. Trypomastigotes were then washed with HBSS, cytospun onto glass slides, and incubated with mouse anti-PGE 2 (1/100) Ab (Cayman Chemical-414013, Ann Arbor, MI, USA) in 0.1% normal goat serum and guinea pig polyclonal anti-mouse perilipin 2/adipose differentiation related protein (PLIN2/ADRP) Ab (Fitzgerald -20R-AP002) (1/ 1000) in 0.1% normal donkey serum simultaneously for 1 h at RT. IgG 1 , kappa monoclonal Ab isotype control, clone MOPC 21 (Sigma Aldrich) and non immune guinea pig serum were used as controls for PGE 2 and PLIN2/ADRP Abs, respectively. Cells were washed twice and incubated with secondary Abs, goat anti-mouse conjugated with AlexaFluor-488 (1/1000) (Molecular Probes), and CY3-conjugated donkey anti-guinea pig (1/1000) (Jackson Immuno Research Laboratories). The slides were then washed (three times, 10 min each) and mounted with Vectashield 1 mounting medium containing DAPI (Vector Laboratories). Cells were analyzed by both phase-contrast and fluorescence microscopy. For colocalization quantitative study, fluorescence images were taken and colocalizations were quantified for 12 cells. Quantifications were determined using ImageJ software by calculating the Pearson's correlation coefficient that measures the degree of linear dependence between the localization of a red signal and the localization of a green signal in the same cell [32].

Statistical analysis
Data were compared using the Student-Neuman-Keuls test and expressed as mean values ± SEM (P<0.05). Analyzes were performed using the software Prism 6.01 (GraphPad software, San Diego, CA, USA).

LBs are formed within the parasite in response to host interaction
LBs are remarkably produced in host cells, mainly in macrophages, in response to T. cruzi infection [33][34][35][36]. We wondered if LBs could also be formed in the parasite cytoplasm in response to the interaction with these inflammatory cells. Metacyclic trypomastigote forms cultured in the presence of peritoneal macrophages for 1 h showed indeed a significant increase of LB numbers compared to trypomastigotes alone (Fig 1A and 1B). This indicates that the contact of T. cruzi infective forms with inflammatory host cells might modulate LB formation in the parasite.
T. cruzi grows and reproduce inside host cells, as amastigotes. Therefore, we next investigated the ultrastructure of LBs in these intracellular forms of the parasite by TEM in both peritoneal (after 24 h of the in vitro infection) and heart inflammatory macrophages (after 12 days of the acute in vivo infection). At these time points of the infection, there is consistent division of amastigotes. Moreover, the parasitism peak in the heart, a target organ of the Chagas' disease, is observed at day 12 of the acute infection in conjunction with an elevated number of infiltrating activated macrophages [24,37]. Histolopathological analysis of the myocardium at this time of the infection confirmed the presence of prominent mononuclear inflammatory processes and amastigote nests (S1 Fig) LBs within amastigotes were observed as round, typical non-membrane bound organelles with varied electron-density (Fig 1C and 1D). Interestingly, LBs in amastigotes growing in vivo within macrophages infiltrated in the myocardium (Fig 1D) showed higher sizes compared to LBs formed within amastigotes living in cultured macrophages (Fig 1C) as shown by morphometric analyses (Fig 1E).
Since LBs formed within host cells such as macrophages change electron-density during inflammatory responses (reviewed in [38]), we next analyzed electron-density aspects of LBs formed within amastigotes. Based on our previous studies [34,38,39] and using a software for analyzing different electron-density gradations [39], LBs were classified as strongly electrondense (SED), electron-dense (ED) or electron-lucent (EL) organelles. Quantitative TEM studies revealed that LBs from amastigotes living in heart macrophages (in vivo infection) were more electron-dense, with elevated numbers of SED and ED LBs compared to LBs from peritoneal macrophages (in vitro infection) (Fig 1F; compare Fig 1C and 1D).
Taken together, these results show that parasite LBs are not inert but dynamic organelles, able to respond to host interaction and inflammatory events.

AA induces rapid LB formation in T. cruzi infective forms
Earlier studies demonstrated that AA is a potent stimulator of LB formation in mammalian cells [40][41] and that these organelles incorporate AA, mainly esterified in phospholipids [42][43][44]. The effect of AA on parasite LB formation was investigated. Four different staining techniques were used for LB visualization: osmium tetroxide, which stains phospholipids and three fluorescent probes (BODIPY, Oil Red O and Nile Red), which are more specific for neutral lipids [5,26]. All methods enabled clear visualization of LBs within the parasite (Fig 2A-2E). Moreover, by enumerating LBs with osmium staining, it was observed a dose-dependent increase in LB numbers in AA-stimulated trypomastigotes compared to unstimulated controls (kept in vehicle) after 1 or 24 h, with a maximum LB formation at the dose of 7.5 μM (Fig 2F,  2G and S2 Fig). Trypomastigote forms of the parasite were also evaluated by TEM (Fig 2H and  2I). LBs were seen as typical, cytoplasmic non-membrane bound organelles (Fig 2I, arrows). AA had no effect on the general morphology of stimulated parasites compared to unstimulated ones as shown by qualitative and quantitative ultrastructural analyses (S3 Fig)

AA is incorporated into LBs
To explore the molecular properties of lipids within the parasites, we next used Raman spectroscopy. This technique provides information about the structure of chemical components present in the biological sample, with the advantages of minimal sample preparation, without need of labeling and free from water interference [45]. The result is showed by the Raman spectrum, where the intensity of the signals is proportional to the relative concentration of a compound [45]. Fig 3 shows Raman spectra of AA (positive control), unstimulated and AAstimulated parasites. The spectrum of stimulated cells differed from the spectrum of unstimulated cells by increased relative intensity of the bands at 3015 and 2929 cm -1 , which are characteristic of AA-lipid spectra [46]. The band exhibited at 3015 cm -1 which corresponds to the = C-H stretching, indicates higher content of unsaturated lipids in stimulated cells (Fig 3). This band is more prominent in AA than in other fatty acids [46]. AA also has a broad and intense band at around 2920 cm -1 region due to C-H 2 stretching vibrations. The observed increase in the relative intensity of the 2929 cm -1 in the stimulated cells may be due to the presence of the 2920 cm -1 broad band of AA. Therefore, the increase in the relative intensity of the bands at 3015 and 2929 cm -1 are indicative of the AA incorporation. These results showed that AA is incorporated into parasites, induces LB formation and that these organelles are likely involved in the AA metabolism.
The Raman spectra obtained from the entire parasite have also contribution from different types of lipids and from other biomolecules, as for example proteins and DNA, which can macrophage (in vivo infection after 12 days) (D). Amastigote LBs are indicated (circles). (E) LB sizes are significantly higher in amastigotes growing in vivo (black bar) compared to amastigotes LBs living in vitro (white bar). (F) LBs from amastigote forms of the parasite within heart macrophages are more electron-dense compared to amastigote LBs from peritoneal macrophages. Cultured peritoneal macrophages and heart macrophages were fixed and processed for TEM [25]. A total of 36 electron micrographs, 50 parasites and 125 LBs were evaluated. LBs were classified as Strongly Electron-Dense (SED) Electron-Dense (ED) or Electron-Lucent (EL) by using the software ImageJ 1 as in materials and methods. Scale bar, 10 μm (B); 5 μm (C, D). * P<0.05.  To overcome this difficulty, we isolated LB using subcellular fractionation from unstimulated and AA-stimulated parasites and used Raman directly on LBs. This technique has been used as a valuable tool to study LBs in physiology and pathology [47]. The results are shown in Fig 4. By using this approach, it was possible to determine that the major contribution to the LB comes from the presence of AA. As shown in Fig 4, the spectrum of AA in solution is similar to the previous published in the literature [47]. It has the more intense bands at 1265 cm -1 (= C-H in plane bending), 1446 cm -1 (C-H bending) and 1659 cm -1 (C = C stretching) [47]. The spectrum of the non-treated LB fraction did not show any spectral characteristic of the AA (Fig 4). However, the spectra of the LB fraction obtained from parasites treated with AA showed the same spectral profile than that of the AA in solution, which demonstrate the presence of the AA in the sample (Fig 4).
DAPI staining in Ai, Bi, Ci and Di. In (E), osmium staining is observed under bright field microscopy. (F, G) Number of LBs within trypomastigotes after stimulation with exogenous AA for 1 h (F) or 24 h (G). Bars represent the mean ± SEM of LB per parasite from 50 consecutively counted parasites from at least 4 independent pools. * P < 0.05 between groups. Cells were enumerated after osmium staining. In (H, I), trypomastigotes are observed by transmission electron microscopy. Typical nonmembrane bound LBs (arrows) are seen in the cytoplasm after stimulation with AA (I). Scale bar, 5 μm (A-E); 1 μm (H, I).
doi:10.1371/journal.pone.0160433.g002 The presence of increased AA in the LB fraction purified from the stimulated group was next confirmed by MALDI-TOF mass spectroscopy (Fig 5). First, a spectrum from a pure AA solution (m/z 304.24) was acquired as a positive control (Fig 5A). Second, this technique identified the m/z 304.24 peak related to the AA in both non-stimulated and AA-stimulated samples. In order to verify the existence of a fluctuation at the m/z 304.24 peak of the non-stimulated and AA-stimulated sample, the yield related to this peak was calculated by dividing the peak area by the total area of the corresponding spectrum, as before [48]. Thus, m/z 304.24 peak from the non-stimulated and AA-stimulated samples showed a related yield of 0.0035 and 0.018 respectively, that is, 5 times higher in the stimulated group (Fig 5B, black line) compared to the value obtained from the non-stimulated sample (Fig 5B, red line).
Therefore, by using different approaches, our results consistently demonstrate that LBs are cytoplasmic sites of AA accumulation in the parasite T. cruzi.

AA-stimulated parasites generate PGE 2
Biologically active eicosanoids such as PGs are derived from AA [8,49]. We next investigated if the parasite would produce PGs upon AA stimulation. Since PGE 2 is consistently synthesized during T. cruzi infections [35,50], we evaluated the production of this PG by trypomastigotes. Incubation of these parasite forms with AA produced high levels of PGE 2 after 24h compared  to controls cultured in the absence of AA (Fig 6A). Incubation of the parasites with OA, which is also able to induce LB formation in trypomastigotes (S4 Fig), did not induce PGE 2 production ( Fig 6B). Thus, we concluded that T. cruzi is capable to generate PGE 2 through the AA metabolism.

Parasites show PGE synthase but not COX-2 expression
We next evaluated by Western blotting whether key eicosanoid-forming enzymes COX-2 and PGE synthase were localized within the parasites. As positive controls, we used cultures of macrophages infected with Mycobacterium bovis BCG, which show both expression for these enzymes and high PGE 2 production in response to the infection [30,51]. Trypomastigotes lysates from both unstimulated and AA-stimulated cells were negative for COX-2 ( Fig 7A). On the other hand, we found a high level of PGE synthase within AA-stimulated parasites but not in unstimulated cells (Fig 7B), indicating an activation of the AA cascade and a possible pathway for PGE 2 synthesis.

Parasite LBs are sites for PGE 2 synthesis
Because parasite stimulation with AA led to quantitative increases in LB numbers, PGE 2 release and PGE synthase expression, we hypothesized that parasite LBs might also serve as domains for compartmentalized PGE 2 synthesis. With the adaptation to a method using carbodiimide to immobilize eicosanoid carboxyl groups to proximate proteins [52], formation of eicosanoids was investigated at its sites of production. PLIN 2/ADRP was used as a marker of LBs for colocalization purpose [31]. With the use of this technique (EicosaCell) [31], PGE 2 formation was demonstrated specifically at LBs from trypomastigotes of T. cruzi stimulated with AA ( Fig  8). AA-stimulated trypomastigotes, exhibited a strong localized punctuated or ring shape staining for PLIN2/ADRP-labeled LBs (compare Fig 8A-8C), indicating formation of LBs in response to the stimulation of AA. AA-stimulated trypomastigotes, but not vehicle-stimulated parasites, showed intense and punctuated immunofluorescent staining for PGE 2 (compare Fig  8Ai and 8Bi). As shown in Fig 8Bii, the PGE 2 intracellular site of production (8Bi) matched PLIN2/ADRP-stained LBs (Fig 8B). Our colocalization quantitative analyses using ImageJ showed that PGE 2 and PLIN2/ADRP significantly colocalized in cells [Pearson's coefficient of 0.84 ± 0.02 (mean ± SEM)]. The specificity of the immunofluorescence for PGE 2 was supported by the absence of immunostaining when an isotype control antibody replaced the anti-PGE 2 monoclonal antibody (Fig 8C-8Cii). Controls in which the anti-PLIN2/ADRP Ab was replaced by guinea pig serum were negative (data not shown). These findings validated the specificity for detecting PGE 2 formed at its formation sites within stimulated trypomastigotes, and place LBs as candidate sites for newly formed PGE 2 during the T. cruzi infection.

Discussion and Conclusions
Here we demonstrate for the first time that the parasite T. cruzi itself produces LBs in response to the host-parasite interaction and that these organelles may be sources of inflammation mediators. Analogous to LB formed in mammalian leukocytes and macrophages during infectious diseases [30,[33][34][35]53], increases in parasite LB numbers and sizes, in conjunction with changes in LB electron-density highlight the fact that parasite LBs are also dynamic and active organelles, able to modify their structure in concert with cell activation.  Evidence that LBs participate in the regulation of the T. cruzi response during host-parasite interaction was provided here during both in vitro and in vivo infection. Interestingly, LBs in amastigote forms growing in macrophages elicited from the in vivo infection during the acute phase showed higher size and electron-density compared to amastigotes LBs living in cultured macrophages after 1h. This observation is likely consequence of the amplified exposure of proliferating T. cruzi forms to the host inflammatory milieu. In mammalian cells, the size and electron-density of newly formed LBs represent an important structural indicative of the participation of these organelles in innate immune responses [34].
LB morphological changes may reflect differences in lipid composition, stages of formation of new LBs, and/or neutral lipids/phospholipids ratio within LBs [34]. In fact, lipid composition affects LB electron-density [54] based on the fact that osmium tetroxide binds preferentially to the unsaturated bonds of fatty acids [55,56]. Here, we demonstrated by Raman spectroscopy that both AA-stimulated parasites and LBs isolated from AA-stimulated-parasites have a higher content of unsaturated fatty acids, as a result of AA incorporation, compared to unstimulated cells (Figs 3 and 4). The presence of increased AA in the LB fraction was also confirmed by MALDI-TOF mass spectroscopy in LBs purified from the stimulated group (Fig 5). This can explain the changes of LB electron-density within the parasites during the in vivo infection. These LBs, formed as a result of host interaction, are more electron-dense which likely reflects the high content of AA, captured by the reaction with osmium during sample preparation for TEM. Accordingly, in macrophages, increase in LB electron-density was also associated with the cascade of events involved in the synthesis of inflammatory mediators within LBs formed in response to cell activation (reviewed in [38]). Under stimulation, AA is released from its sterified pool and acts as a substrate for enzymatic conversion into lipid mediators [49].
Based on our previous findings of production of AA-derived PGE 2 within LBs from macrophages taking part in immune responses during T. cruzi infection (reviewed in [36]), we considered whether parasite LBs might also play a role in inflammation. Evidence for prostaglandin production in parasites including T. cruzi, T. brucei and Leishmania species was provided by studies from Kubata and other groups (reviewed in [57]). Parasite enzymes involved in the AA cascade and PG synthesis were also identified in protozoans [57]. However, these studies did not associate eicosanoid synthesis with LB formation within the parasite.
Both mammalian [42,43] and parasite [58,59] LBs incorporate AA, the common precursor of eicosanoids. As noted, by using Raman spectroscopy, we detected AA incorporation into stimulated parasites (Fig 3). This technique also detected increased AA content directly in LBs isolated from AA-stimulated parasites (Fig 4). In fact, Raman has gained recognition for biomedical applications [60] and has been recently used to characterize molecular details associated with lipid metabolism and LB formation in macrophages [61] and host-parasite interaction [62]. Our results also demonstrated the existence of a pathway associated with the AA cascade. PGE 2 synthase, but not COX-2, was consistently identified, for the first time, within the parasite (Fig 4). The lack of COX expression might be explained by the absence of homologs of mammalian COX in parasitic microbes although a COX-like enzyme has been reported (reviewed in [57]). On the other hand, several PG synthases with different degrees of homology with human have been identified in parasites [57]. combination with labeling for PLIN2/ADRP. (Cii) is the merged image of (C) and (Ci). The nuclei of parasites are observed after DAPI staining (blue). Cells were immunolabeled by using the EicosaCell technique [31]. Colocalization quantitative analyses showed that PGE 2 and PLIN2/ADRP significantly colocalized in cells [Pearson's correlation coefficient of 0.84 ± 0.02 (mean ± SEM)]. Scale bar, 2 μm (all images). Images are representative of 3 independent experiments. doi:10.1371/journal.pone.0160433.g008 In a recent work it was demonstrated that AA stimulation increases both the number of LBs and the release of PGF2α by metacyclic forms of the parasite Leishmania infantum chagasi [14]. Moreover, PGF synthase was localized in LBs formed within this parasite, thus suggesting that these organelles are sites involved in the synthesis of PGF 2 α [14].
Here, we provide direct evidence that parasite LBs are indeed intracellular sites of eicosanoid production. AA-, but not OA-stimulated trypomastigotes released PGE 2 in parallel to a consistent formation of cytoplasmic LBs (Fig 2) and this PG was fully immunolocalized within parasite LBs (Fig 8). Our data are in agreement with a previous work demonstrating high levels of PGE 2 in lysates of T. cruzi when incubated with AA [63]. However, the results presented here are the first association of the PG production with specific intracellular sites, i.e., LBs.
Therefore, from our present results in T. cruzi and evidences provided by studies in other protozoan parasites, as noted, it seems evident that these organisms are able to generate PGs, but what is the role of parasite-derived PGs in the pathogenesis of parasitic diseases? Because PGE 2 is a potent immunomodulator, it could contribute to the immunosuppression observed during T. cruzi infections with implications to the survival of the parasite in its host. Investigations need to be done to explore this possibility and to uncover the PG metabolic pathways in this parasite.
In conclusion, this work demonstrates that LBs are formed in the parasite T. cruzi in response to the host-parasite interaction and exogenous AA stimulation, that AA is incorporated into parasite LBs and that these organelles serve as sites for PGE 2 synthesis, implying a role for T. cruzi-derived PGs in Chagas' disease pathogenesis. Ã P < 0.05 between groups. Cells were enumerated using osmium staining.