Differential expression of polyamine biosynthetic pathways in skin lesions and in plasma reveals distinct profiles in diffuse cutaneous leishmaniasis

Tegumentary leishmaniasis (TL) is a parasitic disease that can result in wide spectrum clinical manifestations. It is necessary to understand host and parasite determinants of clinical outcomes to identify novel therapeutic targets. Previous studies have indicated that the polyamine biosynthetic pathway is critical for Leishmania growth and survival. Despite its importance, expression of the such pathway has not been previously investigated in TL patients. We performed an exploratory analysis employing Systems Biology tools to compare circulating polyamines and amino acid concentration as well as polyamine pathway gene expression in cutaneous lesions patients presenting with distinct TL disease presentations. Diffuse cutaneous leishmaniasis (DCL) was associated with higher concentrations of amino acids, polyamines and its substrate transporters than mucosal cutaneous leishmaniasis or localized cutaneous leishmaniasis. In addition, the RNA expression of polyamine-related genes of patients lesions from two separate cohorts demonstrated that differential activation of this pathway is associated with parasite loads and able to discriminate the clinical spectrum of TL. Taken together, our findings highlight a new aspect of DCL immunopathogenesis indicating that the polyamine pathway may be explored as a novel therapeutic target to control disease burden.

Transcriptomic analyses of tegumentary lesions from publicly available datasets validate the differential expression of the polyamine biosynthetic pathway in distinct clinical forms of leishmaniasis. To validate our findings on the polyamine pathway in TL, we re-analyzed the gene expression data from an independent patient cohort, which was recently published 16 . This approach revealed that skin lesions from TL patients generally exhibit a distinct gene expression profile compared to normal skin from uninfected healthy endemic controls (Fig. 3A). Fold-difference analysis highlighted that DCL patients exhibited higher expression of AMD1, whereas SLC7A2 was down modulated compared to health controls (Fig. 3A). Additional investigation using a principal component analysis (PCA) model revealed that the overall gene expression profile of the transcripts evaluated here was able to effectively segregate the TL clinical forms (Fig. 3B, C). Such PCA model including expression values of all genes validated the results from the fold-difference analysis, with segregation of all clinical forms (Fig. 3B). Indeed, a discriminant analysis using ROC curves of such combination of genes resulted in high accuracy in distinguishing the clinical groups (Fig. 3C).
Scientific RepoRtS | (2020) 10:10543 | https://doi.org/10.1038/s41598-020-67432-5 www.nature.com/scientificreports/ We next used the same dataset described above to examine if the expression profile of the genes from the polyamine pathway was associated with parasite load in patient lesions. In fact, the estimated values of parasite load were correlated with: (i) the Leishmania genes (which included annotated genes and also putative genes) (Fig. 4A) and (ii) host genes from the polyamine biosynthetic pathway (Fig. 4B). Hierarchical cluster analysis of the Leishmania transcripts quantified by the RNAseq indicated that patients with DCL were the ones who displayed the highest levels of parasite transcripts (Fig. 4A). This approach revealed in Fig. 4A that expression values of all the Leishmania genes examined were positively correlated with parasite load in the group of patients with DCL, but not in those with LCL. Indeed, only histone H4 expression level displayed a positive correlation with the parasite load in the group of LCL patients whereas levels of histone H1 putative, 60S ribosomal protein Plasma levels of ARG1, polyamines: Putrescin, Cadaverine, Spermidine and amino acids: Arginine, Ornithine, Citrulline, were compared among patients with localized (LCL; n = 29), mucosal (MCL; n = 13) and diffuse (DCL) leishmaniasis as well as healthy controls (n = 43). (A) A hierarchical clustering analysis (Ward's method) was employed to show the amino acids and polyamines measured using a representative profile of geometric mean values (log2 -transformed) displayed for each clinical group. The color scale of the heatmap represents z-score by row. (B) Univariate analyzes with scatter plots of the comparisons are shown. Data were compared using the Kruskal-Wallis test with Dunn's multiple comparisons ad hoc test (*P < 0.05, ***P < 0.001). Lines represent median values. Grey bars represent the percentiles 25 and 75 from healthy controls (C) Frequency of the indicated polyamines in healthy controls or those with MCL, LCL and DCL was compared the total free polyamines using the chi square test. www.nature.com/scientificreports/ L39 putative, histone h1 like protein, and amino acid transporter (putative) were not statistically correlated and the remaining genes from the list were actually negatively correlated with parasite loads. Finally, we tested if the expression values of genes from the polyamine pathway could be associated with parasite load. First, the overall gene expression profile of the host polyamine pathway could separate the individuals from the distinct clinical groups (Fig. 4B), corroborating with the results from the PCA model (Fig. 3B). Our analyses also demonstrated that while expression values of several genes exhibited similar association profiles with parasite loads between DCL and LCL patient groups, such as ARG1, other targets displayed a divergent pattern, such as NOS1 and NOS3 (Fig. 4B). Intriguingly, the analyses of a publicly available transcriptome dataset of skin samples demonstrated that expression values of both ARG1 and ARG2 were higher in healthy controls than in LCL (Fig. 3A).
In addition, the expression of these molecules was slightly lower, but not statistically significant in DCL patients compared to healthy controls. Furthermore, the transcriptome data also indicated that expression values of ARG1 were inversely correlated with parasite loads in the lesions from both LCL and DCL whereas the those of ARG2 were also negative correlated in LCL but only marginally directly correlated with parasite loads in DCL lesions (Fig. 4B). www.nature.com/scientificreports/

Discussion
The polyamine biosynthetic pathway has been described as critical to promote Leishmania intracellular replication inside host macrophages 6,18,19 . In this exploratory study, we performed a detailed investigation of the expression profiles of components of such pathway in skin lesions and in peripheral blood of patients presenting with distinct clinical forms within the spectrum of TL disease. Our analyses demonstrated that DCL patients, compared with LCL and MCL, presented higher plasma concentration of polyamines (spermidine and cadaverin) and specific amino acids (ornithine and citruline), identifying a distinct biosignature that characterized this anergic form of TL. In mammalian cells, putrescine is the precursor for spermidine formation and probably for this reason its concentrations are usually low [18][19][20] . Indeed, we did not find any difference in the plasma concentrations of putrescine between the clinical forms, although spermidine concentrations were higher in DCL patients. We hypothesize that putrescine may be rapidly converted into other polyamines that promote Leishmania replication in DCL lesions. Interesting, cadaverin was the most abundant polyamine detected in plasma from the TL patients, and especially in those with DCL. Previous studies reported that cadaverin is induced to compensate significant decreases in concentrations of polyamines or their substrates 21 . Studies with Escherichia coli cell-free system showed that cadaverin has the same ability as putrescine and spermine in promoting protein synthesis 21 . The role of cadaverin in parasitic infections is still largely unknown.
Previous work from our group has shown that patients with DCL display elevated plasma protein concentrations of ARG1 and ODC 4 . These enzymes regulate arginine availability in through regulation of its precursor, ornithine 22 . Arginase is also involved in the urea cycle, in which it catalyzes interconversion of arginine-citrullineornithine 22 . l-arginine is considered a conditionally essential amino acid because even though it is synthesized in human body, it still needs dietary supplementation 23 . Interestingly, here we demonstrate that, although arginine www.nature.com/scientificreports/ plasma concentrations were not different among the distinct TL clinical forms, DCL patients had high concentrations of ornithine and citrulline. Although citrulline can be produced from arginine by iNOS/NOS2 22 , our findings led us to suggest that in the context of DCL, this amino acid is potentially being interconverted by arginase, favoring the production of polyamines. In recent years, there has been an increase in the number of studies describing the importance of amino acid transporters in the trypanosomatid metabolism and especially in Leishmania. Most amino acids are involved in osmotic control, metacyclogenesis, establishment of infection, regulation of autophagy and apoptosis, resistance to oxidative stress, and synthesis of polyamines 24 . Regarding the essential amino acids for the biosynthesis of polyamine, such as arginine and ornithine, Leishmania is auxotrophic for arginine and depends on uptake from the external environment through a specific transporter, amino acid permease 3 (AAP3) 17,25,26 . This means that the production of polyamines in the parasites is directly related to their ability to acquire such amino acids from the host cell 26 . In fact, Leishmania sp. has been described to benefit from host-derived polyamines and several studies have indicated that genetic or pharmacological suppression of parasite enzymes involved in polyamine pathways results in impairment of parasite growth and establishment of host cell infection 22,27,28,29 . More recently, it has been shown that during infection, L. amazonensis is able to alter the host metabolism, inducing polyamine production 30 . Although in our analyses we could not find a specific AAP3 gene, due to the lack of genome annotation, we retrieved the functional annotation in other Leishmania putative genes that are involved    www.nature.com/scientificreports/ amino acid transport (shown in Fig. 4 as amino acid transporter, putative). Our results demonstrated a higher expression of the parasite amino acid transporter (putative) in DCL lesions, which positively correlated with the total counts of parasite transcripts, reinforcing the hypothesis that higher expression of amino acid transporters by Leishmania may indeed favor parasite persistence during infection. It is possible that the systemic balance of polyamines and amino acids induced by infection could contribute to the persistence of the high parasite loads observed in DCL lesions. Our results are in agreement with this idea, although a formal confirmation by mechanistic studies is still necessary.
In the present study, we tested if there are differences in expression of the target genes of the polyamine pathway among different clinical forms in skin lesions. Our results demonstrated that among the expression values of ARG1, CAT-2A and SMS were higher in DCL patients. It has been shown that CAT-2 expression values are increased by both Leishmania infection and l-arginine deprivation 10 . Therefore, we hypothesize that parasites from DCL patients positively regulate CAT-2A expression, thereby increasing polyamine production, which is important for their proliferation and maintenance of infection. In previous studies from our group and others, lesions from DCL patients were found to present greater expression of IL-4 and IL-10 transcripts in relation to TNF expression 4 , whereas lesions from patients with LCL and MCL are characterized by high expression of IFNγ mRNA and absence of IL-4 3 . In addition, THP-1 macrophages infected with L. donovani are described to exhibit increased l-arginine uptake by CAT-2, augmented ARG1 activity and higher levels of spermidine, that correlated with increased concentrations of IL-10 and reduced concentrations of IL-12 and TNF-α 10 . Whether the increased expression of ARG1 and CAT-2A in skin lesions from patients with DCL is associated with the immunosuppression warrants further investigation.
The analysis in patient lesions demonstrated that ODC1 was not a part of the biosignature of the polyamine pathway in DCL skin lesions. ODC is the key enzyme implicated in polyamine biosynthesis required for in vitro intracellular proliferation in Leishmania-infected cells 19,31 . The intracellular concentration of polyamines is regulated by several mechanisms, including the synthesis, degradation and efflux/uptake by the polyamine transporters 6 . High concentrations of intracellular polyamines induce the expression of important enzymes involved in polyamine recycling: spermidine/spermineN1-acetyltransferase (SSAT) and peroxisomal n(1)-acetylspermine/spermidine oxidase (PAOX), associated to interconversion and degradation of polyamines. Another important step of the regulation of polyamine production is mediated by the fast turnover of ODC. When cellular polyamine levels are high, they induce the biosynthesis of an ODC inhibitor named ODC antizyme (OAZ), which prevents its dimerization and promotes ODC degradation, through the 26S proteasome 32,33 . OAZ is induced by an excess of polyamines (and has a fast turnover), and besides regulating the degradation of ODC, negatively regulates cellular polyamine transporters 34 . However, the proteins involved in polyamine transport and the exact mechanisms by which polyamines regulate their uptake in the mammalian cells are not well known. Cells in which the OAZ enzyme is expressed to high levels exhibit a marked reduction in polyamine uptake. Our data suggest that high expression of OAZ in the DCL lesion could be responsible for the degradation of ODC, justifying the low levels of this enzyme. However, the exact molecular and cellular mechanisms involved in ODC in lesions or systemic regulation in DCL patients remain to be explored.
Although the RNA transcripts for CAT-2B, SSAT1, SRM, OAZ and PAOX did not show statistical difference between the diverse TL clinical forms (Fig. 2C), we observed a distinct biosignature of the polyamine biosynthetic pathway in DCL individuals in the hierarchical cluster analysis, corroborating the idea that this metabolic pathway is indeed important in parasite proliferation and successful establishment of Leishmania infection. However, a limitation of the present study was that we did not have available sample/data to directly test correlations between polyamine expression and parasite burden in the lesions. Thus, we employed an analytical approach to specifically test this hypothesis in a separate patient cohort.
An important contribution of our study is the validation analysis using an independent patient cohort recently published 16 . In this approach, we tested whether the transcriptomic profile also revealed differential gene expression in the TL disease groups. Indeed, the transcriptional data from skin biopsies from DCL and LCL patients compared to normal skin revealed that the targets from the polyamine pathway are able to discriminate the distinct TL clinical groups. Interestingly, the SLC7A2 gene was down modulated in DCL patients. Arginine transport is mediated by members of the solute carrier 7 family (SLC7), that is divided into two subfamilies: the cationic amino acid transporters (CATs; SLC7A1-4 and SLC7A14) and L-type amino acids transporters (LATs: SLC7A5-13), also known as CATs 35 . Although expression of this gene diverges between the RNAseq and n-counter analyses, being higher in the first and lower in the latter, CAT2A and CAT2B are isoforms of the SLC7A1 gene and by RNAseq technique, we could not identify such isoforms separately. According to our study, the Adenosylmethionine decarboxylase 1 (AMD1) up regulation in DCL patients may be relate to the high concentrations of spermidine found in these patients. AMD1 is an enzyme involved in biosynthesis of spermidine 36 . The importance of this enzyme in tumorigenesis in gastric cancers associated with polyamine synthesis has been recently been demonstrated 36 . Of note, its role in parasitic diseases, including Leishmania infection, is still not clear.
The transcriptome data from skin lesions indicated that expression levels of ARG1 and ARG2 were not statistically distinct between DCL patients and healthy controls. Further analyses also revealed a negative correlation between ARG1 expression levels and parasite loads in both LCL and DCL whereas ARG2 expression was negatively correlated in LCL but slightly positively correlated with parasite loads in DCL lesions. At first glance, such findings may contrast with the idea that arginase-1 is involved in the pathogenesis of DCL, as previously reported 4 . Our experiments in plasma revealed higher levels of arginase-1 in DCL patients compared to controls, and the nanostring experiment unfortunately did not include samples from healthy controls for comparisons. Although the results presented here clearly show a differential modulation of the polyamine pathway in distinct clinical forms of TL, additional studies are still warranted to investigate the direct participation of arginase-1 in cellular responses to Leishmania in skin lesions. www.nature.com/scientificreports/ The transcriptome analyses also revealed that the overall gene expression profile of the polyamine pathway is associated with the parasite loads and that such association is linked to the TL clinical presentation (in either LCL or DCL). Although the expression level of genes from the parasite should be proportional to transcriptionally estimated parasite load, it is known that post-transcriptional, epigenetic regulation can result in discrepancies between abundance of a given gene and number of parasites. The previous report on this transcriptome data has identified the 15 most highly and uniformity expressed parasite genes in each of the patients analyzed with DCL 16 . It is indeed expected that such expression levels of genes should directly follow the parasite loads. However, we detected an important difference in the profile of correlation between expression levels of Leishmania genes and parasite loads between the groups of DCL and LCL. This finding is surprising and reinforce the hypothesis of differential modulation of gene expression occurring in the parasites. Interestingly, among the genes that exhibited divergence in association of expression levels with parasite loads between DCL and LCL are those whose functions are related to amino acids transport. Leishmania putative genes of these transporters were shown here to be highly expressed in patients with DCL and the expression level was related to parasite loads in DCL but not in LCL, supporting the hypothesis that parasites may take advantage of an immunomodulated environment. Additional studies are necessary to test if the activation of the polyamine pathway is a cause or consequence of the higher parasite loads in DCL.
In conclusion, our findings indicate that the differential activation of the polyamine pathway characterizes DCL relative to other clinical forms of TL (summary of findings from the experiments performed here are illustrated in Fig. 5) and open perspectives for future studies testing the manipulation of such pathway to reduce immunopathology in TL.

Methods
Ethics statement. This study was approved by the institutional review board from Instituto Gonçalo Moniz, Fundação Oswaldo Cruz. All clinical investigations were conducted according to the Declaration of Helsinki. Written informed consent was obtained from all participants or legal guardians, and all data analyzed were anonymized. Study design. We performed a cross-section study of patients presenting with different clinical forms of TL as well as uninfected endemic controls. Social-demographic and clinical characteristics of all patients were www.nature.com/scientificreports/ previously described 4,37 . DCL cases (n = 14) were obtained from a study performed between 1980 and 1990 in the state of Maranhão, Northeast of Brazil 4 . DCL diagnosis was based using previously described criteria 1,2 . In addition, we evaluated data from age-and sex-matched patients with MCL (n = 13) or LCL (n = 29) recruited at our reference clinic in Jiquiriçá, BA-Brazil, as previously described 37 . MCL and LCL individuals included in the present study were required to have no previous history of TL and to be treatment naïve. For plasma analyses, we included samples from 43 healthy controls (matched by age and sex to the TL groups) from the same region of endemicity and who had negative results of an anti-Leishmania delayed-type hypersensitivity test. Diagnosis criteria for LCL and MCL were published previously 37 and were based on anti-Leishmania delayed-type hypersensitivity test, detection of anti-Leishmania antibody, or detection of Leishmania parasites in biopsy tissue specimens by either immunohistochemistry or qualitative polymerase chain reaction (PCR) assays.
Polyamine trofiles. The polyamine profiles and content were performed according to a previous study 38 Briefly, serum samples from patients with Leishmaniasis (60 µl) were mixed with cold 5% (v/v) perchloric acid at a ratio of 1:4 (v/v). Then, the samples were submitted to three cycles of freezing (− 20 °C) and thawing (at room temperature), prior to centrifugation at 11,000g for 10 min. The supernatant containing free polyamines was collected. Free polyamines were derivatized with dansylchloride (5 mg ml −1 in acetone), 0.05 mM diaminoheptane-DAH (internal standard) and saturated sodium carbonate. After 50 min incubation in the dark at 70 °C, the excess of dansylchloride was converted to dansylalanine by adding proline (100 mg ml −1 ). After 30 min incubation (room temperature), dansylated PAs were extracted with of toluene 1:1 (v/v), the organic phase containing the polyamines was collected. The toluene phase was evaporated with gaseous nitrogen (40 °C). Dansylated polyamines were resuspended in 100 µl of acetonitrile. Polyamines were separated by high-performance liquid chromatography (HPLC, Shimadzu, Japan) using a C18 reversed-phase column (5 µm × 4.6 mm × 250 mm-Sulpelcosil, Supelco), as described previously 38 . Polyamines were detected at 340 nm (excitation) and 510 nm (emission) wavelengths with an RF-20A fluorescence detector (Shimadzu). Peak areas and retention times were measured by comparison with standard known concentrations of polyamines (Table S1).
Free amino acids profiles. The amino acid content was determined as previously described 39 . Serum samples from patients with Leishmaniasis (60 µl) were extracted in 1.8 ml of 80% ethanol (v/v) and concentrated in 'speed vac' . Samples were resuspended in 0.6 ml Milli'Q and centrifuged at 20,000g for 10 min. The supernatant was filtered through a 20 µm membrane. Amino acids were derivatized before injection with o-phthalaldehyde and separated by HPLC (Shimadzu) on a C18 reverse phase column (as described above). The gradient program was developed as in 40  Arginase-1 protein measurement in plasma. Plasma levels of arginase 1 (Hycult Biotech, Uden, the Netherlands) were measurement using enzyme-linked immunosorbent assay ELISA according to the manufacturer's instructions. nCounter analysis. Tissue samples from which we had high-quality messenger RNA (mRNA) were obtained from a subset of three patients with DCL, four patients with MCL and seven patients with LCL who were also matched for age and sex. Skin tissues were used from DCL and LCL whereas nasal mucosal samples were obtained from MCL patients as previously described 4,37 . Total RNA was extracted from cryopreserved lesion biopsy specimens, using Trizol reagent (Invitrogen, Carlsbad, California), with an additional purification step using RNeasy columns (Qiagen, Venlo, Netherlands). nCounter analysis (NanoString Technologies, Seattle, Washington) was performed based on direct molecular bar coding of target RNA transcripts and digital detection 41 . The chosen targets genes were: ARG1 (Arginase 1), ODC1 (Ornitine Decarboxylase 1), SRM (Spermidine Syntahse), SMS (Spermine Synthase), SAT1 (spermidine-spermine acetyl transferase), PAOX (Peroxisomal oxidase), SLC7A2 (CAT2-cationic aminoacid transporter 2), OAZ1 (Ornithine Decarboxylase antienzyme). To account for differences in leukocyte infiltration between patient lesions, data were normalized for CD45, which encodes the pan-leukocyte marker CD45, detectable at fentomolar range, as previously reported 41 .
RNA-seq analysis. Data Fig. 3A) and they were used in additional analyses to test whether its expression values were able to separate the distinct clinical groups. Differentially expressed genes (DEGs) were examined by edgeR package 44 . The parasite burden from RNAseq data was measured by normalizing the library size 45 . The gene expression values used were the TMM-normalized log CPM values. The correlation between the parasite burden and gene expression values (both human and leishmania) were performed with the Hmisc package. As only a few Leishmania genes have been functionally annotated by biological assays, we collapsed the genes by putative functions and used the most representative expressed genes (Table S2). This strategy was employed to avoid the synonymies between genes with the same function; such approach is commonly used in high throughput analysis (https ://www.ncbi.nlm. nih.gov/pmc/artic les/PMC31 66942 /pdf/1471-2105-12-322.pdf). Changes in gene expression were considered significant when statistical test values (false discovery rate [FDR] adjusted p-value) were lower than 0.05 and the fold-difference higher than ± 1.5. Principal Component Analysis (PCA) was performed using the TMM-normalized log-transformed CPM values with plotPCA function from Deseq2 package 46 to verify if the expression values of the genes from the polyamine biosynthetic pathway were able to classify the samples within the groups, regardless of the fact that some of these genes were not differentially expressed between the clinical groups. The receiver operator characteristics (ROC) curves with area under the curve were measured and plotted with pROC package 47 . The ROC curve analysis was performed with the same gene expression values which were inputted in the PCA algorithm, to assess the sensibility and specificity of this classification. Moreover, the ROC curve analysis used a multinomial model, in which the outcomes (HC, DCL and LCL) were binarized. Thus, this approach allowed us to compare the power of all the genes from polyamine biosynthetic pathway described above to discriminate between the following groups: (