FUM Gene Expression Profile and Fumonisin Production by Fusarium verticillioides Inoculated in Bt and Non-Bt Maize

This study aimed to determine the levels of fumonisins produced by Fusarium verticillioides and FUM gene expression on Bt (Bacillus thuringiensis) and non-Bt maize, post harvest, during different periods of incubation. Transgenic hybrids 30F35 YG, 2B710 Hx and their isogenic (30F35 and 2B710) were collected from the field and a subset of 30 samples selected for the experiments. Maize samples were sterilized by gamma radiation at a dose of 20 kGy. Samples were then inoculated with F. verticillioides and analyzed under controlled conditions of temperature and relative humidity for fumonisin B1 and B2 (FB1 and FB2) production and FUM1, FUM3, FUM6, FUM7, FUM8, FUM13, FUM14, FUM15, and FUM19 expression. 2B710 Hx and 30F35 YG kernel samples were virtually intact when compared to the non-Bt hybrids that came from the field. Statistical analysis showed that FB1 production was significantly lower in 30F35 YG and 2B710 Hx than in the 30F35 and 2B710 hybrids (P < 0.05). However, there was no statistical difference for FB2 production (P > 0.05). The kernel injuries observed in the non-Bt samples have possibly facilitated F. verticillioides penetration and promoted FB1 production under controlled conditions. FUM genes were expressed by F. verticillioides in all of the samples. However, there was indication of lower expression of a few FUM genes in the Bt hybrids; and a weak association between FB1 production and the relative expression of some of the FUM genes were observed in the 30F35 YG hybrid.


INTRODUCTION
Maize (Zea mays L.) is one of the world's major agricultural crops serving as a staple food for millions, with an annual average production and consumption of 989.2 million metric tons (Mt) and 953.9 Mt (2013953.9 Mt ( -2014, respectively. Brazil is currently the world's third largest producer, after United States and China, with an average production of 79.3 Mt over the last year (United States Department of Agriculture [USDA], 2014).
Susceptible to fungal and mycotoxin contamination, numerous toxic fungal secondary metabolites can be found in maize, though, fumonisins occur with greatest frequency. Members of the Fusarium fujikuroi species complex, mainly F. verticillioides (Sacc.) Nirenberg and F. proliferatum (Matsush.) Nirenberg, are capable of producing these mycotoxins consistently (Desjardins, 2006). F. verticillioides is one of the most important species associated with maize throughout the world and produces high levels of fumonisins. Contamination with these mycotoxins has a significant impact, compromising the quality of maize products (de la Campa et al., 2005).
Based on chemical structure, fumonisins have been classified into A, B, C, and P groups; the most common are the B analogs and fumonisin B 1 (FB 1 ) is the most prevalent and toxic of the group (Stepien et al., 2011). Fumonisin B 1 is known to cause toxicity in animals and humans due to the inhibition of sphingolipid metabolism and cell cycle regulation, resulting in diverse and complex effects. Examples include leukoencephalomalacia in horses, pulmonary oedema, and hydrothorax in swine; fumonisin B 1 is additionally carcinogenic to rodents and exhibits nephrotoxic as well as hepatotoxic activity in rats and rabbits (Desjardins, 2006). Epidemiological studies have suggested that fumonisins may also be associated with oesophageal cancer and neural tube birth defects in humans (Marasas et al., 2004).
Globally, fumonisin contamination in food is a concern. The FDA (the United States Food and Drug Administration) has established a limit of 3-4 ppm of fumonisin contamination (FB 1 + FB 2 + FB 3 ) in human foods and 5-100 ppm in animal feeds. The Commission Regulation (EC) of the European Union has established a maximum level of 4 ppm for FB 1 + FB 2 in unprocessed maize and 1 ppm in maize and maize-based products intended for human consumption (European Commission Regulation [EC], 2007). However, the legislation is distinct around the world, depending on the concern that the authorities have about the potential toxic effects of fumonisins on animals and their implications for industry (Food and Drug Administration [FDA], 2001;Schatzmayr and Streit, 2013).
Low levels of fumonisins can occur even in intact maize kernels, since F. verticillioides is found in both asymptomatic and diseased plants. Environmental conditions, water availability and the genetic background of the plant and the pathogen are significant factors in disease development and mycotoxin production (Oren et al., 2003;Bowers et al., 2013). It has been shown that physically injured kernels increase the propensity for fungal contamination and mycotoxin production Parsons and Munkvold, 2012;Bowers et al., 2013). Kernel injuries caused by ear feeding insects are particularly important. These pests act as vectors for fungal spores, increasing the severity of fungal disease and mycotoxin production (Dowd, 2000;Wu, 2006;Ferreira-Castro et al., 2012).
The damage caused by European corn borer (ECB, Ostrinia nubialis Hübner), Southwestern corn borer (SWCB, Diatraea grandiosella Dyar), corn earworm (CEW, Helicoverpa zea Boddie) and fall armyworm (FAW, Spodoptera frugiperda, J.E. Smith) has been shown to favor mycotoxin contamination in the field and to contribute to mycotoxin accumulation during storage (Sinha, 1994;Dowd, 2000;Wu, 2006). Bt (Bacillus thuringiensis) hybrids have been effective in reducing injuries by these insects, indirectly controlling plant susceptibility to fungal infection and mycotoxin contamination (Munkvold et al., 1997;Ostry et al., 2010;Bowers et al., 2013;Foresti et al., 2013) It is widely known that Bt maize presents a lower risk of fumonisin contamination compared with non-Bt hybrids when exposed to lepidopteran insects (Duvick, 2001;Pazzi et al., 2006;Abbas et al., 2011;Bowers et al., 2013Bowers et al., , 2014. A recent study has also demonstrated that Bt maize exhibited reduction in deoxynivalenol and zearalenone contamination in harvested maize kernels (Ostry et al., 2010). Although many studies have shown the relationship between kernel injuries by insects and fumonisin contamination in maize during harvest, none has focused on the fumonisin accumulation in Bt and non-Bt maize after this period.
The aim of this study was to verify the levels of fumonisins produced by F. verticillioides on Bt and non-Bt maize, post harvest, in different periods of incubation under controlled conditions. The second objective of this study was to compare FUM gene expression between Bt and non-Bt hybrids and to study the association between fumonisin production and FUM gene expression by F. verticillioides for each of the hybrids.

Maize Grain Samples
Maize cultivar samples of Bt 2B710 Hx and 30F35 YG and their isogenic non-Bt 2B710 and 30F35 were provided by the Agronomic Institute of Campinas (Instituto Agronômico de Campinas, APTA-São Paulo, Brazil). These samples were sown in November/2010 and harvested in March/2011 in Cruzália, State of São Paulo, Brazil.
Sampling was conducted according to methodology proposed by Delp et al. (1986) with modifications. Bt and non-Bt crop fields were sampled by dividing them into four sectors of uniform size containing eight rows each. Ten samples were randomly collected from each sector, for a total of 40 samples per hybrid. The hybrids were harvested manually in the experimental field by the technical support staff of the Agronomic Institute of Campinas. Thirty subsamples for each hybrid were then selected for this experiment, totalling 120 tests.
The injuries caused by S. frugiperda in 30F35 YG, 2B710 Hx and their isogenic hybrids were previously analyzed from sowing to harvest by The Agronomic Institute of Campinas in the same experimental field used in the current study (Michelotto et al., 2011).
All samples were sterilized by gamma radiation at a dose of 20 kGy to eliminate the natural mycoflora at the Nuclear and Energetic Research Institute (Instituto de Pesquisas Energéticas e Nucleares-IPEN-CNEN/SP). A GammaCell irradiator equipped with a cobalt 60 source was used and a dose rate ranging from 4.74 to 4.84 kGy/h was applied. The mycoflora was verified in order to control the irradiation procedure (Berjak, 1984). Samples were previously analyzed for the presence of fumonisins according to the methodology described by Visconti et al. (2001), because gamma radiation does not eliminate fumonisin from grains (Ferreira et al., 2007). The values (for each sample) were considered to be a basal FB 1 or FB 2 contamination, and were used to calculate the final fumonisin concentration produced by F. verticillioides after each period of incubation (total concentration minus basal concentration).

Fusarium verticillioides Strain
Fusarium verticillioides isolate 13BA was obtained from the culture collection of the Institute of Biomedical Sciences, ICBII, University of São Paulo, Brazil. The isolate was maintained on SNA (Spezieller Nährstoffarmer Agar) by monosporic isolation and it was used in all experiments (Rocha et al., 2011).
Before beginning the experiments, the ability of fumonisin production by F. verticillioides strain 13BA was re-evaluated according to methodology proposed by Ross et al. (1990) and Sydenham et al. (1990), with concentrations of 120and 36 μg/g for FB 1 and FB 2 , respectively.

Inoculation of Corn Samples with Fusarium verticillioides Spore Suspension
The F. verticillioides fungal spore suspension was prepared according to Ferreira et al. (2007). The final suspension was adjusted to 5 × 10 5 spores/mL and inoculated into 5 g of maize bran (previously treated with 20 kGy gamma radiation). This mixture was added to Petri dishes containing 25 g of grains used for fumonisin and FUM gene cluster expression analyses. The samples were stored in a plastic container at a relative humidity and temperature controlled by a previously calibrated thermo hygrometer. The container was incubated in a BOD incubator at 25 • C and a relative humidity of 97.5 to 99.0% obtained with 30% K 2 SO 4 solution as described by Winston and Bates (1960) for 10, 20, and 30 days. After each incubation period, 10 samples of each hybrid were analyzed, with 20 g for fumonisin analyses and 5 g for gene expression experiments.
Fumonisins were quantified based on a calibration curve using standard solutions of FB 1 and FB 2 (Sigma-Aldrich). The coefficient of correlation was 0.993 for FB 1 and 0.995 for FB 2 . The limit of quantification (LOQ) for FB 1 was 0.015 μg/g, with a mean recovery of 92.38% and standard deviation of 13.68% (five replicates). For FB 2 , LOQ was 0.015 μg/g, with a mean recovery of 85.39% and standard deviation of 6.87% (five replicates). LOQ was determined via recovery tests. The lowest validated spike (0.015 μg/g) level with a percentage of recovery between 70 and 120% and a standard variation lower than 20% was defined as the method LOQ in this study (European Commission Regulation [EC], 2007).
The final fumonisin concentration was calculated through the subtraction of the total FB 1 or FB 2 for each analyzed sample (basal FB 1 or FB 2 concentration plus the concentration of FB 1 or FB 2 produced by F. verticillioides during the respective incubation periods) minus the basal FB 1 or FB 2 contamination of each sample that came from the field.

FUM Gene Cluster Expression
Isolation of mRNA and Reverse Transcription RNA was extracted from 5 g of corn samples corresponding to each inoculation period (10, 20, and 30 days) using the Plant RNA Isolation Aid Kit (Ambion R , Life Technologies, Carlsbad, CA, USA) followed by purification utilizing RNAqueous Kit (Ambion R , Life Technologies, Carlsbad, CA, USA), according to the manufacturer's instructions. cDNA was synthesized with the High Capacity cDNA Reverse Transcription kit according to the manufacturer's instructions (Applied Biosystems, Life Technologies, Carlsbad, CA, USA). cDNA was synthesized in a Gene Amp R PCR System 9700 thermal cycler (Applied Biosystems, Life Technologies) for 1 h at 37 • C. The samples were stored at −20 • C.
The Platinum SYBR Green qPCR SuperMix-UDG reagent (Invitrogen, Life Technologies) was used as the reaction mixture, according to the manufacturer's instructions. Each sample was analyzed in duplicate. An appropriate negative control containing no template was used in all reactions to exclude possible contaminations. The quantification of mRNA was normalized using the average of TUB (β-tubulin) and CALM (calmodulin) genes (Bustin et al., 2009), with cDNA amplifications ran on the same plate. The following reference genes were tested prior to the experiments: CALM, TUB, EF-1α (elongation factor 1 α) and 18s rRNA (18s ribosomal RNA). The Ct values of TUB and CALM were similar; therefore, the mean values of these genes were used for normalizing the data.
For validation of the method, serial dilutions of the target genes and of the reference genes were prepared in triplicate. The primers' efficiency was calculated based on the curve generated by the software. FUM2, FUM10, and FUM21 were excluded from the analysis since qPCR parameters were not acceptable according to the method of analysis proposed by Pfaffl (2001) and Schmittgen and Livak (2008).

Statistical Analysis
The results were analyzed using Gamlss R 2.9 package and Statistical Analysis Software (SAS) version 9.1. The Gamlss model (5% level of significance) with Weibull distribution was used to evaluate differences of FB 1 and FB 2 production between Bt and non-Bt maize during the period of incubation (10, 20, and 30 days; Rigby and Stasinopoulos, 2005). The Weibull distribution was chosen using quantile residual plots (Dunn and Smyth, 1996). The analysis was conducted based on the overall data (from the first to 30th day). The Gamlss model was used to verify the interaction between hybrids (Bt and non-Bt) and period of incubation (10, 20, and 30 days). As these interactions were not significant (P > 0.05), further analyses based on each period of incubation (10, 20, and 30 days) were not conducted.
The differences of FUM gene expression between Bt and non-Bt maize were evaluated using the Mann-Whitney test, since the data was not normally distributed (5% level of significance) for each period of incubation (Gibbons and Chakraborti, 2003

RESULTS
Production of FB 1 and FB 2 in Bt and Non-Bt Maize Samples Artificially Contaminated with F. verticillioides During the Periods of 10, 20, and 30 days FB 1 and FB 2 were quantified during 10, 20, and 30 days of incubation. The Bt hybrid 30F35 YG presented FB 1 levels between 0.11 to 4.99 μg/g (mean: 2.38 μg/g) and FB 2 from 0.02 to 2.66 μg/g (mean: 1.53 μg/g). 30F35 hybrid presented FB 1 levels from 1.49 to 5.34 μg/g (mean: 3.17 μg/g) and FB 2 from 0.015 to 2.83 μg/g (mean: 1.7 μg/g) ( Table 2). Statistical analysis showed that FB 1 contamination was significantly lower in 30F35YG when compared to its isogenic hybrid 30F35 (P = 0.0095). However, there was no statistical difference between Bt and non-Bt maize samples for FB 2 production (P = 0.827). Interaction between hybrids (Bt and non-Bt) and period of incubation (10, 20, and 30 days) was not significant (P > 0.05), therefore there was no evidence that the mean difference of FB 1 (P = 0.26) and FB 2 (P = 0.82) production between Bt and non-Bt maize changed over the period of incubation. 2B710 Hx presented levels of FB 1 and FB 2 between 0.19-9.71 μg/g (mean: 6.42 μg/g) and 0.02-6.75 μg/g (mean: 3.98 μg/g), respectively. The isogenic 2B710 hybrid presented FB 1 levels from 3.33-11.55 μg/g (mean: 8.49 μg/g) and FB 2 from 0.8 to 8.31 μg/g (mean: 4.67 μg/g) ( Table 2). FB 1 contamination was significantly lower in 2B710Hx when compared to its isogenic hybrid 2B710 (P = 0.0001). There was no statistical difference between Bt and non-Bt maize samples for FB 2 production (P = 0.382). There was no evidence that the mean difference of FB 1 (P = 0.95) and FB 2 (P = 0.61) production between Bt and non-Bt maize changed over the period of incubation.

Optimization of qPCR Reactions and FUM Gene Expression in Bt and Non-Bt
Maize Samples Artificially Contaminated with F. verticillioides During the Periods of 10, 20, and 30 days

The comparative
Ct method was chosen for FUM gene expression analysis, since qPCR efficiencies were similar. Ct relative quantification is used to compare the gene expression between one sample and another (control). In this study, the control was considered the samples with the lowest Ct value for all of the genes for each hybrid (Supplementary Tables S1 and S2).
The efficiencies (E) obtained in the qPCR analysis were between 78 and 110% for all of the studied genes, with slopes between −4 (FUM21) and −3.1 (FUM8, FUM14, FUM15) and r 2 from 0.97 (FUM2, FUM6, FUM10) to 0.99 (all the other genes; Supplementary Figure S1). According to the European Network of GMO Laboratories, acceptable slopes for standard curve range from −3.1 to −3.6. In this study, FUM2, FUM10, and FUM21 presented values outside this range (European Network of Gmo Laboratories [ENGL], 2008). The melt-curve analysis resulted in one single peak for all of the genes used in the analysis (Supplementary Figure S1).
The average between the Ct values of the reference genes β-tubulin (TUB) and CALM were selected to normalize the Ct values of the samples. The average of CALM and TUB showed better repeatability and less variability in Ct values when compared to other tested genes (EF-1α and 18s rRNA). The Ct values were plotted against the log dilutions, and the slopes ranged from −0.24 (FUM21 in relation to TUB and CALM) to 0.02 (FUM 1 in relation to TUB and CALM). FUM2, FUM10, and FUM21 reactions did not present acceptable validation parameters once slope values for compatibility tests were not within −0.1 and 0.1; therefore they were excluded from the analysis. Even though r 2 for FUM6 was below 0.99, this gene was not excluded from the analyses, since efficiency and slope values were satisfactory.
Amplification of the FUM genes was observed in all of the samples. Statistical analysis showed that there was an indication of higher FUM1 expression during the three periods of incubation (P < 0.02), higher FUM13 expression during the 10th day of incubation and higher FUM7, FUM8, and FUM14 expressions during the 30th day of incubation (P < 0.04) and higher levels of FUM19 expression in the 20 and 30th days of incubation (P < 0.04) in the non-Bt 30F35 hybrid. There was an indication of higher FUM19 expression during the 10th day of incubation in the non-Bt 2B710 hybrid (P < 0.03).

DISCUSSION
Samples representing the non-Bt hybrids 2B710 and 30F35 that came from the field presented damaged kernels as a result of insect injuries, which could possibly facilitate fungal penetration and consequently fumonisin production. 2B710 Hx and 30F35 YG samples were virtually intact (Michelotto et al., 2011). It has been shown that wounded plants and seeds facilitate fungal contamination, increasing the chance of diseases and mycotoxin production (Wallin, 1986;Gradziel and Wang, 1994;Mahoney and Rodrigues, 1996;Imathiu et al., 2008;Mohammadi and Hokmabadi, 2010). The Bt hybrids 30F35 YG and 2B710 Hx were chosen due to their worldwide utilization for controlling insect pests. 30F35 YG expresses cry1Ab, an efficient protein for reducing injuries from lepidopteran pests' feeding (Bowers et al., 2014). Although cry1Ab is not highly effective against CEW (Bowers et al., 2013) and FAW, important lepidopteran insects found in Brazil (Farias et al., 2014), this protein has demonstrated efficacy in preventing insect injuries and therefore aiding in the reduction of fungal contamination and mycotoxin production (Munkvold, 2003;Ostry et al., 2010;Leslie and Logrieco, 2014).
2B710 Hx expresses cry1F protein, and it has been shown to be ineffective against CEW (Bowers et al., 2013), however, hybrids expressing this protein have demonstrated temporary efficacy in controlling FAW in Brazil, since cry1F resistant populations of S. frugiperda have emerged from maize field where cry1F protein has been extensively used (Farias et al., 2014). This insect is native to the tropical regions of the western hemisphere from the United States to Argentina. The larvae feed primarily on maize leaf tissue, causing extensive defoliation (Farias et al., 2014); in addition, S. frugiperda can migrate to corn ear, causing damage to kernels (Gorman and Kang, 1991), creating infection sites for toxigenic fungi, such as Fusarium and Aspergillus (Munkvold, 2003(Munkvold, , 2014. Complementary research conducted by Michelotto et al. (2011), using the field and maize samples corresponding to those FIGURE 1 | Mean relative expression of FUM1, FUM3,FUM6,FUM7,FUM8,FUM13,FUM14,FUM15,and FUM19 in 30F35 and 30F35 YG hybrids during 30 days of incubation.
FIGURE 2 | Mean relative expression of FUM1, FUM3,FUM6,FUM7,FUM8,FUM13,FUM14,FUM15,and FUM19 in 2B710 and 2B710 Hx hybrids during 30 days of incubation. used in this study, has demonstrated that the damage caused by S. frugiperda was significantly lower in 30F35 YG and 2B710 Hx when compared to their respective isogenic hybrids from sowing to harvest. This may explain the pattern of fumonisin production in the samples under controlled conditions, since damaged kernels can promote F. verticillioides invasion and therefore facilitate fumonisin production in non-Bt maize. In fact, insects are able to cause injuries to the external protection of grains and plant tissues in the field, allowing the fungal hyphae to readily penetrate, have access to nutrients and produce mycotoxins (Wu et al., 2004;Wu, 2006;Jouany, 2007). Since Bt maize reduces insect damage and the risk of fungal contamination, it could indirectly control mycotoxin accumulation during storage. However, it is important to emphasize that a deeper investigation regarding the differences between Bt and non-Bt maize in F. verticillioides contamination and fumonisin production as well as the use of different cry proteins to effectively control insect pests are warranted to aid in the management strategies and to better understand the correlation between insect pest control and fungal invasion/mycotoxin production in maize.
The current study also compared the fumonisin production by F. verticillioides with FUM1, FUM3, FUM6, FUM7, FUM8, FUM13, FUM14, FUM15, and FUM19 gene expression for each of the studied hybrids, and also compared the FUM gene expression between Bt and non-Bt hybrids.
FUM genes were expressed by F. verticillioides in all of the corn samples, with an indication of lower gene expression for some of the studied genes in the Bt hybrids. There was a weak correlation between FB 1 and FB 2 production and the relative expression of some of the studied FUM genes were observed in 30F35 YG. Among these genes, FUM1, FUM14, and FUM19 were better correlated with fumonisin production. Interestingly, it was observed differences in the expression of these same genes in Bt and non-Bt hybrids.
Studies have shown a correlation between some of the FUM genes and fumonisin production (Lopez-Errasquín et al., 2007;Jurado et al., 2010;Rocha et al., 2011;Lazzaro et al., 2012;Medina et al., 2013). FUM19 encodes a protein that is highly similar to the ATP-binding cassette of multidrug resistance transporters, which act as efflux pumps to reduce cellular concentrations of toxins and thereby confer protection (Desjardins, 2006). Although Lopez-Errasquín et al. (2007), Jurado et al. (2010) and Rocha et al. (2011), have found a positive correlation between FUM19 and fumonisin production, our results have shown a weak association between this gene expression and fumonisin production in 30F35 YG hybrid. FUM14 also had a weak association with fumonisin production; this gene encodes a protein involved in the formation of tricarballylic ester function. Indeed, a previous study has demonstrated that FUM14, as well as FUM1 and FUM19 exhibit a significant effect on the synthesis of FB 1 and FB 2 (Medina et al., 2013). Further studies regarding the expression of these genes should be conducted to verify whether or not they could be useful for diagnostics.
It is relevant to emphasize that previous studies have used synthetic media to produce fumonisins and analyze FUM gene expression (Lopez-Errasquín et al., 2007;Jurado et al., 2010;Rocha et al., 2011;Lazzaro et al., 2012;Medina et al., 2013), which increases the quality and concentration of RNA. Fungal RNA extraction from maize can be limited, once kernel consists of 90% starch and 10% proteins (Gibbon and Larkins, 2005). Even though the experimental Petri dishes containing maize grains were covered in F. verticillioides mycelium, the final RNA consisted of plant and fungal RNA, which likely decreased the performance of qPCR reactions, thereby interfering in the correlation between FUM gene cluster and fumonisin production. The rationale for isolating fungal RNA from maize grains is due to the fact that F. verticillioides can persist in undamaged plants and produce fumonisins; thus, identifying an adequate genetic marker for assessing fumonisin production during interaction between F. verticillioides and symptomless plants could aid in the prevention of fumonisin accumulation in the field and storage (Presello et al., 2007;Duncan and Howard, 2010;Bowers et al., 2013).

CONCLUSION
Non-Bt maize samples 30F35 and 2B710 presented damaged kernels caused by insects in the field. In this study, wounded grains have possibly facilitated F. verticillioides penetration, leading to higher fumonisin production in non-Bt hybrids during the period of incubation due to the increase of fungal biomass. Insects cause physical damage, disruption of nutrients and grain injuries to several crops. Management of insects through the use of Bt maize can greatly reduce insect damage, thus indirectly controlling fungal and mycotoxin contamination in the field and the accumulation of mycotoxins during storage. Despite the known effect of Bt maize on fumonisin content in the field, due to the control of insect pests, there is a lack of information regarding the accumulation of mycotoxins in Bt maize during a post-harvest period. Highlighting that, further investigations using different Bt hybrids could possibly aid in the controlling strategies. The gene expression analyses demonstrated that further studies should be conducted to determine an acceptable qualitative genetic marker for identifying fumonisin production in early stages of infection by F. verticillioides in maize.

AUTHOR CONTRIBUTIONS
LR, BC, and AD conceived the study. AD and MM provided the maize samples. MM performed field sampling and initial tests for grain damage. VB and LA conducted fumonisin analysis. FF-C conducted gene expression tests. GP performed statistical analyses. LR and GP analyzed the overall data for the manuscript. LR wrote the manuscript. All authors approved the final version of the manuscript.

FUNDING
This research was funded by FAPESP (Fundação de Amparo a Pesquisa do Estado de São Paulo).