Introduction

Bilirubin toxicity to the CNS has been extensively studied for decades and has been shown to be linked to the activation of multiple complex signal cascades, and affects potential toxic/adaptation mechanisms in the brain through gene expression modulation. Examples include oxidative stress and the antioxidant response, excitotoxicity, inflammation, intracellular trafficking, protein degradation, apoptosis, as well as bilirubin transport and bilirubin oxidization (reviewed in1).

Epigenetic processes, such as histone acetylation and DNA methylation, regulate the expression of genes through modifications of DNA structure and accessibility. These regulatory mechanisms often contribute to the onset and progression of human neurological disorders, and are altered by toxic compounds (e.g.: cocaine, alcohol)2,3,4,5,6,7,8. Indeed, histone acetylation is considered an integral part of brain development and differentiation, synaptic plasticity, learning, memory, and neuron maintenance and survival9,10,11,12. Notably, it is reported that temporal changes in gene expression by acetylation/deacetylation of gene promoters induce persistent changes in the cell (e.g. cell fate), changes in the neurological behaviour8, as well induction of excitotoxicity, calcium overload, oxidative stress, inflammation and apoptosis13, with the last five described mechanisms in hyperbilirubinemic animals and humans. This suggests the possibility of a link between the hyperbilirubinemic phenotype and the epigenetic. On this basis, we decided to investigate the effect of hyperbilirubinemia on the epigenetic control of the Cll hypoplasia.

Results

Serum bilirubin and cerebellar development

To evaluate the possible correlation between serum bilirubin and the levels of H3K14Ac, we quantified total and free bilirubin in the blood, and the Cll weight in hyperbilirubinemic pups (jj) and normobilirubinemic littermates (control: ctrl) from 2 days after birth (P2) until the adult age. At every post-natal age, the total serum bilirubin (TSB, Fig. 1A) was statistically higher in jj animals compared to ctrl (Σ8.5 lifelong, one-way ANOVA: p ≤ 0.0001, followed by Tukey post-test, p ≤ 0.001). At P2, the TSB was about of 190 µM, peaking at P17 (Σ256 µM), and stabilizing in the adulthood (Σ126 µM), (ever significantly higher than in ctrl, one-way ANOVA: p ≤ 0.0001, followed by Tukey post-test, p ≤ 0.001).

Figure 1
figure 1

Total Serum Bilirubin (TSB), calculated free bilirubin (cBf) in the blood, cerebellar weight, and Western blot analysis of the level of histone 3 acetylation (H3K14Ac) P: post-natal age in days, Adult: more than 1-year-old. Black bars jj rats, White bars: ctrls. (A) TSB (µM); (B) cBf (nM), (C) Cll weight (mg/animal). Results are expressed as mean ± S.D. of 6–15 animals each group and post-natal age. One way ANOVA followed by Tukey post-test: ***p < 0.001. (D) H3K14Ac levels in the Cll of jj animals vs. ctrl. Results are as mean ± S.D. of 3–6 animals each group and post-natal age. Unpaired t-test with Welch correction, *p < 0.05 vs. age matched ctrl.

Full size image

Free bilirubin is the moiety able to cross the blood-brain interfaces leading to neurological damage. In the absence of a reliable method for a direct quantification in the rat, free bilirubin was calculated as previously described14. Differently from TSB, the calculated Bf (cBf, Fig. 1B) level in jj pups dropped during development (P2 Σ150 nM, P9 Σ120 nM, P17 Σ35 nM, ever significantly higher than in ctrl, one-way ANOVA: p ≤ 0.0001, followed by Tukey post-test, p ≤ 0.001), felling to the levels not statistically different from those in ctrl in the adult age (adult jj Σ7 nM; One way ANOVA, followed by Tukey post-test, p > 0.05).

Cll weight (Fig. 1C) was similar in jj and ctrl littermates up to P9, becoming significantly different at P17 (Σ30% weight loss vs. age-matched ctrl, one way ANOVA followed by Tukey post-test: p < 0.001), and increasing later on (Adult: Σ40%, one way ANOVA followed by Tukey post-test: p < 0.001).

Western blot analysis of global acetylation of histone H3K14

The follow the level of H3K14Ac in the developing cerebellum of jj and controls rats by Western blot, we used the 07-353 anti-H3K13Ac antibody. At P2, no significant difference was observed in the level of H3K14Ac in the Cll of jj animals compared to age-matched ctrl (Fig. 1D) (unpaired t-test with Welch correction, p = 0.2687). The level of H3K14Ac in jj was significantly increased (1.65 ± 0.54 fold, unpaired t-test with Welch correction, p < 0.0222) at P9 and significantly decreased at P17 (0.67 ± 0.18 fold, unpaired t-test with Welch correction, p < 0.0187). In the adults there was no difference in the level of H3K14Ac between jj and ctrl (unpaired t-test with Welch correction, p = 0.4508).

ChIP-Seq analysis

To link the effect of hyperbilirubinemia on H3K14Ac with the genes controlled by this epigenetic mechanism, the 07–353 anti-H3K13Ac antibody used for Western blot analysis was also used to perform chromatin immunoprecipitation, followed by DNA sequencing (ChIP-Seq – full result available on GEO repository # GSE109145). After removal of duplicate DNA fragments and DNA fragments present in both jj and ctrl (physiological genes), 1884 unique DNA sequences were identified. Since variations in the level of histone acetylation in the promoter region positively correlate with gene transcription9,15, we focused on peaks identified by ChIP-Seq on the promoter regions (Table 1: 255 genes). As shown in Fig. 2, the functional annotation analysis of the corresponding genes16,17,18 revealed an enrichment for genes involved in CNS development (Σ45%), metabolism & homeostasis (Σ31%), signalling (Σ13%), response to stimuli & communication (Σ5%), transport (Σ5%), and binding (Σ2%).

Table 1 Full list of ChIP-Seq TSS-Promoter genes.
Full size table
Figure 2
figure 2

Biological function of the identified Chip-Seq chromatin sequences (A) GeneCodis analysis (on genes with peaks found in their TSS-promoter regions) for enriched biological functions. (B) List of the 94 (45% of the total found) genes enriched for functions related to the CNS development. In red, genes confirmed by RTqPCR. Hypergeometric p-value ever <0.00005, Corrected (FDR) Hypergeometric p-value < 0.05.

Full size image

Morphological features of the Gunn rat Cll

Since our results strongly suggested an impact of bilirubin on the genetic program of CNS maturation, we systematically followed the histological development of the cerebellum of jj rats in the attempt to interpret the genetic results. No morphological alterations between jj and ctrl were obvious at P2 (Fig. 3A,B). In both jj and ctrl animals, Purkinje cells were organized in 3–5 layers, with a round/oval shape and a reticulated cytoplasm (Fig. 3B). At P9, in spite of a conserved architecture, signs of cellular sufferance/death, microgliosis, extracellular matrix abnormalities and edema were evident in jj pups. PCs in ctrl displayed a clear definition of the plasma-membrane, cytoplasm, and nuclear areas, and a round/drop shape, and were organized in 3/1 layers. On the contrary, in jj pups, PCs were largely present in 4/2 layers, with an undefined, irregular shape. At P17, microgliosis and signs of cellular sufferance were still present in jj rats. PCs in ctrl were well differentiated, with a drop shape, and almost completely organized in a single layer, diffusely in 2/1 layers and still presenting the altered morphology described at P9 in jj. In the adult animal, the effect of Cll hypoplasia was well appreciable, with a less developed structure characterized by large spaces between the folia (Fig. 3A). Microgliosis was reduced but still present. No PC’s neurites were visible in jj rats, where PCs appeared atrophic and apoptotic (Fig. 3B).

Figure 3
figure 3

Histological finding (A) Full Cll images (scale bar 400 µm) showing the normal development (ctrl, upper series of pictures) and the progression of the Cll hypoplasia in jj animals (lower series of pictures). (B) Details (scale bar 100 µm) of the major histological alterations in the developing Cll of jj rat vs. age matched ctrl. P: post-natal age in days, Adult: more than 1 year old. *Purkinje cells (PCs); >PC’s neurites; ∆ microgliosis; [] extracellular matrix alteration; → oedema. 2–3 animals each genotype/age have been used. Miniatures: Nissl stain. Larger pictures: Haematoxylin & Eosin.

Full size image

RTqPCR analysis of selected genes

Due to the surprising percentage of enrichment for genes involved in CNS development, we decided to confirm and quantify the epigenetic control of a selected panel of genes, by assessing their expression by RTqPCR (selected genes are those in red in Fig. 2B, in which their biological functions based on the Gene Ontology analysis are indicated. RTqPCR results are in Fig. 4). Ptk2 (protein tyrosine kinase 2 beta, considered a key gene in neurite outgrowth and elongation, synapses formation, and actin reorganization19), was significantly down-regulated in P2 jj pups (Σ2 fold vs. age-matched ctrl, unpaired t-test with Welch correction, p < 0.047), normalizing thereafter. Mag (myelin-associated glycoprotein), barely detectable immediately after birth, was highly expressed in ctrl and Σ2.5 fold down-regulated in jj pups at P9 (unpaired t-test with Welch correction, p < 0.0402), reversing to a Σ1.2 fold up-regulation at P17 (unpaired t-test with Welch correction, p < 0.0306). Icam1 (intracellular adhesion molecule 1, expressed mainly by the endothelial cells forming the blood-brain barrier, involved in cell adhesion, leucocytes20 and monocytes extravasation21, and morphogenesis) was up-regulated 1.6 fold in P17 jj rats (unpaired t-test with Welch correction, p < 0.0416). Similarly, we observed a Σ2.2 fold increase (unpaired t-test with Welch correction, p < 0.0315) of Chmp1a (charged multi-vesicular body protein 1a, regulating the neural progenitor cell proliferation22). In adult jj Cll, Col4a3 (collagenase 4a3, the major structural component of the basal membrane, involved in the extracellular matrix remodeling23, providing the functional compartmentalization of the brain by clustering of growth factors, neurotransmitters/ions receptors, as well contributing to migration and differentiation24), Casp6 (caspase 6 - proliferation and morphogenesis – Fig. 2B), and Arghap4 (Rho GTPase-activating protein, inhibiting the cell motility and axon outgrowth via regulating the cytoskeleton dynamics25) were upregulated Σ2.5fold (unpaired t-test with Welch correction, p < 0.00547), Σ1.9fold (unpaired t-test with Welch correction, p < 0.0287) and Σ1.6 fold (unpaired t-test with Welch correction, p < 0.0142) respectively. No modulation of Anxa2 (annexin2), Agrn (Agrin), and Tubb2b (Tubulin2b) was detected at any post-natal age in jj rats (data not shown). Il6 (intron region segment resulting from ChIP-Seq analysis) was also investigated. In ctrl animals the Il6 level rapidly decreases from P2 to P9, stabilizing thereafter. In jj pups, a significant down-regulation of Il6 was present immediately after birth compared to ctrl animals (Σ2.9fold, unpaired t-test with Welch correction, p < 0.0315), while a 1.65 fold up-regulation was noticed at P9 (unpaired t-test with Welch correction, p < 0.0248), normalizing later on.

Figure 4
figure 4

Analysis of the expression of selected genes involved in CNS development Arghap4: Rho GTP-ase activating protein 4; Casp6: Caspase 6; Chmp1a: Charged multi-vesicular body protein 1a; Col4a3: Collagenase 4 a3; Icam1: Intracellular adhesion molecule 1; Mag: Myelin-associated glycoprotein; Ptk2: Protein tyrosine kinase 2 beta; Il6: Interleukin 6. P: post-natal age in days, Adult: more than 1-year-old. White bars: ctrl; Black bars: jj. Results are expressed as mean ± S.D. of 6 animals each genotype/age. Unpaired t-test with Welch correction, *p < 0.05; **p < 0.05; ***p < 0.005 vs. age-matched controls.

Full size image

Discussion

Cll hypoplasia is a hallmark of hyperbilirubinemia in rodents26,27,28,29, and cerebellar involvement with morphological and behavioral abnormalities has also been reported in severely hyperbilirubinemic neonates30,31,32. Inflammation and oxidative stress are considered the major mechanisms of bilirubin neurotoxicity, whereas the impact of hyperbilirubinaemia on CNS development has been only marginally envisaged, and evaluated mostly by in vitro experiments33,34.

Unexpectedly, the known inflammatory or oxidant effectors of bilirubin neurotoxicity have been not identified in our data (ChIP-Seq, followed by Gene Ontology analysis), revealing that 45% of genes displaying a Histone 3 lysine 14 acetylation are related to CNS development. Indeed, only 3 genes among all the 255 identified TSS- Promoter sequences have been previously reported in the literature for their association with hyperbilirubinemia, namely myelin28,31,32,34, tubulin35, and Icam136.

The down-regulation of Mag has been reported in in vitro studies, in agreement with the defective myelination observed both in bilirubin neurotoxicity models28,34 and neonates32. Mag down-regulation is also a known consequence of bilirubin-induced perturbation of the oligodendrocytes maturation. A possible additional link between what has been previously described and the present results is the fact that histone acetylation is a known mechanism controlling oligodendrocyte differentiation and myelin production, both in physiological CNS development and in repair processes after demyelination6,10.

Our data are in agreement with the literature also in relation to Il6, whose intron sequence was identified by ChIP-Seq analysis. Il6 is a well-known effector of bilirubin neurotoxicity and possibly linked with the reported defective myelination. In fact, apart from the possible inflammatory activity, Il6 is involved in oligodendrogenesis37,38, a process active up to P45 in rodents and 2 years in humans39, and reactivated in pathological conditions. During reactivation, injured neurons and oligodendrocytes may reactivate myelin synthesis by overexpressing Il6 and its receptor (Il6r/CD126), restoring normal behavior in injured animals10,40.

Both Mag and Il6 present a fluctuating behavior, being significantly down-regulated in the early post-natal life, and reverting thereafter to the level of age-matched controls (Fig. 4). Notably, in our work, IL6 modulation (P9) precedes Mag increase (P17), supporting the inductor role of Il6 in myelination described in the literature10,40. The fluctuating expression of Il6 and Mag (firstly up-, then down regulated), is present also for H3K14Ac levels, increasing at P9, and reverting under the level of age-matched controls at P17, and normalizing in the adult age.

The regulation of the other genes is more difficult to be analyzed since they are very new in the bilirubin field and no data are provided by literature. While we still have to confirm the role of the various genes identified in this study through methods such as gene silencing in vitro, our work suggests that the epigenetic impairment of neurodevelopmental processes in hyperbilirubinemia may be a relevant mechanism of bilirubin neurotoxicity. It is worth mentioning that Chmp1a, Arghap4, Casp6, Ptk2, Col4a3 are genes involved in key steps of brain development as proliferation, migration, morphogenesis, neurite outgrowth and elongation, synaptogenesis, extracellular matrix formation and compartmentalization, as well the pathological axonal degeneration and apoptosis observed19,22,25,41,42 in jj rats. By adding epigenetic dysregulation to the list of the mechanisms related to bilirubin-induced neuronal damage, we can confirm and expand the concept of a widespread toxic effect of the pigment on the CNS43, improving our understanding of the cellular and molecular mechanisms of bilirubin induced damage to CNS.

Materials and Methods

Animals

Gunn rats (Hds Blue:Gunn-UDPGTj, P2, 9, 17; P ± 1 day. Adult = more than 1 year old) were obtained from the SPF animal facility of CBM S.c.a.r.l. (AREA Science Park, Basovizza). Ages were selected based on previous evidence26,44. Animals were housed in a temperature-controlled environment (22 ± 2 °C), on a 12 hours light/dark schedule, and ad-libitum access to food and water. The study was approved by the animal care and use committee of the CBM Scarl and the competent Italian Ministry. All procedures were performed according to the Italian Law (decree 87-848) and European Community directive (86-606-ECC). Maximal effort to minimize the number of the animals used and their sufferance was done.

TSB, cBf and Cerebellum weight quantification

Serum and Cll were collected as previously described26,45. In brief, blood samples were collected during the sacrifice (decapitation under urethane anaesthesia 1.0–1.2 g/kg IP) and centrifuged at 2000 rpm, 20 min RT. Total serum bilirubin (TSB) was quantified by the diazo reaction, as previously described26. Free bilirubin was calculated (cBf) by applying the formula and the albumin-bilirubin dissociation constants for Gunn pups detailed in literature14. Cerebellum was dissected immediately after the sacrifice, and the weight recorded by a precision balance.

Western blot analysis of the levels of H3K14Ac

Western blot was performed as previously described44,45. In brief, Cll were mechanically homogenized by glass-glass Dounce (in 0.25 M sucrose, 40.2 mM KH2PO4, 9.8 mM K2HPO4, 1 mM EDTA, 0.1 mM DTT, pH 7.4), and total protein concentration quantified by the Bicinchoninic Acid Protein Assay following the supplier instruction (B-9643 and C2284, Sigma, Missouri, USA). 25 μg of Cll whole extract proteins were denatured (10% of β-mercaptoethanol -Sigma Chemical, St. Louis, MO, USA, plus 5 min boiling), separated by 12% SDS-PAGE by electrophoresis in a Hoefer SE 250 System (Amersham BioSciences, UK), and electro-transferred onto immune-blot PVDF membranes (0.2 μm; Whatman Schkleicher and Schuell, Dassel, Germany) at 100 V for 60 min (Bio-Rad Laboratories, Hercules, CA, USA). Efficiency of the transfer was assessed by lack of Coomassie blue coloration of the gel after blotting, and Ponceau staining of the PVDF membrane (both chemicals: Sigma, St. Louis, MO, USA). After blocking (1.5 hrs, RT in blocking solution: 3% defatted milk in 0.2% Tween 20; 20 mM Tris-HCl pH 7.5; 500 mM NaCl), membranes were incubated O/N at 4 °C with the polyclonal anti-acetyl histone H3 (lys14) antibody (07-353, Merck Millipore, Temecula, CA, USA; final concentration 0.7 μg/mL). The day after, membranes were washed 3 × 5 min in blocking buffer, then incubated 2hrs with the secondary antibody anti-rabbit IgG peroxidase (Dako, Agilent Technologies, Santa Clara, CA, USA, final concentration 0.0625 μg/mL) in blocking solution. The signal was revealed by chemiluminescence (ECL-Plus Western blotting Detection Reagents, GE-Healthcare Bio-Science, Italy) and visualized on X-ray films (BioMax Light, Kodak Rochester, NY, USA). The results were normalized vs. the actin signal, visualized incubating the same membrane used for revealing the H3K14Ac with the anti-actin antibody A2066 (sigma- Chemical, St. Louis, MO, USA; final concentration 0.07 μg/mL, MW 42KDa). Bands intensity was quantified by the Scion Image software (GE Healthcare Europe GmbH, France).

ChIP-Seq analysis

The 07-353 anti-H3K13Ac antibody used for Western blot analysis was also used to perform chromatin immunoprecipitation, followed by DNA sequencing (ChIP-Seq – full result available on GEO repository # GSE109145). Chromatin immunoprecipitation (ChIP) was performed following the Magna ChIPTM G Tissue Kit (#17-20000, Merck Millipore, Temecula, CA, USA) procedure and applying the same Ab used in Western blot. Cll tissue (60 mg) was homogenized, DNA sheared (average size of 100–400 bp, by Sonopuls HD 3100, Bandelin, Germany, sonicator. Power 50%, 15″ × 18 cycles, 10″ pause between each cycle, on ice), cross-linked with 1% formaldehyde (5′, RT), and protein-DNA complexes immune-precipitated (5 μL, 07-353 Ab, Merck Millipore, Temecula, CA, USA) by G magnetic beads on the magnetic rack (LSKMAGS08 Pure ProteomeTM Magnetic Stand, Merck Millipore, Temecula, CA, USA). Protein-DNA crosslink was reversed (proteinase K, 62 °C, 2 h; plus 95° C × 10′), and DNA stored at −20 °C until use. As suggested by the manufacturer, the efficiency and specificity of the ChIP procedure were assessed by Western blot, and Real Time PCR (RTqPCR). Samples were quantified by Quant-iTTM PicoGreen® dsDNA Kits (Thermo Fisher Scientific, Waltham, MA, USA), according to manufacturer’s instruction.

Libraries were prepared by using the NEBNext® UltraTM II DNA Library Prep Kit from Illumina® (E7645, New England BioLabs®Inc, MA, USA), following the manufacturer’s instructions starting from 10 ng of fragmented DNA. After end repair and adaptor ligation, adaptor-ligated DNA clean-up (without size-selection, Agencourt AMPure XP magnetic beads, Beckman Coulter Life Sciences, CA, USA), library enrichment (98°C × 30 sec; 98°C × 10 secplus 65°C × 75 min × 10 cycles; 65°C × 5 min, in a Bio-Rad thermal cycler, Bio-Rad, Richmond, CA, USA), and PCR clean up (Agencourt AMPure XP magnetic beads, Beckman Coulter Life Sciences, CA, USA), the libraries were quantified using the PicoGreen fluorescent dye, as reported above, and stored at −20 °C. Before sequencing, libraries were denatured and diluted to a final concentration of 15 pM with 10% PhiX (Illumina, New England BioLabs®Inc, MA, USA) control. Paired-end sequencing was performed using the MiSeq reagent kit v3 2 × 150 in the Illumina® MiSeq® system (Illumina, San Diego, CA, USA). A total of 4 P9 jj Cll (2 runs) and 3 P9 control Cll (1 run) were used. Reads were mapped to the Rattus norvegicus (rn4) genome using bowtie246. Duplicate reads were filtered. The quality of the sequences was evaluated using fastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Peaks were called using MACS247 and annotated using HOMER software48. Functional enrichment study was determined using GeneCodis (http://genecodis.cnb.csic.es/, hypergeometric test, FDR corrected)16,17,18.

Histology and morphometric analysis

Immediately after animals sacrifice, the brain was removed from the skull and fixed in 4% formalin buffered solution (4% formaldehyde 37%, 33 nM NaH2PO4, 46 mM Na2HPO4), then embedded in paraffin. Sagittal sections of the brain (3–5 μm) were obtained by a microtome (Microm-hm 340 e- BioOptica, Milan, Italy), affixed on the glass slides and dried at 60 °C for 1 hour. Hematoxylin and eosin stain (H&E) was performed by a Leica ST5020 Multistainer (Leica Microsystem, Milan, Italy). Cresyl violet (Nissl) staining was performed manually on hydrated sections (xylol 3 × 5 min; 100% ethanol 2 × 2 min; 95% ethanol 2 × 2 min; 80% ethanol 1 × 2 min; 70% ethanol 1 × 2 min; H2O 2 × 5 min) by incubating the slices for 1 hr in cresyl violet solution (0.1% cresyl violet powder, 10 drops glacial acetic acid in H2OmQ). After washing (twice H2OmQ), differentiation (75% ethanol, 95% ethanol plus 5% chloroform, 3 drops glacial acetic acid) and dehydration (100% ethanol 2 × 5 min; xylol 2 × 5 min), slices were mounted (Eukitt 03989, SIGMA Aldrich). Pictures were collected by a D-Sight plus image digital microscope & scanner (Menarini Diagnostics, Firenze, Italy). Histology was read by 3 independent pathologists, blinded to experimental design.

RTqPCR on selected genes

RTqPCR was performed as previously described26,43. Total RNA extraction (Eurogold RNA Pure reagent, Euroclone, Milan, Italy) and retro-transcription (1 μg RNA, High Capacity cDNA Reverse Transcription Kit, Applied Biosystems, Monza, Italy) were performed following the manufacturer instruction in a thermal cycler (Gene Amp PCR System 2400, Perkin-Elmer, Boston, MA, USA) at 25 °C for 5 min, 37 °C for 120 min, and 85 °C for 5 min. The final cDNA was stored at 20 °C until use. Primers were designed using the Beacon designer 8.1 software (Premier Biosoft International, Palo Alto, CA, USA) on rat sequences available in GenBank (Table 2). RtqPCR was performed in an iCycler iQ thermocycler (Bio-Rad Laboratories, Hercules, CA, USA) in presence of 25 ng of cDNA, sense and antisense gene-specific primers (250 nM each), in SSoAdvance SYBER green supermix (Bio-Rad Laboratories, Hercules, CA, USA). Amplification protocol was 95 °C × 3 min, 40 cycle of 95° C × 20 sec; 60 °C × 20 sec and 72 °C × 30 sec, followed by 72 °C × 5 min. Melting curve analysis was performed to assess product specificity. The relative quantification was made using the iCycler iQ Software, version 3.1 (Bio-Rad Laboratories, Hercules, CA, USA) by the Pfaffl modification of the ΔΔCT equation, taking into account the efficiencies of the individual genes49,50. The results were normalized to the housekeeping genes and the levels of mRNA were expressed relative to a reference sample50,51.

Table 2 Primers specification.
Full size table

Statistics

The statistical analysis was performed by GraphPad InStat for Windows (GraphPad Software, Inc, La Jolla, CA, USA). The ANOVA test, followed by Tukey-Kramer multiple comparison tests, was used to analise TSB, cBf, and Cll weight during the development. The unpaired two-tailed Student’s t-test, based on unequal variance, was applied to evaluate the difference between jj and controls at the same age (Western blot, RTqPCR). All data are expressed as mean ± S.D. of multiple biological repetition. A p-value lower than 0.05 was considered statistically significant.