Targeted Metabolomic Analysis of a Mucopolysaccharidosis IIIB Mouse Model Reveals an Imbalance of Branched-Chain Amino Acid and Fatty Acid Metabolism

Mucopolysaccharidoses (MPSs) are inherited disorders of the glycosaminoglycan (GAG) metabolism. The defective digestion of GAGs within the intralysosomal compartment of affected patients leads to a broad spectrum of clinical manifestations ranging from cardiovascular disease to neurological impairment. The molecular mechanisms underlying the progression of the disease downstream of the genetic mutation of genes encoding for lysosomal enzymes still remain unclear. Here, we applied a targeted metabolomic approach to a mouse model of PS IIIB, using a platform dedicated to the diagnosis of inherited metabolic disorders, in order to identify amino acid and fatty acid metabolic pathway alterations or the manifestations of other metabolic phenotypes. Our analysis highlighted an increase in the levels of branched-chain amino acids (BCAAs: Val, Ile, and Leu), aromatic amino acids (Tyr and Phe), free carnitine, and acylcarnitines in the liver and heart tissues of MPS IIIB mice as compared to the wild type (WT). Moreover, Ala, Met, Glu, Gly, Arg, Orn, and Cit amino acids were also found upregulated in the liver of MPS IIIB mice. These findings show a specific impairment of the BCAA and fatty acid catabolism in the heart of MPS IIIB mice. In the liver of affected mice, the glucose-alanine cycle and urea cycle resulted in being altered alongside a deregulation of the BCAA metabolism. Thus, our data demonstrate that an accumulation of BCAAs occurs secondary to lysosomal GAG storage, in both the liver and the heart of MPS IIIB mice. Since BCAAs regulate the biogenesis of lysosomes and autophagy mechanisms through mTOR signaling, impacting on lipid metabolism, this condition might contribute to the progression of the MPS IIIB disease.


Introduction
Lysosomal storage diseases (LSDs) are a group of hereditary metabolic disorders which are inherited in an autosomal recessive manner, except for some that are X-linked [1]. They are caused by mutations in gene encoding for enzymes involved in the degradation of various macromolecules. The substrate(s) of the defective enzyme(s) building up over time leads to excess cellular storage of this material, which results in altered cellular functions, tissue damage, and organ dysfunctions [2]. As well as the accumulation of undigested or partially digested macromolecules in lysosomes, pathological

Metabolomic Profiles of Heart and Liver of MPS IIIB Mice
A targeted metabolic analysis was performed to evaluate changes in the concentration of Ala, Val, Xle (Leu and/or Ile), Met, Phe, Tyr, Asp, Glu, Gly, Orn, Cit, and Arg amino acids and saturated, unsaturated, hydroxylated and branched acylcarnitines in the heart and liver of NAGLU −/− mice compared to age-matched wild-type (WT) mice.
Firstly, in order to test the power of the comparative studies performed, we carried out Spearman's rank correlation analysis to evaluate sample reproducibility within the groups based on nine animals for each genotype for each tissue. As expected, the overall analysis displayed a positive intragroup correlation and a negative intergroup correlation in both the analyzed tissue-metabolomes (Figure 1a,b). Moreover, the datasets, corresponding to the 48 analyzed metabolites (Tables S1 and S2), were processed by principal component analysis (PCA) to evaluate the difference rate and variance occurring between the genotype groups (NAGLU −/− and WT) in the metabolic profiles characterizing the heart and liver. The PCA score plot revealed a clear separation between the analyzed data groups for each tissue (heart: principal component (PC) 1 variance 40.8%, PC2 variance 21.9%; liver: PC1 variance 44.0%, PC2 variance 9%). Outliers were not found in the heart and liver metabolomes (Figure 2a,b).
To further understand the alterations in NAGLU knockout-related metabolites in the heart and liver, we used a heatmap to visually represent the differentially abundant metabolites in NAGLU −/− and WT genotypes for each tissue; the concentrations of the metabolites were ranked by the t-test (p < 0.05 cut-off). The results showed correct sample group clusterization, and the dendrogram structure, using Euclidean distance, resulted in two main clusters, relating to the compared samples (NAGLU −/− and WT) and metabolites, respectively (Figure 3a,b).
The heatmaps exhibit distinct patterns of metabolites between MPS IIIB mice and age-matched WT mice in both the analyzed tissues (heart and liver). Thus, the detected metabolic abnormalities are likely to be associated with the MPS IIIB phenotype, allowing the identification of a pathological metabolomic signature of the MPS IIIB disease. Hierarchical cluster analysis and heatmap of the metabolites in the (a) heart and (b) liver tissues from NAGLU −/− and WT mice. The concentrations of the metabolites from NAGLU −/− and WT heart and liver were normalized according to protein tissue contents, log(2) transformed, and Pareto scaled. The color code in the heatmap represents the relative metabolite abundance: red and blue colors mean increased and decreased levels of each metabolite in NAGLU −/− versus WT, respectively.

Heart-and Liver-Specific Metabolomic Changes in MPS IIIB Mice
With a univariate analysis, we detected significant differences in bot amino acid and acylcarnitine levels in the liver and heart of MPS IIIB mice when compared to WT.

Pathway Analysis and Discriminant Metabolites Identification
In order to identify the principal impaired metabolic processes affected by the absence of NAGLU in the heart and liver, the quantitative data obtained by the univariate analysis were used to explore dysregulated metabolic pathways. The comparative analysis of NAGLU −/− versus the control revealed the deregulation of three main metabolic pathways in the heart: the "oxidation of branched-chain fatty acids", "beta oxidation of very-long-chain fatty acids", and "valine, leucine and isoleucine degradation" pathways ( Figure 8a). In the liver, the absence of NAGLU seems to cause dysregulation of the following pathways: "urea cycle", "glycine and serine metabolism", "phenylalanine and tyrosine metabolism", "glucose-alanine cycle," "arginine and proline metabolism", "aspartate metabolism", "valine, leucine and isoleucine degradation" and "alanine metabolism" (Figure 8b). Finally, in order to identify the metabolites most closely associated with NAGLU deficiency, a supervised classification through a partial least squares-discriminant analysis (PLS-DA) was performed (Figure 9a,b, upper panels). The supervised analysis allowed us to define the most discriminant metabolites between NAGLU −/− versus WT in heart and liver tissue, extracting a combination of variables that can be considered to have predictive value. Among them, the critical metabolites in heart and liver tissue were then estimated according to their Variable Importance in Projection (VIP) score. From the PLS-DA, the differentiation between NAGLU −/− and WT genotypes occurred mainly along component 1, measured as 34.2% and 44.1% in the heart and liver tissues, respectively. The most important features identified by the PLS-DA and the VIP score plot (threshold > 1.5) included six metabolites in the heart, C5:1, C6:1, C4, C18:1, C16:1, and C10:2, and eight key metabolites in the liver, C4, C12OH, C3, C6DC, C4OH, C8, C6, and C4DC, as potentially closely associated with NAGLU deficiency (Figure 9a,b, lower panels).

Discussion
This study represents the first metabolomics-based investigation of the heart and liver of MPS IIIB-affected mice. The results obtained demonstrate that lysosomal HS storage triggers a metabolic imbalance in these organs, which is likely to contribute to disease progression. Given the resemblance of the MPS IIIB mouse model to the human disease, we anticipate that the metabolic impairment observed in the liver and heart of mice may also occur in patients with MPS IIIB, thus accounting for clinical manifestations affecting these organs.
The present study shows that NAGLU −/− mice had increased levels of BCAAs (Leu, Ile, and Val), Phe and Tyr amino acids, free carnitine, and acylcarnitines in both liver and heart tissue. The augmented levels of BCAAs concomitant with the accumulation of short-chain acylcarnitines (C3, C5), involved in BCAA metabolism [33,34], suggest an impaired catabolism of BCAAs in the liver and heart of MPS IIIB-affected mice. BCAA metabolism occurs in the liver; however, it is also active in other tissues, including the heart, adipose tissues, kidney, and skeletal muscle [35]. Furthermore, the increased level of most of the acylcarnitines in the liver also suggests a deregulation of the β-oxidation of long-chain fatty acids in MPS IIIB mice, in agreement with the primary function of acylcarnitines of serving as carriers to transport long-chain fatty acids to mitochondria for subsequent β-oxidation, thus providing energy for cell activities.
In the liver tissue of MPS IIIB mice, as well as the amino acids discussed above, we also found elevated levels of the Ala, Met, Glu, Gly, Arg, Orn, and Cit amino acids. These results suggest a deregulation of the glucose-alanine cycle and the urea cycle in the liver of MPS IIIB mice, as well as the altered BCAA and other amino acid metabolisms. These results are consistent with a previous metabolite analysis of the liver of MPS I mice showing decreased levels of lipids, simple carbohydrates, and nucleotides, whereas most amino acids, amino acid derivatives, dipeptides, and urea resulted in being increased, suggesting an increase in protein catabolism [36]. Although some results have been reported in previous global metabolomic profiling of MPS IIIB mice and patients, where a reduction of all key amino acids was observed in the serum of affected subjects as compared to controls, probably due to either differences in the methodological strategies or metabolic differences in the tested tissues [29,37], our findings are definitely in agreement with previous metabolic phenotyping performed on the urine of MPS III (A-D) and MPS VI patients [31,32].
The enhancement of protein catabolism in MPS IIIB mouse tissue at least in part may fulfill the intermediary metabolism in response to carbohydrate and lipid deficiencies. Indeed, inside the cells, BCAAs can be stored in amino acid pools, integrated into proteins, or shuttled to mitochondria for oxidation. They serve as signaling molecules regulating glucose and lipid metabolism as well as protein synthesis through the phosphoinositide 3-kinase/protein kinase B/mammalian target of the rapamycin (PI3K/AKT/mTOR) signal pathway [38,39]. In particular, the final metabolites produced in the catabolism of BCAAs provide an energy supply. The terminal metabolites of the catabolism of Leu are acetoacetate and acetyl-CoenzymeA (CoA), which are ketogenic; Ile yields to propionyl-CoA and acetyl-CoA, which are both glucogenic and ketogenic, and Val produces succinyl-CoA, which is glucogenic. Acetyl-CoA can enter the tricarboxylic acid cycle (TCA), while succinyl-CoA and acetoacetate are TCA intermediates. These findings are reinforced by previous studies on correlated disease models. In MPS I, MPS IIIB, MPS VII, Niemann-Pick type A/B, and infantile neuronal ceroid lipofuscinosis mouse models, a significantly reduced adiposity was detected, suggesting that the reduction of lysosomal stored material recycling results in an energy imbalance with an overall impact on autophagy levels and metabolism [36,40,41].
Multiple reports show that an impairment of autophagy by engulfed lysosomes drives the progression of the disease in LSDs [3][4][5][42][43][44][45]. Autophagy is a cellular process that targets cytoplasmic material such as long-lived proteins and organelles to lysosomes for degradation into simpler metabolites [46]. Autophagy activation is regulated by the mechanistic target of rapamycin complex 1 (mTORC1), whose activity is dependent on a variety of inputs from energy, growth factors, oxygen, and nutrients [47,48]. Upon activation, mTORC1 promotes anabolic processes while blocking autophagy and other catabolic processes. Experimental studies, showing that amino acid withdrawal from cells suppresses mTORC1 signaling and triggers autophagy, have highlighted the regulatory role of cellular amino acids in mTORC1 activity [49]. These findings are consistent with the identification of a group of small GTPases, Rags, as central mediators of amino acid signaling for mTORC1 [50,51]. Recently, a definite role has emerged for the amino acid/mTORC1 pathway centered at the lysosomes in the control of autophagy [52].
Amino acids promote the localization of mTORC1 at the lysosomes and induce its activation through both Rag-dependent and Rag-independent pathways [51,53,54]. Lysosomal localization and the subsequent activation of mTORC1 involves two concerted events: sensitization by priming amino acids and subsequent activation by activating amino acids [55]. The BCAAs Leu, Ile and Val belong to the group of activating amino acids, and Leu is among the main positive regulator of mTORC1 signaling. When activated, mTORC1 stimulates protein, lipid, and nucleotide biosynthesis by phosphorylating several downstream targets, while inhibiting cellular catabolism through the repression of autophagy [56]. Hyperactivation of mTORC1 associated with the suppression of autophagy has been recognized as the main pathological mechanism through which lysosomal storage impairs chondrocyte function and bone growth in MPS VII [45]. On the other hand, defective autophagic flux has been demonstrated in the brain of mouse models of multiple sulfatase deficiency (MSD), MPS IIIA, and Nieman-Pick type C (NPC) disease, and in human fibroblasts from patients affected by MPS VI, Gaucher disease, Fabry disease, and other LSDs [57]. Inhibitors of mTORC1, which stimulate autophagy, have been successfully tested in models of some LSDs as a potential therapeutic intervention [4,5]. Thus, the imbalance of BCAAs and other amino acid levels in the NAGLU −/− mouse heart correlates well with our previous findings demonstrating an impaired lysosomal autophagic flux associated with cardiac phenotype in MPS IIIB mice [9].
Notably, lysosomal mTORC1 and Rag GTPases have been shown to regulate the localization of the transcription factor EB (TFEB) involved in the lysosomal biogenesis and autophagy, by phosphorylating it on several serine and threonine residues [58]. When phosphorylated by mTORC1, TFEB is retained in the cytoplasm. On the contrary, mTORC1 inactivation induces the dephosphorylation of TFEB with consequent relocalization to the nucleus, where TFEB promotes the expression of lysosomal and autophagic-related genes, thus contributing to the promotion of autophagic flux, the biogenesis of new lysosomes, and the clearance of storage material [59]. The amino acid pathway is the main orchestrator in regulating the activity of TFEB, and its impaired signaling has been shown to highly contribute to the pathogenesis of LSDs [60]. Thus, we can speculate that the altered levels of BCAAs, other amino acids, and acylcarnitines in the heart and liver of MPS IIIB mice are likely to deregulate mTORC1 and its downstream target TFEB, thus perturbing the autophagic process in these tissues.
An impairment of autophagy is usually associated with the dysregulation of mitochondrial quality control pathways and the accumulation of damaged mitochondria within the cells. In LSDs, reduced autophagic flux has been shown to lead to the persistence of dysfunctional mitochondria with a consequent increase in reactive oxygen species production, altered calcium homeostasis, and enhanced pro-apoptotic signals [43,61,62]. Moreover, many studies have shown that defective mitochondrial activity contributes to neuroinflammation and cognitive defects in different MPS III mouse models [63,64]. These findings on the diffuse presence of defective mitochondria in many MPS diseases support our results relating to impaired β-oxidation-related pathways in MPS IIIB mice.
In conclusion, the metabolic phenotype observed in the liver and heart of MPS IIIB mice supports the view that autophagy and mitophagy deregulation may play a prominent role in the pathogenesis of the MPS IIIB disease. This may be relevant for the development of effective therapies for curing MPS IIIB.

Animals
The animal model of MPS IIIB (NAGLU knockout mice, NAGLU −/− ) was created by Prof. Elizabeth Neufeld, UCLA, by interrupting exon 6 of the NAGLU gene on the C57/BL6 background [21]. NAGLU −/− and WT mice were genotyped by PCR [9]. Mice (4 per cage) were maintained on a 12 h light/dark cycle, identical temperature conditions (21 ± 1 • C) and humidity (60 ± 5%), and free access to normal mouse chow [65,66]. The experimental protocols were carried out following ARRIVE guidelines and EU Directive 2010/63/EU for animal experiments. All mouse care and handling procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the Biotechnology Center, AORN Cardarelli (Naples, Italy), project identification code C6AC4.4 (19 February 2016). The sacrifice of the animals was performed in the morning to avoid variation in sample collection due to time [67,68]. Firstly, 8-month-old animals were euthanized, and then the thoracic and abdominal cavities were opened to expose the heart and liver. Organs were perfused by PBS in order to eliminate blood residue in the vessels of the collected tissues. The organs, liver and heart, were rapidly removed, washed with ice-cold PBS, and stored at −80 • C.

Extraction and Derivatization of the Metabolites
Metabolite extraction from the heart and liver tissue of nine MPS IIIB and nine WT male mice for liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis was performed according to published methods, with little modification [69,70]. The frozen tissues were homogenized in 1000 µL of 50:50 cold methanol/0.1 M hydrogen chloride (HCl) (Sigma-Aldrich, St. Louis, MO, USA). Mechanical homogenization was performed using a TissueLyser LT homogenizer (Qiagen, Duesseldorf, Germany) at high speed, shaking in 2 mL microcentrifuge tubes and with stainless-steel beads. The mixture was centrifuged at 13,000 rpm for 60 min at 4 • C to isolate metabolites from the protein pellet. Proteins were extracted from the pellet to estimate protein concentration, as explained [71]. The supernatant, containing metabolites, was treated with potassium hydroxide (Sigma-Aldrich, St. Louis, MO, USA) until it reached pH 7-8, then centrifuged at 13,000 rpm for 40 min, cleaned by the pellet, and dried under nitrogen. The dried metabolite mixture was resuspended in 1.0 mL of methanol and analyzed by tandem mass spectrometry to determine the contents of amino acids (AAs) and acylcarnitines (ACs). The metabolomic platform was developed to perform the fast identification of biomarkers of inherited metabolomic diseases and was adapted for the purpose of the present analytical work. The chosen analytical method (neutral loss scan and precursor ion scan) is ideal for the simultaneous, fast, and quantitative measurement of a family of metabolites, such as amino acids or acylcarnitines [68,71,72].

Metabolite LC-MS/MS Measurements
A group of metabolites constituted of 12 AAs and 35 ACs was measured employing a targeted MS-based metabolomic platform [73]. In detail, tandem mass spectrometric analysis was carried out with an API 4000 triple quadrupole mass spectrometer (Applied Biosystems-Sciex, Toronto, Canada) coupled with an 1100 series Agilent high-performance liquid chromatography system (Agilent Technologies, Waldbronn, Germany).
MS Quantitative analysis of the data was performed with ChemoView v1.2 software through the comparison of the analyte and its corresponding internal standard areas. Using stable isotope-labeled internal standards improves correction for the matrix effect. The MS method's accuracy and precision were estimated by analyzing quality control (QC) samples provided by ERNDIM (European research network for evaluation and improvement of screening, diagnosis, and treatment of inherited disorders of metabolism, Manchester, UK; www.erndimqa.nl), prepared at low, mid, high, and very high concentrations.

Metabolite Statistical Analysis and Feature Selection
The NAGLU −/− and WT mouse tissue metabolite measurements were processed by univariate and multivariate analysis. Metabolite concentrations, expressed as µM, were normalized to the protein content of the analyzed tissues and referred to as 1 mg of total protein content [74].
Univariate statistics were calculated with GraphPad Prism 8.0, and the results are presented as the mean ± standard error of the mean (SEM). The statistical significance of the difference in metabolite tissue concentrations between genotype conditions (NAGLU −/− and WT) was evaluated by parametric (unpaired t-test) or non-parametric (Mann-Whitney test) tests when data failed the Shapiro-Wilk normality test. Two outlier samples were removed from the metabolome analysis.
Multivariate statistical analysis and metabolite set enrichment analysis were performed using MetaboAnalyst 4.0 (http://www.metaboanalyst.ca) [75]. The normalized metabolic dataset was log(2)-transformed and scaled according to the Pareto scaling method. Dataset homogeneity was evaluated by principal component analysis (PCA) to get an overview of the data and identify potential severe outliers. Partial least squares-discriminant analysis (PLS-DA) was used as a supervised method to maximize the covariance between the independent variables (metabolites) and the corresponding dependent variable Y for predictive feature identification. The variable importance on projection (VIP) was estimated. VIP is a weighted sum of squares of the PLS loadings taking into account the amount of explained variation in each dimension. Finally, the interpretation of the metabolomic dataset was carried out using metabolite set enrichment analysis by MetaboAnalyst 4.0 (http://www.metaboanalyst.ca). The Human Metabolome Database was used to assign the correctly numbered ID (HMDB ID) to each differentially abundant metabolite.

Conclusions
In summary, targeted metabolomic profiling demonstrates elevated levels of BCAAs, other amino acids, and acylcarnitines in the liver and heart of the MPS IIIB mouse model. We also identified impaired metabolic pathways such as BCAA catabolism and phenylalanine and tyrosine metabolism in both the liver and heart; branched-and long-chain fatty acid β-oxidation in the heart; and mainly the urea cycle and the glucose-alanine cycle in the liver of MPS IIIB mice. We speculate that the secondary accumulation of BCAAs and other amino acids in the liver and the heart of MPS IIIB mice affects autophagy, which in turn leads to altered cellular metabolism, thus contributing to the progression of the disease.