Morquio a Syndrome: Identification of Differential Patterns of Molecular Pathway Interac-Tions in Bone Lesions

1. Introduction

Mucopolysaccharidosis type IVA (MPS IVA; Morquio A syndrome) (OMIM 253000) is a rare autosomal recessive lysosomal storage disease (LSD) caused by deficiency of a hydrolase enzyme, N-acetylgalactosamine-6-sulfate sulfatase (GALNS, EC 3.1.6.4) [1,2], which results in the accumulation of glycosaminoglycans (GAGs) such as keratan sulfate (KS) and chondroitin-6-sulfate (C6S) in multiple tissues, primarily bone, cartilage, heart valves, and cornea, leading to devastating systemic skeletal dysplasia with incomplete ossification and consequent growth impairment [3,4].

The prevalence of MPS IVA ranges from 1 per 76,000 births in Northern Ireland to 1 per 640,000 births in Western Australia [5]. Excessive accumulation of KS and C6S in bone, cartilage, and their extracellular matrix (ECM) causes a unique skeletal dysplasia in patients with MPS IVA. Although most MPS IVA patients generally appear healthy at birth, skeletal deformities tend to manifest in the first few years of life. Common features in severe MPS IVA include skeletal dysplasia with short neck and trunk, cervical spinal cord compression, tracheal obstruction, pectus carinatum, joint laxity, kyphoscoliosis, coxa valga, and genu valgum [6,7]. MPS IVA patients often become severely disabled and require a wheelchair in adolescence. Marked non-skeletal manifestations, including respiratory disease, spinal cord compression, cardiac disease, impaired vision, hearing loss, and dental problems, are also described in MPS IVA patients [8,9]. Untreated patients with severe disease typically die in the second or third decade of life due to respiratory problems, cervical spinal cord complications, or heart valve disease [7,10].

While musculoskeletal manifestations predominate in MPS IVA, there are certain similarities in the underlying processes with more common bone diseases. Alterations observed at the cartilage level include abnormal structure and/or degradation of the intra- and extracellular matrix, organelle malfunction, vacuolated cells, and cell death, leading to reduced numbers of chondrocytes, erosion of articular cartilage, and bone growth impairment. These changes are associated with an inflammatory response and altered gene and protein expression. This response is similar to that seen in arthritis (AR), and osteoarthritis [11,12].

Although the clinical presentation of MPS IVA is well documented, the mechanisms that account for the associated skeletal pathology are not completely elucidated. While some aspects of the pathology are described [13,14], analyses are complicated by difficulties in acquiring the necessary samples, which require invasive bone biopsy techniques. In recent years, proteomics approaches using human bone cells have been used to identify biomarkers and/or bone lesion patterns in other lysosomal diseases [15,16]. However, until now, no such studies have analyzed bones from MPS IVA patients or animal models. Using two proteomic technologies, we investigated differential patterns of molecular pathway interactions by analyzing femur bones from age- and sex-matched MPS IVA and wild-type (WT) mice. The mice examined were aged 6–8 weeks, which represents an intermediate stage of disease progression in which KS deposits are already detectable in the bone. Only male mice were selected, as the hormonal changes that occur in female mice represent potential confounding factors.

2. Results

Femur bones from untreated UNT mice [17] and from WT mice were extracted and analyzed using two precise and sensitive proteomic techniques: DDA (data-dependent analysis) and SWATH (sequential window acquisition of all theoretical fragment ion spectra), an information-dependent acquisition (IDA) technique.

DDA identified 1056 and 1227 proteins in UNT and WT mice, of which 129 were exclusive to the UNT group and 300 to the WT group (Supplementary Figure S1a and Supplementary Table S1). SWATH-MS identified 900 proteins and quantified 618 proteins, ultimately identifying 184 differentially expressed proteins (DEPs) (Supplementary Figure S1b and Supplementary Table S2).

We initially focused on proteins identified by DDA for which SWATH-MS revealed differential expression between UNT and WT mice.

2.1. Section I: Dysregulation of Metabolic Pathways Implicated in Energy Production

Cells produce ATP using 3 main interconnected metabolic pathways: glycolysis, the Krebs cycle (TCA cycle), and oxidative phosphorylation (OXPHOS).

We identified around 100 proteins related to energy production (Figure 1a). String analysis revealed 3 large clusters of proteins related to glycolysis, gluconeogenesis, TCA cycle, amino acid metabolism, fatty acid β-oxidation, and the electron transport chain (Figure 1b). Together, SWATH-MS and DDA identified 77 proteins. In comparison, DDA alone identified 22 (Figure 1c).

DDA identified 15 proteins involved in glycolysis (Supplementary Table S1). SWATH-MS showed that 8 were significantly upregulated in the UNT group (p <0.05). Aldoa, Eno 3, and Ldha showed a 3-fold increase relative to the WT group (Figure 2a; Supplementary Table S3, Section 1.1). Only Eno 1 was downregulated more than 3-fold in UNT mice. Of 14 proteins involved in the TCA cycle (Table S1) that DDA identified, Pdhb was downregulated in UNT versus MT mice, while Idh3a and Mdh2 were upregulated with a fold change (FC) >3 (Figure 2b; Supplementary Table S3, Section 1.2).

Mitochondria are key players in cellular energy regulation and can produce most of the required adenosine triphosphate molecules via the oxidative phosphorylation system (OXPHOS). DDA identified several proteins involved in OXPHOS (Supplementary Table S1): 30 corresponded to complex I, including Ndufa6, Ndufa13, Ndufb9, Ndufc2, and Ndufv, all of which were upregulated with an FC >3 in UNT versus WT mice (Figure 2c). Ndufb4 was detected in the UNT group only. Three proteins corresponding to complex II, Sdha and Sdhb, were significantly upregulated in UNT, Sdhb with an FC >3 (Figure 3a). Of the 5 complex III proteins identified by DDA analysis, only 1 (Uqcrfs1) was significantly upregulated in UNT versus WT mice. DDA identified 9 proteins corresponding to complex IV (Supplementary Table S1). Of these, 4 were significantly upregulated in UNT mice, with an FC >3 for Cox5a, Cox6a2, and Cox7a2 (Figure 3b). Mitochondrial (mt) ATP synthase or Complex V uses energy from the electrochemical proton gradient to phosphorylate ADP, producing ATP. DDA identified 14 proteins (Table S1) related to this process: 6 were significantly upregulated in the UNT group (Atp5f1d, Atp5mf, and Atp5po with an FC >3), while Atp5mk was downregulated (Figure 3c; Supplementary Table S3, Section 1.3).

Of proteins related to β-oxidation that were identified by DDA (Supplementary Table S1), Acadvl and Fabp3 were significantly upregulated in UNT with a FC > 3.0, while Eci1 was downregulated (Figure 3d; Supplementary Table S3, Section 1.4).

In summary, the upregulation of several proteins involved in energy production in UNT mice suggests high energy production in this group.

2.2. Section II: Proteins Related to Mitochondria Function: Oxidative Stress, Voltage-Dependent Channels, Apoptosis, Compensatory Mechanisms of Oxidative Stress, and DNA Damage

DDA identified 43 proteins related to oxidative stress, voltage channels, apoptosis, DNA damage and repair (Figure 4a and Supplementary Table S1). String analysis revealed proteins related to apoptosis, transport, iron transport, and DNA damage (Figure 4b). SWATH-MS and DDA identified 32 proteins, while DDA alone identified 11 (Figure 4c).

DDA identified 15 proteins related to Reactive oxygen species (ROS) (Supplementary Table S1). On SWATH-MS, 8 were significantly upregulated in UNT versus WT mice (FC >3 for Sod1, Prdx1, Prdx2, Prdx3, Prdx5 and S100a9; Figure 5a; Supplementary Table S3, Section 2.1). Six proteins related to voltage-dependent channels in mitochondria were identified by DDA, none of which were significantly dysregulated in UNT mice, although all showed minor expression in UNT mice. DDA detected five proteins related to apoptosis in both groups, 4 of which showed no significant differences between UNT and WT mice. Casp6 was identified in the WT group, suggesting a defect in apoptosis in UNT mice (Supplementary Table S1). Of 15 detected proteins related to iron transport (Supplementary Table S1), 6 (Tf, Cisd1, Fth1, Mb, Hbb-b1, and Hpx) were significantly upregulated in UNT versus WT mice, with FC >3 (Figure 5b and Table S3, Section 2.2). DDA identified 3 proteins (TPi1, Uba, Ube2n) related to the prevention of DNA damage (Supplementary Table S1), all of which were upregulated in UNT versus WT mice with FC >3 (Figure 5c; Table S3, Section 2.3).

Our findings indicate a marked elevation of oxidative stress in UNT mice, mainly via Prdx3/Prdx5. UNT mice showed reduced expression of proteins related to voltage-gated channels and no activation of the apoptotic pathway via caspases. However, UNT mice also showed marked upregulation of proteins involved in iron capture and increased expression of proteins that participate in DNA damage prevention.

2.3. Section III. Histones, Ribosomes, Proteasomes, Vesicular Transport and Lysosomes

We identified 186 proteins related to histones, ribosomes, proteasomes, vesicular transport, and lysosomes (Supplementary Table S1). The Reactome analysis (Figure 6a) depicts proteins related to lysosomes (part I) and vesicle transport (part II). A string analysis revealed clusters of proteins related to ribosomal organization and biogenesis, proteasomes, lysosomes, and signal translation (Figure 6b). Together, SWATH-MS and DDA identified 83 proteins, while DDA alone identified 103 proteins (Figure 6c).

DDA identified 2 integrins that regulate the nucleus to cytoplasm trafficking of macromolecules (Kpna2, Kpnb1) in WT mice only (Supplementary Table S1). On SWATH-MS, 5 histones were upregulated in UNT versus WT mice with a FC >3 (H1-0, H1-2, H1-4, H2az2, and H4c1) (Figure 7a; Table S III, Section 3.1).

Ribosomal proteins: DDA identified 22 proteins from the 40S ribosomal subunit (Supplementary Table S1), of which 5 were significantly upregulated in the UNT group, 3 with a FC >3 (Rps5, Rps18, and Rps25) (Figure 6b; Table S III, Section 3.2). Of the 27 detected proteins related to the 60S region, only 1 (Rpl12) showed a FC >3 in UNT versus WT mice (Figure 7b; Table S III, Section 3.2). Seven ribosomal proteins were detected only in UNT mice (Rpl6, Rpl18, Rpl18a, Rpl24, Rpl27, Rpl28, and Rpl38) (Supplementary Table S1).

Proteasomes: DDA identified 22 proteasome proteins (Supplementary Table S1), only one of which was significantly upregulated in the UNT group with a FC <3 (Table S3, section 3.3). Psmd8 was expressed only in UNT mice, while 12 proteins were expressed only in WT mice (Psma2, Psma3, Psma6, Psma7, Psmb1, Psmb2, Psmb3, Psmb5, Psmb6, Psmb8, Psmb9, and Psmd2) (Supplementary Table S1).

Vesicular transport: DDA identified 30 proteins related to vesicular transport (Supplementary Table S1), 3 of which were significantly upregulated in UNT mice, 2 with a FC >3 (Rab5b, Tpd52l2) (Figure 7c; Supplementary Table S3, Section 3.4). Sorl1 was detected only in UNT mice. Of 13 detected proteins related to the lysosomal membrane (Supplementary Table S1), 4 were significantly upregulated in UNT mice, 3 with an FC > 3 (M6pr, Lgals1, and Lgals3) (Figure 7d; Supplementary Table S3, Section 3.5). Cd63, Cd74, and Atp6v1a were expressed only in WT mice. Of the 30 proteins identified, 22 were in the lysosome (Supplementary Table S1), 1 of which was downregulated in UNT mice with an FC <3 (Asah1) (Figure 7e; Supplementary Table S3, Section 3.5). It should be noted that Galns (of which a deficit is observed in MPS IVA) and other related proteins (Gba1, Pla2g15, and Naaa) were identified only in WT mice.

DDA identified 44 cytoskeleton-related proteins (Supplementary Table S1), 17 of which were identified as DEPs on SWATH-MS: 5 were upregulated in WT versus UNT mice, 4 with a FC <0.3 (Calm3, Acta1, Tnnc2, and Gdpd2) (Figure 8a; Supplementary Table S3, Section 3.6). Eleven proteins were upregulated in UNT versus WT mice, 7 with FC >3 (Myl2, Myfpl, Lasp1, Rhoa, Arpc3, Arpc5, and Arpc5l) (Figure 8b; Supplementary Table S3, Section 3.6).

In summary, in UNT mice, we observed an increase in protein biosynthesis via histones and ribosomes, a partial lack of proteasome proteins, a lack of calcium necessary for cytoskeletal proteins, an increase in proteins related to vesicular trafficking, the presence of proteins related lysosomal membrane damage repair, and an absence of certain membrane receptors. Crucially, we observed a total absence of the Galns enzyme in UNT mice, which may result in the accumulation of KS and C6S in lysosomes (KS and C6S are sugar chains, which were not measured in the present study), a characteristic feature of MPS IVA disease.

2.4. Section IV. Extracellular matrix.

We identified 99 proteins related to cell adhesion (Figure 9a, Supplementary Table S1). SWATH-MS and DDA identified 60 proteins, while DDA alone identified 39 (Figure 9b).

To facilitate interpretation of the large numbers of matrix components identified, proteins were ordered by families, as indicated below.

Hyaluronan-related proteins: a single protein (Cemip2) was identified (Supplementary Table S1) but showed no significant difference in expression between WT and UNT mice.

Proteoglycans

Hyalectans: 3 proteins were identified. Hapln1 was significantly upregulated in UNT versus WT mice (FC <3).

• Small leucine-rich proteoglycans: DDA identified Class I, II, and IV proteins (Supplementary Table S1).

○

Class I proteins: Dcn was significantly upregulated in UNT mice (FC >3; Figure 9c, Supplementary Table S3, Section 4.1). Aspn was detected in UNT mice only.

○

Class II proteins: DDA identified 4 proteins (Fmod, Lum, Prelp, Kera; Supplementary Table S1), all of which were significantly upregulated in UNT mice with FC >3 (Figure 9c, Table S III, Section 4.1)

○

Class IV: Only Chad was significantly upregulated in UNT mice, with a FC <3 (Table S3, Section 4.1).

• Cell surface proteoglycans: DDA identified proteins related to the cell surface (e.g., Cspg4; Supplementary Table S1), none of which were differentially expressed in UNT versus WT.

DDA identified various collagen proteins, which we classified according to physiological function (Supplementary Table S1).

• Fibrillar collagen: DDA identified 8 proteins related to fibrillar collagen, 5 of which were significantly upregulated in UNT mice, 4 with an FC <3 (Col1a1, Col1a2, Col11a1, and Col11a2) (Figure 9d, Supplementary Table S3, Section 4.2). Col5a1 was identified only in WT mice.
• Networking-forming collagen: 3 proteins were identified, one of which was present only in WT mice (Col4a1).
• FACITs, Beaded-filament-forming collagens: 3 proteins identified.
• MACITs, anchoring fibrils, multiplexin: 1 protein identified.

Laminins: 5 proteins were identified, none of which differed significantly between groups.

Matrix cellular proteins: DDA identified 13 proteins related to the ECM, 2 of which were significantly upregulated in UNT mice, 1 with an FC >3 (Thbs1) (Figure 10a, Supplementary Table S3, Section 4.3). Cnn3 was detected only in WT mice (Supplementary Table S1).

Cellular interface: DDA identified 2 proteins, which were present only in WT mice.

Extracellular matrix proteases: DDA identified 3 proteins that can be classified in the Metzincin family. Mmp9 was significantly upregulated in UNT mice, with an FC >3 (Figure 10b, Supplementary Table S3 Section 4.4). Adam10 was detected only in UNT mice.

Plasminogen/plasmin system: DDA identified 15 proteins (Supplementary Table S1), 3 of which were significantly upregulated in UNT mice, 1 with a FC >3 (Serpinf1) (Figure 10c; Supplementary Table S3, Section 4.4). Plaur and Serpina1 were identified only in WT mice.

Cathepsin proteases: DDA identified 2 proteins in the Aspartic proteases family (Supplementary Table S1), 1 of which was upregulated in UNT mice, with an FC >3 (Ctsd) (Figure 10d; Supplementary Table S3, Section 4.4).

Serine proteases: 1 protein identified (Ctsg).

Cysteine proteases: DDA identified 3 proteins, 1 of which (Ctsh) was upregulated in UNT mice with an FC >3 (Figure 10d; Supplementary Table S3, Section 4.4).

Extracellular receptor matrix: DDA identified 14 proteins (Supplementary Table S1) belonging to the Integrin family, 4 of which were significantly up- or downregulated in UNT mice, but only 2 with a FC >3 (Itgam, Itgav) (Figure 10e; Supplementary Table S3, Section 4.5).

Discoidin domain receptors: DDA identified 2 proteins, of which only Cdc42 was upregulated in UNT mice (FC <3; Supplementary Table S3, Section 4.6). DDR2 was identified only in WT mice.

Another extracellular receptor: DDA identified 3 proteins, 1 of which (Dpep1) was upregulated with a FC >3 (Figure 10f; Supplementary Table S3, Section 4.7).

In summary, our analysis revealed significant upregulation in UNT mice of several proteoglycans (N-linked glycosylation), mainly related to KS (Fmod, Lum), metalloproteases, cathepsins, and collagens (some fibrillar collagens).

2.5. Section V. Other Proteins of Interest

We identified 5 other proteins of interest (Rtn2, Fetub, Alb, Ckm, and Ckmt2; Supplementary Table S1), 4 of which were significantly upregulated in UNT mice with a FC >3 (Figure 11; Supplementary Table S3, Section 5.1).

3. Quantitative Validation of DEPs

3.1. Mouse DEPs Validation by Different Proteomic Technology

To perform cross-validation of DEPs identified by quantitative SWATH-MS (MS2-based quantification), we used Scaffold software to obtain a semiquantitative value for proteins identified by label-free assay based on spectral count. This label-free method enabled MS1-based quantification (spectral count). We observed a correlation between the two quantitative methods of approximately 18.90% (Table S4), suggesting the quantitative reliability of the proteins considered DEPs in our study. The number of DEPS identified by SWATH-MS was identical to that identified by spectral count (121 DEPs). Using the 2 quantitative methods, we validated 38 proteins (Figure 12a; Table S4). Among these proteins, we identified lactate dehydrogenase (LDH) Ldha and Ldhb as potential biomarkers (Figure 12b; Table S4). String analysis of the validated DEPs identified proteins related to ATP metabolic process, glycolysis and gluconeogenesis, TCA cycle, LDH, OXPHOS, and ROS as a potential biomarkers.

3.2. In Silico Validation of Mouse DEPs Using Data Sets from Human Samples

To validate our quantitative proteomic data obtained by SWATH-MS in mice, we will compare the quantitative proteomic data obtained by SWATH-MS in mice with human data previously obtained by our group using the same proteomic technology [18,19]. As shown in Figure 13A, several detected proteins were common to both analyses, including proteins involved in pathways such as glycolysis, oxidative stress, and vesicle trafficking and proteins involved in the cellular cytoskeleton. Perhaps most interestingly, Ldha and Ldhb were common to all analyses, supporting the potential of these proteins as candidates for diagnosis, prognoses disease monitoring biomarkers.

It should be noted that although Morquio A is a rare disease with a fewer studies, our findings point to multiple similarities between Morquio A patients and those with other common osseous diseases of adults, including arthritis. Therefore, the proteomic approach used here to identify potential therapeutic targets for this rare disease could also help further our understanding of other bone diseases. For this reason, proteomic data obtained from human osteoarthritis bone samples were selected for in silico validation of our findings.

The presence of multiple proteins common to our mouse data set and the 2 human data sets validates the approach used in the present study. The first data set consisted of bone marrow sample data taken from Hennrich et al. [20] and corresponded to 59 human subjects (45 male, 14 female). In that study relative quantification was achieved using tandem mass tag (TMT) labeling, SWATH-MS. The second data set consisted of data from primary cultures of human articular chondrocytes, from both healthy and osteoarthritic patients, taken from a study by Konstantinos et al. [21]; in that paper, samples were analyzed using label-free quantification and scaffold software used for comparative analysis, as in the present study. Comparison of both data sets with our data confirmed that the protein alterations found in our mouse model resembled many of those observed in a bone disease such as osteoarthritis.

Comparison of our findings with those of the 2 aforementioned human studies revealed 30 proteins common to all 3 analyses (Figure 13B), as well as comparable alterations in pathways implicated in glycolysis, ROS, DNA damage, vesicle transport, lysosomes, the cellular cytoskeleton, and the extracellular matrix (Moreover, Ldha and Ldhb were between the common proteins across all 3 studies. Other candidate biomarkers detected in all 3 data sets included collagen type II and MMP9.

In summary, our “in silico” validation results indicate that the DEPS found in mice in the present study are also observed in human samples, particularly MMP9, Col2 and both Ldha and Ldhb.

4. Discussion

The overall goal of the present study was to investigate the differential expression of proteins involved in distinct molecular pathways in MPS IVA and to explore the pathogenic relevance of the observed differences in systemic skeletal dysplasia in this disease. To this end, we investigated the differential expression of proteins of interest in the bone ECM of MPS IVA mice. Due to the rarity of this disease, it is challenging to obtain sufficient numbers of bone samples from affected human patients and published studies using proteomics of human bone samples are very scarce. Therefore, we based our analysis on samples from an MPS IVA mouse model and compared the results obtained with SWATH-MS proteomic data previously published by our group corresponding to fibroblasts and blood cells from MPS IVA patients, and with published data from proteomic analyses of bone marrow and bone (primary chondrocytes)[20,21] samples from human. Moreover, we linked the altered proteins and pathways observed in MPS IVA mice with the multiple causes affecting MPS IVA patients. In our analysis of the data obtained, we sought to identify candidate diagnostic and prognostic biomarkers for MPS IVA that are validated in human in our previous paper and in other human proteomic data sets.

Mass spectrometry-based proteomics is an indispensable tool for biomarker discovery. We used two distinct but complementary proteomic technologies, SWATH and DDA analysis, to better understand MPS IVA. The development of SWATH mass spectrometry (MS) has led to a new era of accurate and reproducible label-free quantification of different proteomes [22,23]. Our group previously used this technology to characterize cells and plasma/serum samples from mice with mucopolysaccharidoses [18,19,24]. Unlike conventional DDA-MS, which is based on specific selection and fragmentation of a limited number of (typically the most abundant) precursor ions in a survey scan, SWATH-MS is based on the cyclical acquisition of precursor ions with specific isolation windows that cover the entire m/z range. All ionized precursor peptides within a sample are fragmented, and subsequent analysis of their fragmentation spectra enables the quantification of peptides of interest using spectral libraries. This technology combines the advantages of high reproducibility and sensitivity of targeted methods with the increased proteome depth typically provided by DDA-MS. The spectral count is a commonly used label-free proteome quantification method because sample preparation is simple: no labeling is required for data generation. However, this technique can introduce spectral count variation in cases of low-abundance proteins in multiplicative LC-MS/MS analysis, hampering sensitive proteome quantification [25]. As many low-abundance proteins play essential roles in cellular processes, quantitatively reliable identification of these proteins is crucial for better understanding these processes.

Using DDA and SWATH-MS, we characterized differentially expressed proteins in bone ECM in MPS IVA mice. Our findings demonstrate alterations in specific cellular functions in the ECM of bone samples from UNT mice. The proteins analyzed include those related to metabolic pathways (i.e., cytoskeleton proteins and proteins involved in mitochondrial function, autophagy, interaction of calcium, iron, and other ions, ribosomes, lysosomes, and the ECM). We observed altered expression of proteins involved in energy-generating pathways, including glycolysis, the Krebs cycle, and OXPHOS, which produce high oxidative stress. Furthermore, we detected altered expression of proteins such as Rtn2 and Sorl1 that facilitate glucose entry into cells via GLUT-4 [26,27]. Similar alterations in osteoarthritis in humans are described, in which GLUT-1 receptor expression is augmented, thereby increasing cellular glucose levels [28]. These high glucose levels result in significant alterations in the expression of other proteins in UNT mice, in which higher levels of cellular hypoxia are observed [29]. Alterations in metabolic pathways also increase lactate production, as evidenced by the upregulation of LDH observed in UNT mice. LDH plays a vital role in generating oxidative stress in rheumatoid arthritis. Using mRNA sequencing in primary chondrocyte cultures from a mouse model of osteoarthritis, Arra et al. [30] detected significant increases in the expression of genes involved in glycolysis, including Ldha. We recently identified candidate biomarkers of MPS IVA [30,31] using the same proteomic technologies in two different sample types: fibroblasts and leukocytes from MPS IVA patients. Those studies showed that LDH expression is upregulated in MPS IVA cells and that levels decrease in response to enzyme replacement treatment [18,24], providing strong evidence to suggest a role for LDH proteins as biomarkers of disease prognosis and progression.

Another important finding is the general upregulation of proteins related to the Krebs cycle and OXPHOS in UNT mice. We detected many upregulated proteins pertaining to mitochondrial complexes I, II, III, IV, and V (ATP-synthetase), including proteins such as Ndufa6 with an FC 86.51, corroborating previous reports of increased ROS levels [30]. Furthermore, we observed elevated expression of ROS compensatory proteins such as Prdx3/Prd [32]. In line with previous findings in an animal model of MPS IVA, upregulation of DNA damage repair proteins has been reported in UNT mice [33]. In that paper, Donida et al. analyzed blood samples from 17 MPS IVA patients and 14 healthy controls using molecular biology techniques and observed increased DNA damage of oxidative origin in patients versus controls. Moreover, they detected increased interleukin 6 (IL-6) expression in patients, suggesting a possible link between inflammation and oxidative stress in MPS IVA. Although we detected proteins related to inflammation in our study, the proteomic techniques used did not allow the detection of low molecular weight proteins, such as IL-6, despite its likely presence in our samples. These alterations in inflammation pathway result in activation of signalling pathways such as STAT3 [34] a transcription factor activated by members of the IL-6 and IL-10 family, and by IL-21, IL-27, G-CSF, leptin, and IFN-I, among other signalling molecules [35]. In their mass spectrometry studies of an AEP^-/- (Asparaginyl endopeptidase Knock Out) mouse model and cell lines, Martínez-Fábregas et al. [36] concluded that an increase in lysosomal cysteine proteases (CtsD) in the kidney reflects the accumulation of undegraded material and/or increased expression of lysosomal proteins. Moreover, they reported an increase in endocytic substrate load, independent of the transcription factor EB (TFEB) and triggered by STAT3 activation downstream of lysosomal oxidative stress. Similar lysosomal adaptations are seen in mice and cells expressing a constitutively active form of STAT3 [36]. Therefore, Martínez-Fabregas J et al. reported that STAT3 signalling is activated by the ROS effect due to substrate accumulation in lysosomes and deficiencies in lysosomal proteases [36] which also occurs in MPS IVA.

The protein MTORC1-TFEB acts as a transcriptional regulator of autophagy and is the most relevant marker of lysosomal activity [37,38] owing to its ability to activate lysosomal biosynthesis. Our data indicate that this protein appears to be inhibited by overexpression of cathepsin D and LAMP1 in MPS IVA. These proteins also contribute to increased inhibition of the autophagy pathway via the lysosome. Proteosomes play a key role [39,40] in this autophagy pathway, mainly via Psma6 and Psmb6, which participate in the active nucleus of the proteasome [41]. However, we did not detect proteins related to proteasomes in UNT mice. Activation of STAT3 [39] in response to lysosomal accumulation stimulates compensatory biosynthesis of cytokines and interleukins. The cytokines and interleukins detected in the present study were the same as those described in other MPS IVA studies[42] used liquid chromatography-tandem mass spectrometry (LC-MS/MS) to measure the 4 disaccharides produced from GAGs in serum or plasma from 34 MPS IVA patients, and identified 9 biomarkers that were significantly elevated in untreated MPS IVA patients: EGF, IL-1β, IL-6, TNF-α, MIP-1α, MMP-1, MMP-2, and MMP-9. In UNT mice we observed activation of protein biogenesis, which explains the high levels of cytoplasmic histones (some of which have been previously detected in proteomic studies of other non-bone pathological cells [18,19]) and upregulation of ribosomal proteins and the exclusive presence of specific proteins such as Rps19 and Rpl24, in agreement with previous reports [43]. This increase in protein biosynthesis normally generates more proteins with endosomal/lysosomal functions [44], and increases ribosome expression in the bone, which usually occurs in the context of low TFEB activity [45].

Analysis of proteins involved in vesicular transport revealed alterations in the endocytosis pathway in UNT mice, as evidenced by upregulation of Rab5c, which participates in vesicular trafficking [38], and Tpd52l2 [46]. This pathway is important because increases in M6pr can induce the expression of many lysosomal enzymes [47]. In the UNT group, we detected overexpression of proteins related to lysosomal membrane repair and autophagy functions [48]. Conversely, other lysosomal membrane proteins, such as Atp6v1a, Cd63, and Cd74, were absent from UNT mice.

In analyzing apoptosis activation, we found that caspases that would normally be elevated in response to high ROS levels [49] were not expressed at high levels in UNT mice. Moreover, the calcium voltage potential between the internal and external membrane was reduced” (in UNT mice), which can prevent activation of apoptosis functions [50,51], suggesting a problem affecting calcium pathways. In the cytoplasm, we observed clear evidence of dysregulation of actin (Acta 1), myosin (Myosin-1), calmodulin (Calm3), and troponin (tnn3) [52,53] in UNT mice. These alterations may have consequences for cytoskeletal protein function. Moreover, ROS may also contribute to the failure of repair mechanisms by reducing the capacity of chondrogenic precursor cells to migrate and proliferate within an injured area: NO has been shown to inhibit chondrocyte migration and attachment to fibronectin via modification of the actin cytoskeleton [54]. This is particularly important in the context of MPS IVA because bone resorption proteins that form the osteoclast podosome require proteins such as F-actin-rich domain, protein 2/3 complex (Arp2/3), and myosin light chain to attach to the bone area for bone degradation (in UNT group we found several proteins elevated)( Arpc3, Arpc5, Arpc5l ,Capza2, Myfpl, Myl2) [55,56]. Actin binding points must be in good condition to enable the transport of lysosomes to the matrix. This does not occur the UNT mice, in which filamin B (Flnb) expression is altered, as described previously by Qiming Xu et al. [57]. Those authors reported that decreased expression of this protein could induce skeletal malformations, including delayed ossification in long bone growth plates; reduced bone mineral density; dysregulation of muscle differentiation; intervertebral disc ossification; altered chondrocyte proliferation, differentiation and apoptosis; impairment of angiogenesis; and hypomotility of osteoblasts, chondrocytes, and fibroblasts [57]. Moreover, lysosomes can transport calcium phosphate [58], obtained by collecting calcium from other organelles such as the endoplasmic reticulum, Golgi, cytoplasm, and mitochondria. Normally, this transport occurs via kinesin motor and dynein pathways [59]. In the present study, the proteins related to the dynein pathway were detected only in WT mice. The transport of collagen and calcium phosphate, which was also altered in UNT mice, is vital to support the role of osteoblasts in bone mineralization.

In our study in UNT mice, the evident decompensation ROS production induces degradation of the bone ECM. Bake et al.[60] reported that exposure of chondrocytes to H₂O₂ inhibits proteoglycan. Findings reported by Taskiran et al. [61] in rabbit articular cartilage and by Häuselmann et al. [62] (in human articular chondrocytes) suggest that interleukin-1 suppresses the synthesis of the cartilaginous matrix. Similarly, Hickery et al. [63] showed that IL-1 synthesis causes inhibition of proteoglycan sulfation in human articular chondrocytes and reported that IL-1-induced NO inhibits sulfation and alters the sulfation pattern of newly synthesized glycosaminoglycan chains [54]. Increased glycolysis induces high levels of lactate excreted to the ECM, decreased pH, and increased levels of metalloproteases such as MMP9 and ADAM10 [64]. However, ROS also contributes to cartilage degradation by mediating the activation of latent collagenase and upregulating the expression of genes coding for matrix metalloproteinases (MMP) [54,65], secreted by lysosomes and activated by Hpx or Hpx domains [64,66,67]. Moreover, an increase in the levels of cathepsins excreted by lysosomes [54] results in significant changes in various collagen types. Collagens I, II, and XI [42,68], as well as the conversion of collagen II to I [12,42,69], have been described in MPS IVA. In our study, we observed high levels of expression of proteoglycans, including KS (fibromodulin (Fmod) and Lumican (Lum)). In the context of increased ROS expression, the sorting receptor Sorl1 (detected only in UNT mice) induces more lysosomal endocytosis activity while the hydrolysis of sphingomyelin to ceramide increases. This effect may be reflected in our data by the low levels of Asah1 enzyme detected in UNT mice [27,70]. Proteins that participate in collagen endocytosis, including fibromodulin, integrins, and caveolin-1, all showed altered expression in UNT mice, in line with other findings (reviewed in Rainero, 2016) [71].

To validate our results, we used a new proteomic quantitative measure to estimate the difference in protein levels between UNT and WT mice. This validation was performed using DDA data that were transformed into a spectral count using Scaffold software. Next, proteins for which differential expression was observed were analyzed using DDA and SWATH-MS, resulting in the validation of 20% of proteins. Thus, 38 proteins were considered differentiated expressed in UNT versus WT mice. These DEPS were related to ATP/metabolic process, glycolysis and gluconeogenesis, the TCA cycle, and LDH expression in the mitochondrial membrane. Interestingly, previous studies have reported alterations in some proteins (Ldha and Ldhab)[18,19] that we found to be differentially expressed (upregulated in UNT) and validated by both techniques, suggesting that elevation of LDH may constitute an early biomarker of MPS IVA.

These potential biomarkers are related to KS accumulation, as their upregulation is linked to an increase in ROS levels [30]. Our findings identify several proteins that may constitute an excellent candidate to diagnostic, prognosis, and disease monitoring biomarkers, including MMP-9, collagen type II, and LDH.

It should be noted that our analysis was performed using only 4 bone samples per group. Further studies with a larger sample size using a different technology (i.e., other omics technologies) will be necessary to corroborate the present findings and better understand the pathogenic mechanisms that occur in the bones of children with MPS IVA.

The literature contains abundant information about the iNOS pathway, which generates large amounts of NO, and studies in chondrocytes indicate that this pathway is altered in MPS IVA, which activates IL-1 and acts on the extracellular matrix. In the present study, we could not quantify alterations in IL-1 or IL6 due to the proteomic technique used. Further studies using more significant numbers of samples will be required to identify the proteins affected in this pathology.

Proteome maps summary of deregulation proteins in MPS IVA.

5. Material and Methods

5.1. Experimental Design and Statistical Analysis

Analyses were performed using samples from 4 untreated MPS IVA (UNT) and 4 WT male mice. The mice were provided by Dr. Tomatsu of Nemours Children’s Health (Wilmington, Delaware, USA), with prior authorization from that institution’s animal ethics committee (RSP19-12482-002). Mice were not subjected to any intervention and were euthanized at 6 weeks of age. The mice used carried the specific GALNS gene disruption in exon 2 [17].

All experiments were conducted using the TripleTOF 6600+ system (Sciex, Framingham, MA, USA). All mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the data set identifier PXD042166. Additional information on the samples analyzed, data acquisition, and processing procedures can be found below.

SWATH analysis was performed using MarkerView 1.3.1, GraphPad Prism 6.01 software (GraphPad Software, San Diego, CA, USA) was used to generate SWATH blox plots, and Spectral counting analysis was performed using Scaffold™ 5.3. In the SWATH analysis, the summed averaged area of all transitions for each protein in each sample were analyzed using a Student’s t-test to determine the extent to which each variable could distinguish between the two groups, and the result was reported as a p-value. Fisher’s exact test and the Benjamini–Hochberg multiple correction test were used for quantification and fold change (FC) determination for spectral counting comparison between groups. In both cases, a protein was considered differentially expressed when p was <0.05 and the FC >1.5 or <0.6, although only proteins with an FC >3 are shown in the figures.

In SWATH analysis MLR (most likely ratio) normalization was performed after statistical analysis to control for possible uneven sample loss across the different samples during the sample preparation process [72,73]. The mean Mass Spectrometry peak area for each protein was determined for each sample, and Student’s t-test (MarkerView software, Sciex, Redwood City, CA, USA) was performed to compare between samples based on the averaged total area of all transitions for each protein. The t-test result (p-value) indicates how well each variable distinguishes between the two groups. Candidate proteins were selected based on t-test results (p <0.05 and FC [increase or decrease] >1.5 or <0.6).

5.2. Animal Model and Bone Protein Extraction

Analyses were performed using 4 untreated MPS IVA (UNT) and 4 WT male mice. Mice were provided by Dr. Tomatsu of Nemours Children’s Health (Wilmington, Delaware, USA), with prior authorization from The Institutional Animal Care and Use Committee (IACUC) approval by Paul T. Fawcett, Ph.D (RSP19-12482-002). Mice were not subjected to any intervention and were euthanized at 6 weeks of age. Femurs with intact femoral caps were collected, and samples were placed on dry ice. To prepare samples, a radioimmunoprecipitation assay (RIPA) buffer was added, and the resulting mixture was placed in tubes containing 2-mm porcelain beads and homogenized for 30 sec using a PowerLizerco. After centrifugation for 30 min at 4°C, the pellet and beads were separated. Each sample was analyzed for total BCA protein and N-acetylgalactosamine-6-sulfate sulfatase enzymatic activity. The remaining samples were sent for analysis at the proteomic analysis platform of the investigation department (IDIS) of the University Clinical Hospital of Santiago de Compostela.

Samples were analyzed using 2 distinct but complementary s technologies: DDA (data-dependent analysis) and SWATH-MS (IDA, information-dependent acquisition) (Supplementary Figure 1). One of the advantages of utilizing a label-free LC-MS/MS workflow such as SWATH-MS is the sensitivity of protein detection and quantification offered. Traditional LC-MS/MS workflows based on DDA increase the overall identification of proteins but suffer from a reduction in the number of identifiable peptides across multiple runs [74,75]. Therefore, SWATH-MS constitutes an attractive solution to improve label-free quantification and has been used for biomarker identification in other rare diseases [18,19].

To understand the interactions between proteins with altered expression in bone samples, we analyzed all SWATH-MS and DDA data, grouping proteins according to known functions, as follows: (i) energy metabolic pathways (e.g., glycolysis, tricarboxylic acid cycle, β-oxidation, and OXPHOS); (ii) mitochondrial function (i.e., oxidative stress, calcium-regulated mitochondrial electrical potential channels, apoptosis, compensatory mechanisms of oxidative stress, and DNA damage prevention); (iii) protein production (ribosomes), cytoskeleton, proteosomes, vesicular trafficking, and lysosomes; (iv) involvement in ECM function; and (v) other proteins of interest.

Proteomic Analysis by TripleTOF 6600 LC-MS/MS (see Supplementary Materials and Methods for further information)

Protein Digestion For protein identification, equal amounts of protein from each femur sample (n=4 per group) were loaded onto a 10% SDS-PAGE gel. The resulting condensed protein bands underwent gel digestion using Trypsin and were processed as described in the Supplementary Materials and Methods and previously by our group [76].

LC-MS/MS in Data-Dependent Acquisition Mode, Shotgun Analysis, and Spectral Count Transformation Digested peptides from each mouse femur sample were separated using reverse phase chromatography as described previously [76] and in the Supplementary Materials and Methods. Semi-quantification by spectral counting was performed using Scaffold software (version 5.0.1; Proteome Software Inc., Portland, OR), as described previously [77]. MS/MS-based peptide and protein identifications performed in Protein Pilot 5.0.1 software were transformed in. mzid format and validated using Scaffold software. Peptide identifications were accepted if they could be established at >95.0% probability by the Percolator posterior error probability calculation. Protein identifications with a probability <99.0% and at least two identified peptides were accepted. Protein probabilities were assigned by the Protein Prophet algorithm. Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony.

Protein Quantification by SWATH-MS Analysis To build the MS/MS spectral libraries, peptide solutions were analyzed by shotgun data-dependent acquisition (DDA) using micro-LCMS/MS, as described in the Supplementary Materials and Methods and previously by our group [18,76]. The MS2 spectra (MS/MS spectra) of the identified peptides were then used to generate the spectral library for SWATH peak extraction using the add-in for PeakView Software (version 2.2, Sciex), MS/MSALL with SWATH Acquisition MicroApp (version 2.0, Sciex). Peptides with a confidence score >99% (obtained from the Protein Pilot database search) were included in the spectral library. For relative quantification by SWATH-MS analysis, SWATH-MS acquisition was performed on a TripleTOF 6600 LC-MS/ MS system (Sciex) using SWATH mode. The acquisition mode consisted of a 250 ms survey (MS scan) MS1 scan from 400 to 1250 m/z, followed by an MS2 (MS/MS) scan from 100 to 1500 m/z (25 ms acquisition time) of the top 65 precursor ions from the survey scan, for a total cycle time of 2.8 s. The fragmented precursors were then added to a dynamic exclusion list for 15 sec. Any singly charged ions were excluded from the (MS/MS analysis) MS2 analysis. Targeted data extraction from the SWATH MS runs was performed by PeakView v.2.2 (ABSciex, USA) using the SWATH MS Acquisition MicroApp v.2.0 (ABSciex, USA). Data were processed using the spectral library created from DDA. SWATH MS quantization was attempted for all proteins in the ion library that were identified by ProteinPilot^TM 5.0.1 with a false discovery rate (FDR) <1%. PeakView computed an FDR and a score for each assigned peptide based on the chromatographic and spectra components: only peptides with an FDR <1%, 10 peptideand 7 transitions per peptide were used for protein quantization. The integrated peak areas were processed by MarkerView software version 1.3.1 (AB SCIEX, USA) for a data-independent method for relative quantitative analysis. A most likely ratio normalization was performed to control for possible uneven sample loss across the different samples during the sample preparation process. Unsupervised multivariate statistical analysis using PCA was performed to compare data across samples.

5.3. Functional and Pathway Analysis

Pathway analysis was performed using Reactome (https://reactome.org/), which applies a statistical (hypergeometric distribution) test to determine whether specific pathways are over-represented (enriched), and produces a probability score, which is corrected for FDR using the Benjamini–Hochberg method. The most enriched pathways were represented using Reactome pathway diagrams. Protein interactions were evaluated using String (https://string-db.org/), applying a minimum required interaction score of PPI=0.9 (protein-protein interaction) and an FDR <0.05. Venn diagrams were generated using http://www.interactivenn.net/ and box plots using GraphPad Prism 9. Statistical analyses were performed using Markerview or Scaffold software. Volcano plots and box plots were generated using GraphPad Prism 9.

5.4. In Silico Analysis

Proteins for “in silico” analysis were selected using two distinct approaches.

The first approach was based on SWATH-MS analyses of leukocytes and fibroblasts previously published by our group [18,19]. The second approach involved validating proteomic data sets taken from published analyses of human bone [20,21]. Few such analyses have been published; most correspond to bone marrow samples. For our validation, we selected 2 data sets, corresponding to bone and bone marrow sample data sets, respectively. The data set that was extracted for our analysis fulfilled the following 3 quality assessment criteria: (1) identification and description of the extraction method used; (2) description of the sampling process used; (3) use of a proteomic technology similar to SWATH_MS.

6. Conclusions

The pathological features of MPS IVA resemble those of other skeletal dysplasias and bone diseases that occur at later ages. However, MPS IVA can also involve ROS accumulation due to excessive glycolysis, upregulation of LDH, and accumulation of substrates in the lysosome. This increase in ROS levels triggers the activation of pathways such as STAT3 signaling, which compensates for deficient autophagy and activates cytokines and interleukins. In MPS IVA, this process becomes chronicled due to the accumulation of deposits that are not eliminated in the lysosomes.

In MPS IVA, dysfunction at the level of the mitochondria and cytoskeleton can be due to calcium deficiency, possibly due to high bone resorption, poor bone mineralization, and excessive ROS activation, which facilitates active-matrix degradation. These alterations can lead to a matrix composition change, mainly due to defective bone regeneration.

Our proteomic technique analyses identify MMP-9 and collagen type II as candidate biomarkers of MPS IVA, suggesting that LDH may constitute a new candidate biomarker. LDH is upregulated in various human cell types (see our studies) [18,19] and downregulated in response to enzyme replacement therapy. All these candidate biomarkers detected in mouse bone samples in the present study have also been found to be altered using proteomic techniques in bone and bone marrow human samples. This biomarker may constitute a useful biomarker for diagnosis and monitoring disease progression of therapeutic efficacy in MPS IVA disease.