A new UHPLC-MS/MS method for the screening of urinary oligosaccharides expands the detection of storage disorders

Background Oligosaccharidoses are storage disorders due to enzymatic defects involved in the breakdown of the oligosaccharidic component of glycosylated proteins. The defect cause the accumulation of oligosaccharides (OS) and, depending on the lacking enzyme, results in characteristic profiles which are helpful for the diagnosis. We developed a new tandem mass spectrometry method for the screening of urinary OS which was applied to identify a large panel of storage disorders. Methods The method was set-up in urine and dried urine spots (DUS). Samples were analysed, without derivatization and using maltoheptaose as internal standard, by UHPLC-MS/MS with MRM acquisition of target OS transitions, including Glc4, the biomarker of Pompe disease. The chromatographic run was < 30 min. Samples from patients with known storage disorders were used for clinical validation. Results The method allowed to confirm the diagnosis of oligosaccharidoses (sialidosis, α-/β-mannosidosis, fucosidosis, aspartylglucosaminuria) and of GM1 and GM2 (Sandhoff type) gangliosidosis, by detecting specific OS profiles. In other storage disorders (mucolipidosis II and III, mucopolysaccharidosis type IVB) the analyisis revealed abnormal OS excretion with non-specific profiles. Besides Pompe disease, the tetrasaccharide Glc4 was increased also in disorders of autophagy (Vici syndrome, Yunis-Varon syndrome, and Danon disease) presenting cardiomuscular involvement with glycogen storage. Overall, results showed a clear separation between patients and controls, both in urine and in DUS. Conclusion This new UHPLC/MS-MS method, which is suitable for rapid and easy screening of OS in urine and DUS, expands the detection of storage disorders from oligosaccharidoses to other diseases, including the novel category of inherited disorders of autophagy.

accounts for over 65 different inherited disorders [1]. Oligosaccharidoses are due to defects of lysosomal enzymes involved in the catabolic pathway for the breakdown of the oligosaccharidic component of glycosylated proteins [2]. The glycosidic groups, composed of fucose, mannose, sialic acid, galactose, and N-acetylglucosamine residues, to form glycoproteins are either N-linked (to asparagine) or O-linked (to serine or threonine) [3]. Enzymatic defects of this catabolic pathway cause the accumulation of oligosaccharides (OS) and, depending on the lacking enzyme, result in characteristic profiles of urinary OS, which are helpful for the diagnosis [4]. The different oligosaccharidoses share common clinical features, which include facial dymorphisms, dysostosis multiplex, hepato/splenomegaly, developmental delay and neurological signs, making difficult the differential diagnosis [2]. Biochemically, OS analysis is the first step for the diagnosis of oligosaccharidoses [5]. The initial methods to analyze urinary OS were based on thin layer chromatography (TLC), but this assay has limited analytical sensitivity and specificity due to interfering compounds derived, especially in early infancy, from the diet or from medications [5][6][7].
A more powerful tool for the analysis of OS has been provided thanks to the introduction in diagnostic laboratories of mass spectrometry (MS), which allows to characterize the different OS species through specific multiple reaction monitoring (MRM) transitions. The analysis with a triple quadrupole after sample derivatization enabled the detection of characteristic OS profiles in the urine of patients with different types of oligosaccharidoses and with other storage disorders, and was also suitable for prenatal diagnosis in amniotic fluid [3,8,9]. Other methods, utilizing capillary high performance anion-exchange chromatography mass spectrometry (HPAEC) or matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF/TOF), with or without sample derivatization, have been developed for structural studies and for disease screening [10,12]. Piraud et al. adapted a MALDI-TOF/TOF based method [12] to a triple quadrupole analyzer, and built a powerful technique suitable in diagnostic laboratory for the screening in urine of a large number of oligosaccharidoses and for prenatal diagnosis in amniotic fluid [4]. More recently, an ultra-high performance liquid chromatography mass spectrometry (UHPLC-MS/MS) method for urinary OS analysis requiring sample derivatization has been reported [13]. The rapid evolution of all these techniques has potentially expanded the list of identifiable diseases to other storage disorders which share with oligosaccharidoses an abnormal excretion of compounds related to this catabolic pathway [3,4,[8][9][10][11][12][13]. Furthermore, in Pompe disease, a glycogen storage disorder due to deficiency of the lysosomal enzyme acid α-glucosidase, the profile of urinary OS by TLC analysis shows the presence of a large band which was characterized by UHPLC-MS/MS analysis as the tetrasaccharide 6-α-Dglucopyranosyl-maltotriose (Glc α 1-6Glc α 1-4Glc α 1-4Glc, designed as Glc4) [14,15]. More recently, Glc4 has been suggested as a target biomarker for diagnosis, monitoring disease progression and to evaluate the response to enzyme replacement therapy in Pompe disease [16,17]. Indeed, an abnormal OS profile by TLC analysis [18,19] as been reported in Yunis-Varon syndrome, an inherited disorder of autophagy, a novel disease group listed among storage disorders [1].
In this study, we report a new UHPLC-MS/MS method, not requiring sample derivatization, for the screening of OS in urine and in dried urine spots (DUS) which was applied to identify a large panel of storage disorders.

Urine samples treatment
Urine samples were ultra-filtered with an Amicon Ultra filter 0.5 mL 3 K 96 PK (Merck KGaA, Darmstadt, Germany) and centrifuged for 9 min at 13.000 rpm. Ultrafiltered urine were diluted with milli-Q water (Milli-Q Advantage A10 System, Merckmillipore, Merck KGaA, Darmstadt, Germany) to obtain a final creatinine concentration of 1 mM. The creatinine concentration was determined with the Jaffé method. Samples with a creatinine concentration < 1 mmol/L were not diluted and were adjusted at a creatinine concentration of 1 mM after the analysis.

Dried urine spot (DUS) samples treatment
To prepare DUS samples 2 ml of urine were spotted, both in the laboratory and in the clinic, on the entire absorbing part of the card commonly used for newborn screening (EBF Eastern Business Forms, INC., SC, US) and dried at room temperature for at least 2 h. DUS were shredded and mixed in deionized water for 30 min for the extraction. The entire absorbing part of the newborn screening card was used for the extraction. The extract was then diluted with Milli-Q water to obtain a creatinine concentration of 1 mM. The extract creatinine concentration was determined with the Jaffé method. Extracts with a creatinine concentration < 1 mmol/L were not diluted and were adjusted at a creatinine concentration of 1 mM after the analysis. Fifty μL of the normalized extract from DUS were mixed in a glass tube with 20 µl of the IS working solution, 130 μL of reconstitution buffer, and vortexed.

UHPLC-MS/MS
Five μL of the finally urine or DUS samples respectively prepared as described in "Urine samples treatment" and "Dried urine spot (DUS) samples treatment" sections, were injected in the UHPLC system Agilent 1290 Infinity II (Agilent Technologies, CA, US) for the chromatographic analysis. The chromatography was performed with a Luna Omega SUGAR 100 column Å, 150 × 2.1 mm (Phenomenex, CA, US) at a flow rate of 0.5 mL/min. The mobile phase was a mixture of ammonium formate (Sigma-Aldrich Steinheim, Germany) 5 mM + 0.05% (v/v) FOA (> 96% purity, reagent grade Sigma-Aldrich Steinheim, Germany) dissolved in water (A) and ammonium formate 5 mM + 0.05% (v/v) FOA dissolved in ACN (≥ 99.9% purity for HPLC, gradient grade, Sigma-Aldrich Steinheim, Germany)/ water 90/10 (B). The gradient program is showed in Table 1. The total run time was 27.50 min (15.50 min for detection and 12.00 min for column reconditioning) at a controlled column temperature of 40 °C.

Oligosaccharides detection
MRM was used for spectra acquisition in positive and negative modes, switching the polarity within a single run. Data acquisition and chromatographic peak The positive iQC was treated with the same procedure of samples and added to each batch series to control chromatography quality and the sensitivity of the tandem mass spectrometer. Figure 1 shows the extract ion chromatogram (XIC) of a selection of storage disorders (sialidosis, α-and β-mannosidosis, fucosidosis, aspartylglucosaminuria, GM1 and GM2 gangliosidosis, Pompe disease, Vici syndrome, Yunis-Varon syndrome and Danon disease) presenting a characteristic OS profile [4,15].

Assay validation
For each MRM transition, linearity was estimated with calibration curves created using three samples at different urine concentrations and calculating the correlation coefficient (R 2 ). The differently concentration samples were prepared by mixing and vortexing in a glass tube, as follow: 50 μL of normalized ultra-filtered urine, 20 µL of the IS working solution and 130 μL of reconstitution buffer; 100 μL of normalized ultra-filtered urine, 20 µl of the IS working solution and 80 μL of reconstitution buffer; 150 μL of normalized ultra-filtered urine, 20 µL of the IS working solution and 30 μL of reconstitution buffer. In fresh urine, the R 2 ranged from 0.8039 to 0.9999 in the positive mode and from 0.7270 to 0.9999 in the negative mode. In DUS, R 2 ranged from 0.7701 to 0.9996 in the positive mode and from 0.8340 to 0.9995 in the negative mode.
For each MRM transition, the intra-day precision, expressed as CV%, was assessed by injecting and analyzing 10 times in the same run 5 µL of the finally urine or DUS samples respectively prepared as described in "Urine samples treatment" and "Dried urine spot (DUS) samples treatment" sections; the inter-day precision, expressed as CV%, was assessed by injecting and analyzing 10 times for three different days 5 µL of the finally urine or DUS samples prepared respectively as described in "Urine samples treatment" and "Dried urine spot (DUS) samples treatment" sections.
The intra-day precision ranged from 6 to 24% in the positive and negative mode in fresh urine, from 5 to 25% in the positive mode and from 2 to 24% in the negative mode in DUS. The inter-day precision ranged from 9 to 25% in the positive mode and from 8 to 23% in the negative mode in fresh urine, from 7 to 26% in the positive mode and from 5 to 24% in the negative mode in DUS. Validation data of linearity, intra-day and inter-day precision in urine and DUS are reported in Additional file 1: A and B.

Detection of oligosaccharides in storage disorders
The study allowed to confirm the diagnosis in all disorders presenting known OS profiles. Additional file 2: A and B illustrate the MoM results obtained from all analyzed samples. Each sample was considered positive for a specific disorder when the MoMs of all qualifying and quantifying transitions were increased compared to controls at least 5 times for fresh urine and 2 times for DUS. Some disorders presented with a specific OS profile while in other diseases, the analysis revealed abnormal OS excretion but not-specific profiles. Overall, DUS samples showed lower background interferences when compared to urine samples. Figure 2 shows the scatter charts of the most characteristic transitions in disorders showing specific OS profiles [4].

Disorders with a specific OS profile
Sialidosis Three negative MRM transitions were selected as the most characteristic for the sialyl-OS NeuAc-Hex3-HexNAc2 and one was used for quantitative analysis. We analyzed 1 urine and 1 DUS sample from one patient and both quantitative and qualitative
α-Mannosidosis Eight positive MRM transitions for the three mannosyl-OS, Hex2-HexNAc, Hex3-HexNAc and Hex4-HexNAc were selected as the most characteristic, and 4 were used for quantitative analysis. We analyzed 2 urine samples and 3 DUS obtained from 2 patients and all quantitative and qualitative MRMs were highly increased compared to controls, with a 110-118 fold elevation in urines and 29-245 fold in DUS for the most characteristic transition.
β-Mannosidosis Nine positive and eight negative MRM transitions for Hex-HexNAc and its derivatives were selected as the most characteristic, and 5 were used for quantitative analysis. We analyzed 1 urine sample and 1 DUS from a patient and all quantitative and qualitative MRMs were highly increased compared to controls, with 720 fold elevation in urine and 155 fold in DUS for the most characteristic transition.
Fucosidosis Six positive MRM transitions for the two fucosyl-OS, Fuc-HexNAc-Asn and Fuc-HexNAc2-Hex3, were selected as the most characteristic, and 3 were used for quantitative analysis. We studied 4 patients and analyzed 3 urine samples and 3 DUS and all quantitative and qualitative MRMs were highly increased compared to controls, with 220-470 fold elevation in urine and 152-373 fold in DUS for the most characteristic transition.
Aspartylglucosaminuria Five positive MRM transitions for aspartylglucosamine GlcN-Asn and the glycoasparaginyl-OS Hex-HexNAc-Asn and ten negative MRMs transitions for aspartylglucosamine GlcN-Asn and the glycoasparaginyl-OS NeuAc-Hex-HexNAc-Asn were selected as the most characteristic, and 3 positive and 3 negative MRMs were used for quantitative analysis. We studied 1 patient with aspartylglucosaminuria and analyzed 1 urine sample and 1 DUS and all quantitative and qualitative MRMs were highly increased compared to controls, with 71 fold elevation in urine and 80 fold in DUS for the most characteristic transition.
GM1 gangliosidosis Five positive MRM transitions for the two galactosyl-OS, Hex3-HexNAc2 and Hex5-HexNAc3 were selected as the most characteristic, and 3 were used for quantitative analysis. We studied 7 patients and analyzed 14 urine samples and 4 DUS and all quantitative and qualitative MRMs were highly increased compared to controls, with 9-1160 fold elevation in urine and 69-883 fold in DUS for the most characteristic transition.

GM2 gangliosidosis (Sandhoff & Tay-Sachs diseases)
Seven positive MRM transitions for the three N-acetylglucosaminyl-OS, Hex2-HexNAc2, Hex3-HexNAc3 and Hex3-HexNAc4 were selected as the most characteristic of GM2 gangliosidosis O variant (Sandhoff disease) and 4 were used for quantitative analysis. We studied 4 Sandhoff patients and analyzed 6 urine and 1 DUS. In all samples, quantitative and qualitative MRMs were highly increased compared to controls, with 99-2102 fold elevation in urine and 79 fold in DUS for the most characteristic transition.
In the 3 patients with GM2 gangliosidosis B variant (Tay-Sachs disease) the OS analysis in urine and DUS did not display abnormalities. Figure 3 shows the scatter charts of disorders presenting an increased excretion of Glc4 which included, besides Pompe disease, also the autophagy related disorders Vici and Yunis-Varon syndromes, and Danon disease.
Pompe disease (glycogen storage disease type II) Two negative MRM transitions of the tetrasaccharide Glc4 and of its isomer maltotetraose (M4) were selected as the most characteristic. We studied 3 untreated patients, two with infantile-onset and one with late-onset Pompe disease, and analyzed 4 urine and 2 DUS. Glc4 was increased compared to controls, with a 10-72 fold elevation in urine and 11-13 fold in DUS, respectively. Lower levels were detected in the patient with milder phenotype. Yunis-Varon syndrome The DUS sample from the patient with Yunis-Varon syndrome showed a 51 fold elevation of Glc4.

Danon disease (Glycogen storage disease type II B)
We studied 1 patient with Danon disease and analyzed 1 urine and 1 DUS. Glc4 was increased compared to controls, with a 45 fold elevation in urine and 75 fold in DUS.

Disorders with a not-specific OS profile
Mucolipidosis type II, and type III We studied 2 patients with mucolipidosis type II and analyzed 3 urines and 1 DUS. In all urine samples OS analysis revealed mixed non specific profiles with 8/8 positive transitions of α-mannosidosis and 5/5 positive transitions of GM1 gangliosidosis.
We studied 2 patients with mucolipidosis type III and analyzed 2 urines and 2 DUS. In urine 5/5 positive MRM transitions of GM1 gangliosidosis were variably increased, however to a lesser extent than in GM1.
Mucopolysaccharidosis IVB DUS from 5 patients were available and all samples displayed increased levels of Glc4 however to a lesser extent than the positivity threshold.

Discussion
The diagnosis of storage disorders is often challenging due the presence of common clinical features and great variability in symptoms, and requires a complex approach which includes the analysis of target biomarkers in biological fluids, the measurement of enzymatic activities in leukocytes and/or fibroblasts and the confirmatory diagnosis by mutation analysis [20]. The availability of novel and specific therapies for a large number of diseases, including oligosaccharidoses, has increased the medical demand of reliable diagnostic techniques to  Table 2).
All examined oligosaccharidoses (i.e. sialidosis, α-/βmannosidosis, fucosidosis, aspartylglucosaminuria) and both GM1 and GM2 (Sandhoff type only) gangliosidosis displayed specific OS profiles, as shown by the presence of single or multiple qualifying and quantifying MRM transitions [4]. In other diseases, such as mucolipidosis type II-III, and mucopolysaccharidosis type IVB the analysis revealed abnormal (non-specific) OS excretion, with mixed patterns combining MRM transitions of different disorders. Each OS was identified through its specific MRM transitions in positive and in negative modes [4]. Given the chemical similarity, we included in the OS analysis also the negative MRM transition of Glc4, the urinary biomarker of Pompe disease [15,17]. This implementation allowed to confirm the diagnosis of Pompe disease, with some variations in Glc4 excretion related to disease phenotypic severity.
Besides Pompe disease, Glc4 levels were markedly increased also in three disorders belonging to the autophagy machinery [21][22][23]. These include Vici syndrome and Yunis-Varon syndrome, two diseases characterized by a complex multisystem phenotype, and Danon disease, a disorder with clinical manifestations very similar to Pompe disease [24][25][26][27][28]. As seen in glycogen storage disorders, all these conditions share a cardiomuscular involvement with increased glycogen storage and variable vacuoles accumulation detectable on light microscopy at the level of skeletal and cardiac muscle,. In previous studies on Yunis-Varon syndrome, abnormalities of urinary OS by TLC analysis have been reported [18,19]. However, given the technical limitation of this method, a precise identification of these compounds was missed in the original reports. Our UHPLC-MS/MS method confirmed a relevant increase of OS excretion in Yunis-Varon syndrome and allowed to identify Glc4 as the disease-target compound. As a novel finding, upon evaluation of samples from patients with Vici syndrome and Danon disaese we detected a striking increase of Glc4 in urine. The structural and pathogenetic similarities with Pompe disease indicate that the increased urinary excretion of Glc4 in these three autophagy-related disorders reflects the abnormal muscle glycogen breakdown [25,26], as also seen in other muscle disorders causing glycogen storage such as GSD type 3 and 6 [15,29]. Glc4 could therefore serve as a biomarker to screen a wider range of storage disoders, including autophagyrelated diseases, presenting cardiomuscular sings with glycogen storage. From a technical perspective, most of recently reported methods using triple quadrupole for the analysis of OS require sample derivatization with 1-phenyl-3-methyl-5-pyrazolone, making the preanalitycal phases of these methods time consuming [3,8,9]. Two MALDI-TOF/ TOF mass spectrometry methods, lacking internal standard, and suitable for disease screening have been reported [11,12], with one not requiring sample derivatization [12].
The advantages of our screening method for OS analysis include a short preanalytical phase, not requiring sample derivatization, the use of UHPLC-MS/MS platform, a more widely available apparatus in diagnostic laboratories than MALDI-TOF/TOF mass spectrometry, and the possibility to perform OS analysis in urine and DUS, with comparable diagnostic results, thus making simpler and easier sample shipment and storage. The chromatography was performed with a HILIC phase column, which was chosen for its ability to improve retention and selective separation of sugar-related compounds. Unlike the more commonly used aminic columns, the HILIC column showed an enhanced lifetime because of the stationary phase composed with highly robust and thermally modified fully porous particles. The gradient was optimized to obtain the elution of all compounds in about 15 min, with a residual 12 min for column reconditioning. The assay validation of the method in urine and DUS, showed that the correlation coefficient and the intra/inter-day precision of positive and negative MRM transitions were comparable in the two biological matrices. Moreover, in the control population, we noted that most of quantitative transitions showed significant differences for age below 6 month (12) versus eldest age (63) (see Additional file 3).
For each MRM, results obtained in patients and controls showed a clear separation of the two populations, both in urine and in DUS, without overlapping.

Conclusions
This new UHPLC/MS-MS method, not requiring sample derivatized and allowing the rapid and easy detection of OS in urine and DUS, expands the screening of storage disorders from oligosaccharidoses to other diseases, including Pompe disease and the novel category of inherited disorders of autophagy causing abnormal muscle glycogen breakdown.