Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Feb 3.
Published in final edited form as: J Am Soc Mass Spectrom. 2020 Dec 10;32(2):436–443. doi: 10.1021/jasms.0c00317

A clinically viable assay for monitoring uromodulin glycosylation

Milani Wijeweera Patabandige 1, Eden P Go 1, Heather Desaire 1,*
PMCID: PMC8541689  NIHMSID: NIHMS1748585  PMID: 33301684

Abstract

Uromodulin, also known as the Tamm-Horsfall protein or THP, is the most abundant protein excreted in human urine. It is associated with the progression of kidney diseases; therefore, changes in the glycosylation profile of this protein could serve as a potential biomarker for kidney health. The typical glycomics analysis approaches used to quantify uromodulin glycosylation involve time-consuming and tedious glycoprotein isolation and labeling steps, which limit their utility in clinical glycomics assays, where sample throughput is important. Herein we introduce a radically simplified sample preparation workflow, with direct ESI-MS analysis, enabling the quantitation of N-linked glycans that originate from uromodulin. The method omits any glycan labeling steps but includes steps to reduce the salt content of the samples, thereby minimizing ion suppression. The method is effective for quantifying subtle glycosylation differences of uromodulin samples derived from different biological states. As a proof of concept, glycosylation from samples that differ by pregnancy status were shown to be differentiable.

Keywords: glycan, uromodulin, glycomics, mass spectrometry, biomarker, THP

Graphical Abstract

graphic file with name nihms-1748585-f0001.jpg

INTRODUCTION

Protein glycosylation, where glycans are covalently attached to the proteins through the side chains of certain amino acid residues, is one of the most abundant post-translational modifications (PTMs) found in nature. This modification can modulate both protein structure and function.1 Additionally, protein glycosylation is highly sensitive to changes in the cellular environment, and as a result, proteins become aberrantly glycosylated during the progression of many diseases, such as cancers,2–4 kidney diseases,5–6 arthritis,7 and Parkinson’s disease.8 Therefore, the relative abundances of these altered glycans can be used as a biomarker, indicating a either a healthy state or a disease state.6–7 Glycans on proteins that are only highly expressed in one organ are particularly attractive biomarker targets, since a protein-specific glycosylation analysis would afford the opportunity to probe the disease state of the organ from which the protein was expressed. One important example of a protein whose glycosylation could serve as a biomarker is uromodulin, a protein that is only highly expressed in the kidneys.

Uromodulin, also known as the Tamm-Horsfall protein or THP, is the most abundant glycoprotein excreted in human urine with a daily excretion rate of 50 – 100 mg.5, 9–10 It is important for preventing kidney stone formation5 and urinary tract infections.11 Uromodulin is 94 kDa in size, and glycans represent approximately 25 – 30% of its weight. This glycoprotein contains eight potential N-linked glycosylation sites, of which 7 are reported to be glycosylated; these sites are mainly occupied by various complex-type di-, tri-, and tetra-antennary glycans in addition to the minute level of high mannose-type glycans.10, 12 One unique feature of the uromodulin glycosylation profile is the acidic nature of many of the reported glycans; these glycans can contain sialic acids and/or sulfate substituents, such as 3-O-sulfated galactose (Gal3S) and/or 4-O-sulfated N-acetylgalactosamine (GalNAc4S).9, 13–14 Analysis of these different glycans of uromodulin is important because of their significance in distinguishing samples of various biological states; for examples, reduced levels of overall glycosylation and sialylation of uromodulin glycoprotein was reported in patients with interstitial cystitis10 and kidney stones.5 Therefore, development of efficient methods for the sensitive detection of uromodulin glycans is important in clinical studies, as the glycans may be useful biomarkers for various diseases.

While analysis of uromodulin glycosylation has the potential to improve diagnosis and treatment of a variety of kidney-related conditions, a simple, and accurate assay that would be clinically viable is not currently available. The purification of the protein from urine is currently done using complex sample preparation procedures using either diatomaceous earth,5, 15 or salt precipitation.10, 12 Furthermore, once the glycans are purified and released, general application of existing glycomics assays introduces many more additional steps; these steps typically include glycan labeling and post-sample clean-up steps, which are laborious and time-consuming. To resolve the challenges of laborious sample preparation methods described above, and to provide kidney researchers with the opportunity to readily assess uromodulin glycosylation changes for improving the diagnosis and prognosis of kidney diseases, we developed a clinically viable procedure for the analysis of uromodulin glycans.

Because uromodulin is, by far, the most abundant protein in urine,5, 9–10 it is possible to develop a radically simplified procedure to generate highly enriched uromodulin samples without the need for purification from diatomaceous earth, which is the most common protocol. In the protocol described herein, all proteins below 50 kD are removed using a molecular weight cut-off filter, removing many potential low molecular weight interferents. IgG is the next most abundant glycoprotein in urine after uromodulin,16–20 but its concentration is still ~ 100X lower than that of uromodulin,16 so its glycosylation in general would minimally impact this assay. To further reduce the minimal impact of IgG, the uromodulin analysis is exclusively conducted in the negative ion mode, which is optimal for uromodulin glycans but a poor choice if the goal is to detect IgG glycans, since IgG’s main glycoforms are neither sialylated nor sulfated.21–23 Overall, this enrichment procedure and analysis in negative ion mode optimizes the balance between the need for samples that are highly enriched in uromodulin with the need for a radically simplified workflow that could be applied on large sample sets.

Aside from simplifying the protein enrichment step, the other aspect of the work described here, which is necessary to advance researchers’ ability to analyze uromodulin on large banks of clinical samples, is to address the labeling steps that are generally thought to be necessary prior to a quantitative glycomics analysis. These steps usually involve either permethylation, which requires anhydrous conditions, or reductive amination, a two-step reaction that generates hazardous waste. After sample clean-up, the analysis of labeled glycans is typically accomplished by HPLC with fluorescence detection or MALDI-MS.5, 10, 12, 24

We hypothesized that since uromodulin is available in abundant quantities, the typical glycan labeling and enrichment steps would not be necessary if the glycans were detected by ESI-MS in the negative ion mode. While these types of glycans have been detected by (−)ESI-MS previously, at relatively high concentrations, this approach would not be possible if high salt concentrations are present. Thus, the salts that are present would have to be removed or minimized. We therefore developed an efficient strategy for salt removal that could be applied on large sample sets. The method was first developed using fetuin as a model glycoprotein, due to its availability in large quantities and its glycosylation similarity to uromodulin; then it was applied to purified uromodulin, followed by uromodulin that was enriched from human urine samples of two different biological states. The method proved to be highly reproducible for multiple samples of the same biological state, delivering very tight within-group clustering by PCA, while showing its utility in classifying samples based on uromodulin-specific glycans from different biological states.

EXPERIMENTAL/METHODS

Materials and Reagents

Bovine fetuin and human uromodulin standards were purchased from Sigma Aldrich (St. Louis, MO) and BioVendor (Asheville, NC), respectively. PNGase F (500, 000 units/mL) was from New England Biolabs (Ipswich, MA). Single donor human urine from a de-identified healthy female and a de-identified third trimester pregnant female were purchased from Innovative Research (Novi, MI). All the chemical reagents used for this study were of analytical grade or better.

PNGase F Enzyme Preparation

PNGase F (1 μL, 500 units) was diluted to 100 μL (for fetuin) or PNGase F (2 μL, 1000 units) was diluted to 200 μL ( for uromodulin and urine samples) with NH4HCO3 buffer (10 mM, pH 7.5) and concentrated in a pre-rinsed 10 kD MWCO filter (14000 g × 15 min) to approximately 50 μL of final volume. Then, the concentrated enzyme solution was diluted by a factor of 10 with the buffer (10 mM NH4HCO3, pH 7.5) followed by another concentrating step (14000 g × 15 min) to obtain approximately 35 μL of final enzyme solution. Finally, the PNGase F concentrate was collected through reverse spin (1000 g × 2 min).

N-linked Glycan Release from Fetuin and Uromodulin Standard

Glycoprotein samples (50 μg) were obtained by transferring appropriate volumes of fetuin (5 mg/mL in 10 mM NH4HCO3 buffer) and uromodulin (2 mg/mL in water) glycoprotein stock solutions. Then, the final glycoprotein solution concentration was adjusted to 2 mg/mL by diluting with the buffer (for fetuin). The PNGase F enzyme solution (35 μL), prepared as described in above, was added to each glycoprotein solution and incubated for 24 hours at 37 °C. After the incubation period, samples were diluted up to 500 μL with 50:50 MeOH: H2O, transferred to pre-rinsed 10 kD MWCO filters, and centrifuged (14000 g × 15 min) to collect the filtrate with released N-linked glycans. The resultant filtrates were then concentrated in a centrivap vacuum concentrator, to yield a final volume of 15 μL, and stored at −20 °C. Prior to the direct ESI-MS analysis, N-linked glycans’ concentrate solutions of fetuin and uromodulin were diluted approximately 70 and 17 times with 50:50 MeOH: H2O, respectively.

Enrichment of Uromodulin from Human Urine and N-linked Glycan Release

Uromodulin was enriched from the crude urine samples obtained from a healthy and a pregnant individual. Urine samples, which were stored at − 80 °C, were thawed to room temperature, vortexed for 10 seconds, and then aliquoted into 10 mL fractions. Three of the 10 mL aliquots of each urine sample (healthy and pregnant) were vortexed for 10 seconds, followed by transferring the vortexed urine samples into pre-rinsed Amicon 50 kD molecular weight cut-off filters (Millipore), separately. The samples were centrifuged at 3500 rpm for 15 mins (at 4 °C) to reduce the volume approximately down to 300 μL. Then the protein concentrate was washed by adding the buffer up to 15 mL and the samples were centrifuged at 3500 rpm for 15 mins. The washing step was repeated another two times to obtain the final concentrate (~ 300 μL) for each triplicate samples; the retentate was carefully transferred to a pre-rinsed 50 kD MWCO filter (0.5 mL); centrifuged at 7000 rpm for 30 mins to further reduce the sample volume to 50 μL. Then the final concentrate was collected by performing reverse spin at 1000 g for 2 min. After that, the N-linked glycan release was performed by adding salt-reduced PNGase F enzyme (1000 units), which had been subjected to 100 times of dilution with the buffer, centrifugation, and second-buffer exchange step, as described above. Thereafter, the samples were incubated, N-linked glycans were collected, samples were concentrated, and stored as described for the protein standards. Finally, the concentrated N-linked glycan solutions, which were generated from triplicate samples of a healthy and a pregnant urine, were diluted 17 times with 50:50 MeOH: H2O (v/v), prior to the direct ESI-MS analysis.

Direct ESI-MS Analysis of Released N-linked Glycans

Direct ESI-MS analysis of the released N-linked glycans was performed using an Orbitrap Fusion Tribrid mass spectrometer (Thermoscientific, San Jose, CA). The mass spectrometer was operated in negative ion mode with a sample injection flow rate of 5 μL/min. The heated-electrospray source was held at 2.4 kV while the ion transfer tube temperature, sheath and auxiliary gas flow rates were set at 300 °C, 10, and 8 Arb units, respectively. The full MS scans for the m/z range of (750 – 1600) were acquired in the Orbitrap with a resolution of 120 k (at m/z 200). The AGC target value for the full MS scan was 4×105, and the maximum injection time was 100 ms. For full MS data of fetuin and uromodulin standards, 30 scans were averaged, while 100 scans were averaged during the analysis of uromodulin N-linked glycans extracted from urine samples.

Both CID and HCD data for the N-linked glycans derived from fetuin and uromodulin standards were generated to confirm the glycan compositions in the samples, and additional MS/MS data were acquired on the uromodulin extracted from normal and pregnant urine samples to confirm the glycan compositions that were not observed in the uromodulin standard samples. For MS/MS, the isolation window for the precursor ions were set as 2 Da, activation time, and activation qz were 10 ms and 0.25, respectively. The selected precursor ions were fragmented by applying appropriate normalized collision energies ranging between 30% – 35%. All the MS data were acquired by using Xcalibur software, version 4.2 (Thermoscientific, San Jose, CA).

Data Analysis

Supplementary Table S1 represents a list of 40 potential uromodulin N-linked glycans tabulated from previous reports.5, 9, 12–14, 25 The resultant full MS data were searched against the N-linked glycans listed in Supplementary Table S1 for glycan assignment; these assignments were done by comparing the theoretical monoisotopic masses of listed N-linked glycans with their experimental m/z’s, and identified the N-linked glycans within 5 ppm mass error. These assignments were further confirmed by manually analyzing the resultant MS/MS data.

Quantifying the glycans:

Full MS scans; 30 and 100 scans for N-linked glycan samples derived from standard glycoproteins and urine samples, respectively, were averaged. The first four isotopic peaks’ raw abundances of each N-linked glycan were summed over all the identified charge states and adducts (protonated forms and sodiated adducts). Then, the percent of each glycan, based on its peak intensity was calculated by dividing a particular glycan peak intensity by the total N-glycan peak intensity of the analyzed sample, multiplied by 100.26

Classification of sample groups:

Principal component analysis was conducted in R, version 3.5.1, and the data was plotted using the package “factoextra”. The data were centered and scaled prior to the PCA transformation.

RESULTS AND DISCUSSION

Overview of the Label-free N-linked Glycan Quantitation Approach

Sample preparation is one critical step in the clinical biomarker discovery field that affects the final throughput of the method. Therefore, development of simple and efficient sample preparation strategies is necessary in quantitative glycomics analysis. Figure 1 shows a schematic diagram of a simple N-linked glycan preparation protocol that enables efficient release and direct ESI-MS quantitation of N-linked glycans without labeling. In this protocol, as the first step, N-linked glycans are released from the glycoproteins of interest by incubating them with salt-reduced PNGase F enzyme. In direct ESI-MS analysis, the presence of salts can increase the ion suppression, reduce the stability of electrospray, and affect the sensitivity of the analysis.27–29 Thus, reducing the amount of salt present in the samples being analyzed is important prior to the direct ESI-MS analysis. Both urine samples and PNGase F used in this study contained salts; thus, these samples needed to be desalted prior to any other sample processing steps. Urine samples typically contain high salt concentration;30 thus, they were desalted by following several washing steps as shown in Figure 1. Additionally, PNGase F, which contains 50 mM of NaCl, was also buffer exchanged several times to minimize the initial salt concentration by at least three orders of magnitude before adding to the glycoprotein solution. Then, after incubation, the glycans were directly extracted, concentrated and analyzed with direct ESI-MS in the negative ion mode without prior labeling.

Figure 1.

Figure 1.

Open in a new tab

Experimental workflow for N-linked glycan profiling of fetuin and uromodulin standard glycoproteins and uromodulin, extracted from urine.

Figure 1 indicates the time requirements for the three longest steps, including a 24-hour incubation with PNGase F. All the steps except for the final MS analysis can be done in parallel, so the time-per-sample goes down as the size of the sample set goes up, and if minimizing total time were critical, the incubation could be done much more rapidly at the cost of more enzyme per sample.

This protocol differs from other standard glycomics approaches because it omits a labeling step. Glycan labeling prior to the MS analysis improves glycan’s ionization efficiency;1, 31–34 however, the labeling process and post-sample clean up steps are time-consuming and potentially introduce additional variability into the analysis. The developed protocol is more rapid, while allowing highly reproducible quantification of the relevant N-linked glycans derived from standard glycoproteins and uromodulin, enriched from human urine samples via direct ESI-MS in the negative ion mode.

The ultimate goal of this study was to develop a simple, label-free, direct ESI-MS approach to effectively profile N-linked glycosylation of uromodulin, which is mainly glycosylated with negatively charged glycans that ionize well in the negative ion mode. We performed the initial method development on a standard glycoprotein, fetuin, that has a similar glycosylation profile and is more affordable, facilitating method development. After method optimization, reproducibility and instrument precision was demonstrated with fetuin before moving on to the more costly protein, uromodulin.

Reproducibility of the Method

Higher reproducibility of a method generally permits enhanced sensitivity towards differentiating minor changes across multiple samples with a higher confidence.35 If the method is reproducible, small differences that are introduced during the sample preparation steps, such as PNGase F release, extraction of N-linked glycans, and dilution of concentrated N-linked glycan samples prior to the MS analysis, should not affect the final quantitation results. Therefore, to test the reproducibility of the quantitative sample preparation workflow, fetuin (50 μg) obtained from the same stock solution was subjected to N-linked glycan release protocol on three different days and analyzed by (−)ESI-MS under identical instrumental parameters, as described in the experimental section. Supplementary Figure S1 represents a direct (−)ESI-MS N-linked glycan profile derived from fetuin; the resulting N-linked glycans were assigned across multiple charge states and multiple adducts (protonated and sodiated). After obtaining the N-linked glycan profile of fetuin, the relative glycan peak percent of the individual glycans were calculated as described in the data analysis section. Quantitative data for four different complex-type N-linked glycans of fetuin over three separate sample preparations is shown in Figure 2A. The quantitative data were highly reproducible for the major N-linked glycans.

Figure 2.

Figure 2.

Open in a new tab

Method reproducibility (A) and instrument precision (B) calculated for four different fetuin N-linked glycans. The relative glycan peak percent for each N-linked glycan composition is plotted for three different sample preparations from the same stock (A) and for the same sample analyzed over week 0, week 22, and week 24 (B). The N-linked glycans are rank ordered from the largest percentage to the smallest percentage. Less than 8% CV was recorded for 3 major fetuin N-linked glycans and about 24% CV was calculated for the least abundant fetuin N-linked glycan in both A and B.

More detailed information about the quantitative results can be found in Supplementary Table S2. It summarizes the raw abundances, mean relative glycan peak percentages, and coefficients of variation values calculated for four fetuin N-linked glycans: H6N5S3, H5N4S2, H6N5S4, and H6N5S2. The glycan abundances were 67%, 19%, 14%, and 0.067% for the four fetuin N-linked glycans, respectively. Additionally, the coefficient of variation values were 1.2%, 7.7%, 7.1%, and 24%, respectively. The results showed that the method is highly reproducible for all major fetuin N-linked glycans, except for the least abundant H6N5S2 glycan peak, which represented less than 1% of the sample. While the CV for the least abundant sample is considerably higher than the others, the variability observed for this species would likely be much lower than the biological variability that would be obtained when analyzing large sets of clinical samples. Thus, we anticipate that glycans whose abundance are down to the level of about half a percent of the total glycan pool could potentially serve as biomarkers in some circumstances, for example, if they are at low abundance in a healthy state but much more abundant in a disease state.

Instrument Precision

Label-free quantitative assays performed with mass spectrometers can be subjected to reproducibility issues over lengthy time periods, as a result of slight changes occurring in the instrument conditions.26 Therefore, the instrument precision over the time period of the analysis was tested by analyzing a fetuin sample on three different days: at week 0, week 22, and week 24. After the first analysis performed on week 0, the released N-linked glycans from the sample were stored at −20 °C and re-analyzed in week 22 and week 24 under identical ESI-MS conditions.

Supplementary Table S3 shows the recorded raw abundances, the mean relative glycan peak percentages, and the coefficients of variation calculated for four fetuin N-linked glycans. The rank order recorded for the fetuin N-linked glycans was consistent over the analysis at different time points, and the coefficients of variation of relative glycan peak percentages calculated across all the N-linked glycans were lower than 6%, except for the least abundant glycan peak: H6N5S2, which showed about 24% of coefficient of variation. Figure 2B illustrates the relative glycan peak percentages recorded for four fetuin N-linked glycans across three time points, and these data clearly show that the instrument performance remained unchanged during the time period of the study.

Based on the initial quantitative data obtained with the fetuin N-linked glycans, the method showed to be highly reproducible over multiple sample preparations and under the MS conditions used for the study. Therefore, we tested the applicability of the developed label-free direct ESI-MS method towards efficient quantitation of N-linked glycans derived from uromodulin.

Quantitation of Human Uromodulin N-linked Glycans

The N-linked glycans of human uromodulin were released and extracted from 50 μg of a purchased glycoprotein standard; the released glycans were concentrated, diluted, and analyzed directly by (−)ESI-MS. A representative mass spectrum is in Figure 3. The glycans were assigned by comparing the high-resolution MS data to the masses of glycans that had been assigned from uromodulin previously, and assignments were confirmed with MS/MS data as described in the experimental section. This procedure resulted in the assignment of twenty-eight uromodulin glycans of a possible forty that had been reported across multiple investigations previously.5, 9, 12–14, 25 See Supplementary Table S1 for the full glycan library.

Figure 3.

Figure 3.

Open in a new tab

A representative uromodulin N-linked glycome profile recorded with direct ESI-MS in negative ion mode. The eighteen most abundant glycan compositions of the 28 identified glycans are shown. Glycan signals’ deprotonated and sodiated adducts ([deglycosylated glycan mass+Na+-4H+]3−) are assigned. Monosaccharide units: blue square (N-acetylglucosamine), green circle (mannose), yellow circle (galactose), purple diamond (N-acetyl neuraminic acid), red triangle (fucose), yellow square (N-acetylgalactosamine), and white star (sulfate groups).

Why were only 28 of the 40 glycans detected? The undetected glycans may be present at very low abundance, <1/200th of the glycan pool, or they may be truly absent. Further optimizing the method to identify these potentially present species was not considered because all 40 glycoforms could be detected in the uromodulin sample from a pregnant donor, even though the sample was injected at a lower concentration. This experiment is elaborated on in the urinalysis section, below. Because all 40 glycoforms are detectable in another sample at lower concentration, we expect that the reason they were not detected in the uromodulin standard is likely because of biological differences in the samples: the uromodulin standard is derived from male urine, and female/pregnant urine was used when these forms were detected. This hypothesis is the subject of a future investigation.

Three analyses of three different lots of uromodulin showed consistent glycosylation profiles. Figure 4 illustrates the relative distribution of 28 N-linked glycan compositions quantified for the three uromodulin standards, and Supplementary Table S4 contains the glycans’ abundances for each replicate. Among the quantified N-linked glycans, G1 glycan with H7N6F1S4 composition, which is reported to be a complex-type, tetra-antennary glycan, was the most abundant in all the analyzed uromodulin standard samples, while contributing about 49% to the total glycan pool. Successively, G2 glycan with H7N6S4 composition followed the G1 glycan, while contributing about 12% to the total glycan pool. Among the other glycan compositions quantified for uromodulin standards, 12 glycan compositions (G3 – G14) showed relative glycan peak percentages lower than 10%, but higher than 1%, while the rest of the 14 glycan compositions (G15 – G28) contributed less than 1% to the total glycan pool.

Figure 4.

Figure 4.

Open in a new tab

Uromodulin standards’ N-linked glycans’ relative peak percent recorded across 28 glycan compositions. G1 to G28 glycan labeling is in consistent with the N-linked glycan list provided in Supplementary Table S1. Ustd1, Ustd2, and Ustd3 are three different uromodulin standards generated from three different stocks; prepared in three different days; analyzed under identical negative ESI-MS conditions.

Lot-to-lot reproducibility was high for the major glycoforms. As shown in Supplementary Table S4, the method yielded less than 8% of coefficients of variation for the relative glycan peak percentages calculated for the three most abundant glycan compositions; these three glycans: G1, G2, and G3 contributed about 49%, 12%, and 8.6% to the total glycan pool. However, the coefficients of variation for all the other glycan compositions, except for G5 and G12, showed relatively higher CV values; this might be a result of lot-to-lot variation of the uromodulin standards used for this study. The higher variation in the quantities of the less-abundant glycans is likely a result of fluctuation in uromodulin glycosylation among different individuals. Clearly biological variability in the glycosylation profile would also be expected in a large clinical study. In order for the glycosylation on uromodulin to be useful as a biomarker for kidney disease, the within-group variance (biological variability) must be lower than the between-group variance (variation between the healthy state and the disease state). Therefore, this aspect would need to be carefully addressed in any clinical assay of kidney disease.

Quantitation of Uromodulin N-linked Glycans Extracted from Human Urine

We next extended the developed approach to analyze N-linked glycans of uromodulin, enriched from human urine and derived from two different biological states. As uromodulin is 100 times more abundant than other glycoproteins in urine, we performed direct filtration to enrich uromodulin from 10 mL aliquots of urine samples of a pregnant and a nonpregnant, healthy woman. Briefly, urine samples were passed through 50 kD MWCO filters, removing all lower molecular weight proteins; then, the resulting uromodulin-enriched urinary proteins were desalted by performing multiple washing steps, all prior to the N-linked glycan release and quantitation. See Figure 1 for uromodulin isolation and N-linked glycan quantitation workflow.

Supplementary Table S5 includes all the N-linked glycans quantified for uromodulin extracted from triplicate samples of a normal urine (NU1, NU2, and NU3) and a pregnant urine sample (PU1, PU2, and PU3), along with three uromodulin standard samples. The method allowed quantitation of a total of 31 and 40 N-linked glycan compositions for NU and PU samples, respectively. The increased number of identified glycoforms in these samples is not likely due to higher starting amounts of sample from the urine isolates; the uromodulin standard produced peaks with higher ion counts for the major glycoforms, indicating it was the more concentrated sample. Rather, the additional glycoforms detected in the samples from urine might be attributable to sex-based differences in glycosylation. Both the normal and pregnant urine were from females, while the uromodulin standard was derived from male urine. In all three samples, the G1 glycan, with H7N6F1S4 composition, is the highest intense glycan peak, although the abundance of this peak varies. The mean relative peak percentage calculated for the G1 glycan is about 49%, 24% and 18% for uromodulin standard, uromodulin, extracted from normal urine, and from pregnant urine samples, respectively. Apart from the most abundant glycan composition, the rest of the glycans quantified for both NU and PU samples contributed less than 12% each to the total glycan pool, while about half of these glycans contributed less than 1% each to the glycan profile.

When the quantitation reproducibility of the developed label-free (−)ESI-MS method is considered; the method proved to be highly reproducible for quantifying N-linked glycans of uromodulin extracted from human urine. For both NU and PU sample groups, a very high within-group reproducibility was observed; the CV values obtained for all the N-linked glycans of NU sample group were below 9%, except for very low abundant G32 and G37 glycans; and even a lower (<3%) CV’s were reported for glycan compositions that contributed more than 1% to the total glycan pool. Similarly, for pregnant urine triplicates, less than 13% of CV values were reported for all of 40 N-linked glycans, except for one glycan composition. Overall, these results clearly showed that the presented approach is highly reproducible while showing its applicability towards targeted quantification of N-linked glycans derived from complex biological matrices.

Quantitation of Glycosylation Differences Among Three Sample Groups

Figure 5 represents the relative distribution of uromodulin N-linked glycans derived from triplicates of the three sample groups in this study: the uromodulin standard, a single donor, nonpregnant female sample, and a single donor, pregnant female sample. Among the quantified N-linked glycans on these three different sample types, many of the glycans showed subtle glycosylation differences between each group, but the differences were larger than the variability within each group. For an example, G22-G24 glycans in NU samples contributed less than 1.5% to the total glycan pool, but the within group reproducibility obtained for these three glycans were very high, while showing CV values below 5%. Similarly, these three glycans contributed less than 0.6% to the total glycan pool of the PU samples; but, still the within-group reproducibility was high and reported CV values less than 8%. This example, along with the data presented in the Supplementary Table S5, clearly show that the developed method provides highly reproducible data over multiple sample preparations and over multiple analyses, and that this reproducibility allows for the detection of subtle glycosylation changes from one sample type to the next.

Figure 5.

Figure 5.

Open in a new tab

N-linked glycans’ Relative peak percentages reported for three separate uromodulin standards (Ustd1, Ustd2, Ustd3), uromodulin, extracted from triplicate samples of a normal urine (NU1, NU2, NU3) and a pregnant urine sample (PU1, PU2, PU3). Glycosylation differences observed in each group were very subtle; however, the data showed high within-group reproducibility.

Figure 6 represents PCA data generated for three different sample groups; normal urine (group 1), pregnant urine (group 2), and uromodulin standards (group 3). These data clearly show that these three groups are unique and clearly separable, even though the glycosylation differences observed among the groups are subtle. For both group 1 and group 2 samples, within-group clustering was very tight; this is because of the high reproducibility provided by the developed approach. Even for group 3, variability within the group was not very broad. The greater spread in this group was very likely to be a result of lot-to-lot variability of the uromodulin standards, yet, this variability is very small compared to the biological variability among the three groups.

Figure 6.

Figure 6.

Open in a new tab

Classification of three sample groups based on their glycosylation data by using Principal component analysis (PCA). Group 1, group 2, and group 3 represent samples of normal urine (NU), pregnant urine (PU), and uromodulin standards (Ustd), respectively. The data clearly show that all three sample groups are unique, even though the glycan differences within them are subtle.

CONCLUSION

We developed a higher-throughput, direct ESI-MS approach to quantify N-linked glycans that ionize well in the negative ion mode. The method is straightforward and omits any glycan labeling steps, which typically require additional post-sample clean up steps prior to the analysis. The developed (−)ESI-MS method was applied to quantify N-linked glycans of a standard glycoprotein, fetuin; it proved to be highly reproducible across multiple sample preparations and multiple analyses. Then, the method was extended to quantify N-linked glycans of uromodulin standards and uromodulin enriched directly from human urine samples of two different biological states. The observed glycosylation differences in uromodulin were subtle between each group; however, within-group reproducibility provided by the method was very high. Moreover, all the analyzed samples were clearly separable into distinct, sample-related clusters by PCA. Therefore, this method can be applied in quantitative glycomics studies, as it is a simple, straightforward one, which effectively permits highly reproducible quantitation data and allows for samples to be discriminated from each other even when the glycosylation differences between different sample types are subtle.

Supplementary Material

Supplementary legend and figs

Supporting Information

Figure S1. Negative ion mode direct ESI-MS spectrum of released label-free N-linked glycan profile of fetuin sample.

Table S1. The list of 40 potential N-linked glycans for uromodulin.

Table S5

Table S5. Data from urinary uromodulin samples and uromodulin standards: Raw abundances, mean relative glycan peak percentages and coefficients of variation.

Table S4

Table S4. Data from three uromodulin standards: Raw abundances, mean relative glycan peak percentages and coefficients of variation.

Table S2 and S3

Table S2. Data from reproducibility study: Raw abundances, mean relative glycan peak percentages and coefficients of variation for fetuin.

Table S3. Data from instrument precision study: Raw abundances, mean relative glycan peak percentages and coefficients of variation for fetuin.

ACKNOWLEDGEMENTS

This work was funded by NIH grant R35GM130354 to Heather Desaire. We are very grateful for the generous support.

REFERENCES

  • 1.Liu S; Cheng L; Fu Y; Liu BF; Liu X, Characterization of IgG N-glycome profile in colorectal cancer progression by MALDI-TOF-MS. J Proteomics 2018, 181, 225–237. [DOI] [PubMed] [Google Scholar]
  • 2.Jia X; Chen J; Sun S; Yang W; Yang S; Shah P; Hoti N; Veltri B; Zhang H, Detection of aggressive prostate cancer associated glycoproteins in urine using glycoproteomics and mass spectrometry. Proteomics 2016, 16 (23), 2989–2996. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Kamiyama T; Yokoo H; Furukawa J; Kurogochi M; Togashi T; Miura N; Nakanishi K; Kamachi H; Kakisaka T; Tsuruga Y; Fujiyoshi M; Taketomi A; Nishimura S; Todo S, Identification of novel serum biomarkers of hepatocellular carcinoma using glycomic analysis. Hepatology (Baltimore, Md.) 2013, 57 (6), 2314–25. [DOI] [PubMed] [Google Scholar]
  • 4.Tan Z; Yin H; Nie S; Lin Z; Zhu J; Ruffin MT; Anderson MA; Simeone DM; Lubman DM, Large-scale identification of core-fucosylated glycopeptide sites in pancreatic cancer serum using mass spectrometry. J. Proteome Res 2015, 14 (4), 1968–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Argade S; Chen T; Shaw T; Berecz Z; Shi W; Choudhury B; Parsons CL; Sur RL, An evaluation of Tamm-Horsfall protein glycans in kidney stone formers using novel techniques. Urolithiasis 2015, 43 (4), 303–12. [DOI] [PubMed] [Google Scholar]
  • 6.Vivekanandan-Giri A; Slocum JL; Buller CL; Basrur V; Ju W; Pop-Busui R; Lubman DM; Kretzler M; Pennathur S, Urine glycoprotein profile reveals novel markers for chronic kidney disease. Int. J. Proteomics 2011, 2011, 214715. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Reiding KR; Bondt A; Hennig R; Gardner RA; O’Flaherty R; Trbojevic-Akmacic I; Shubhakar A; Hazes JMW; Reichl U; Fernandes DL; Pucic-Bakovic M; Rapp E; Spencer DIR; Dolhain R; Rudd PM; Lauc G; Wuhrer M, High-throughput Serum N-Glycomics: Method Comparison and Application to Study Rheumatoid Arthritis and Pregnancy-associated Changes. Mol. Cell. Proteomics 2019, 18 (1), 3–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Russell AC; Simurina M; Garcia MT; Novokmet M; Wang Y; Rudan I; Campbell H; Lauc G; Thomas MG; Wang W, The N-glycosylation of immunoglobulin G as a novel biomarker of Parkinson’s disease. Glycobiology 2017, 27 (5), 501–510. [DOI] [PubMed] [Google Scholar]
  • 9.van Rooijen JJ; Voskamp AF; Kamerling JP; Vliegenthart JF, Glycosylation sites and site-specific glycosylation in human Tamm-Horsfall glycoprotein. Glycobiology 1999, 9 (1), 21–30. [DOI] [PubMed] [Google Scholar]
  • 10.Argade SP; Vanichsarn C; Chenoweth M; Parsons CL, Abnormal glycosylation of Tamm-Horsfall protein in patients with interstitial cystitis. BJU International 2009, 103 (8), 1085–1089. [DOI] [PubMed] [Google Scholar]
  • 11.Bates JM; Raffi HM; Prasadan K; Mascarenhas R; Laszik Z; Maeda N; Hultgren SJ; Kumar S, Tamm-Horsfall protein knockout mice are more prone to urinary tract infection: rapid communication. Kidney Int. 2004, 65 (3), 791–797. [DOI] [PubMed] [Google Scholar]
  • 12.Parsons CL; Stein P; Zupkas P; Chenoweth M; Argade SP; Proctor JG; Datta A; Trotter RN, Defective Tamm-Horsfall protein in patients with interstitial cystitis. J. Urol 2007, 178 (6), 2665–2670. [DOI] [PubMed] [Google Scholar]
  • 13.Hard K; Van Zadelhoff G; Moonen P; Kamerling JP; Vliegenthart FG, The Asn-linked carbohydrate chains of human Tamm-Horsfall glycoprotein of one male. Novel sulfated and novel N-acetylgalactosamine-containing N-linked carbohydrate chains. Eur. J. Biochem 1992, 209 (3), 895–915. [DOI] [PubMed] [Google Scholar]
  • 14.van Rooijen JJ; Kamerling JP; Vliegenthart JF, Sulfated di-, tri- and tetraantennary N-glycans in human Tamm-Horsfall glycoprotein. Eur. J. Biochem 1998, 256 (2), 471–487. [DOI] [PubMed] [Google Scholar]
  • 15.Serafini-Cessi F; Bellabarba G; Malagolini N; Dall’Olio F, Rapid isolation of Tamm-Horsfall glycoprotein (uromodulin) from human urine. J. Immunol. Methods 1989, 120 (2), 185–189. [DOI] [PubMed] [Google Scholar]
  • 16.Deckert T; Kofoed-Enevoldsen A; Vidal P; Nørgaard K; Andreasen HB; Feldt-Rasmussen B, Size- and charge selectivity of glomerular filtration in Type 1 (insulin-dependent) diabetic patients with and without albuminuria. Diabetologia 1993, 36 (3), 244–251. [DOI] [PubMed] [Google Scholar]
  • 17.Kanauchi M; Nishioka H; Hashimoto T; Dohi K, Diagnostic significance of urinary transferrin in diabetic nephropathy. Jpm. J. Neprhol 1995, 37 (11), 649–654. [PubMed] [Google Scholar]
  • 18.Talks BJ; Bradwell SB; Delamere J; Rayner W; Clarke A; Lewis CT; Thomas OD; Bradwell AR, Urinary Alpha-1-Acid Glycoprotein Is a Sensitive Marker of Glomerular Protein Leakage at Altitude. High Alt. Med. Biol 2018, 19 (3), 295–298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Song T; Aldredge D; Lebrilla CB, A Method for In-Depth Structural Annotation of Human Serum Glycans That Yields Biological Variations. Anal. Chem 2015, 87 (15), 7754–7762. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Sun S; Hu Y; Ao M; Shah P; Chen J; Yang W; Jia X; Tian Y; Thomas S; Zhang H, N-GlycositeAtlas: a database resource for mass spectrometry-based human N-linked glycoprotein and glycosylation site mapping. Clin. Proteomics 2019, 16, 35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Shah B; Jiang XG; Chen L; Zhang Z, LC-MS/MS peptide mapping with automated data processing for routine profiling of N-glycans in immunoglobulins. J. Am. Soc. Mass Spectrom 2014, 25 (6), 999–1011. [DOI] [PubMed] [Google Scholar]
  • 22.Stadlmann J; Pabst M; Kolarich D; Kunert R; Altmann F, Analysis of immunoglobulin glycosylation by LC-ESI-MS of glycopeptides and oligosaccharides. Proteomics 2008, 8 (14), 2858–2871. [DOI] [PubMed] [Google Scholar]
  • 23.Zhang Y; Go EP; Desaire H, Maximizing coverage of glycosylation heterogeneity in MALDI-MS analysis of glycoproteins with up to 27 glycosylation sites. Anal. Chem 2008, 80 (9), 3144–3158. [DOI] [PubMed] [Google Scholar]
  • 24.Hong C; Abdullah M; Wong N, Evaluation of glycan profiles of Tamm-Horsfall glycoprotein and uromodulin. Int. J. Pharm. Sci 2013, 5 (3), 385–389. [Google Scholar]
  • 25.Smagula RM; Van Halbeek H; Decker JM; Muchmore AV; Moody CE; Sherblom AP, Pregnancy-associated changes in oligomannose oligosaccharides of human and bovine uromodulin (Tamm-Horsfall glycoprotein). Glycoconj. J 1990, 7 (6), 609–624. [DOI] [PubMed] [Google Scholar]
  • 26.Rebecchi KR; Wenke JL; Go EP; Desaire H, Label-free quantitation: a new glycoproteomics approach. J. Am. Soc. Mass Spectrom 2009, 20 (6), 1048–1059. [DOI] [PubMed] [Google Scholar]
  • 27.Jackson AU; Talaty N; Cooks RG; Van Berkel GJ, Salt tolerance of desorption electrospray ionization (DESI). J. Am. Soc. Mass Spectrom 2007, 18 (12), 2218–2225. [DOI] [PubMed] [Google Scholar]
  • 28.Karki S; Shi F; Archer JJ; Sistani H; Levis RJ, Direct Analysis of Proteins from Solutions with High Salt Concentration Using Laser Electrospray Mass Spectrometry. J. Am. Soc. Mass Spectrom 2018, 29 (5), 1002–1011. [DOI] [PubMed] [Google Scholar]
  • 29.Gonzalez-Dominguez R; Castilla-Quintero R; Garcia-Barrera T; Gomez-Ariza JL, Development of a metabolomic approach based on urine samples and direct infusion mass spectrometry. Anal. Biochem 2014, 465, 20–27. [DOI] [PubMed] [Google Scholar]
  • 30.Johannesson N; Pearce E; Dulay M; Zare RN; Bergquist J; Markides KE, On-line biological sample cleanup for electrospray mass spectrometry using sol-gel columns. J.Chrom. B 2006, 842 (1), 70–74. [DOI] [PubMed] [Google Scholar]
  • 31.Wei L; Cai Y; Yang L; Zhang Y; Lu H, Duplex Stable Isotope Labeling (DuSIL) for Simultaneous Quantitation and Distinction of Sialylated and Neutral N-Glycans by MALDI-MS. Anal. Chem 2018, 90 (17), 10442–10449. [DOI] [PubMed] [Google Scholar]
  • 32.Tsai TH; Wang M; Di Poto C; Hu Y; Zhou S; Zhao Y; Varghese RS; Luo Y; Tadesse MG; Ziada DH; Desai CS; Shetty K; Mechref Y; Ressom HW, LC-MS profiling of N-Glycans derived from human serum samples for biomarker discovery in hepatocellular carcinoma. J. Proteome Resch 2014, 13 (11), 4859–4868. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Yang S; Wang M; Chen L; Yin B; Song G; Turko IV; Phinney KW; Betenbaugh MJ; Zhang H; Li S, QUANTITY: An Isobaric Tag for Quantitative Glycomics. Sci. Repts 2015, 5, 17585. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Zhou S; Hu Y; Veillon L; Snovida SI; Rogers JC; Saba J; Mechref Y, Quantitative LC-MS/MS Glycomic Analysis of Biological Samples Using AminoxyTMT. Anal. Chem 2016, 88 (15), 7515–7522. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Hong Q; Ruhaak LR; Stroble C; Parker E; Huang J; Maverakis E; Lebrilla CB, A Method for Comprehensive Glycosite-Mapping and Direct Quantitation of Serum Glycoproteins. J. Proteome Resch 2015, 14 (12), 5179–5192. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary legend and figs

Supporting Information

Figure S1. Negative ion mode direct ESI-MS spectrum of released label-free N-linked glycan profile of fetuin sample.

Table S1. The list of 40 potential N-linked glycans for uromodulin.

NIHMS1748585-supplement-Supplementary_legend_and_figs.docx (702.6KB, docx)
Table S5

Table S5. Data from urinary uromodulin samples and uromodulin standards: Raw abundances, mean relative glycan peak percentages and coefficients of variation.

NIHMS1748585-supplement-Table_S5.xlsx (28.6KB, xlsx)
Table S4

Table S4. Data from three uromodulin standards: Raw abundances, mean relative glycan peak percentages and coefficients of variation.

NIHMS1748585-supplement-Table_S4.xlsx (18.4KB, xlsx)
Table S2 and S3

Table S2. Data from reproducibility study: Raw abundances, mean relative glycan peak percentages and coefficients of variation for fetuin.

Table S3. Data from instrument precision study: Raw abundances, mean relative glycan peak percentages and coefficients of variation for fetuin.

NIHMS1748585-supplement-Table_S2_and_S3.xlsx (15.1KB, xlsx)

RESOURCES