Author: laurenmd (Page 5 of 9)

NOVEL ASPECT: Direct comparison of trypsin and nonspecific methods using new software for analyzing glycopeptides with an improved scoring algorithm.

INTRODUCTION:
The determination of site-specific glycosylation in proteins remains a difficult task.  Two methods currently in use employ specific proteases such as trypsin or a non-specific cocktail of proteases.  Trypsin yields large well-defined glycopeptides.  These large glycopeptides can be difficult to analyze by tandem MS.  Additionally, glycopeptides may be produced containing two glycosylation sites.  Nonspecific proteases yield smaller glycopeptides that vary in length and are more difficult to predict.  Presented here is a comparison between the trypsin approach and that using nonspecific digestion.  A direct comparison is now possible because we have developed software to analyze both tryptic and non-tryptic glycopeptides.  We find that the nonspecific protease method often has higher sensitivity and more complete coverage of site-specific glycan heterogeneity.

METHODS:
In-gel digestions were used to produce the glycopeptides.  Briefly, 10 μg of a target protein was reduced with dithiothreitol and alkylated with iodoacetamide prior to running on SDS-PAGE.  Gel bands were cut, destained, and dried in a speed-vac.  To the dried gels, an appropriate amount of trypsin or pronase was added in ammonium bicarbonate buffer.  After digesting overnight the glycopeptides were extracted.  Samples were dried and resuspended in 20 μL prior to analysis on an Agilent HPLC-Chip/TOF MS.  The resulting data was analyzed with in-house software utilizing accurate mass and fragment scoring to confirm identification.

ABSTRACT:
A systematic comparison was performed involving several proteins.  Proteins were selected due to their abundance in human serum, applications to biotechnology, or complexity.  Included in this group are plant glycoproteins, which often contain α(1-3) linked fucose on the core glucoseamine.  The lack of biochemical tools to cleave glycans with this linkage makes plant glycoproteins difficult to characterize by methods requiring enzymatic glycan release.  For brevity, only the results of immunoglobulin M and α-2-macroglobulin from human serum are provided.In IgM, tryptic digestion yielded 17 glycan compositions at three sites, while nonspecific digestion yielded 26 compositions across all five potential glycosylation sites.  Of the five sites, two were not observed with tryptic analysis because they fell within the same peptide.  When IgM was digested and analyzed using the nonspecific digestion, more compositions were monitored and each result was achieved with higher confidence due to the higher mass accuracy afforded by the lower mass peptides and the presence of peptide-specific scoring ions.In α-2-macroglobulin, tryptic digestion yielded six glycan compositions at one site while nonspecific digestion yielded 28 glycan compositions at all eight sites.  This protein is an even more striking example of the difference between these methods.  Of the eight glycosylation sites half were inaccessible to tryptic analysis due to the fact that several sites shared a tryptic peptide.  Only the peptide bearing site1424N was detected.  Glycopeptides corresponding to the other sites were not observed, possibly due to their large sizes.  The shorter peptides derived from nonspecific digestion afforded greater sensitivity and likely aided the search due to the reduced probability of unanticipated secondary posttranslational modifications.

NOVEL ASPECT: This tool will greatly advance the understanding of the mechanisms of action of enzymes, and the prediction of bioactive peptides.

INTRODUCTION:
For many years, MS has been widely employed for the identification and characterization of proteins and more recently peptides. The sheer volume of data generated by MS warrants the need for sophisticated data handling systems prediction, and visualization.In this work we focus on the peptides produced in a hydrolysate. We have constructed ‘EnzymePredictor’ a software that allows the visualization of the hydrolysate and the prediction of sets of enzymes that have been used or can be used to generate the current hydrolysate’s identified peptides. The software has a friendly visualization output that allows the user to simultaneously view the positional information of the peptides on their source protein, and the possible enzymes that have been used to produce them.

METHODS:
The software EnzymePredictor is coded in perl. To assess the software’s performance we experimentally generated a hydrolysate. We used human breast milk as the raw material, and performed three independent digestions. We chose to use human milk as opposed to a “clean” protein digestion because of its complex nature, which represents the reality of a food hydrolysate.  We carried out a digestion with trypsin; one with chymotrypsin; and a digestion using the combination of both trypsin and chymotrypsin. The latter digestion is to examine if the software is sensitive to the usage of a combination of enzymes. The resulting hydrolysates were then passed through mass-spectrometry and the MassHunter software (Agilent Technologies Inc.) to yield the list of peptides.

ABSTRACT:
We have developed a tool to rapidly evaluate the evidence for which enzymes are most likely to have cleaved the sample. EnzymePredictor, a web-based software, has been developed to (i) identify the protein sources of fragments found in the hydrolysates and map them back on it, (ii) identify enzymes that could yield such cleavages, and (iii) generate a colored visualization of the hydrolysate, the source proteins, the fragments, and the predicted enzymes. It tabulates the enzymes ranked according to their cleavage counts. The provision of odds ratio and standard error by the software permits users to evaluate how distinctively particular enzymes may be favoured over other enzymes as the most likely cleavers of the samples. Finally, the method displays the cleavage not only according to peptides, but also according to proteins, permitting evaluation of whether the cleavage pattern is general across all proteins, or specific to a subset. We illustrate the application of this method using milk hydrolysates, and show how it can rapidly identify the enzymes or enzyme combinations used in generating the peptides.The software successfully identified in each of the three cases the enzymes responsible for the cleaving of the human breast milk proteins.

The approach developed here will accelerate the identification of enzymes most likely to have been used in hydrolyzing a set of mass spectrometrically identified peptides derived from proteins. This has utility not only in understanding the results of mass spectrometry experiments, but also in choosing enzymes likely to yield similar cleavage patterns. EnzymePredictor can be found at http://bioware.ucd.ie/∼enzpred/Enzpred.php

NOVEL ASPECT: We constructed a serum-based library containing 164 structures with relative abundances, retention times, accurate masses, and tandem MS spectra.

INTRODUCTION:
Glycosylation is an abundant and common posttranslational modification of proteins with functions related to stabilization of protein conformation, secretion through membranes, protein turnover control, proteinase hydrolysis protection, increase of the protein solubility, and cell recognition. They are also an important new target for biomarker development. Here we present the construction of a serum-based N-glycan library with serial digestion using exoglycosidases monitored by nanoLC MS. The library will be the core of a method that employs nanoLC MS with accurate mass, reproducible retention times, and tandem MS spectra for the rapid throughput glycomics.

METHODS:
Human pooled serum N-glycans were released using PNGase-F digestion and chemically reduced with NaBH4. Released N-glycans were fractionated off-line by HPLC using a Hypercarb PGC column, and fractions were screened for glycan compositions using MALDI-FTICR-MS. Serial exoglycosidase treatments, which were monitored using nanoLC-Chip-Q-TOF-MS, allowed determination of N-glycan structures and linkages.

ABSTRACT:
In a typical serum N-glycan nanoLC MS profile, more than 300 compounds are observed over five orders of magnitudein abundances. We have now constructed a library containing 164 N-glycan structures, of which 28 are fully elucidated, more than 80 partially elucidated, and the remainder are putative structures. Included in the library are the intensities in pooled human serum, retention time, accurate mass and fragmentation profile. Previously, our group developed an N-glycan database that was based on commercially available standards of the five most abundant glycoproteins in serum.  We expand on this work by annotating glycans released directly from serum.We created a systematic name for the N-glycans. For example, N5402a, where “N” means N-glycans, and the compositions for Hex:HexNAc:Fuc:NeuAc, followed a letter according to their relative abundances in serum, “a” being the most abundant. The two most abundant N-glycans in serum are isomers of the biantennary disialylated structure (N5402a and N5402b), where the structure of highest abundance has the sialic acid (NeuAc) linked α2-3 to the 1-6-antenna and α2-6 to the 1-3-antenna, while the structure of second highest abundance has the NeuAc residues on both antenna linked α2-3. Similarly two isomeric structures of the monosialylated N-glycan (N5401a and N5401b) could be resolved. Both structures contained NeuAc with α2-6 linkages, but N5401a has the NeuAc on the 1-3-antenna, while N5401b has the NeuAc on the 1-6-antenna.In conclusion, we constructed a serum-based library containing 164 structures with identifiers, relative abundances, retention times, accurate masses, and tandem MS spectra that will be used for the rapid throughput glycomics and for biomarker discovery with glycan structures.

NOVEL ASPECT: Monitoring the degree of glycosylation with the absolute protein quantification using MRM methods

INTRODUCTION:
The studies aimed towards glycan biomarker discovery have focused on glycan characterization and profiling of released glycans. Site-specific glycosylation analysis is less developed but may provide a new type of biomarkers with higher sensitivity and specificity. Quantitation of peptide-conjugated glycans directly facilitates the differential analysis of distinct glycoforms associated with specific proteins at distinct sites. We have developed a method using MRM to monitor protein glycosylation normalized to absolute protein concentrations to examine quantitative changes in glycosylation at a site-specific level. The remarkable sensitivity and selectivity of MRM have enabled the detection of low abundant glycopeptides from serum directly.

METHODS:
The glycosylation pattern of protein standards was first determined using both trypsin and pronase digestion. Proteins were treated with DTT and IAA before an overnight digestion in a 37°C water bath. Glycopeptides were profiled using an Agilent 6520 HPLC-Chip/QTOF MS, and identified using an in-house-software tool, GPFinder. Once the site specific glycosylation was determined, quantification was performed with the tryptic peptides using an Agilent 6490 QqQ MS coupled with an Agilent 1290 UPLC system. The absolute amount of proteins was determined by the peak area of peptides, while the degree of glycosylation was normalized to the protein content, thus allowing quantification of glycoforms on the site-specific and protein-specific level. The method was applied to glycoproteins in 11 healthy human sera.

ABSTRACT:
We determined the site-specific glycosylation of immunoglobulin G, A, and M, transferrin, alpha-2-macroglobulin, and haptoglobin using nonspecific proteolysis. Tryptic digestion yielded limited information compared to nonspecific protease analysis, but was necessary to obtain a consistent peptide sequence for MRM.

The CID fragmentation of the tryptic glycopeptides was determined to find the optimal transition for the MRM experiments. Quantitation experiments were performed on a triple quadrupole optimized for glycopeptide analysis with MRM transitions from molecular ions to either m/z 203 (HexNAc) or m/z 366 (HexNAc-Hex).

MRM is not typically used for glycopeptide quantitation. The problem is the lack of glycopeptide standards. In this research, we use the absolute protein abundances to normalize the glycopeptide abundances. In this way, relative changes in protein glycosylation can be measured at the site-specific and protein-specific level.
As an example, a peptide that is common to IgG1, IgG2, IgG3 and IgG4 was quantified to obtain the absolute amount of IgG protein. The result showed a wide dynamic range (>1000) and low LOD (6.01fmole). The average IgG concentration for a set of 11 samples was 5.25mg/ml, with a relative standard deviation (RSD) of 21.8%. Peptides that were unique for each subclass were quantified to monitor the variations in abundances for the four IgG subclasses.

Using this method, we showed that specific glycoforms can be monitored for IgG (25 tryptic glycopeptides), transferrin (3), IgA (6), alpha-2-macroglobulin (5), and haptoglobin (3) directly from serum. Some sites from some proteins were not observed due to the problems associated with tryptic digestion of glycoproteins. Our normalized glycopeptide abundances, which remove the contribution of protein concentration, gave results allowed the simultaneous monitoring of glycosylation across several proteins and several sites. This method will have implications in the discovery of glycan biomarker by associating the glycans to specific proteins.

This course is designed for scientists who want to learn specific techniques for the MS and MS/MS characterization of glycans and glycoproteins. The course will address fundamental aspects of glycobiology, sample preparation and handling, mass spectrometry (hardware and software), and bioinformatic tools for interpretation of results. Real-world examples of the application of these techniques will include characterization of intact glycoproteins, characterization of released glycans, analysis of complex mixtures of glycoproteins and glycans. The role of MS-based methods in interdisciplinary efforts to solve these complex problems will also be addressed. Problem sets will be used to emphasize practical aspects of glycoprotein, glycans characterization, glycosylation site identification, and the application of MS techniques to a variety of circumstances commonly encountered in the lab.

NOVEL ASPECT: A comprehensive MS method was used to analyze endogenous peptides from human milk that were shown to be antimicrobial.

INTRODUCTION:
A novel, streamlined and high-throughput analytical approach has been developed to capture and identify a comprehensive set of peptides produced by the in vivo proteolytic digestion of human milk.
By employing an iterative exclusion list on the tandem MS analysis and multiple injections, the number of unique endogenous peptides identified increased by nearly 5-fold compared to a single tandem identification run. An annotated library of endogenous peptides sequences and their PTMs has been constructed to streamline yield a method for the rapid and comprehensive analysis of peptides in milk.
The results indicate that the digestion of proteins in the mammary yield specific peptide sequences from distinct proteins that exert protective influence on both the infant and the mother.

METHODS:
Milk peptide purification was performed through centrifugation, protein precipitation with TCA and C18 solid phase extraction.
Peptide samples were analyzed in the positive mode on an Agilent nanoLC-Q-TOF with a C18 chromatographic chip. Automated precursor selection based on ion abundance was employed for tandem fragmentation. After each analysis, newly identified peptides were added to an exclusion list for subsequent repeat experiments. Molecular ions on the exclusion list were ignored by the instrument and hence were not fragmented again. This approach allowed deeper exploration of the sample.
Peptide identification was accomplished using two searchers, MS-GFDB and X!Tandem against a milk protein library (1472 entries) based on a query to the Uniprot database. Peptide results were accepted with p-values lower than 0.01.

ABSTRACT:
Extensive peptidomic analysis of milk allowed the identification of 508 unique peptide sequences from 27 different proteins including β-casein (62% of the total), polymeric immunoglobulin receptor, butyrophilin, αs1-casein, osteopontin, κ-casein and mucin-1. However, neither the number of peptides found nor their relative signal intensities appears to follow the natural abundances of their respective protein origin. For example, highly abundant proteins in milk such as α-lactalbumin and lactoferrin are not represented.

Peptides identified ranged in length from 6 to 37 amino acids. 163 peptides (32% of the total) were phosphorylated. By comparison with the database Phosphosite it was determined than 35 of these sequences had a previously unknown phosphorylation site.

Interestingly, the peptides identified were not randomly distributed across the overall protein sequence but were focused on specific sites for butyrophilin, κ-casein, osteopontin, mucin-1, perilipin and polymeric immunoglobulin receptor. β-casein is unique in this regard as almost the entire sequence was represented.  These results indicate that proteolysis in the mammary gland is selective—released peptides were drawn only from specific proteins and typically from selected parts of the parent sequence.

Selective hydrolysis suggests specific functional roles for the endogenous peptides. A large number (62) of the endogenous peptides were found to share at least 57% of their sequence with known functional peptides,based on a comparison with functional peptide databases (BIOPEP, PeptideDB, CAMP and APD2). Fifty-five of the peptides matched known antibacterial sequences, while seven peptides matched immunomodulatory sequences.
The antimicrobial activity of the milk peptides was tested in a radial diffusion assay against a Gram-negative (E. coli) and a Gram-positive (S. aureus) species.  We show for the first time that the endogenous peptides in milk are antimicrobial. S. aureus is a bacteria commonly found in mastitis suggesting that the peptides not only protect the infant but the mother as well.

NOVEL ASPECT: HMO extraction and purification at the 96-well level now allows for large-scale biomarker studies.

INTRODUCTION:
Human milk is known to be the best nutritional source for newborns. Human milk oligosaccharides (HMOs) are a highly abundant constituent in human milk, and its protective and prebiotic effects are what drive studies today. Further investigating these properties and future biomarker studies will require large patient sample sets, where high-throughput sample preparation and analysis are favored. The development of a sample preparation method at the 96-well level would increase the repeatability of the process and reduce batch-to-batch sample variations. Furthermore, it will allow shorter handling times, giving us the ability to process larger amounts of samples. In this study, we have developed and optimized a high-throughput method to rapidly profile human milk oligosaccharides at a 96-well level.

METHODS:
Five batches of breast milk samples from the same donor were processed to find the optimal separation method by using the traditional HMO extraction method under different conditions (exclusion of ethanol, Folch, and precipitation times). Samples were reduced and purified using an automated liquid handler to prevent variation in solid phase extraction (SPE). Then the optimized method was repeated using smaller volumes, and SPE on a 96-well plate with graphitic carbon was administered via centrifugation to test the saturation limit of the columns. Glycan profiles and isomer information were obtained by HMO analysis on a chip-based nano-LC (Agilent Chip/TOF-MS) using acetonitrile/water and formic acid for separation. For data-processing, an in-house library of HMO masses was used.

ABSTRACT:
Based on the HMO extraction and purification method that was previously established in our lab, we first optimized whether we could omit certain steps that may be difficult to process at the 96-well plate level. Traditionally, human milk is centrifuged to remove fats, treated with chloroform/methanol (Folch) to remove additional lipids, and treated with ethanol to remove excess proteins prior to reduction and SPE. After we excluded the Folch method, the overall HMO profile was still comparable to samples treated with Folch using 500uL of sample. However, excluding both the Folch and ethanol addition resulted in lower analytical signal and different HMO profiles compared to the traditional batch.We then evaluated a commercially available 96-well graphitic carbon SPE plate and compared it to graphitized carbon cartridges used for individual sample analysis to test the saturation limit of the 96-well plate. To accommodate the 96-well plate, the sample volumes were reduced to 10, 25, 35, and 50uL and SPE was performed using centrifugation. While solvent amounts were reduced for plate-based SPE, no additional contaminants were observed during the analyses. Samples were reconstituted to 50uL to show a gradual increase in glycan signal between the different volumes and compositional glycan profiles were obtained by Agilent Chip/TOF-MS and identified using an in-house library of accurate glycan masses found in humans. It could be concluded that the 96-well graphitic carbon plate is not saturated using the milk volumes used.With this procedure, we are able to process 96 samples with a handling time of around 3 hours. The application of 96-wells allows for reduction of batch-to-batch sample variations, and increases repeatability of the sample preparation procedure. This procedure will allow the processing of large sample cohorts to further investigate the biological activity of HMOs and how it may possibly be changing in case studies.