| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Molecular & Cellular Proteomics 6:638-647, 2007.
© 2007 by The American Society for Biochemistry and Molecular Biology, Inc.
,**
From the Department of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark, Department of Molecular Biology, University of Aarhus, DK-8000 Aarhus C, Denmark, ¶ Department of Surgical Gastroenterology, Hvidovre University Hospital, DK-2650 Hvidovre, Denmark, and || Institute of Veterinary Pathobiology, The Royal Veterinary and Agricultural University, DK-1870 Frederiksberg C, Copenhagen, Denmark
 |
ABSTRACT |
|---|
 TOP ABSTRACT EXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
At present, the complete characterization of N-glycan structures (determination of glycosylation site(s), monosaccharide composition, and linkage type) requires a combination of techniques, reducing the overall sensitivity. One standard approach to characterize glycans and their linkage type is to isolate glycopeptides or free glycans by chromatographic methods and subsequently treat these with specific endo-/exoglycosidases while mapping the mass changes by MS (6). However, this approach requires a relative large amount of starting material and has only rarely been compatible with the limited amounts available from gel-separated glycoproteins (7, 8).
Other strategies for the characterization of N-glycoproteins have been reported, including a sensitive approach that characterizes gel-separated glycoproteins using sequential specific and nonspecific enzymatic treatment followed by enrichment of glycopeptides using graphite powder packed as microcolumns and final characterization by MS and MS/MS (9). Some recent investigations have focused on the selective identification of N-glycosylated proteins. For example, a technique for glycoprotein/peptide isolation and enrichment from complex mixtures involving double lectin chromatography prior to identification using LC-ESI MS has been introduced (10).
Although hydrophilic interaction liquid chromatography (HILIC)1 was developed more than 30 years ago for the separation and analysis of polar compounds such as sugars and oligosaccharides (11, 12), it has until lately received relatively limited attention. However, with the increasing use of MS coupled to LC and the need to analyze polar compounds in samples of increasing complexity, HILIC has attracted more focus recently. The hydrophilic stationary phase in HILIC generates an environment enriched with an aqueous layer. Due to the relative hydrophobicity of the mobile phase (40–97% acetonitrile in water) this aqueous layer is the major contributor to retaining hydrophilic compounds through hydrogen bonds and dipole-dipole interactions (13). For further information about the function and the availability of various HILIC materials, a recent and extensive review by Hemström and Irgum (14) is recommended. The retention characteristics of HILIC are approximately opposite of that observed in reversed phase separations, meaning that the technique is ideal for the analysis of hydrophilic compounds, which are difficult to analyze by separations based on hydrophobicity (13). In relation, several studies have showed that HILIC is useful for the analysis of released glycans (15–17) and glycopeptides (18). Additionally HILIC has been used for sample preparation for desalting of N-linked glycans (19) and enrichment of glycosylphosphatidylinositol-anchored peptides (20). Furthermore a method was introduced recently in which HILIC was used for enrichment of glycopeptides from complex mixtures prior to identification of glycoproteins and determination of their glycosylation site(s) using LC-ESI MS (21). However, no determination of glycosylation structure was performed.
In this study, we present an extension of this technique for site-specific characterization of low amounts of glycopeptides originating from N-glycosylated proteins separated by gel electrophoresis. Gel-separated glycoproteins were proteolytically digested, and a small fraction of the extracted peptides was used for the identification of the glycoprotein, while the rest was applied to custom made HILIC columns at nanoscale levels for glycopeptide enrichment. The peptide mass fingerprinting (PMF) and the characterization of the glycan structures were performed using MALDI-TOF and -Q-TOF MS and MS/MS. The method for glycoprofiling was developed and tested using the doubly N-glycosylated recombinant protein tissue inhibitor of metalloproteinases-1 (TIMP-1) from two different sources, produced by Chinese hamster ovary cells (TIMP-1CHO) and NS0 cells (TIMP-1NS0). As plasma TIMP-1 (TIMP-1Plasma) is present in appropriately low amounts (50–80 ng/ml) (22) it was used to test the sensitivity of the method. TIMP-1 is a key player in the regulation and development of various cancers and serves as a strong cancer marker (23–28) making its characterization of high biological relevance. The TIMP-1Plasma glycoprofiles of five healthy individuals were determined to test the sensitivity of the method and to monitor the person-specific variation in glycosylation. In addition, the glycoprofile of TIMP-1 derived from platelets (TIMP-1Platelets) was determined to establish the variation in glycosylation of natural TIMP-1 from different sources.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Collection of Blood Samples Containing TIMP-1Plasma and TIMP-1Platelets—
Blood (450 ml) was donated by each healthy volunteer and mixed with 63 ml of citrate-phosphate-dextrose anticoagulation solution. Plasma (230–250 ml) was purified from this mixture and kept at –38 °C until use.
In addition, a mixture of platelet concentrates was purified from whole blood of four healthy volunteers donated in the blood bank. The purified platelets were thawed and frozen five times to disrupt platelets and release intracellular TIMP-1 and centrifuged at 3300 rpm for 15 min at 4 °C. Various protease inhibitors were added to the supernatant: 10 µg/ml aprotinin, 1 µg/ml pepstatin A (both from Roche Applied Science), 1 µg/ml leupeptin (Sigma), and 0.1 mM Pefablock (Roche Applied Science). The samples were kept at –20 °C until use.
Purification of TIMP-1 Using Immunoaffinity Chromatography—
Following centrifugation (3000 rpm for 10 min) the supernatant of TIMP-1Plasma or TIMP-1Platelets samples was diluted in equal amounts of 20 mM NaH2PO4, pH 7, containing 2 mM EDTA and loaded onto a Protein G-Sepharose 4 Fast Flow column (Amersham Biosciences) equilibrated in the same buffer. The column was washed in 20 mM NaH2PO4, pH 7, containing 2 mM EDTA, and the flow-through was collected. The bound IgG was eluted with 0.1 M glycine, pH 2.7, and discarded. The collected fraction was reapplied to further reduce the IgG concentration. The IgG-depleted sample was then applied to an anti-TIMP-1-Sepharose 4B affinity column (Sigma) containing the immobilized antibodies VT-1 and VT-5 (29). The column was washed in 20 mM HEPES, 100 mM NaCl, pH 7.4. Bound protein was eluted using 0.1 M glycine, pH 2.7, dialyzed against 10 mM NH4HCO3, and lyophilized. To test whether the anti-TIMP-1 affinity column showed preference for specific glycoforms, a sample of TIMP-1CHO was carried through the same procedure.
SDS-PAGE, Western Blots, and In-gel Digestion—
The dried samples were boiled in SDS sample buffer (non-reducing conditions) and were resolved by SDS-PAGE (5–15% gradient gel). A 160-V potential was applied for 1 h, and the gel contents were visualized using Coomassie Brilliant Blue staining. The distinct bands below the 31-kDa marker were excised and used for in-gel digestion.
The gel pieces were covered with water and ACN (1:1) and washed 5–10 min with rotation, and the liquid was subsequently discarded. This procedure was repeated. 100% ACN was added, and when gel pieces were dehydrated the ACN was removed. The gel pieces were rehydrated in 10 mM DTT, 0.1 M NH4HCO3, pH 7.8, and incubated for 45 min at 56 °C to reduce the protein. The liquid was removed and quickly replaced with the same volume (
60 µl) of iodoacetamide (10 mg/ml) in 0.1 M NH4HCO3, pH 7.8, and incubated for 30 min in the dark. The iodoacetamide solution was removed, and the gel pieces were washed as described above with a final wash in 100% ACN to dehydrate the gel pieces. The gel pieces were rehydrated in 70 µl of digestion buffer containing 50 mM NH4HCO3, pH 7.8, in water and 0.5 µg of trypsin (sequencing grade, modified; Promega, Madison, WI) in 5 µl of 0.01% HCl and incubated for 45 min at 4 °C. The remaining solution was then removed and replaced with 40 µl of the same digestion buffer without trypsin and placed at 37 °C overnight. A small fraction of the supernatant was saved for PMF; the rest was dried and used for glycan analysis.
Another gel was run under the same conditions for Western analysis. The protein bands were transferred to Immobilon-P membrane (PVDF) incubated with an established and characterized primary anti-TIMP-1 antibody, VT-5 (29), and an anti-mouse IgG (whole molecule) peroxidase conjugate (Sigma) and visualized using an ECL detection kit (Amersham Biosciences)
R2 Poros and HILIC Microcolumns—
Microcolumns were prepared by packing either Poros R2 (Applied Biosystems, Framingham, MA) or HILIC material (ZICTM-HILIC, 200 Å, 10 µm, zwitterionic sulfobetaine functional group; kindly provided by SeQuant AB, Umea, Sweden) into GELoader tips (Eppendorf GmbH, Hamburg, Germany) as described previously (30).
To identify the protein, a small fraction of the in-gel digested sample was applied to a Poros R2 microcolumn equilibrated with 5% formic acid (FA), washed twice with 10 µl 5% FA, and eluted directly onto the MALDI target with 0.8 µl 10 mg/ml 
-cyano-4-hydroxycinnamic acid in 70% ACN, 0.1% trifluoroacetic acid.
To perform the glycan analysis, the dried fraction was redissolved in 15 µl of 80% ACN, 2% FA and loaded onto a HILIC microcolumn equilibrated with 80% ACN, 2% FA. The bound analyte was washed twice in the same buffer and eluted with 0.8 µl of 2% FA directly onto a MALDI target and mixed with 0.5 µl of matrix (2,5-dihydroxybenzoic acid or -cyano-4-hydroxycinnamic acid).
Mass Spectrometry—
Peptide mass fingerprinting and glycan analyses were performed with MALDI-TOF MS using either a Voyager-DE STR (Applied Biosystems), a Bruker Ultraflex (Bruker Daltonics, Bremen, Germany) with TOF-TOF technology, or a MALDI-Q-TOF Ultima (Waters/Micromass, Manchester, UK). Reflector mode was activated, and all samples were analyzed in positive polarity mode. Species were fragmented using the MALDI-Q-TOF Ultima. The collision energy during the MS/MS experiments was 70–140 eV, and argon was used as collision gas. The instrument was calibrated by multipoint calibration using polyethylene glycol 2000.
Software—
MALDI-TOF MS data were viewed with the programs MoverZ (Genomic Solutions, Ann Arbor, MI), Dataexplorer Version 4.5 (Applied Biosystems), or FlexAnalysis Version 2.0 (Bruker Daltonics). Internal two-point calibrations were made when possible using known masses of tryptic peptides or other components. External calibrations using tryptic lactoglobulin were alternatively used.
The acquired MALDI-Q-TOF data were viewed in MassLynx Version 4.0 (Waters/Micromass). The mass accuracy was generally below 30 ppm for all MS analyses. Sequence handling for mass spectrometric data was carried out using the GPMAW software (Lighthouse Data, Odense, Denmark) (31).
|
RESULTS |
|---|
|
TOPABSTRACTEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
|
Fragmentation of Glycopeptides—
The glycopeptides were not only characterized by their mass but also fragmented using MALDI-Q-TOF MS/MS to verify their structures (Fig. 3). The fragmentation resulted primarily in the loss of carbohydrate residues facilitating the interpretation of the spectra. The fragmentation was mainly limited to backbone cleavages of the glycan moiety; however, a few high abundance cross-ring fragmentation products were consistently observed, i.e. m z 1570.83 for Asn-78 glycopeptides and m/z 1835.95 for Asn-30 glycopeptides (Fig. 3). The fragmentation allowed for the determination of the composition of carbohydrate residues, and the appearance of several y-ions in the lower mass region verified the peptide part of the glycopeptide. Generally it was possible, in combination with the fact that all N-glycans feature the same core structure, to propose the glycan type by analyzing the fragmentation pattern.
|
|
As expected, the glycoprofiles of the three samples were rather different as a result of the various organisms producing the glycosylated TIMP-1. Although many of the glycans appeared in all three TIMP-1s, the intensities were significantly different.
Comparison of Individual-specific Glycoprofiles of TIMP-1Plasma and TIMP-1Platelets—
Using the described method, the TIMP-1 glycoprofiles were determined from five purified plasma samples and a sample containing a pool of platelets from four healthy individuals (Fig. 5). Very few differences were observed in the glycoprofiles among the five plasma samples. Few differences were also observed when comparing these glycoprofiles to the platelet-derived TIMP-1 sample. The intensities of the individual glycopeptide signals varied slightly among the samples; however, part of this minor variation can be ascribed to the analytical method. A few unique signals appeared in some spectra; however, as these did not correspond to glycopeptides, these contaminants did not contribute to the variation among the glycoprofiles.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
The principle of the method uses a robust in-gel digestion procedure combined with Poros R2 micropurification, facilitating the protein identification. A one-dimensional gel approach is included to get a separation step into the approach and to have a uniform sample platform from where the identification and glycoanalysis can be performed with high reproducibility from sample to sample unaffected by prior sample handling. The glycopeptide enrichment in the last part of the method takes advantage of the very glycoselective behavior of the HILIC material. Leaving the glycan on the peptide part is an important feature of the method, ensuring a site-specific glycan analysis, which is essential when dealing with proteins glycosylated at multiple sites. Furthermore the site specificity rules out the presence of contaminating glycoproteins that could lead to erroneous conclusions, which can be the case if the glycan moiety is cleaved off the protein part.
Based on the results presented, it is clear that the method for site-specific glycoanalysis feature some advantages, which challenge or complement existing techniques for solving glycan structures. The method is simple and robust, and the limited number of steps makes the analysis fast. Most of the reagents and materials for the glycoanalysis, with the exception of HILIC material, are standard supplies used on a daily basis in most mass spectrometric laboratories. Moreover the high sensitivity allows an extensive glycoanalysis from a well resolved spot in a one-dimensional gel.
One of the drawbacks of the presented approach is that no direct evidence for linkage type could be obtained. Usually linkage type analysis and determination of glycan type require alternative techniques, e.g. methylation MS analysis, glycosidase digestions, or nuclear magnetic resonance, involving considerably larger amounts of glycoprotein. However, by analyzing the fragmentation pattern of a particular glycopeptide along with the knowledge of the core unit structure, it was possible to propose glycan type in addition to the carbohydrate composition. Another drawback of the approach is associated with the limited mass range of the MALDI-Q-TOF MS and MS/MS analysis (m/z < 4500). This aspect can usually be circumvented by using appropriate proteolytic enzymes, which yield glycopeptides with a relatively small peptide moiety. Furthermore it is important to choose an enzyme that will separate different glycosylation sites into isolated peptides. Here trypsin satisfies these requirements; however, for many other proteins it may be necessary to use a second proteolytic enzyme (e.g. chymotrypsin) following trypsin or alternatively use proteinase K to narrow the peptide sequence around the glycosylation site to 3–5 residues.
The method for glycoanalysis was developed using recombinant human TIMP-1CHO and TIMP-1NS0. From small amounts of starting material on a gel (3 µg) (Fig. 1), extensive glycan characterizations could be performed. In agreement with a recent report (21), the HILIC material demonstrated remarkable selectivity toward glycopeptides (Fig. 2) by eliminating the dominating signals for regular peptides and thus facilitating the glycoanalysis. The functional group of the HILIC material used, a zwitterion containing a quaternary amine and a sulfonic acid group (sulfoalkylbetaine), is believed to be responsible for the selectivity toward glycopeptides (14). This is possibly elicited by the charged environment of the functional group that generates a water-enriched liquid layer that retains glycopeptides by forming hydrogen bonds to the hydroxyls of the glycans. In addition, dipole-dipole interactions and weak electrostatic forces are known contribute to the retention (33).
Distinct differences in the glycoprofiles were observed for the two recombinant TIMP-1 species, demonstrating that the HILIC material was not selective toward a specific set of certain glycans. However, it has been reported previously that TIMP-1CHO (from the same manufacturer) contains a minor fraction of high mannose glycans (<5% of Asn-78 glycopeptides) (32). Because these structures were not observed in this study, it can be speculated that the HILIC material deselected this type of oligosaccharide structure or alternatively that the high mannose structures were not present in the investigated product.
High reliability of the glycoprofiling was shown by good agreement between the quantities of each characterized glycan of TIMP-1CHO when compared with a reference study (Supplemental Table S-1). The relative abundances of glycans in this study are based on the assumption that the efficiency of the ionization process for a glycopeptide is strongly dependent on the nature of the peptide moiety and only weakly dependent on the neutral oligosaccharide (32). Charged carbohydrate units (e.g. sialic acids) were excluded from this assumption as sialylated glycopeptides and molecules carrying a net negative charge in general give rise to weaker mass signals than neutral or positively charged molecules due to lower ionization efficiency. In addition, it should be stressed that this approach for relative quantitation of glycans is only valid for glycopeptides with a peptide part of a reasonable size (>8–10 residues).
When carbohydrates are analyzed using MS, there is often the question of whether some of the observed microheterogeneity is caused by MS fragmentation. Glycosidic bonds have been reported to undergo in-source fragmentation resulting in the appearance of minor glycan fragments in the MALDI spectrum. However, this kind of fragmentation is normally associated with the loss of sialic acid residues and only to a minor extent for other carbohydrate units (34). Given the fact that a few core structures and the non-glycosylated peptides were observed in the glycoanalysis, Supplemental Table S-1) (i.e. m/z 1487.8, 1752.9, 1836.9, and 2040.0) it is likely that the structures undergo MS fragmentation to a minor extent as these observed components most likely are insufficiently hydrophilic to be retained in the HILIC micropurification. However, the extent of this fragmentation is estimated to be of limited importance to the glycoanalysis in general as only a few low abundance, small glycopeptides were observed. Due to this fragmentation, it is likely that some sialylated structures were observed in reduced amount or were totally absent in the MS analysis compared with the actual contents of the sample. The amounts of nonsialylated structures were consequently estimated too high. Taking this into account, the quantitative analysis presents only a rough estimation of the distribution of glycans.
In summary, it was shown that a comprehensive site-specific glycoanalysis of a low amount of glycoprotein was feasible using the presented technique. The potential of the method was illustrated by the fact that significantly different glycoprofiles were characterized of TIMP-1 originating from two different cell lines. The speed of analysis for the method is limited by the interpretation of data. When sufficient bioinformatics tools are developed, the method has the potential of a reasonably high throughput thus enabling its use for glycoproteome analysis.
Glycoprofiles of TIMP-1Plasma originating from five healthy individuals were subsequently characterized with the aim to test the sensitivity of the method and to determine the person-specific glycan variation of TIMP-1. The resulting glycoprofiles showed very high similarity (Fig. 5), indicating that healthy individuals display only a minimal variation in the glycosylation pattern of this protein. The glycoprofiles of TIMP-1 isolated from purified platelets was also determined. Here it was found that plasma- and platelet-derived TIMP-1 display the same glycans and in roughly the same intensity. The variation between platelet-derived and plasma-derived TIMP-1 was in the same range as the TIMP-1Plasma 1–5 variation. To rule out the possibility that the affinity column, used for purification of these naturally occurring TIMP-1s, was selecting a specific group of glycan structures and thereby creating an artificial similarity of the glycoprofiles, the TIMP-1CHO glycoprofile was determined before and after it was applied to the anti-TIMP-1 affinity column. Because no differences could be observed between these two samples (data not shown), it was concluded that the anti-TIMP-1 antibodies used for the purification were not selectively retaining TIMP-1 molecules carrying certain glycostructures.
It is widely accepted that carbohydrates linked to proteins can play a central role in determining the physicochemical properties of these conjugates. In addition, many protein-protein interactions are determined by glycosylation. Unfortunately very little is generally known on the biological role of specific glycans and populations of glycoforms on particular proteins. Such investigations are still to be performed for TIMP-1, meaning that it is difficult to predict the function of the observed glycoforms. However, it can be speculated that the highly conserved glycosylation pattern of TIMP-1 among healthy individuals is important for specific functions and thus has to be tightly regulated.
TIMP-1 regulates the proteolytic activities of the matrix metalloproteinases in a precise manner in healthy individuals by the formation of 1:1 matrix metalloproteinase-TIMP-1 complexes. Furthermore TIMP-1 can stimulate cell proliferation and inhibit apoptosis (for a review, see Ref. 35); the latter functions could facilitate tumor progression. In line with this hypothesis, high tissue or plasma levels of TIMP-1 have consistently been associated with poor patient survival (35). As it has been known for decades that glycoprofiles may change as a consequence of disturbed protein processing in cancer tissue (36) it can be hypothesized that this also could be true for TIMP-1. Following this line of thought, cancer cell-induced variations in the glycosylation patterns of TIMP-1 would represent a valuable new tool because colorectal cancer patients have been shown to have significantly increased amounts of total TIMP-1 in the plasma compared with the background level produced by healthy individuals (27, 28). The opportunity to identify possible glycosylation variants of TIMP-1 in cancer patients would increase both the sensitivity and specificity of TIMP-1 as a biomarker for the early detection of colorectal cancer; e.g. the increased amount of plasma TIMP-1, which is associated with colorectal cancer disease, might display glycan alteration that is likely to be present in sufficiently high concentrations to avoid suppression from the non-malignant glycans. The method presented here allows for these types of investigations.
To the best of our knowledge, this is the first study that has characterized the glycan composition of naturally occurring TIMP-1. The glycan composition of TIMP-1 constitutes a potential marker for cancer, and based on the individualized analyses, it can be concluded that the glycoprofile of plasma TIMP-1 is ideal as a reference for healthy individuals due to the minimal person-specific glycan variations.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Published, MCP Papers in Press, January 6, 2007, DOI 10.1074/mcp.M600407-MCP200
1 The abbreviations used are: HILIC, hydrophilic interaction liquid chromatography; FA, formic acid; PMF, peptide mass fingerprinting; TIMP-1, tissue inhibitor of metalloproteinases-1; CHO, Chinese hamster ovary. 
* This work was supported by a grant from the Oticon Foundation (Hellerup, Denmark) (to M. T.-A.) and by the Danish Cancer Society and the Danish Medical Research Counsel. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
S The on-line version of this article (available at http://www.mcponline.org) contains supplemental material.
** To whom correspondence should be addressed. Tel.: 45-6550-2371; Fax: 45-6550-2467; E-mail: php{at}bmb.sdu.dk
|
REFERENCES |
|---|
|
TOPABSTRACTEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
variants separated by gel electrophoresis.
Electrophoresis
17, 925
–931[CrossRef][Medline]
-casein macropeptide glycosylation.
J. Chromatogr. A
929, 151
–163[CrossRef][Medline].
J. Mass Spectrom.
31, 1109
–1118[CrossRef][Medline]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  L. Dissing-Olesen, M. Thaysen-Andersen, M. Meldgaard, P. Hojrup, and B. Finsen The Function of the Human Interferon-{beta}1a Glycan Determined in Vivo J. Pharmacol. Exp. Ther., July 1, 2008; 326(1): 338 - 347. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y.-S. Kim, S. Y. Hwang, H.-Y. Kang, H. Sohn, S. Oh, J.-Y. Kim, J. S. Yoo, Y. H. Kim, C.-H. Kim, J.-H. Jeon, et al. Functional Proteomics Study Reveals That N-Acetylglucosaminyltransferase V Reinforces the Invasive/Metastatic Potential of Colon Cancer through Aberrant Glycosylation on Tissue Inhibitor of Metalloproteinase- Mol. Cell. Proteomics, January 1, 2008; 7(1): 1 - 14. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| All ASBMB Journals | Journal of Biological Chemistry |
| Journal of Lipid Research | ASBMB Today |
