cancer biomarker discovery: lectin-based strategies targeting...

Disease Markers 33 (2012) 1–10 1DOI 10.3233/DMA-2012-0904IOS Press

Review

Cancer biomarker discovery: Lectin-basedstrategies targeting glycoproteins

David Clark and Li Mao∗Department of Oncology and Diagnostic Sciences, University of Maryland, School of Dentistry, Baltimore, MD,USA

Abstract. Biomarker discovery can identify molecular markers in various cancers that can be used for detection, screening,diagnosis, andmonitoring of disease progression. Lectin-affinity is a technique that can be used for the enrichment of glycoproteinsfrom a complex sample, facilitating the discovery of novel cancer biomarkers associated with a disease state.

Keywords: Lectins, lectin-affinity, biomarkers, glycosylation, glycoproteins

1. Cancer biomarker discovery

A biomarker, as defined by the National Instituteof Health, is “a characteristic that is objectively mea-sured and evaluated as an indicator of normal biolog-ic processes, pathogenic processes or pharmacologicresponses to a therapeutic intervention”, thus summa-rizing the role they can play in identifying the occur-rence or progression of a pathogenic disease [4]. Withthis in mind, discovery of biomarkers related to cancerwould be beneficial in the scope of detection, diagnosis,screening, and monitoring disease progression. Devel-opment of cancer biomarkers is a challenging process,best described as occurring in five consecutive phases:(1) Preclinical exploratory studies, (2) clinical assay fordevelopment for clinical disease, (3) retrospective lon-gitudinal repository studies, (4) prospective screeningstudies, and (5) cancer control studies [31].

Biomarkers related to cancer may also be used to di-rect drug therapy, serving to explain clinical results inresponse to chemotherapy treatment. Such biomarkers,termed predictive biomarkers, may be used not only to

∗Corresponding author: Li Mao, MD, Department of Oncologyand Diagnostic Sciences, University of Maryland School of Den-tistry, 650 W. Baltimore St., Baltimore, MD 21201, USA. E-mail:[email protected].

direct which agent of combination of agents should beused in treating a disease state, but also identify thepopulation of patients who may show the best responseto treatment (as well as identify those who may showadverse effects) [14]. In clinical trials requiring long-term intervention strategies, which require long peri-ods of observation and large patient groups, biomarkerscan be substituted for clinical responses or clinical endpoints [4]. Furthermore, incorporating a pattern-basedbiomarker, constructed of several single biomarkersthat when evaluated together may be used in detection,risk assessment, screening, and diagnosis, may be moreinformative in predicting responses to individual thera-peutic regimens [21]. These biomarkers will be an im-portant integrated component of personalizedmedicinestrategy in cancer treatments.

Sources of biomarkers can include tumor tissue,as well as bodily fluids including blood (serum andplasma), urine, sputum, saliva and cerebrospinal flu-id (CSF). Development of biomarkers that can be ob-tained via minimal or non-invasive methods are ide-al, and blood fulfills this description [35]. Not onlyis blood easily accessible for sampling, but becauseof the network of arteries, veins, and capillaries thatcome in contact with organs, it can collect molecularindicators such as proteins secreted, shed, or otherwisereleased by tissues [36]. The challenge in biomarkerdiscovery using blood is that the plasma proteome is

ISSN 0278-0240/12/$27.50 2012 – IOS Press and the authors. All rights reserved

2 D. Clark and L. Mao / Cancer biomarker discovery: Lectin-based strategies targeting glycoproteins

the most complex human proteome, consisting of pro-teins whose function is dependent on their presence inplasma and transient proteins which are secreted andshed into the bloodstream [13]. It is this latter groupwhere candidate biomarkers would most likely be clas-sified, as proteins of interest would be secreted, shed, orleaked from the affected tissue. Further complicatingbiomarker discovery is the dynamic range of proteinconcentrations found in plasma, varying up to twelveorders of magnitude, wherein highly abundant proteins(i.e. albumin and IgG) compose the majority of theplasma proteome, making low-abundant proteins thatmay serve as biomarkers difficult to identify [12]. Onestrategy to increase the identification of low abundantproteins is to deplete the serum sample of highly abun-dant proteins using immune-affinity, thus reducing thedynamic range. However, several drawbacks of thisstrategy include the loss of proteins of interest due tononselective binding or depletion of albumin, whichfunctions as a carrier protein to assist in the circulationand transport of other proteins [13]. With these chal-lenges in mind, prior to a biomarker found in plasmabeing considered for clinical use in a cancer setting,several criteria must be met including: (1) release intothe circulation in detectable amounts by a small, non-symptomatic tumor, (2) exhibit specificity for the tis-sue or organ of origin, and (3) exhibit specificity forthe cancer state, meaning it has no association with anoncancer disease [7]. Of note, the target biomarkermay not exist in the tumor proteome itself, but insteadmanifests in the tumor microenvironment as a responseto the presence or progression of the tumor [27].

There have been many studies focusing on biomark-ers discovery, based on various type of biomoleculessuch as DNA, mRNA, ncRNA, lipids, carbohydrates,and proteins. Alterations in DNA and mRNA expres-sion ultimately result in aberrant expression or modifi-cation of protein products, indicative of a disease state,whichmake proteins attractive targets in biomarker dis-covery [22]. This review will focus on a sub-populationof proteins, glycoproteins, as potential biomarkers andpresent challenges and promises of lectin-affinity basedin biomarker discovery for oncology applications.

2. Targeting glycoproteins as biomarkers

Glycosylation is the most commonpost-translationalmodification (PTM) of proteins, involving the covalentattachment of a glycan, or sugar structure, to the peptidebackbone [34]. Approximately 50% of all human pro-

teins are glycosylated, functioning in many biologicalprocesses, including development, immune response,cell division, inflammation, and cellular regulation [3,30]. In addition to glycosylation’s involvement inmanycellular functions, reliant on a specific interaction thatoccurs with the extended glycan structure, it is also as-sociated with proper protein folding. There are sever-al types of glycan attachments in glycoproteins, withthe two most common occurring as N-linked glycansor O-linked glycans. N-linked glycosylation involvesthe production of an oligosaccharide precursor and anen bloc transfer onto the side chain amide nitrogen ofan asparagine residue of a newly synthesized proteinin the endoplasmic reticulum (ER). O-linked glycosy-lation occurs at a later stage of protein processing inthe Golgi apparatus, and involves the attachment of aglycan structure to a serine or threonine residue at thehydroxy oxygen of the amino acid [17].

Targeting glycosylation in biomarker discovery canbe construed as a more direct approachbased on the fol-lowing thoughts. Aberrant glycosylation has been as-sociated with a cancer disease state for many years [8].Alterations in N-linked glycans that have a biologicalrole in cell adhesion have been associated with can-cer cell invasion and metastasis [34]. The majorityof proteins localized at the cell membrane or secret-ed from the cell surface are glycosylated, thus may bedetectable in the blood stream [36]. The structure ofglycans, or the sites of glyco-post-translational modi-fications could be altered during cancer initiation andprogression [15]. Glycoproteomic studies that analyzeprofile comparisons, of normal and disease states, pro-vide potential new cancer biomarkers that can be usedin clinical setting in the future [24]. In fact, the major-ity of the U.S. Food and Drug Administration (FDA)approved cancer biomarkers currently used in a clini-cal setting are glycoproteins or glycosylated (Table 1).Furthermore, targeting the differences in glycosylation,specifically alterations in the glycan structure, may pro-vide opportunities to identify biomarkers that exhibitimproved detection sensitivity and specificity [8]. Can-cer biomarkers that detect specific glycoforms of a pro-tein, resulting in increased sensitivity and specificity,may allow for earlier detection of a disease state [10].Earlier detection of cancers may provide an improvedtherapeutic window and an increase in overall survival.

3. Lectin-based glycoprotein biomarkerdevelopment

Methodologies to enrich glycoproteins have beenemployed previously and include the use of hydrazine

D. Clark and L. Mao / Cancer biomarker discovery: Lectin-based strategies targeting glycoproteins 3

Table 1Current US Food and Drug Adminstration (FDA) approved cancer biomarkers. Abbreviations: CA: cancer antigen, GIST: gastroinstestinalstromal tumor, NMP22: nuclear matrix protein 22

Biomarker Type Glycosylated Cancer type Source Clinical use

α−Fetoprotein (AFP) Gylcoprotein Yes Nonseminnomatous Serum Stagingtesticular

Human chorionic Glycoprotein Yes Testicular Serum Staginggonadotropin-βCA 19-9 Glycan Yes Pancreatic Serum MonitoringCA 125 Glycoprotein Yes Ovarian Serum MonitoringCarcinoembryonic antigen (CEA) Glycoprotein Yes Colon Serum MonitoringEpidermal growth Protein Yes Colon Colon Therapy selectionfactor receptor (EGFR)KIT Protein Yes GIST Gastrointestinal

tumorDiagnosis, therapyselection

Thyroglobulin (Tg) Protein Thyroid Serum MonitoringProstate-specific antigen (PSA) Glycoprotein Yes Prostate Serum Screening,

monitoringCA 15-3 Glycoprotein Yes Breast Serum MonitoringCA 27-29 Glycoprotein Yes Breast Serum MonitoringOestrogen receptor and Protein Yes Breast Breast tumor Hormonal therapy

selectionprogesterone receptorHER2/NEU Protein Yes Breast Breast

tumor/SerumMonitoring,prognosis,Therapy selection

NMP22 Protein Bladder Urine Screening,monitoring

Fibrin/ Fibrin degradation protein Protein Yes Bladder Urine MonitoringBladder tumor-associated antigen Protein Bladder Urine Monitoring

chemistry, carbohydrate-reactive antibodies, and stud-ies that focus on analysis of glycan structures after re-lease from the associated protein [8]. Most of thesestrategies incorporatemass spectrometry (MS) analysisafter initial separation or enrichment to identify targetedglycoproteins or glycan structures. One methodologythat has been employed extensively is the use of lectinaffinity, and may hold promise in targeted glycoproteinanalysis for cancer biomarker discovery.

Lectins are glycan-binding proteins that bind withgreat specificity to selected sugar moieties, or function-al groups. The first lectin recognized, ricin, was iso-lated almost a century ago from castor seeds and addi-tional lectins have been discovered since that are usedto isolate sugar-boundmolecules in biological systems.They are found both in animal and plant systems aswell as lower organisms such as viruses, which uselectins to attach to host organism cells. A current listof lectins and the glycan structures they recognize aresummarized in Table 2. With the multitude of lectinsavailable to isolate sugar-containing molecules, theiruse has been employed in several studies both as singlereagents for purification and enrichment, as well as incombination to identify a larger profile of glycosylatedmolecules. More recently, protein microarrays that in-

corporate lectins to serve as screening tools have alsobeen developed.

The use of multi-lectin affinity has the benefit of ob-taining a global glycoproteome, as the different affini-ties of each lectin includedwill yield a wider net of cap-tured glycoproteins [45]. Development of multi-lectinchromatography (M-LAC) to study the plasma pro-teome has been detailed by one group, who developedsingle use agarose-bound lectin columns containingthe mixture of lectins concanavalin A (Con A), jacalin(JAC), and wheat germ agglutinin (WGA) [33]. In ad-dition to M-LAC, the authors also included depletion ofalbumin and IgG, two abundant plasma proteins, in anattempt to increase identification of low abundancepro-teins. To increase the robustness of this approach,mak-ing it more suitable to clinical studies, another groupimmobilized the same lectins, Con A, WGA, and JAC,onto a column [16]. Termed high-performance multi-lectin affinity chromatography (HP-M-LAC), this studysought to validate the approach of M-LAC and im-munodepletion, while also generating a column thatwould allow for good reproducibility when analyzingmultiple samples. One study that has employed theuse of the M-LAC platform yielded several potentialcandidate proteins including thrombospondin-1 and 5,


Table 2Lectins and their corresponding recognized glycan moieties

Lectin (common name) Abbv. Glycan specificity

α/β-D- Galactosy general specificityAgarius bisporous (common mushroom) ABA Galβ1-3GalNAc-serine/threonineAllomyrina dichotoma (Japanese rhinoceros beetle) AlloA NeuAcα2-3Galβ1-4GlcNAcArtocarpus integrifolia/Jacalin (jackfruit seeds) ALL, JAC Galβ1-3GalNAc, Galα1-6GalArachis hypogaea (peanut) PNA Galβ1-3GalNAcRicinus communis (castor bean) RCA120 Galβ1-4GalNAc

D-N-Acetylgalactoasiminyl general specificityDolichos biflorus (horse gram) DBA GalNAcα1-3GalNAc

GalNAcα1-3Gal

Glycine max (soybean) SBA GalNAcα1-3GalNAcGalNAcαvβ1-3/4Gal

Helix pomatia (common snail) HPA α−GalNAc, GalNAcβ1-4GalPhaseolus lunatus (lima bean) LBA GalNAcα1-3(Fucα1-2)Galβ

GalNAcα1-2Galβ

Vicia villosa (hairy vetch) VVA,VVL

GalNAc-Ser/Thr

D-N-Acetylglucosamine general specificityDatura stramonium (jimson weed) DSA,

DSL(GlcNAcβ1-4)2−4 ,Galβ1-4GlcNAc

Griffonia simplicifolia II GS-II GlcNAcα1-4Galβ1-4GlcNAcLycopersicon esculentum (tomato) LEL (GlcNAcβ1-4)1−4

Triticum vulgare/Wheat Germ Agglutinin (wheat germ) WGA (GlcNAcβ1-4)2−5 , Neu5AcManβ1-4GlcNAcβ1-4GlacNAc

α-L-fucosyl general specificityAleuria aurantia (orange peel fungus) AAL Fucα1-6/3GlcNAcAnguilla anguilla (European eel) AAA α−L-FucLotus tetragonolobus (winged pea) LTL Fucα1-2Galβ1-4(Fucα1-

3)GlcNAcUles europaeus I (common gorse) UEA I Fucα1-2Galβ

D-Mannosyl general specificityCanavalia ensiformis/Concanavalin A (jack bean) Con A Branched N-linked hexa-saccharideGalanthus nivalis (snowdrop) GNA Manα1-3ManLens culinaris (lentil) LCA Fucα1-6GlcNAc-Asn containing

N-linked oligosaccharides

Pisum sativum (pea) PSA Fucα1-6GlcNAc-Asn containingN-linked oligosaccharides

Neu5Ac general specificityMaackia amurensis agglutinin II MAA II NeuAcα2-3Galβ1-4Glc/GlcNAcMaackia amurensis leukoagglutinin MAL Neu5Ac/Gcα2,3Galβ1, 4 Glc(Nac)Maackia amurensis hemoagglutinin MAH Neu5Ac/Gcα2,3Galβ1,

3(Neu5Aca2,6)GalNAc

Sambucus nigra (black elderberry) SNA NeuAcα2-6Gal/GalNAc

Lectins with heterogeneous specificitiesCicer arietinum (chick pea) CPA Complex structure, binding

inhibited by human IgM, fetuin

Euonymus europaeus (Eurpean spindle) EEA Galα1-3(Fucα1-2)Galβ1-3/4GlcNAc

Phaseolus vulgaris PHA-E N-linked bi-antennaryerythroagglutinating (common bean)

Phaseolus vulgaris PHA-L N-linked tri-tetra-antennaryleukoagglutinating (common bean)


alpha-1B-glycoprotein, serum amyloid P-component,and tenascin-X from the serum of breast cancer patientsto be investigated in future studies [44]. In addition tothe use of a multi-lectin affinity column,which incorpo-rated the mixture of lectins Con A, JAC, WGA immobi-lized onto the column resin, the authors also employedimmunodepletion, iso-electric focusing (IEF) fraction-ation, and LC-MS identification. Further analysis ofthe plasma proteome following a similar methodolo-gy, of immunodepletion of albumin and IgG followedby M-LAC established the reproducibility of this plat-form [33].

To continue with the theme of a multi-lectin ap-proach, another study employed the use of biotinylatedlectins E-PHA, AAL, and DSL incubated with ovar-ian tumor tissue lysate, leading to glycoprotein cap-ture [1]. The biotinylated lectins, bound with glyco-proteins, were then removed from the tissue lysate withthe aid of paramagnetic streptavidin beads, and thebound glycoproteins eluted. In a previous study, theauthors identified changes in glycotransferase expres-sion in ovarian tissue, the results of which influencedtheir lectin selection. In doing so, the authors wereable to identify 47 potential lectin-reactive markers inovarian tumors. An additional aspect of the previouslymentioned study includes the validation of several ofthe identified candidate proteins in serum using a sin-gle labeled lectin as an antibody and Western Blotting.Their results indicated some of their candidate markerproteins were bound to a specific lectin, implying thatit was the glycan structure, and not the protein alone,that could be used to distinguish between malignantand normal patient serum. After assessing the valueof several strategies to enrich and identify glycopro-teins, one group decided to perform a “double” multi-lectin capture approach,wherein patient serumwasfirstpassed through a multi-lectin column containing ConA, WGA, and SNA, the resulting eluate was then tryp-tic digested, and the digested peptides were, again, en-riched using the multi-lectin column [5]. This two stageenrichment of glycoproteins, and then glycopeptides,yielded approximately 45 proteins that were differen-tially expressed in the sera of breast cancer patients andnon-disease patients, and when combined with ESI-LC-MS, allowed for glycosylation-site mapping of theidentified proteins.

Multi-lectin capture of glycoproteins allows for aglobal analysis of changes in glycosylation patterns ofproteins between disease and non-disease states. Asthe previously described experiments illustrate, enrich-ment of glycoproteins with lectins allows for identifi-

cation for putative biomarkers, especially when thosemethodologies employ the analysis of patient serum.All of these studies identify potential candidates thatwould need further validation studies, but establish agroundwork of which future studies can be based on.

A single lectin enrichment approach of glycoproteinsappears in greater frequency in the literature, where-in capturing glycosylated proteins of a specific moietymay allow for the identification of candidate markersthat can be validate in future studies. In doing so, theresulting glycoprotein fraction is reduced in complex-ity compared to those derived from multi-lectin affini-ty. This reduction in complexity may also help for theidentification of proteins expressed in low abundancethat may not be detected in any other methodology.One study focused on analyzing samples from normalindividual and patients with lung adenocarcinoma withthe use of the lectin WGA in attempt to identify gly-coprotein biomarkers [11]. First, serum samples fromnormal individuals and patients with lung adenocarci-noma were screened using FITC-labeled lectin stainingmethod, and identified the lectinWGA as exhibiting thehighest specificity for bound glycoproteins in the serumsamples. Glycoproteins in serum samples were thenenriched using the WGA lectin, and the enriched frac-tions were then loaded onto 2-DE gels and analyzed via2D DIGE to observe expression differences betweenthe two sample groups. Interestingly, the authors notedthat WGA incubation not only enriched for the glyco-proteins that bound specifically, but also resulted in theremoval of high abundant serum proteins, that wouldnormally mask potential markers of disease. Using the2-DE methodology, the study yielded the identificationof 39 differentially expressed protein spots. Anotherstudy that analyzed differences in glycosylation in lungcancer used the lectin, Con A, and employed 2-DE todetect changes in protein expression [40]. In this study,serum samples from healthy patients and lung cancerpatients were enriched with Con A, and resulting gly-coprotein samples separated on a 2D gel, spots excised,and analyzed via ESI-LC-MS. Several candidate pro-teins were identified, and again, the use of lectin affin-ity resulted in enrichment of glycoproteins, while alsoremoving highly abundant serum proteins such as IgGand albumin. Analysis of colon cancer serum was at-tempted to reveal the potential link of β-haptoglobinand colon cancer. Increased β-haptoglobin expressionhas often been associated with various pathobiologicalprocesses, thus one group sought to determine whetheraltered glycosylation of β-haptoglobin could serve as adisease marker in colon cancer [28]. The lectin Aleuria


aurantia (AAL) was used to compare the level of fuco-sylation between colon cancer patients, irritable-bowelsyndrome (IBS) patients, and normal subjects, and re-searchers observed that increased fucosylation of β-haptoglobin correlates not only with a cancer state, butwas also indicative of the cancer stage.

To evaluate changes in glycoprotein expression in amodel of breast cancer progression, one group studiedthe membrane fraction derived from two breast cancercell line phenotypes: a normal, premalignant cell lineand a malignant, metastatic cell line [42]. The lectinsWGA and Con A bound to agarose were individuallypacked into spin columns,and themembrane fraction ofeach cell line incubatedwith agarose-bound lectins,andunbound proteins washed away. The resulting boundfraction was tryptically digested and analyzed via LC-MS/MS. To quantitate the glycoprotein differences be-tween the two cell lines, label-free spectral countingwas employed, and 27 membrane glycoproteins weredetected and determined to be differentially expressed.Although this study analyzed cell lines, the approachcould be applied to more relevant clinical samples suchas tumor tissue biopsies. An additional study that usedindividual, agarose-bound lectins in spin columns, ex-amined the difference in expressed glycoproteins be-tween normal and pancreatic cancer serum [46]. Inthis study, not only were expression changes in gly-coprotein analyzed and identified via LC-MS, but alsothe changes in expression of the glycan structures viaMALDI. Interestingly, the latter aspect revealed thatnot only will a change in abundance of a protein relateto a disease state, but also a change in the glycosylationmodification.

In one study, researchers overexpressed the gly-cotransferase glycotransferase cosaminyltransferase V(GnT-V) in WiDr cells to mimic the conditions wit-nessed in colon cancer cells, and analyzed glycosyla-tion differences between the control and experimentalcell lines using the lectin phytohemagglutinin-L4 (L-PHA) [2]. The authors also employed the mass spec-trometry based technique of multiple-reactionmonitor-ing (MRM) to quantify the glycosylation changes ex-pressed between the control and GnT-V overexpressedcells. The two glycoproteins, TIMP1 and PTPκ, bothof which are target proteins of GnT-V, were selectedfor MRM quantitation to assist in a proof-of-conceptexperiment that MRM quantitation could be employed,in conjunction with lectin affinity enrichment, to detectdifferential expression of glycosylated proteins.

An interesting proof-of concept experiment devel-oped incorporated lectin affinity enrichment combined

with an immunodepletion step involving the use of an-tibodies raised against all proteins in the sub-proteome(derived from enriched glycosylated protein fraction)of healthy humans [25]. This methodology, termedTandem Affinity Depletion, employed the use of thelectin, WGA, and antibodies derived from llamas, asthe authors detailed that old world camels generate an-tibodies containing only a heavy chain, which are morerobust in terms of pH and temperature. Using a knowncancer marker, CEA, spiked into the plasma samplefrom several colon cancer patients, the authors wereable to enrich the glycoprotein by a factor of 600–800,validating their approach. The technique, in contrastto immunodepletion experiments which often serve toidentify as many low abundance proteins as possible,instead focused on reducing the complexity of the plas-ma proteome to better differentiate between normal anddisease states.

The development of methodologies that allow forquantifying of glycoproteins in serum has been the fo-cus of several recent studies. One group studied thechanges of sialylation, after enrichment of sialylatedglycoproteins with SNA lectin, in serum of patientswith prostate cancer [37]. Termed lectin-directed tan-dem labeling, distinct sample group enriched glyco-proteins were isotopically labeled with either light orheavy forms of acetic anhydride, and then quantitfiedvia nano-ESI-LC-MS/MS.Researcherswere able to de-termine that it was an increase in sialylation of severalglycoproteins, but not the protein concentration (as de-termined by analysis of non-lectin enriched fractions),that correlated with a disease state. Another study at-tempted to identify N-linked glycoproteins associatedwith hepatocellular carcinoma (HCC) in plasma with-out extensive enrichment steps, and quantify the differ-ences via iTRAQ labeling [18]. Knowing that identifi-cation of glycoproteins via MS are often masked by thesignal of nonglycosylated proteins, researchers gener-ated an exclusion list of peptide masses from nongly-cosylated proteins, thus training the MS instrument toacquire spectra from glycosylated peptides only. Toincrease the pool of potential identified glycosylatedpeptides, the inclusion of lectin enrichment was em-ployed. After identifying several glycopeptides in theserumof HCC patients, iTRAQ labelingwas coupled tothe exclude peak list technique, to elucidate abundancedifferences between normal and HCC plasma samples.Two glycoproteins, vitronectin and antithrombin III,were identified via this method.

Covalent attachment of lectins to magnetic beadsto enrich for glycopeptides or glycoproteins was de-


scribed, and the resulting bound fractions analyzed us-ing MALDI [38]. To facilitate a high-throughput au-tomated procedure, one group employed the use of asingle lectin coupled to magnetic beads, resulting in aone-step glycoprotein purification platform [20]. Useof a robotic liquid handling system and 96-well plateformat, three lectins, Con A, JAC, and WGA, were cou-pled to commercial magnetic beads individually, andthe resulting pulled-down protein fractions were ana-lyzed via 2-DE and the spot identified using MS. Thebenefit of this methodology is increased reproducibil-ity as well application of a high-throughput biomark-er and glycoproteomic analysis technique. Combinedwith bioinformatic analysis, the resulting datasets ex-hibit the potential to reveal and identify alterations inglycosylation profiles of disease states.

Previous studies have employed the use of lectin con-jugated to beads or immobilized onto a column. Onestudy that differs in the approach of lectin affinity isthe use of free lectins to bind to glycoproteins, andthen recover the enriched glycoproteins by either am-monium sulfate precipitation, or the use of a membranefilter unit and centrifugation [43]. Using a multitudeof lectins individually in parallel, including Con A,WGA, Ricinus communis agglutinin (RCA120), Sam-bucus sieboldiana lectin (SSA), and Maackia amuren-sis agglutinin (MAM), the two distinct methodologieswere employed and evaluated. The first, which em-ployed ammonium sulfate precipitation oligosaccha-rides derived from albumin, revealed that the recoveredoligosaccharides moiety recovered in each lectin frac-tion was specific to the lectin employed. The second,incorporating free lectins incubated with tryptically di-gested glycopeptides from glycoprotein standard on topof a centrifugation unit membrane, showed an increaseof enrichment for glycopeptides in each of the frac-tions, with individual lectins showing more specifici-ty for individual, distinct proteins found in the proteinstandard. The results of these two methodologies illus-trate that the use of a free lectin solution approach wascomparable to those results obtained in other affinitychromatography experiments.

Many of the previously mentioned studies requiresubstantial amounts of sample, and in instances wheresample amount is limited, one group developed ananoscale Con A-chelating monolithic capillary [9].The use of this monolithic capillary would allowfor complex, biological samples occurring in minuteamounts to be analyzed and enriched for glycoproteins.To assess the value of their platform, urine samplesfrom healthy individuals and bladder cancer patients

Fig. 1. Flowchart of generalized lectin-affinity strategies. Incorpo-ration of quantitation reagents (i.e. iTRAQ or SILAC) would occurprior to cell lysis. Spiking of internal standard prior to analysis wouldallow for quantitation of tissue, plasma, or serum samples. Immun-odepletion is an optional inclusion (as is other fractionation stepssuch as SEC, SAX, or iso-electric focusing), based on the complexityand composition of lectin-bound fractions.

were tested, using only 10 μg of initial proteins. Whencompared to a standard lectin column, the nanoscalemonolithic column resulted in the identification ofmoreglycoproteins.

As these studies indicate, use of single lectin affinityhas the benefit of reducing the complexity of a pro-teome, in most cases the plasma, or serum, proteome,thus identifying potential candidate biomarkers. A gen-eral workflow diagram for lectin-affinity based method-ologies is illustrated in Fig. 1. Single lectins appear tonot only enrich for glycoproteins, but also indirectlyresult in an immunodepletionof highly abundant serumproteins. The multitude of studies incorporating sin-gle lectin affinity generates a lengthy list of potentialbiomarkers that would need further validation in larg-er patient groups, and the use of methodologies thatemphasize a high-throughput platform may be the bestapproach in doing so.

Several drawbacks of the lectin affinity approach arethat overall, they are quite time-consuming with themultiple stages including enrichment, protein identifi-cation, and data analysis, and require a great deal ofexpertise regarding mass spectrometry analysis. In ad-dition, the use of lectin affinity is based on a pull-downapproach, thereby introducing the occurrence of addi-tional proteins complexed with the lectin-target pro-tein. Although this latter aspect has not been reportedto interfere significantly with the selection of candidateproteins derived from lectin affinity experiments, it is


a concern that could arise when validating potentialcandidate biomarkers.

The development of protein microarrays incorporat-ing lectins has also been employed to study changesin glycosylation. As noted, the use of lectin microar-rays are beneficial as they are more suitable for high-throughput and systematic analysis of protein glyco-sylation, in contrast to more time consuming method-ologies that incorporate affinity chromatography andMS analysis [46]. In a brief communication, the de-velopment of lectin microarrays was detailed, and ex-perimentally shown to provide a distinct, glycosylationpattern for selected protein standards, comparable toprevious methods [32].

To develop potential diagnostic tools in identifyingaggressive prostate cancer, lectin microarrayswere em-ployed [19]. First, the authors used PSA, a knownglycosylated protein found in prostate cancer to eval-uate their methodology, and then expanded by apply-ing their platform to differentiate glycosylation pat-terns between non-aggressive prostate cancer and ag-gressive prostate cancer. Using lectin microarrays, theresearchers were able to detect several isoforms ofPSA and membrane metallo-endopeptidase (MME),both are which are highly expressed in prostate can-cer. Lectin microarrays were also used to determinesurface glycan structures, and thus identify potentialbiomarkers of breast cancer stem-cells [39]. In attemptto generate additional cell surface markers that wouldbe used to differentiate between cancer stem-like cellsand cancer cell lines, the cell lines were analyzed withthe use of lectin microarrays. The researchers revealedthat the breast cancer cell line, MCF7, which can begrown under standard and sphere culture conditions,displayed distinct glycosylation patterns in response tothe culture conditions. To identify differences in gly-can expression between normal and tumorigenic breastcancer cell lines, researchers used lectin arrays [6]. Theinclusion of lectin arrays in the study attempted to re-veal cell surface expression changes in carbohydratesbetween invasive and noninvasive cell lines, and in do-ing so, the study also revealed a glycosylation patternthat related to tissue-specific homing in metastasis. Asthe authors noted, the differences revealed could befunctionally relevant in the development of diagnos-tic and therapeutic targets for detecting and treatingmetastasis.

One study incorporated both multi-lectin affinity andlectin microarrays to analyze the glycosylation patternin serum from patients with pancreatic cancer in anattempt to identify potential serum biomarkers [29].

First, glycoproteins from serum were enriched using ageneral lectin column, and the resulting fraction wasseparated via reversed-phaseHPLC to reduce complex-ity. The researchers selected lectins that exhibited agreat deal of specificity to avoid non-specific proteinbinding, and the resulting array results indicated therewere differences in the glycosylation patterns betweenthe two sample groups. As detailed by the authors,this approach not only allowed for the identification ofglycosylation pattern differences, but also allows forthe identification protein glycosylation differences re-lated to the disease state. To identify carbohydrate-based tumor biomarkers, comparison of serum frompatients with lung cancer and healthy individuals in aplatform using lectin-coupled ProteinChip arrays wasestablished [41]. To reduce interference from glycosy-lated serum proteins, immunodepletion was used priorto enrichment of the remaining glycoproteins with thelectins JAC and SNA coupled to ProteinChip arrays.The researchers discovered the specific glycosylationsite for the protein apoC-III, and two, distinct glycanstructures on this protein, one of which was elevated inthe serum of lung cancer patients.

4. Concluding remarks

Biomarker research has been ongoing for manyyears now, and although perhaps hundreds of candi-date biomarkers have been suggested, only a few havebeen selected for clinical validation and eventually ap-proved for use in clinical settings. There are a numberof factors such as the complexity of cancer biology,the selection bias in the biomarker discovery processes,and the difficulty to conduct prospective clinical vali-dation studies, which all contribute to this low rate ofbiomarker development but are beyond the scope ofthis review. As discouraging as that may be, the ben-efit of biomarkers in the field of medicine and healthmanagement is immeasurable. Specifically in cancer,biomarkers may not only lead to early detection or ver-ify a disease state, but may also have a role in diag-nosing tumor stage, and classifying cancers based ontheir genetic/biologic properties. It is this latter as-pect, wherein cancer biomarkers could play the biggestrole in the development of personalized medicine. Nolonger would treatment be a trial and error process,but rather a directed-therapy based on the molecularmakeup of the diseased individual.

With the advancement in technologies, we are nowable to explore the entire molecular patterns in cells


or tissues. Biomarker discoveries using these globalapproaches were criticized as “fishing expeditions” inthe past, but now show great potential in biomarkerdiscovery. Using proteomics, we are able to comparethe proteome profile between a disease and non-diseasestate, revealing a treasure trove of valuable informa-tion, however, limitations of the technologies used inproteomics currently makes deep analysis of biologi-cal fluids and tissues a challenge. Deep-mining of theproteome could potentially yield clues as to changes innormal biological functions due to a disease state, andwith the approaches currently in use, generation of sub-proteomes may be the best method to accomplish thisgoal. One such sub-proteome, the glycoproteome, ap-pears to be the most beneficial to study, for the reasonspreviously mentioned in this review, and lectin-basedtechniques may prove to be a preferred approach. Asmultiple studies have revealed, the use of lectin-basedtechniques to analyze the glycoproteome has yieldedseveral proteins of interest for various cancers fromdifferent biological sources. These lectin-based strate-gies can be viewed as analyzing the sub-proteome of asub-proteome, resulting in a deep-mining of the humanproteome.

Identifying differences in protein profiles betweennormal and disease statesmay reveal potential biomark-ers, but to determine increased expression of a pro-tein(s) due to the disease state requires precise quantifi-cation. In this review, we discussed studies using 2D-DIGE, iTRAQ, and label-free methods to assess andquantitate relative protein expression differences, how-ever, these methods are limited in the detection of slightprotein expression differences, especially in sampleswith a large dynamic range. In contrast, stable isotopelabeling by amino acids in cell culture (SILAC) is sim-ple, highly sensitive, accurate in samples with a largedynamic range, and its workflow inherently reduces ex-perimental error, allowing for the quantification of evenminute protein expression differences [23]. Surpris-ingly, the incorporation SILAC with lectin-affinity israre, and may be due to many of these studies analyzingserum or tissue samples, wherein SILAC requires liv-ing organisms or autotrophic samples. One techniqueto address shortcoming of the SILAC methodology issuper-SILAC. The super-SILAC method involves theuse of several SILAC-labeled cell lines that match thetissue of interest, serving as an internal standard forquantitation of proteins derived from the tissue sam-ple [26]. The super-SILAC approach is sensitive andaccurate, thus inclusion of this quantitation method forfuture biomarker discovery experiments may better re-

veal subtle differences in the proteomes of normal anddisease states.

Although it appears many of the current lectin-basedtechnologies are limited in their ability to be high-throughput, specific, sensitive, and reproducible, thecombination of these distinct methodologies may beenough to overcome the limitations of each individualtechnique. The future may hold the development ofa biomarker discovery/validation platform that can beapplied in the detection candidate markers in many dis-ease states. The development of such a platform wouldresult in the transition of the identified biomarkers intoclinically relevant applications.

References

[1] K.L Abbot, J. Lim, L. Wells, B.B. Benigno, J.F. McDon-ald, M. Pierce, Identification of candidate biomarkers withcancer-specific glycosylation in the tissue and serum of en-dometroid ovarian cancer patients by glycoproteomic analysis.Proteomics 10 (2010), 470-481.

[2] Y.H. Ah, Y. Kim, E.S. Ji, J.Y. Lee, J. Jung, J.H. Ko, J.S. Yoo,Comparative quantitation of aberrant glycoforms by lectin-based glycoprotein enrichment coupled with multiple-reactionmonitoring mass spectrometry. Anal Chem 82 (2010), 4441-4447.

[3] H.J. An, S.R. Kronewitter, M.L A de Leoz, C.B. Lebrilla,Glycomics and disease markers. Current Opinion in ChemicalBiology 13 (2009), 601-607.

[4] Biomarkers Definitions Working Group, Biomarkers and sur-rogate endpoints: Preferred definitions and conceptual frame-work. Clin Pharmacol Ther 69 (2001), 89-95.

[5] C.D. Calvano, C.G. Zambonin, O.N. Jensen, Assessment oflectin and HILIC based enrichment protocols for characteriza-tion of serum glycoproteins by mass spectrometry. Proteomics71 (2008), 304-317.

[6] S. Chen, T. Zheng, M.R. Shortreed, C. Alexander, L.M. Smith,Analysis of cell surface carbohydrate expression patterns innormal and tumorigenic human breast cell lines using lectinarrays. Anal Chem 79 (2007), 5698-5702.

[7] E.P. Diamandis, Cancer biomarkers: can we turn recent fail-ures into success? J Natl Cancer Inst 102 (2010), 1462-1467.

[8] P.M. Drake, W. Cho, B. Li, A. Prakobphol, E. Johansen, N.L.Anderson, F.E. Regnier, B.W. Gibson, S.J. Fisher, Sweeten-ing the pot: adding glycosylation to the biomarker discoveryequation. Clin Chem 56 (2010), 223-236.

[9] S. Feng, N. Yang, S. Pennathur, S. Goodison, D.M. Lubman,Enrichment of glycoproteins using nanoscale chelating con-canavalin a monolithic capillary chromatography. Anal Chem81 (2009), 3776-3783.

[10] G.W. Hart and R.J. Copeland, Glycomics hits the big time.Cell 143 (2010), 672-676.

[11] P. Hongsachart, R. Huang-Liu, S. Sinchaikul, F. Pan, S.Phutrakut, Y. Chaung, C. Yu, S. Chen, Glycoproteomic anal-ysis of WGA-bound glycoprotein biomarkers in sera from pa-tient with lung adenocarcinmoa. Electrophoresis 30 (2009),1206-1220.

[12] G.L. Hortin, D. Sviridov, The dynamic range problem in theanalysis of the plasma proteome. Proteomics 73 (2010), 629-636.


[13] J.M. Jacobs, J.N. Adkins, W. Qian, T. Liu, Y. Shen, D.G.CampII, R.D. Smith, Utilizing human blood plasma for proteomicbiomarker discovery. J Proteome Res (2005), 1073-1085.

[14] S. Khleif, J.H.Doroshow, W.N.Hait, AACR-FDA-NCI cancerbiomarkers collbaborative consensus report: advancing theuse of biomarkers in cancer drug development. Clin Can Res16 (2010), 3299-3318.

[15] E.H. Kim and D.E. Misek, Glycoproteomics-based identifica-tion of cancer biomarkers. Int J Proteomics (2011), 84-94.

[16] M. Kullolli, W.S. Hancock, M. Hincapie, Preparation of ahigh-performance multi-lectin affinity chromatography (HP-M-LAC) adsorbent for the analysis of human plasma glyco-proteins. J Sep Sci 31 (2008), 2733-2739.

[17] A. Larkin and B. Imperiali, The expanding horizons ofasparagine-linked glycosylation. Biochemistry 50 (2011),4411-4426.

[18] H. Lee, K. Na, E. Choi, K.S. Kim, H. Kim, Y. Paik, Simplemethod for quantitative anaysis of n-linked glycoproteins inhepatocellular carcinoma specimens. J Proteome Res 9 (2010),308-318.

[19] Y. Li, S. Tao, G.S. Bova, A.Y. Liu, D.W. Chan, H. Zhu, H.Zhang, Detection and verification of glycosylation patterns ofglycoproteins from clinical specimens using lectin microar-rays and lectin-based immunosrobent assays. Anal. Chem. 83(2011), 8509-8516.

[20] D. Loo, A. Jones, M.M. Hill, Lectin magnetic bead array forbiomarker discovery. J Proteome Res 9 (2010), 5496-5500.

[21] J.A. Ludwig and J.N.Weinstein, Biomarkers in cancer staging,prognosis and treatment selection. Nature Reviews Cancer 5(2005), 845-856.

[22] S. Makawita and E.P. Diamandis, The bottleneck in the can-cer biomarker pipeline and protein quantitation through mass-spectrometry-based approaches: current strategies for candi-date verification. Clin Chem 56 (2010), 212-222.

[23] M. Mann and N.L. Kelleher, Precision proteomics: The casefor high resolution and high mass accuracy. PNAS 105 (2008),18132-18138.

[24] D.L. Meany and D.W. Chan, Aberrant glycosylation associ-ated with enzymes as cancer biomarkers. Clin Proteomics 8(2011), 7-21.

[25] N. Mortezai, S. Harder, C. Schnabel, E. Moors, M. Gauly,H. Schluter, C. Wagener, F. Buck, Tandem affinity depletion:a combination of affinity fractionation and immunoaffinitydepletion allows the detection of low-abundance componentsin the complex proteomes of body fluids. J Proteome Res 9(2010), 6126-6134.

[26] T.A. Neuber and P. Tempst, Super-SILAC for tumors andtissues. Nature Methods 7 (2010), 361-362.

[27] S. Ogino, J. Galon, C.S. Fuchs, G. Dranoff, Cancerimmunology-analysis of host and tumor factors for personal-ized medicine. Nat Rev Clin Oncol (2011), doi: 10.1038/nr-clinonc.2011.122.

[28] S. Park, S. Lee, N. Kawasaki, S. Itoh, K. Kang, S.H. Ryu, N.Hashii, J. Kim, J. Kim, J.H. Kim, α1-3/4 fucosylation at Asn241 of β-haptoglobin is a novel marker for colon cancer: acombinatorial approach for development of glycan biomark-ers. Int J Cancer (2011), doi: 10.1002/ijc.26288.

[29] T.H. Patwa, J. Zhao, M.A. Anderson, D.M. Simeone, D.M.Lubman Screening of glycosylation patterns in serum usingnatural glycoproteinmicroarrays andmulti-lectinfluorescencedetection. Anal Chem 78 (2006), 6411-6421.

[30] T. Patwa, C. Li, D.M. Simeone, D.M. Lubman, Glycopro-tein analysis using protein microarrays and mass spectrometry.

Mass Spectrometry Reviews 29 (2010), 830-844.[31] M.S. Pepe, R. Etzioni, Z. Feng, J.D. Potter, M.L. Thompson,

M. Thornquist, M. Winget, Y. Yasui, Phases of biomarkerdevelopment for early detection of cancer. J Natl Cancer Inst93 (2001), 1054-1061.

[32] K.T. Pilobello, L. Krishnamoorthy, D. Slawek, L.K. Mahal,Development of a lectin microarray for the rapid analysis ofprotein glycopatterns. ChemBio Chem 6 (2005), 985-989.

[33] T. Plavina, E. Wakshull, W.S. Hancock, M. Hincapie, Com-bination of abundant protein depletion and multi-lectin affin-ity chromatography (M-LAC) for plasma protein biomarkerdiscovery. J Proteome Res (2007), 662-671.

[34] C.A. Reis, H. Osorio, L. Silva, C. Gomes, L. David, Alter-ations in glycosylation as biomarkers for cancer detection. JClin Pathol (2010), 322-329.

[35] J.M. Rhea and R.J. Molinaro, Cancer Biomarkers: Survivingthe journey from bench to bedside. MLO Med Lab Obs 43(2011), 10-18.

[36] R. Schiess, B. Wollscheid, R. Aebersold, Targeted proteomicstrategy for clinical biomarker discovery. Mol Oncol 3 (2009),33-44.

[37] V. Shetty, Z. Nickens, P. Shah, G. Sinnathamby, O.J. Semmes,R. Philip, Investigation of sialylation aberration in n-linkedglycoproteins by lectin and tandem labeling (LTL) quantitativeproteomics. Anal Chem 82 (2010), 9201-9210.

[38] K. Sparbier, S. Koch, T. Wenzel, M. Kostrzewa, Selective Iso-lation of glycoproteins and glycopeptides for maldi-tof ms de-tection supported by magentic particles. Journal of Biomolec-ular Techniques 16 (2005), 407-413.

[39] S. Tao, Y. Li, J. Zhou, J. Qian, R.L. Schnaar, Y. Zhang, I.J.Goldstein, H. Zhu, J.P. Schneck, Lectin microarrays identifycell-specific and functionally significant cell surface glycanmarkers. Glycobiology 18 (2008), 761-769.

[40] T.T. Thanh, N.T.M Phuong, N.B. Nhi, P.V. Chi, Changes ofglycoproteins in lung cancer patients. J Proteomics Bioinform1 (2008), 11-16.

[41] K.Ueda, Y. Fukase, T.Katagiri, N. Ishikawa, S. Irie, T. Sato, H.Ito, H. Nakayama, Y. Miyagi, E. Tsuchiya, N. Kohno, M. Shi-wa, Y. Nakamura, Y. Daigo, Targeted serum glycoproteomicsfor the discovery of lung cancer-associated glycosylation dis-orders using lectin-coupled ProteinChip arrays. Proteomics 9(2009), 2182-2192.

[42] Y. Wang, X. Ao, H. Vuong, M. Konanur, F.R. Miller, S. Good-ison, D.M. Lubman, Membrane glycoproteins associated withbreast tumor cell progression identified by a lectin affinityapproach. Journal of Proteome Research 7 (2008), 4313-4325.

[43] M. Yodoshi, T. Oyama, K. Masaki, K. Kakehi, T. Hayakawa,S. Suzuki, Affinity entrapment of oligosaccharides and gly-copeptides using free lectin solution. Analytical Sciences 27(2011), 395-400.

[44] Z. Zeng, M. Hincapie, S.J. Pitteri, S. Hanash, J. Schalkwi-jk, J.M. Hogan, H. Wang, W.S. Hancock, A proteomics plat-form combining depletion, multi-lectin affinity chromatogra-phy (M-LAC), and isoelectric focusing to study the breastcancer proteome. Anal Chem 83 (2011), 4845-4854.

[45] L. Zhang, H. Lu, P. Yang, Specific enrichment methods forglycoproteome research. Anal Bioanal Chem 396 (2010), 199-203.

[46] J. Zhao, D.M. Simeone, D. Heldt, M.A. Anderson, D.M. Lub-man, Comparative serum glycoproteomics using lectin select-ed sialic acid glycoproteins with mass spectrometric analysis:application to pancreatic cancer serum. Journal of ProteomeResearch 5 (2006), 1792-1802.

Submit your manuscripts athttp://www.hindawi.com

Stem CellsInternational

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014


MEDIATORSINFLAMMATION

of


Behavioural Neurology

EndocrinologyInternational Journal of



Disease Markers


BioMed Research International

OncologyJournal of



Oxidative Medicine and Cellular Longevity


PPAR Research

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

Immunology ResearchHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Journal of

ObesityJournal of



Computational and Mathematical Methods in Medicine

OphthalmologyJournal of


Diabetes ResearchJournal of



Research and TreatmentAIDS


Gastroenterology Research and Practice


Parkinson’s Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttp://www.hindawi.com

cancer biomarker discovery: lectin-based strategies targeting...

Documents