And Glycoarrays

DEVELOPMENTAND APPLICATION OF PEPTIDE- AND GLYCOARRAYS

A THESIS SUBMITTEDTO THE UNIVERSITYOF MANCHESTER FORTHEDEGREEOF DOCTOROFPHILOSOPHY (PHD) IN THE FACULTY OF ENGINEERINGAND PHYSICAL SCIENCES

DIPL.-CHEM.MARTIN WEISSENBORN,MSC

SCHOOLOF CHEMISTRY 2012 CONTENTS

Declaration5

Copyright6

Acknowledgements7

Abstract9

1 Thesis Structure 10

2 Enzymatic Reactions on Solid-Support 11

3 Enzymatic Glycosylations on Arrays 12

4 Glycoarrays on Gold Surfaces 13

5 Objectives of this Thesis 14 5.1 Array Formation...... 14 5.2 Analysis of Arrays...... 15 5.3 Application of Microarrays...... 16

6 Methodologies applied in this Thesis 17 6.1 Chemical Synthesis...... 17 6.2 Arrays on Gold...... 17 6.2.1 Coupling into SAMs...... 17 6.2.2 MALDI-ToF MS analysis of SAMs...... 19 6.3 Surface Plasmon Resonance (SPR) on SAMs...... 20 6.4 Arrays on Polystyrene...... 22

7 Preparation of aminoethyl glycosides for glycoconjugation 23 7.1 Supporting Information...... 24

8 Oxo-ester mediated native chemical ligation 25 8.1 Supporting Information...... 26

2 CONTENTS

9 MALDI-ToF MS Analysis on Glass and Polystyrene 27 9.1 Supporting Information...... 28

10 Dual purpose S-trityl-linkers for glycoarray fabrication on both polystyrene and gold 29 10.1 Supporting Information...... 30

11 High-Throughput Screening of Protein Glycosylation Using Lectin-Binding Biophotonic Microarray Imaging 31

12 Crystal structure of a soluble form of human CD73 with ecto-5’-nucleotidase activity 32 12.1 Supporting Information...... 33

13 Chemoenzymatic Synthesis of O-Mannosylpeptides 34 13.1 Supporting Information...... 35

14 Conclusion and Outlook 36 14.1 Conclusion...... 36 14.2 Outlook...... 37

15 Bibliography 39

Final word count: 69800

3 LIST OF FIGURES

6.1 Formation of SAMs on gold...... 18 6.2 Formation and functionalisation of self-assembling monolayers on gold.. 18 6.3 The mechanism of the activation of the linker 2 by EDC to form the activated ester 3...... 19 6.4 Maleimide functionalised SAMs followed by the formation of thioethers. 19 6.5 Schematic illustration of direct MALDI-ToF MS analysis of SAMs.... 21 6.6 MALDI-ToF MS spectra of unmodiﬁed and modiﬁed SAMs...... 21 6.7 Illustration of the surface plasmon resonance (SPR) technique...... 22

4 DECLARATION

The University of Manchester PhD by published work Candidate Declaration

Candidate Name: Martin Weissenborn

Faculty: Engineering and Physical Sciences

Thesis Title: Development and Application of Peptide- and Glycoarrays

Declaration to be completed by the candidate:

I declare that no portion of this work referred to in this thesis has been submitted in support of an application for another degree or qualiﬁcation of this or any other university or other institute of learning.

Signed: Date: December 7, 2012

5 COPYRIGHT

The author of this thesis (including any appendices and/or schedules to this thesis) owns any copyright in it (the "Copyright")1and he has given The University of Manchester the right to use such Copyright for any administrative, promotional, educational and/or teaching purposes. Copies of this thesis, either in full or in extracts, may be made only in accordance with the regulations of the John Rylands University Library of Manchester. Details of these regulations may be obtained from the Librarian. This page must form part of any such copies made. The ownership of any patents, designs, trade marks and any and all other intellectual property rights except for the Copyright (the "Intellectual Property Rights") and any reproductions of copyright works, for example graphs and tables ("Reproductions"), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property Rights and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property Rights and/or Reproductions. Further information on the conditions under which disclosure, publication and ex- ploitation of this thesis, the Copyright and any Intellectual Property Rights and/or Repro- ductions described in it may take place is available from the Head of School of Chem- istry(or the Vice-President) and the Dean of the Faculty of Engineering and Physical Sci- ences, for Faculty of Engineering and Physical Sciences candidates.

1This excludes material already printed in academic journals, for which the copyright belongs to said journal and publisher. Pages for which the author does not own the copyright are numbered differently from the rest of the thesis.

6 ACKNOWLEDGEMENTS

First and foremost I would like to thank my supervisor, Prof Sabine Flitsch, for giving me the opportunity to work on this great project in a fascinating and highly interdisciplinary network. Thanks also goes to my former supervisor Prof Thisbe Lindhorst for recommending me to Prof. Flitsch and for the fruitful collaboration.

I am furthermore grateful for the generous support of the European Commission for the Marie Curie research fellowship. This made many collaborations possible.

Thanks to these collaborators for hosting me in their labs and giving rise to many interesting and successful projects.

Mein Dank an meine Familie ist nicht in einem Satz zusammenzufassen. Ich danke euch für eure Unterstützung in allen Lebenslagen, die Idee Chemie zu studieren und das Studium zu ﬁnanzieren. Und noch so vieles mehr.

I am also very proud of being a part of the Turner/Flitsch group. It was great to work in such a nice atmosphere for three years. Special thanks goes to my colleagues Dr Robert Sardzik, Dr Josef Voglmeir, Dr Damien Debecker, Dr Mark Corbett, Paul Kelly, Roberto Castangia, Dominique Richardson and Christopher Gray.

Dank an an meine Freunde aus Kiel, für eine tolle gemeinsame Zeit in der ich viel Hilfe von euch erfahren habe, and to my friends in Manchester for a great time.

Lastly, I would like to thank Dr Ivan Vilotijevic, Dr Mark Corbett and Paul Kelly for their fantastic and numerous proof readings — including this section.

7 Ich widme diese Arbeit in Liebe und Dankbarkeit meiner Familie. Abstract

Microarrays enable high throughput analysis with minute amounts of analyte. They are widely used in the ’omics’ fields both as diagnostic and analytical tools. Their ability to dramatically impact an entire field of research has focused our attention on the development of novel methods for the formation, analysis and applications of microarrays to study carbohydrate-protein interactions and the analysis of glycosylation patterns of biomolecules. Availability of appropriately modified ligands is often a limiting factor in the preparation of microarrays. To address this issue robust routes for the synthesis of nine aminoethyl glycosides were developed that can be employed for microarray formation. The syntheses of more complex ligands typically deliver small quantities of material despite the requirements for special skills, equipment and long preparation times. Considering the number of complex oligosaccharides that are necessary for systematic microarray studies, the problem of availability of these complex structures is difficult to address solely with synthetic ligands. A modified native chemical ligation (NCL) strategy, in which a surface bound oxo-ester is used instead of a thioester, was optimised and used for efficient chemoselective immobilisation of sugars and peptides carrying N-terminal cysteines. The reaction proceeds under physiological conditions and has the potential to become a valuable tool for immobilisation of N-terminal cysteine-containing molecules from biological samples. The new NCL coupling methodology was developed on gold surfaces and analysed by MALDI-ToF MS. The majority of array systems, however, rely on secondary protein interactions on glass or polystyrene surfaces. A direct, more accurate analytical tool could ease the analysis and significantly improve the quality of data read-out from glass microarrays. MALDI-ToF MS that is applicable to gold microarrays cannot be used on surfaces that do not provide the necessary electrical conductivity. The undertaken experiments indicated that application of conductive tape to the back of glass or polystyrene slides made MALDI-ToF analysis on poorly conducting surfaces possible. Furthermore, the triphenylmethyl (trityl) groups attached to the surface-molecules were shown to act as ’internal-matrix’ and enable the direct MALDI analysis. Once the new array formation and analysis techniques were developed, we turned our attention towards the application of microarrays to analyse carbohydrate-protein interactions. The tools for analysis of glycosylation of biomolecules are laborious and can only be used in specialised labs. As glycosylated biomolecules gain prominence in research, clinical and industrial settings, high throughput analysis of glycosylation patterns is becoming a requirement for quality control. A technique for screening of glycosylation patterns in glycopeptides on microarrays was developed based on biophotonic scattering. This technique enables the detection of glycosylation patterns by screening immobilised glycoproteins with a range of lectins. To study the interactions between enzymes and carbohydrates, a chemoenzymatic synthesis of a mannopeptide, which consisted of four carbohydrate units, was shown in solution and on chip. Three different glycosyltransferases were successfully employed. New methods for microarray formation and analysis were developed and applied to carbohydrate-protein interaction studies. This yielded a new technique to determine protein glycosylation patterns and to produce complex glycans by enzymatic synthesis.

9 CHAPTER ONE

THESIS STRUCTURE

This thesis describes our work on the development of peptide- and glycoarrays. It consists of the articles that have already been published and the manuscripts that have been submitted for publication.

Two review articles serve as a general introduction to the thesis. The ﬁrst article (chapter2) provides an overview of microarray platforms—microarray surfaces and the techniques for their analysis—and introduces enzymatic reactions with substrates immobilised on solid-support. The following chapter3 focuses on enzymatic glycosylations on array based systems and provides a comparison of direct and indirect techniques used for their analysis. An overview of the glycoarrays on gold surfaces developed in the Flitsch group (chapter4) precedes the chapters about the speciﬁc objectives of the work described herein (chapter5) and the methodologies utilised in our experiments (chapter6).

The original research articles presented in chapters7-13 describe four aspects of our work: (i) the synthesis of ligands that were used on the microarrays (chapter7), (ii) the development of a new method for immobilisation of ligands on microarrays (chapter8), (iii) the development of a novel glycoarray and its analysis by MALDI-ToF MS (chapter 9 and 10), and (iv) the use of this array for detection of interaction of carbohydrates with lectins (chapters 11 and 12) and enzymes (chapter 13). A detailed description of author contributions is provided at the beginning of each chapter.

The conclusions and future outlook are discussed in the ﬁnal chapter 14. The numbering of the molecules is consistent within each chapter. The compound numbers cannot be cross-referenced between different chapters due to the format in which this thesis is presented. The methodologies and experiments developed by the author are described in detail in chapter6. Additional supporting information is provided after each article for all experiments that were independently carried out by the author. A link or a reference to the complete supporting information including the work carried out by the collaborators is provided where appropriate.

10 CHAPTER TWO

ENZYMATIC REACTIONS ON SOLID-SUPPORT

This invited review for Chemical Society Reviews is ready for submission. Copyrights for the implemented ﬁgures will be obtained for the publication. M. J Weissenborn, C. J. Gray, C. E. Eyers, S. L. Flitsch, Enzymatic Reactions on Solid-Support.

The first draft of this review was written by C. J. Gray and M. J. Weissenborn. This consisted of finding relevant publications and summarising them. C. J. Gray focused on the methods and surfaces part and M. J. Weissenborn on the enzymatic reactions sections. The graphics and tables in the first draft were made by C. J. Gray and M. J. Weissenborn. The formatting of all references, figures, tables and text in the typesetting programme LATEX was done by M. J. Weissenborn. The draft was then finalised by C. E. Eyers and S. L. Flitsch.

11 Enzymatic Reactions on Immobilised Substrates†

‡ ‡ Martin J. Weissenborn, Christopher J. Gray, Claire E. Eyers∗ and Sabine L. Flitsch∗

Received Xth XXXXXXXXXX 20XX, Accepted Xth XXXXXXXXX 20XX First published on the web Xth XXXXXXXXXX 200X DOI: 10.1039/b000000x

1 Introduction that chemical and enzymatic reactions on surfaces will become more and more important. A number of reviews have The study of enzymatic reactions on substrates which are im- been previously published on specific aspects of enzymatic mobilised on surfaces is a relatively new field compared to the reactions on surfaces, but to our knowledge this is the first ’reverse’ system involving immobilised enzymes and soluble review that attempts to give the reader an overview of the area substrates (Fig. 1). and discuss scope and limitations of the technology. 19–23 A wide range of enzyme substrates have been immobilised to surfaces ranging from large to small molecules and include antibodies, 24 proteins, 25 peptides, 26 glycans, 12,27 glycopeptides, 10 DNA, 6 RNA, 28 and a variety of small substrates. 29 These studies have demonstrated significant applications in diagnostics, 24 ’omics’ studies, 30 sensors, 31 plant sciences, 32 enzyme kinetics, 33 enzyme specificity, 34 cell biology, 35 and organic (synthetic) chemistry. 10,36 Enzymes have also been shown to be important tools for solid-phase synthesis, since they are often regio- and stere- oselective and can catalyse reactions under mild biocompatible conditions. Solid-phase chemical synthesis for peptides, DNA and RNA is now well developed, but solid-phase synthesis of other biomolecules, such as carbohydrate, remains challenging and enzymatic glycosylations have become important Fig. 1 Two different ways of enzyme reaction with surfaces are tools for the synthesis of oligosaccharides. For example, Mrk- shown.(Top) Enzyme immobilisation and conversion of substrates in sich and Ban (2008) have used enzymes to prepare a series of solution phase and (bottom) enzymatic reactions on immobilised disaccharides and trisaccharides chemoenzymatically on gold substrates. from 11 building blocks (5 galactosides, 5 glucosides and a glucoaminoside). 37 This review is organised to cover three The interest in this area of research has been driven by ap- important aspects of the area: analytical techniques, choice plications in bionanotechnology 1,2 as biocompatible tools and of surfaces and the range of enzymatic reactions that have by applications in high-throughput screening systems particu- been studied. Despite many examples of successes, it is un- larly on nanoparticles 3 and microarrays. 4 An important appli- likely that every enzyme is able to act on surfaces, in particular cation of microarrays is the elucidation of post-translational where active sites are buried deep in the enzyme structure and modifications (PTMs) such as phosphorylations, glycosyla- tethers to solid supports may not be tolerated by the enzyme. tions and proteolysis which are invariably catalysed by en- In Section 4, we have put together a list of successful enzy- zymes and increasingly provide important drug targets. 5–18 matic reactions on solid support reported so far and discussed Given the drive towards automation and miniaturisation in these reactions in more detail to give the reader information chemistry, biotechnology,biology and medicine, it is expected on the scope of the technology. The right choice of surface is an integral part of the reaction Electronic Supplementary Information (ESI) available: [details of any system shown in Fig. 1. A wide range of enzyme-compatible † supplementary information available should be included here]. See DOI: surfaces have been reported including resin beads, 38 glass, 39 10.1039/b000000x/ gold 10 and polystyrene. 40 It should be noted that reaction 131 Princess Road, Manchester, M1 7DN, United Kingdom. Fax: 44- 161- 275-1311; Tel: -44- 161-306-5172; E-mail: sabine.fl[email protected] characteristics can change considerably when moving from These authors contributed equally to this work soluble to immobilised substrates in terms of kinetics, 14,41 ‡

1–32 1 | Fig. 2 The different analytical techniques subdivided in label-free (left) and label-required (right). The label-free analytical techniques can be performed directly on the surface or after cleavage in solution. The analytical techniques that require a label are generally analysed directly on the surface. thermodynamics 42 and stereospecificity. 19,43 Given that teth- 2 Analytical Techniques ering to surfaces changes the substrate structure and can influence accessibility of the substrate by the enzyme, such differ- Enzymatic transformations will always result in the change of ences are perhaps not surprising. the chemical composition of surfaces and the challenge for analysis is to monitor such changes. A wide range of analyt- The choice of surface in the reaction system shown in Fig. ical techniques to study surface composition have now been 1 depends on particular applications – for example, commer- described. These techniques can be classified as those requir- cially available glass slides are commonly used in microarray ing labelling (such as radioactive or fluorescent labelling) and applications, whereas plastic coated microtiter wells would be label-free methods (Fig. 2). the format of choice for Enzyme-linked immunosorbent assays (ELISA). A particularly important problem that needs to be overcome for enzyme-catalysis on immobilised substrates 2.1 Radioactivity is non-specific protein adsorption 44 which can quickly lead to the inactivation of enzymes. This review will discuss the range The use of radioactive isotopes to follow biochemical reac- of successful surfaces and attachment chemistries 12,45–47 in tions has been used for a long time. For surface reactions, detail in Section 3. any incorporation of radioactivity through enzymatic transformations can be monitored by exposure to a photogenic film. There are now a number of analytical techniques avail- The advantage is high sensitivity, 99 retention of native sub- able to monitor enzyme activity on surfaces (Fig. 2). These strate structure and does not require an affinity tag. Smith techniques include matrix-assisted laser desorption/ionisation 10,35,37 et al. demonstrated the use of radioactively labelled ATP to time-of-flight mass spectrometry (MALDI-ToF MS), monitor the activity of protein kinase A (PKA) and compared fluorescence, 48–50 radioactive radiation, 17,26,51 electrochem- 52 53 that to an anti-phospho-PKA antibody and found the antibody ical techiques, surface plasmon resonance (SPR) and produced false negatives (Fig. 3). atomic force microscopy (AFM). 54,55 The methods for analy- A disadvantage of this method is the need for suitably sis can dictate the choice of surface and will be discussed in labelled substrates, thus radioactive labelling is often used more detail in Section 2. when the common biochemicals are commercially available, An overview of the available analytical techniques on the such as 32P (or 33P)-ATP. Other examples are acetyltrans- different surfaces is given in table 1. ferases, fucosyltransferases and methyltransferases, which

2 1–32 | Table 1 Publications of employed analytical techniques (left column) on the different surfaces (top row).

Glass Nitrogel Nitrocellulose/ Gold Non- Other on Glass Cellulose Immobilised on Glass Polymer

MALDI ToF MS 12 56 9,10,27,30,35–37,57–65 12 45

Fluorescence 7,16,29,48,66–80 49,81,82 5,26,83–86 3,59,87–89 90–93 94,95

QCM 33,89,96,97

Radiation 98 25,26,51,99–103 87,104 40,105,106

Surface plasmon optics 28,53,87,89,104,107–114

AFM 55,114,115 54,116

Electrochemical 65,117–120 52,54 have been studied using 14C-AcCoA, 17,98 14C-Fucose 106 and self-assembled monolayers on gold (SAMs). These systems 3H-AdoMet, 101,105 respectively. Sulphur (35S) radiolabels are important for SPR and MALDI based studies. 44,126 Al- have also been used. 40 These radiolabels are all beta emitters ternatively the surface can be blocked with a protein solution with relatively long half lives (14C 5730 years, 3H 12 years, such as bovine serum albumin (BSA) or milk proteins, pre- 32P 14 days, 33P 25 days and 35S 87 days). 121∼ venting further adsorption of labelled proteins. 127 This is typ- ∼ ∼ ∼ ically a simpler approach than coating the surface with PEG; 2.2 Fluorescence however, care must be taken to ensure the blocking proteins are not substrates for the labelled ligand of interest (for exam- The use of fluorescent labels and subsequent spectroscopic ple, some blocking proteins are glycosylated and can cross- analysis is now the most commonly used method of visual- react with lectins). 128 ising enzymatic action on a high-throughput microarray. Flu- orescent detection is a very sensitive technique exemplifiedby If a suitable product-binding protein is not available, an al- the fact a single molecule of Cy3 or Cy5 dye can be detected ternative strategy can be the incorporationof generic biochem- in 10µm2 of protein immobilised to a surface. 122 Multiplex- ical labels such as biotin into the reaction products. These ing≈ of different reactions on the same surface is possible us- labels are then detected by labelled binding proteins such as ing multiple dyes allowing multiple quantification in a sin- streptavidin. 5,50,71,82 The incorporation of such labels can sig- gle experiment. 123 There is now a huge range of fluorescent nificantly distort substrate structures 122 and hence small la- dyes commercially available such as Cy3, Cy5, RhodRed and bels are often very useful. Particularly useful have been azide FITC-Fluorescent labels are cheaper and safer than radioac- and alkyne labels which are generally biochemically inert and tive labels and allow for ’real-time’ monitoring. 124 There are can be coupled subsequently after the enzymatic reactions several different strategies of measuring enzyme activity by by using ’click’ chemistry (Huisgen cycloaddition). 129 Where fluorescence. larger labels are tolerated, fluorescent groups can be directly A very popular way of incorporating fluorescent labels into incorporated into the product during the reaction and be mon- an enzymatic reaction scheme is by detection of product with itored by fluorescence spectroscopy. 6,32 This has been suc- specific binding proteins, such as antibodies or lectins. 48,84 cessfully used to study transglycosylase activity of plant crude This is a useful method where such product binding proteins extracts on a glycoarray. Oligosaccharides labelled with sul- are available. 56 The method requires careful controls to avoid forhodamine were attached to the glycoarray substrates thus non-specific adsorption of binding proteins to surfaces. Such allowing observation of transglycosylase activity from a plant problems can be overcome by modification of surfaces with extract. 32 This type of approach has advantages such as it is hydrophilic polyethylene glycol (PEG) groups prior to attach- highly sensitive and can produce quantitative data 130 and does ment of the enzyme substrate. 125 A very thoroughand system- not rely on an additional ligand-tag binding reaction thus im- atic study has shown that PEG is most efficient in preventing proving the qualitative value of fluorescence observed. A pos- non-specific protein adsorption when added to the termini of sible issue with this strategy is that the fluorescently labelled

1–32 3 | cium (II). 31,131 An advantage of this techniqueis the reduction of background fluorescence due to the masking of the fluo- rophore (or presence of a quencher). 132 Alternatively, the loss of fluorescence could be used to determine enzymatic activity. 94 The substrate density on an array surface can greatly influence reaction efficiency and it is generally advisable to run reactions at different surface densities. If the substrate layer on the spot is too dense, reaction yields of enzymatic transformations can be very low, possibly due to steric hindrance. 133,134 However, reducing the density of the layer also reduces the sensitivity of the technique. Shimomura et al. have addressed the problem of sensitivity by monitoring the phosphorylation of tyrosine residues on immobilised peptides using a much smaller fluorescently tagged ligand phos-tag. 135 Phos-tag is a dinuclear zinc complex with a vacancy between the two zinc centres capable of binding to phosphorylatedpeptides. 136 This tag produced more quantitative results than using the conventional fluorescently labelled anti-phospho-tyrosine antibody. 135

Fig. 3 Serine was mutated for alanine and compared to with the 2.3 Electrochemical wildtype peptides (framed in red). Phosphorylation by PKA was determined with an anti-phospho-PKA-antibody and incorporation Particularly useful for the design of diagnostic devices is de- of 32P phosphate. The radioactive detection showed the tection by electrochemical methods. This method involves phosphorylation of the fifth peptide which was not detected by the incorporation of a redox active compound during the reac- antibody. 26 tion that can then be picked up by cyclic or differential pulse voltammetry (CV or DPV). Both CV and DPV involving ramping the potential of an electrode and measuring the cur- cosubstrate might not be able to be processed by the enzyme rent. During reduction (or oxidation) of a redox active com- that is being studied. Mazan et al. (2011) found that the effi- pound, the magnitude of current rapidly increases. 137 Song ciency of their oligosaccharide fluorescent acceptor molecule et al. were able to study protein kinase C (PKC) by using a increased with increasing oligosaccharide length with a pen- γ-ferrocene (redox active) labelled ATP co-substrate on a pep- tasaccharide being the smallest possible acceptor. 130 A num- tide array by CV represented in Fig. 4. 119 ber of very ingenious screening protocols have been developed Similarly Kerman et al. studied tyrosine kinase p60c-Src which make use of masked fluorescent systems in the enzyme and protein kinase A (PKA) phosphorylating immobilised substrate which gets ’unmasked’ upon enzymatic transforma- RaytideTMEL and Kemptide. This was achieved in a similar tion, for example by using Fluorescence Resonance Energy fashion to Song et al. except that a biotinylated ATP cosub- Transfer (FRET) analysis. This is a particularly useful method strate was used and in a later step a streptavidin coated AuNP for monitoring proteolytic cleavage of peptides in solution and was bound to this, which could be observed by DPV. 52 Us- on arrays. 66 On gold nanoparticles (AuNP), the masking can ing this approach they were able to study inhibition of these be achieved simply by immobilisation of a fluorescently la- kinases. The addition of AuNP made the technique very sen- belled peptide. The AuNP quenches the fluorescence of the sitive and could detect RaytideTMEL and Kemptide at concen- label prior to proteolytic cleavage. Proteolytic cleavage re- tration of 5 and 10 µM, respectively. Platinum nanoparticles sults in the fluorescent molecule being released thus emitting (PtNP) have also been used as tags. 117 Very recently, rolling fluorescence. 3 Kim et al. also exploited a similar system for circle amplification (RCA) was used to amplify the signal of a lead (II) sensor. The sensor was based on a lead-depending CV allowing the action of PKA on peptides to be monitored at DNAzyme. This sensor allowed detection of lead (II) in wa- a detection limit of 0.5 units/mL,comparableto that of fluores- ter at concentrations as low as 5 nM and could be used to cence (Fig. 5). Peptides were immobilised to a gold electrode rapidly determine lead concentrations in sera. Low levels of and phosphorylated by PKA. Then Zr (IV) was added with lead ( 500 nM) can cause severe neurotoxic effects as it eas- a primer probe that complexed the phosphate group and the ily unspecifically∼ uptaken via calcium channels instead of cal- primer probe. This primer probe was extended using a DNA

4 1–32 | Fig. 4 Cyclic voltammetry (CV) analysis of protein kinase C (PKC) activity with the aid of an redox actively labelled ATP. The PKC incorporates the ferrocene tagged phosphate in the peptide causing a change in the potential detectable by CV. 119 polymerase (aided by a complementary template probe). This extended DNA strand was capable of binding multiple Ru (III) species that could be detected by CV. 120 Nayak et al. reported the use of a masked electrochemically active linker that does not require tagging by metals. 118 This has allowed to study the action of a serine esterase (cuti- Fig. 5 The sensitive detection of protein kinase A (PKA) activity by nase) in real-time on 4-hydroxyphenyl valerate immobilised rolling circle amplification (RCA) as signal amplification for cyclic to a SAM on gold. Enzymatic hydrolysis produced a hydro- voltammetry (CV).(Top) Illustration of the RCA method and 120 quinone that is redox active and could therefore be measured (bottom) analysis bei CV. by CV (also previously confirmed by MALDI-ToF MS 65). With this method they were able to determine the kinetics techniques. 139,140 Typically studies are done in the solution of enzymatic reaction and found that Kcat /KM of the reaction was similar to that of cutinase in the solution phase. 118 Star phase where products of enzymatic reactions can be either et al. were also able to monitor the action of an amyloglu- directly infused or separated by liquid chromatography (LC) cosidase on starch immobilised to a single-walled carbon nan- prior to ionisation, typically by electrospray ionisation (ESI) otube (SWNT) surface without the use of a label. This formed and subsequent mass spectrometric analysis. When substrates something similar to a field-effect transistor, so that when the are immobilised to surfaces, however, ESI is not a feasible starch was hydrolysed down to glucose by amyloglucosidase, ionisation technique unless a separate step is added to cleave the SWNT surface is exposed which can be measured by a the modified substrates from the surface and then analyse them 141 change in current passing through the SWNT. 54 in solution Additional chemical cleavage steps might cause additional modification of the substrates or cleavage of labile modifications therefore adding complexity and also work to 2.4 MALDI-ToF MS the analysis. Mass spectrometry (MS) or tandem mass spectrometry MALDI-ToF MS analysis is an ideal solution as MALDI (MS/MS) techniques are already a widespread tool in studying ionisation is a very soft ionisation technique and hence pre- enzymatic reactions and are routinely used in studying post- serves labile bound species. Also data acquisition is rapid translational modifications such as phosphorylation, glycosy- compared to chromatography coupled instruments and typi- lation and acetylation. 138 Mass spectrometry has the advan- cally only singly charged ions are observed thus reducing the tage of being very sensitive (often sub fmol), allows for la- complexity of spectra. 142 MALDI is usually connected to a bel free detection, can be used to study mixtures and allows time-of-flight (ToF) mass analyser as both are pulsed tech- for more detailed structural elucidation using tandem MS/MS niques so complement one another and the MALDI target

1–32 5 | plate can be placed in line with the ToF reducing ion loss. 143 and mass spectrometric analysis can be done on the same sur- A further advantage of using MALDI-ToF MS is that there face. 110 is no theoretical upper limit of mass that can be analysed and it has a large dynamic range. 144 However ions produced 2.5 Surface Plasmon Resonance (SPR) by MALDI ionisation can metastable decay in the ToF de- vice, although this can be beneﬁcial as it allows MS/MS to Surface Plasmon resonance (SPR) is an analytical method that be performed. 145,146 MALDI-ToF MS requires substrate sur- is widely used for quantitative studies of ligand-protein interfaces that are electrically conductive. 12,147 actions. Ligands are generally immobilised on conductingsur- faces and are interrogated by speciﬁc binding of biomolecules (Fig. 7).

Fig. 6 MALDI-ToF MS analysis of enzymatic reactions on SAMs Fig. 7 Representation of enzymatic reactions and lectin binding immobilised on gold. The coupling of a mannopeptide to the SAMs (top) and the corresponding SPR analysis (bottom). followed by three enzymatic elongation steps by POMGnT, N-acetylglucosamine was immobilised on an SPR gold chip β1,4-GalT and the trans-sialidase from Trypanozoma cruzi (TcTS) enzymatically enlogated (galactosylated and sialylated). The and its analysis by MALDI-ToF MS is shown. 10 sialylation was shown by binding of Maakia amurensis agglutinin (MAA) followed by the removal by a sialic acid solution. 53 Enzymatic modiﬁcations have been successfully studied by MALDI-ToF MS on many occasions and are most studied Although SPR is generally a label free technique, changes on gold surfaces with a few papers reporting studies on as in surface structure upon enzymatic transformation are gen- specially conducting indium tin oxide (ITO) glass slides. 148 erally not large enough to be monitored directly by SPR. As MALDI-ToF analysis on self-assembled monolayers has been a result, the analysis of enzymatic transformations need cou- used to study a range of enzymes such as glycosyltransferases, pling to selective product over substrate recognition by a larger proteases and kinases. 27,46,134,149 The enzyme substrates are biomolecule. 53,111 One examplewas doneby the Peters group. immobilised on gold surfaces via an alkylthiol linker gen- They monitored the action of sialyltransferase (ST3Gal-III) erating a stable sulphur-gold bond that is resistant to many on a glycoarray of natural and synthetic acceptor substrates, chemical and biochemical reaction conditions. However, it by binding of the lectin Maakia amurensis agglutinin (MAA) appears to be labile when subjected to the MALDI laser and followed by SPR (Fig. 7). They also used Erythrina crista- is cleaved in situ selectively from the surface during measure- galli lectin (ECL), a galactose speciﬁc lectin, to determine ment. No additional hydrolysis of the analyte from the sur- how much glycan had not been sialylated. They then further face is hence required. 10 The MALDI-ToF analysis of such desialylated the surface with a sialidase. This way, they were SAM surfaces is becoming increasingly popular, since it al- able to reuse the same SPR chip for several sialylation stud- lows for the direct, label-free quantitative analysis of surface ies. 53 reactions. 8,36,150 The underlying gold surface makes this plat- The Katayama group (2008), monitored the action of the form also compatible with SPR studies, such that both SPR cysteine protease caspase on peptide arrays using SPR. Strep-

6 1–32 | tavidin binding to biotinylated immobilised peptides (on ly- 2.6 Atomic Force Microscopy (AFM) sine) ampliﬁed the SPR signal of the peptide array. After caspase action the biotinylated lysine residue was lost and hencea AFM is a microscopic method that can detect molecular changes on surfaces in a label free manner (schematically large reduction in SPR signal was detected. Studying caspases 154 is important as they are involved in intracellluar signal trans- shown in Fig. 9). duction of apoptosis and hyperaction of the caspases related to many diseases such as autoimmune disorders. 112,151

SPR imaging (SPRi) allows for the high-throughput analysis of an entire surface, that is illuminated at a ﬁxed angle of incidence (and camera position) with plane polarised light and the change in reﬂectance intensity is measured using a high resolution CCD camera (Fig. 8).

Fig. 9 Atomic force microscopy (AFM) for the detection of protease activity. The difference in height of the substrate on the surface before and after protease action was monitored. 116

Its applications to monitoring enzymatic reactions have so far been limited. Star et al. were able to monitor the action of amyloglucosidase on starch immobilised to a silicon slide. 54 Also Foose et al. used AFM (in conjunction with ellipsometry) to monitor proteolytic cleavage (using serine protease subtilisin Carlsberg) on thick and stable ovalbumin protein ﬁlms that were formed by cross-linking the amine group of lysine residues with glutaraldehyde. These ﬁlms were immobilised to silicon wafers via surface amine groups also using glutaraldehyde as the link. 116 The recent development of Fig. 8 SPR imaging detection of caspase-3 proteolytic cleavage on high-speed AFM (HS-AFM) that allows real-time monitoring an optimal peptide resulting in release of a protein tag (EGFP). 107 of enzymatic transformation may increase the popularity of AFM. 155

Both enzyme speciﬁcity and kinetic data can be obtained si- 2.7 Quartz Crystal Microbalance (QCM) multaneously using this imaging method. 152,153 Kim and Kim et al. have monitored the action of caspase-3 on a synthetic Another surface analysis method that is gaining popularity peptides designed to be a substrate for caspase-3. This pep- is the use of Quartz Crystal Microbalances (QCM). Like tide was also synthesised to contain an N-terminal glutathione MALDI-ToF MS, SPR and AFM, QCM is a label free tech- (GST) tag, which was used to immobilise the peptide to a gold nique that is capable of monitoring the real time associa- chip and it also possessed an enhanced green ﬂuorescent pro- tion and dissociation of enzymes to an immobilised substrate. tein (EGFP) label on the C-terminus. Loss of the EGFP label From QCM data enzyme kinetics can be calculated. 33,97,156 could be observed by SPRi. 107 For more information on SPRi In a typical QCM experiment, when only substrate is immo- see the excellent review by Scarano et al. 153 bilised to the electrode and an electric current is applied, a

1–32 7 | shear wave is created along the quartz crystal at a certain fre- 3 Surfaces quency (usually tuned to the fundamental frequency). Then as the enzyme associates with the substrate the mass increases, The choice of surface is often determined by the mode cho- which results in a loss in frequency of the shear wave. When sen to study enzymatic reactions. In this Section we discuss a the enzyme dissociates, the mass massively decreases and the number of surface platforms used in this area. Such surfaces frequency increases again (Fig. 10). 157 Also, if enzymatic ac- are also chosen on the basing of their resistance to protein de- tion results in the formation/destruction of large compounds naturation and accessibility of ligands to proteins in aqueous immobilised to the QCM surface then the change in mass of solutions. The latter is particularly important for porous syn- substrate on the surface can be observed. thetic polymers which need to swell in water and provide accessibility of substrate to the enzyme. 14,19 There is evidence that larger enzymes are not able to penetrate porous polymer beads and hence reactions are limited to bead surfaces and proceed in low overall yields. 160

3.1 Glass

Glass slides are the most commonly used surface to study enzymatic reactions, often in a high-throughput manner. Slides can be cheaply bought and are available prefunctionalised with various groups such as amines, epoxides and aldehydes for covalent immobilisation. 73 However issues can arise due to impurities or the enzyme being studied non-specifically binding to the functionalised glass slides. This can be pre- vented by using a more specific immobilisation strategy such as Diels-Alder reactions and blocking the free linkers with a protein such as BSA or a small molecule. 127 Alternatively molecules can be immobilised to plain glass slides via a silane group or alternative functionality introduced to the glass sur- 7 Fig. 10 The illustration (top) and analysis (bottom) of glycogen face. Another advantage of glass slides is their thermal sta- phosphorylsis on a QCM. (A) Addition of phosphorylase to bility making them perfect for DNA array applications. They immobilised amylopectin (B) Addition of phosphoric acid (Pi) and are also stable to organic and aqueous and acidic and basic (C) Addition of glucose 1-phosphate (G1P). 96 solutions, unlike some polymer materials. 161 Substrates immobilised to glass slides are most commonly analysed by fluorescence spectroscopy or detection of radiolabelling. 122 The Okahata group has extensively used QCM to study The development of microarray printers that are now ca- 33 enzymatic transformations such as glucan hydrolysis and pable of spotting sub nL of substrate to a surface, 162 allows 33,96 glycogen phosphorylsis. In an interesting paper in 2004 high-throughput simultaneous analysis of enzymes on a sin- they extensively studied the kinetics of phosphorylase b (a gle slide in a single experiment. An impressive example is the glycogen phosphorylase) on immobilised amylopectin by de- manufacture of a protein array containing 80% of the yeast activating (T-state phosphorylase b) and activating the en- proteome 163 which Lu et al. used to identify substrates for the zyme (R-state) by addition of AMP, so that it would form ubiquitinating enzyme Rsp5 164 and for the acetyltransferase the substrate-enzyme intermediate and then hydrolyses and NuA4. 17 33 phosphorylates amylopectin, respectively. From these stud- The lack of direct analytical physicochemical techniques of ies they were able to determine the kinetic values kon, ko f f functionalised glass slides is still an issue and recent reports (enzyme association/dissociation), Kd, kcat and KM. A further have described new analytical techniques suitable for glass study on glycogen phosphorylases was performed where the surfaces, such as MALDI-ToF MS on glass slides backed with kinetics of the polymerisation or the phosphorolysis of amy- conductive tape 12. Resonance light scattering (RLS) was used 96 lopectin could be studied (Fig. 10). to monitor phosphorylation of peptides 165 by measuring how QCM is limited however as it cannot be used in high- light is scattered as it pass through a solvent containing aggre- throughput studies. Furthermore, an experienced user is of- gates of particles. 166 The RLS strategy however requires mul- ten required to perform experiments and finally the calibration tiple steps to incorporate a AuNP onto the phosophorylated procedures tend to be long and time consuming. 158,159 peptide which then had to be coated in silver to improve the

8 1–32 | signal. 165 Other technical issues with glass slides are interfer- yellow. 169 Enzymes can then be screened against these sub- ence of fluorescence or radiation from adjacent spots 127 Also strates. 170 Smith and co-workers used this approach to syn- for protein arrays the surface charge on the glass may cause thesise a library of peptides with the consensus sequence for the protein to denature, which may result in it no longer being PKA (R-X-X-S/T) (Fig. 3). Phosphorylation was confirmed a substrate for a particular enzyme thus leading to a false neg- by Western blotting with anti-phospho-PKA antibodies. Then ative. 127 Despite these limitations glass is widely used. Some they arrayed 20-mer peptide sequences (3 amino acid overlap) of the surface coatings used on glass shall now be discussed. from the protein kinases PKD and MARK3 along with the adapter protein RIL, to determine how phosphorylation events 3.1.1 Nitro/cellulose coated. A glass slide coated with regulate further kinase activity and resultant signal response. thin layers (for FASTTMslides 15 µm) of cellulose, or now ∼ They observed phosphorylation of RIL at Ser119, confirmed more frequently used nitrocellulose, is a widely used sur- by mutation to alanine, and when this mutant was produced 17 face to immobilise substrates to especially proteins and gly- in prostate cancer cells, a massive increase in growth rate was 32 167 cans but also DNA. This is because they appear to main- observed suggesting a mutation at this position is linked to tain the protein conformation and oligosaccharides and pro- cancer. Phosphorylation of PKD and MARK3 was found to teins can be readily non-covalently adsorbed to the surface, be purely structural. 26 thus not requiring chemical derivatisation to fix them to the 32 Other advantages of using nitrocellulose arrays is that they surface. However, this creates a problem as the orientation can be stored (dried) for as long as the substrate is stable 127 they are immobilised in cannot be controlled and the surface and lower amounts can be used. 171 Another possible advan- must be blocked to prevent non-specific adsorption of the en- 32 tage is that they have a lower background fluorescence than zyme or other chemical moieties. Substrates are also typi- hydrogel coated slides 5 although this is probably be down cally immobilised in the presence of glycerol, which reduces to a higher sample loading as it is known that nitrocellulose the rate of evaporation of spots, critical when dealing with nL produces quite a high background fluorescence depending on spots, and are also spotted in a high humidity (70-80%) en- the wavelength of light used (particularly bad when common vironment. Glycerol also stabilises protein structures by in- Cy3 dyes used). 127,168 This background fluorescence can be creasing viscosity thus reducing the number of collisions with reducedby makingthe chips with a thinnercoating of nitrocel- the solvent. However, increasing the viscosity also reduced lulose. 168 A major disadvantage compared to uncoated glass the rate the protein diffuses thus increases the time required 127 slides is that overnight immobilisation of substrates to the ni- for immobilisation. The interaction between glycans, pro- trocellulose is required unlike activated glass surface which teins and nucleic acids and the nitrocellulose surface is still require 30 mins. 127 Also larger sample volumes are required not fully understood although it is believed to be due to hy- compared∼ to gold and plain glass surfaces. 104 drophobic and electrostatic interactions. 168 By far the most important application of nitrocellulose 3.1.2 Hydrogel Coated. Hydrogels represented in Fig. coated glass slide was for the SPOT technology (Fig. 11). 169 12, sometimes called biochips, are significantly advantageous over other supports as they have pseudo-aqueoussolution conditions thus providing native reactions conditions for proteins. 81

Fig. 11 The direct on-chip peptide formation via SPOT synthesis.

SPOT technology is essentially the high-throughputparallel Fig. 12 Hydrogel coated glass. The polymer shows a porous surface to solid-phase synthesis of peptides using a microarray printer for substrate presentation. that prints solutions of activated/protected amino acids. These spots act as mini-reaction vessels that build the peptide on the Substrates are also immobilised in a 3D environment un- support in a high-throughput manner. Bromophenol blue in- like the slides where substrates are immobilised in two di- dicator that detects uncoupled amino acids can be used to ob- mensions, which is more biologically similar. Hydrogels are serve visually the coupling as the colour changes from blue to cross-linked polymers that act to immobilise a large amount

1–32 9 | of solvent molecules in the structure thus giving it gel prop- assembled monolayers (SAMs) of alkanethiols that are fairly erties. To emphasise this, Kiyonaka et al. were able to pre- stable to many chemical conditions and are biocompatible pare hydrogels from saccharide-appended amino acetates that (Fig. 13). were capable of immobilising 38000 molecules of water by one molecule of the saccharide-appended amino acetate in the supramolecular hydrogel structure. 81 Hydrogels also are very porous and as a result possess a high surface area to volume ratio compared to most other supports, allowing immobilisation of more substrate. They are also typically chemically robust. Substrates are immobilised to hydrogels covalently, normally by photoinduced copolymerisation of a non-saturated tagged substrate with the gelator. 172 Although other immobilising groups such as N- hydroxysuccinamide (NHS), 165 alkynes 82 or epoxides 49 can be incorporated into the polymer and used to immobilise the substrate. Hydrogels are commonly made from bis/acrylamides polymers and are ther- mally, chemically and mechanically stable. They are also stable in organic, aqueous, acidic, basic and oxidising solutions and can even be sonicated. 56 They also exhibit a low amount of autofluorescence, favourable for fluorescence based studies. Another commonly employed polymer for preparation of Fig. 13 Self-assembling monolayers (SAMs) on gold. The gray part represents the alkyl chains, which are needed for the assembly, the hydrogels is agarose. 82,98 Interestingly, Parker et al. immo- polyethylene glycol units (red) work against unspecific protein bilised peptide substrates to an acrylamide hydrogel coated adsorption and the carboxylic group (blue) is used for chemical glass slide by derivatising cysteine residues with acrylamide modification. and then copolymerising it with the gelator during formation of the supramolecular hydrogel. A UV labile linker was also These SAMs possess similar characteristics to cell surfaces, inserted between cysteine and the acrylamide. Using this, they thus making the environmentthe enzyme acts on more biolog- investigated the specificity of two kinases. Surprising, they ically similar. 173 This was illustrated by the Mrksich group were able to monitor phosphorylation reproducibly and accu- (2009), who immobilised a peptide substrate to the kinase Abl rately by MALDI-ToF MS despite being on a non-conductive on a SAM (monitored by MALDI-ToF MS). Phosphorylation support. The UV pulse used to initiate sublimation of the ma- of this peptideresulted in the formationof a ligand for the SH2 trix (in this case CHCA) was capable of cleaving the UV la- adapter domain on Abl, which when bound together results in bile bond (although optimal scission was observed using a UV Abl phosphorylating neighbouring peptides that it would not lamp before MALDI-ToF analysis) allowing ionisation of the normally. Kinetics of this second phosphorylation is more re- 56 screened peptides and not the hydrogel support. alistic (faster) than it would be in solution as the phosphorylation occurs intramolecularly. 30 SAMs are typically composed 3.2 Gold surfaces for enzymatic studies of molecules possessing a thiol binding group, a hydrophobic alkyl chain that aids ordering of the SAM and then a termi- Gold is also a popular surface and almost universally sub- nal polyethylene glycol (PEG) group to prevent non-specific strates are immobilised to gold surfaces using thiols as the adhesion of proteins (i.e. enzymes) to the alkyl chains, which Au-S bond is relatively stable with strength of about 50 Kcal may also have, for example, a carboxylic acid 8 or alkene 174 1 mol− . Gold can be thinly coated to slides (vapour depo- to covalently link substrates to. Some of the molecules though sition, sputter coaters) or formed into nanoparticles 150 Gold do not contain functionality to allow substrates to bind to it surfaces can be used with a diverse range of analytical tech- and act as dilutors so substrates are not too densely packed, niques such as MALDI-ToF, SPR, AFM, Fluorescence, QCM, which would inhibit enzymatic action or ligand binding. 134 electrochemical techniques and radiation. 3,10,28,33,55,104 Gold Carboxylic acid termini can be easily activated (by NHS surface allows for a combinations of analytical techniques can or pentafluorophenol (PFP)) and amines coupled to it directly be used, for example MALDI-ToF could be used to directly or by native chemical ligation. 47 Maleimides are used to co- probe what is on the surface of a slide, 9 whilst QCM could be valently link thiols to the SAM and alkenes can undergo used to determine the kinetics of enzymatic activities 96 and Diels-Alder cycloadditions very specifically to cis-dienes. 174 AFM to map the topology of the surface. 55 In addition alkynes can undergo a Huisgen cycloaddition with Gold surfaces can also be coated simply with self- azides, which is useful when handling biological samples as

10 1–32 | there are very few alkynes and azides in nature. 129 A compre- clamped down tightly and care must be taken when removing hensive range of covalent immobilisation techniques to SAMs the ’stamp’ to prevent cross contamination. Also the channels was reviewed by Flitsch and co-workers. 46 It is also possi- must be blocked with protein such as BSA to prevent the en- ble to non-covalently immobilise substrates to SAMs normally zyme binding to it. 61 achieved through hydrophobic interactions. For example, Re- Substrates can be directly immobilised to gold surfaces for ichardt and co-workers reported the formation of a SAM of example via terminal cysteine residues in peptides 3 or sub- alkanethiols (no PEG group) on gold and tethered lipid tagged strates can be immobilised to thiol-maleimide/epoxide/alkyne oligosaccharides to it. Using this array, they were able to coated slides. 109 However, care must be taken as Inamori et probe by MALDI-ToF MS not only the action of a range of al. noticed that immobilisation of substrates without a linker purified glycosyl hydrolases on immobilised oligosaccharides, reduced the enzymatic activity of the kinase cSrc on the im- but also endogenous hydrolases from soil, compost and saliva mobilised peptide (efficiency of 20% after 2h without linker samples (Fig. 14). 63 and 60% after 1h with linker compared to the enzymatic reaction in solution). They rationed that the PEG spacer they included moved the substrate out of the proximity of the surface and also increased its mobility thus allowing the kinase to access the substrate. The density of this layer may also be important, however as they noticed a reduction in activity when the density of substrates was reduced, which was solved by cross linking them. In comparison though, the efficiency and rate of phosphorylation for the kinase PKA was not altered upon addition of a linker thus emphasising the requirement to optimise the immobilisation strategy. 104 The application of using nanoparticles to analyse enzymatic reactions is interesting due to the unique intrinsic properties of nanoparticles. They are also advantageous as compared to slides as they possess greater surface areas 175 and intermediate between macroscopic and microscopic (want to infer they are more biologically mimicking). Citrate-reduced gold Fig. 14 Analysis of glycosidases by MALDI-ToF MS on an array colloids (AuNP aggregates), routinely synthesised by reduc- 63 presenting various glycans. tion of chloroauric acid with trisodium citrate (the Turkevich method), 176,177 are widely used because of their stability as In both types of immobilisation care must be taken when citrate not only reduces but caps the colloid and the size and immobilising substrates to surfaces so they are orientated in shape of the particles can be controlled by altering the reac- such a way that the enzyme can still access the site it is mod- tion amounts and conditions). Typically AuNPs are used in ifying. 104 A major issue of SAMs arrays, is that they have a surface-enhance Raman spectroscopy (SERS), which is capa- low thermal stability as heating results in cleavage of the Au- ble of detecting single molecules. 178 Ruan et al. used this S bond. This makes them unsuitable for DNAarray applica- to study the enzymatic action of alkaline phosphatase (ALP), tions, which typically involve heating steps like in polymerase found in human serum and related to bone metastasis, by oxi- chain reaction (PCR). 127 For more information about SAMs, dising the productof ALP forminga dye and analysingit using the reader is recommended to read the Whitesides review. 150 SERS with citrate-reduced gold colloids. 15,179 This is briefly Classically enzyme solutions are directly spotted to the im- mentioned for completion of the use of gold to study enzy- mobilised substrates and placed in a closed and/or humid en- matic reactions although does not fit the scope of this review vironment to prevent the solution evaporating (as most enzy- as the enzymatic reaction is done in solution and then col- matic reactions occur in water), which would cause the en- loids added. Enzyme reactions have, however, been studied zymatic reaction to terminate. This spotting requires only on substrates immobilised to AuNPs. This is typically de- small quantities of precious enzymes. Alternatively, microflu- tected electrochemically 52 or by fluorescence taking advan- idic devices can be constructed that allow enzymes to be tage of the AuNPs ability to act as a chromophore. 3,31) Also flowed over immobilised substrates controlled by a polymer conventional MALDI-ToF analysis can be performed with the ’stamp’ with microchannels punched into it using a small nee- added benefit that no organic matrix is required as the AuNPs dle. Smaller amounts of enzyme can still be employed com- appear to aid the desorption/ionisation process. 175 pared to typical solution phase studies and unlike the spot- One of the main disadvantages of using gold surfaces is that ting method the solution cannot evaporate. The stamp must be they are typically more expensive than glass or polymer sur-

1–32 11 | faces. In addition glass slides can be purchased with the sur- (polyoxyethylene-polystyrene) 180 and PEGA ((polyethylene face prefunctionalised, whereas gold surfaces must be func- glycol)-acrylamide) resins. 11 PEGA developed by Meldal et tionalised before experimentation. al. is made of cross-linked polyacrylamide core and tethered to 2-aminopropyl groups by PEG groups. 184 PEGA is often 3.3 Synthetic Polymers as solid supports favoured due to its large swelling volume that allows enzymes to penetrateinto the poresof the resin 11 (Enzymes of 43 kDa 180 ∼ The ﬁnal most widely used support are synthetic polymers for can penetrate PEGA1900). Inducing a charge on PEGA can the reasons that they are usually very porous therefore have also improve the accessibility of certain enzymes to the im- larger surface areas, are cheap to make and usually very stable mobilised substrate. 181 However, typically non-swelling con- or can be tailored to allow for stability undercertain conditions trolled pore glass (CPG) is now used instead of these poly- (Fig. 15). mers as it can be made to have fairly homogenous pore sizes and can come in a wider range of sizes capable of accommo- dating larger enzymes. 180 For further information on polymer supports see the review from Halling and our group. 19

3.4 Other

There have also been more inventive surfaces to which substrates have been immobilised to. For example Wu, Wong and co-workers formed thin films of aluminium oxide on a glass Fig. 15 Substrate carrying polymers. Enzyme action on substrates support, to which phosphate terminated prefluorinated hydro- is restricted by pore sizes of the differen polymers. Analysis of the carbons were spotted. This formed a monolayer on the sur- product formation can be carried out by radiation, fluorescence, face due to the high affinity phosphates have for aluminium confocal raman microscopy and MALDI-ToF MS. oxide. Glycans tagged with perfluoroalkyl chains were then immobilised to these monolayers exploiting fluorous interac- Polymers can also be easily cast into different shapes most tions. This system was highly advantageous as neither hy- commonly microtiter plates, frequently used for enzyme- drophobic nor hydrophilic compounds would adhere to the linked lectin assays (ELLAs) (or ELISA assays) 91 and beads, surface due to fluorine’s ’non-stick’ property. These immo- often used for solid-phase synthesis. 42 Like for most glass bilised glycans were degraded by cellulases, which was mon- coatings, polymer surfaces need to be blocked with proteins itored by MALDI-ToF MS without the addition of a matrix such as BSA or milk proteins to prevent non-specific hy- and fluorescently tagged lectins. The major limitation is the drophobic adsorption of enzymes and other impurities to the requirement to synthetically tag the substrates with perfluo- surface, thus denaturing enzymes. 127 Similar to gold, poly- ralkyl chains, which is not feasible for screening large sub- mers can be analysed by multiple analytical techniques such strate libraries. 45 As mentioned in Section 2.1, carbon has as fluorescence, AFM and radiation 105 for slides or microtiter also been used as a support. Star et al. were able to mon- plates. Confocal Raman microscopy 180 and UV can also itor the action of amyloglucosidases on starch immobilised be used when analysing substrates immobilised to beads. 181 to SWNT by measuring electronic changes. 54 Also Song et In addition modification of the slides can allow them to be al. monitored phosphorylation of the two well studied pep- analysed directly by MALDI-ToF MS 12 or the products can tides RaytideTMEL and Kemptide, which were immobilised to be cleaved from the polymer and be analysed by LC-MS or screen-printed carbon electrodes through amide chemistry, by NMR. 42 Care must be taken, however, when studying fluo- using a AuNP tagged ATP cosubstrate again by electrochem- rescence on resin beads as the laser can only penetrate a few ical measurements. 52 These were both innovative techniques, microns into the resin beads thus cannot access all dye tagged however were not high-throughput, and hence could not be molecules, although confocal Raman microscopy is less af- used to screen multiple enzyme-substrate relationships. Fi- fected by this. 180,182 Glycans, proteins and peptides have all nally, silicon has also been employedas a supportto study pro- been studied as substrates for enzymatic reaction on these sur- tease action of subtilisin Carlsberg on a glutaraldehyde cross- faces. coupled protein film that was immobilised covalently via sur- Some of the most commonly used polymer supports face amine groups. This was monitored by ellipsometry and to study enzymatic reactions are polystyrene, 183 TentaGel AFM. 116

12 1–32 | 4 Enzymatic Transformations on solid-support tyrosine) to epoxides. 187–189 Peptides containing N-terminal cysteines can be coupled to pentaﬂuorophenyl esters activated Not all enzymes might be suitable for reactions on solid sup- surfaces by native chemical ligation. 47,190,191 In addition to port, in particular where active sites are deeply buried in the taking advantage of natural functional groups in the peptide enzyme structure. Nevertheless, a wide range of reactions substrates, unnatural orthogonal groups have also been used. have been successful and these reactions will be discussed in Examples include dienophilesor azides, which allows for cou- more detail in the following subdivided by substrate classes pling to dienes (via Diels-Alder reaction) or alkyne (’Click- and family of enzymes. An overview of all enzymatic trans- Chemistry’). 192,193 For peptides the non-naturalgroups can be formations and corresponding references is provided in Tables incorporated during solid-phase synthesis. For proteins more 2,3 and 4. controlled chemistries are required or using unnatural amino acids during translation of the protein. 190 Also the storage 4.1 Peptides and Proteins conditions for the different protein arrays is crucial for stable proteins on the surface. 194 Protein and peptides arrays are now the most widely used type Several types of enzymes have been employed on proof array as they can be used for most proteomic studies (Fig. tein/peptide arrays and are listed in table 2 . These will be 16), clinical diagnostics and SPOT synthesis. 185 discussed in more detail in the following.

4.1.1 Transferases. Transglutaminase studies on microarrays have ﬁrst been performed by Kwon et al. 70 They developed an on-chip system where they immobilised N,N’- dimethylcasein and studied the enzyme activity by incorporation of a biotinylated amine substrate. With this system analysed eleven mammalian cell lines towards its transglutaminase activity and were able to show higher sensitivity than by incorporation of radioactively tagged substrates. They applied this system to distinguish between different isozymes of transglutaminase. 50 They were to analyse the different isozymes FXIII and TG2 on ﬁbrinogen by varying the pH, substrate and Ca2+ concentration (Fig. 17). This method was also successfully applied to mixtures of the enzyme in whole differentiated monocytic leukaemia THP-1 cells.

Fig. 16 Peptide and Protein arrays. Immobilisation can be carried out via covalent and non-covalent coupling methodologies.

The presentation of substrate for the enzyme is an important issue and is determined by the method used for attach- Fig. 17 Fluorescence detection of two different types of ment. Both peptides and proteins have been immobilised non- transglutaminases (FXIII and TG2) by transferring biotin labelled covalently to hydrophobic surfaces such as polystyrene, but amines. 50 can cause problems with denaturation of proteins and weak binding. 186 Alternatively proteins or peptides can be immo- Methyltransferases have been suggested to be involved in bilised covalently making use of active functional groups such important cellular processes such as transcription regulation, as amines, carboxyl, hydroxyl and thiol groups (Fig. 16). signal transduction and control of protein-protein interac- Amines are the most commonly used functional group, which tion. 105 The transcription regulation is, amongst others, per- allow coupling to aldehydes or to EDC/NHS activated esters formed by histone methylation which is catalysed by methyl- or . Thiols (in cysteine) react very selectively with maleimides transferases. 101 They catalyse the transfer of a methyl group on surfaces, carboxylic acid (aspartic and glutamic acid) can from S-adenosyl methionine to either an arginine or lysine be coupled to amines and hydroxyl groups (serine, threonine, residue. Methyltransferases have been studied by incorpo-

1–32 13 | Table 2 Table for peptide and protein arrays

Enzyme Reaction Lit. Proteases α-Chymotrypsin Hydrolyses C-terminal to aromatic amino acid 11,16,81 residues Activated complement C1r Hydrolyses after R/K in the consensus sequence 195 -R-K-R- - and R/K-A/S/V/I-K- - | | Activated complement C1s Hydrolyses after K/R with S/A/G in the -2 po- 195 sition Activated complement D Hydrolyses after R/K in the consensus sequence 195 -R-K-R- - and R/K-A/S/T/V-K- - | | Activated protein C (Human) Hydrolyses after K/R in the preferable sequence 195 -Q-K/T/Q-K/R- - | Caspase-3 Hydrolyses after D in the consensus sequence 107,112 -D-E-V-D- - | Caspase-6 Hydrolyses after D in the consensus sequence 112 -V-E-I-D- - | Caspase-8 Hydrolyses after D in the consensus sequence 112 -I-E-T-D- - | Cathepsin B Hydrolyses after R/K in the consensus sequence 195 -X-T/V-R/K- - where X is any basic residue | Cathepsin G Hydrolyses after R/K in the consensus sequence 195 -I/E/V-T/V-K/R- - | Cathepsin H Hydrolyses after R/K with L in the -2 position 195 Cathepsin K Hydrolyses after R/K with L in the -2 position 195 Cathepsin L Hydrolyses after R/K in the consensus sequence 195 -K/R-L/F/V-K/R- - | Cathepsin S Hydrolyses after R/K with L in the -2 position 195 Cathepsin V Hydrolyses after R/K with L/V in the -2 posi- 195 tion Chymopapain Hydrolyses after K/R in the preferable sequence 67,195 -A-F/P/V/A/R-V/T-K/R- - | Dioctyl sodium sulfosuccinate Transfer of N-protected (Boc, Ac and Cbz) F, 196 paired chymotrypsin Y, W, A or L to L residues conducted in 70:30 isooctane:THF Factor aXIIa Hydrolyses after R in the preferable sequence 195 -M-F/T-R- - | Factor IXab Hydrolyses after R in the consensus sequence - 195 L/M/Q/R-X/Z-R- - where X is any basic residue and Z is any aromatic| residue Factor VIIa +TF Hydrolyses after R in the consensus sequence 195 -Q/R/N/P-V/T-R- - | Factor XIa Hydrolyses after K/R in the preferable sequence 195 -F-N-K/R- - | Ficin Hydrolyses after R/K in the consensus sequence 195 -P-L/V/T-K/R- - | Lysyl-endopeptidase (LEP) Hydrolyses C-terminal to K 81 Matrix metalloproteinase-1 Hydrolyses collagenous proteins preferentially 197 (MMP-1) between -G- -L- | Matrix metalloproteinase-2 Hydrolyses collagenous proteins N-terminal 197 (MMP-2) to hydrophobic residues in the preferable sequence -X- -Z-Z-Z- where X is any amino acid and Z is any| hydrophobic amino acid

14 1–32 | Table 2 continued

Enzyme Reaction Lit. Proteases Matrix metalloproteinase-3 Hydrolyses collagenous proteins N-terminal 198 (MMP-3) to hydrophobic residues in the preferable sequence -X- -Z-Z-Z- where X is any amino acid and Z is any| hydrophobic amino acid Matrix metalloproteinase-9 Hydrolyses collagenous proteins N-terminal 197 (MMP-9) to hydrophobic residues in the preferable sequence -X- -Z-Z-Z- where X is any amino acid and Z is any| hydrophobic amino acid Papain Hydrolyses after R/K in the consensus sequence 195 -P-V-K/R- - | Plasma Kallikrein (Human) Hydrolyses after K/R in the preferable sequence 195 -K/R-K/F/Y-K/R- - | Protease factor Xa Hydrolyses after R preferably in the consensus 108 sequence -I-D/E-G-R- - | Rhodesain Hydrolyses after R/K with F/L/Y/V in the -2 po- 195 sition Stem Bromelain Hydrolyses after R/K in the consensus sequence 195 -P-R-K/R- - | Subtilisin Hydrolyses after A/L/V in the sequence -F/D- 67 A/F/V-X-A/L/V- - where X is any amino acid | Subtilisin Carlesberg Hydrolyses after R/K in the preferable sequence 116,195 -R/Q/S/T-I/A/V-R/K- - | Thermolysin Hydrolyses N-terminal to hydrophobic and/or 36,42 aromatic residues unless P is +2 Thrombin Hydrolyses after R/K enhanced if P is -1 34 Trypsin Hydrolyses C-terminal to R/K unless followed 7,195 by P Tryptase Hydrolyses after R/K in the consensus sequence 195 -R/K-S/T/N-R/K- - | V8 protease (glutamyl en- Hydrolyses C-terminal to E 81 dopeptidase) Kinases 3-Phosphoinositide-dependent Phosphorylates S/T in the consensus sequence 25,100 kinase 1 (PDK1) -S/T-A-P-E- or -T-P-E-Y- Abl Phosphorylates Y in the consensus sequence 30,56,58,61,89,111,199 -X-Y-Z-Z-P- where X is any aliphatic amino acid and Z is any amino acid Calmodulin-dependent protein Phosphorylates S/T in the consensus sequence 25,58,111,200 kinase II (CaMK II) -R-X-X-S/T- where X is any amino acid Casein Kinase 2 (CK2) Phosphorylates S/T when there are acidic 25,100 amino acids in the 1 to 4 positions c-Src Phosphorylates Y residues preferentially when 52,58,59,61,72,80,87,104,111,201 L/I -1 position Cycli-dependent kinase 5 Phosphorylates S/T when P is +1 and basic 25 (CDK5) amino acids are +2 to +4 Dual speciﬁcity mitrogen- Phosphorylates S/T (exact sequence require- 25 activated protein kinase kinase ments unknown) 6 (MKK6)

1–32 15 | Table 2 continued

Enzyme Reaction Lit. Kinases Epidermal growth factor recep- Phosphorylates Y in the consensus sequence 111 tor (EGFR) -Z-P-Z-Y-F/L-X-F/V- where Z is an acidic residue and X is any amino acid Insulin receptor (InsR) kinase Phosphorylates Y when acidic residues are -1, 111 -2 or -3 and M is +1, +2, +3 or +4 Janus Kinase 1 (JAK1) Phosphorylates Y preferably in the motif -Y-X- 111 X-L- where X is any amino acid MAPK-activated protein kinase Phosphorylates S in the consensus sequence -X- 111 2 (MAPKAPK-2) X-Z-X-R-X-X-S/T-X-X- where X is any amino acid and Z is a bulky hydrophobic residue Mitrogen-activated protein ki- Phosphorylates S/T in the consensus motif -P- 25,111 nase 1 (MAPK1 or Erk2) X-S/T-P- where X is any amino acid Mitrogen-activated protein ki- Phosphorylates S/T in the consensus sequence 58,61 nase 3 (MAPK3 or Erk1) -P-X-S/T-P- where X is any amino acid Mitrogen-activated protein ki- Phosphorylates T when followed by P 25,111 nase 8 (MAPK8 or JNK1) Never in mitosis A-related ki- Phosphorylates S/T when L/F are -3 position 202 nase 6 (NEK6) p38α Phosphorylates S/T when P is -2 and +1 25,111 p38β2 Phosphorylates S/T (exact sequence require- 25 ments unknown) PknB Phosphorylates S/T residues preferentially 51 when P +1 position Protein Kinase A (PKA) Phosphorylate S/T in the consensus sequence - 25,26,52,58,61,76,86,100,104,109,111,165,199,200 R/K-R/K-X-S/T- where X is any amino acid Protein Kinase B (PKB or Phosphorylates S/T in the consensus sequence 86,111 Akt1) -R-X-R-X-X-S/T- where X is any amino acid Protein Kinase C (PKC) Phosphorylates S/T when basic residues are in 61,76,119,200 positions -3 to -1 and/or +1 to +3 Protein Kinase C α (PKCα) Phosphorylates S/T when basic residues are in 111 positions -2, -3, -4, +2 and +3 Protein Kinase C δ (PKCδ) Phosphorylates S/T when basic residues -1 to -4 111 and F/I/M +1 position Protein Kinase C ζ (PKCζ) Phosphorylates S/T when basic residues -2 to -4 111 and +3 to +5 and F/I/M +1 and +2 position Protein Kinase G (PKG) Phosphorylates S/T when basic residues are in 58,61 the -2 and -3 position Rho-associated kinase α Phosphorylates S/T in the consensus sequence - 25 (ROKα) R-X-X-S/T- or -R-X-S/T- where X is any amino acid β-Isoform of glycogen syn- Phosphorylates S/T when S/T or pS/pT are in 25,100 thase kinase 3 (GSK3β) +4 position Phosphatases Alkaline phosphatase Dephosphorylates peptides and proteins 61 Protein phosphatase 2A, 2C Dephosphorylates phosphorylated threonine 61 and serine residues Protein Tyrosine Phosphatase Dephosphorylates Y when acidic residues are 79,94 1B (PTP1B) in the -1, -4 and -5 position and a hydrophobic residue is in the +3 position

16 1–32 | Table 2 continued

Enzyme Reaction Lit. Phosphatases Protein Tyrosine Phosphatase µ Dephosphorylates Y when hydrophobic residues are in 94 (PTPµ) -2,-1 and +1 position and polar residues in +2 position. Also aromatic residues can occupy position +1. SHP1 Dephosphorylates Y in the consensus sequence - 79 S/I/V/L-X-Y-X-X-I/V/L- where X is any amino acid SHP2 Dephosphorylates Y in the consensus sequence - 79 S/I/V/L-X-Y-X-X-I/V/L- where X is any amino acid TCPTP Dephosphorylates Y when acidic residues are in the - 79 3,-4 and -5 position and hydrophobic residues at the -2 and +3 positions Tyrosine phosphatase Dephosphorylates phosphorylated tyrosine residues 61 ∆SHP1 Dephosphorylates Y when -2 and +3 position are I/V/L. 79 Also favours acidic residues -5, -1, +2 and +4 position ∆SHP2 Dephosphorylates Y when -2 and +3 position are I/V/L. 79 Also favours acidic residues -5, -4, -1, +1 and +5 position and hydrophobic residues at +2 position De/Acetylase Histone (or lysine) Deacetylase Deacetylates K ideally in the sequence -R-K-R/Y- 35,57 2 (H(K)DAC2) Histone (or lysine) Deacetylase Deacetylates K when aromatic amino acid residues are 35,57 3 (H(K)DAC3) 1 H(K)DAC3 + SMRT Deacetlyates K in the consensus sequence -R/Y-K-R/Y- 35,57 Histone (or lysine) Deacetylase Deacetylates K ideally in the sequence -R-K-F- 35,57 8 (H(K)DAC8) Sirtuin 1 (SIRT1) Deacetylates K in the consensus sequence -R-K-R- 35,57 NuA4 17 Lipases Candida antartica lipase Hydrolyses ester (3-(5-hydroxyhexyloxy)benzaldehyde 74 (CAL), acetate group and C2/8 alkyl length) generating C2/C8 alcohol group and carboxylic acid fragment Candida rugosa lipase (CRL) Hydrolyses ester (3-(5-hydroxyhexyloxy)benzaldehyde 74 acetate group and C10 alkyl length) generating C10 alcohol group and carboxylic acid fragment Chromobacterium viscosum Hydrolyses ester (3-(5-hydroxyhexyloxy)benzaldehyde 43,74 lipoprotein lipase (CVL) acetate group and C6/10 alkyl length) generating C6/C10 alcohol group and carboxylic acid fragment Hog pancreatic lipase (HPL) Hydrolyses ester (3-(5-hydroxyhexyloxy)benzaldehyde 74 acetate group and C4 alkyl length) generating C4 alcohol group and carboxylic acid fragment Mucor javanicus lipase (MJL) Hydrolyses ester (3-(5-hydroxyhexyloxy)benzaldehyde 74 acetate group and C8 alkyl length) generating C8 alcohol group and carboxylic acid fragment Pseudomonas ﬂuorescens li- Hydrolyses ester (3-(5-hydroxyhexyloxy)benzaldehyde 74 pase (PSBL) acetate group and C2 alkyl length) generating C2 alcohol group and carboxylic acid fragment Pseudomonas sp. B lipoprotein Hydrolyses ester (3-(5-hydroxyhexyloxy)benzaldehyde 74 lipase (PSBL) acetate group and C2 alkyl length) generating C2 alcohol group and carboxylic acid fragment Pseudomonas sp. lipoprot li- Hydrolyses ester (3-(5-hydroxyhexyloxy)benzaldehyde 74 pase (PSL1) acetate group and C10 alkyl length) generating C10 alcohol group and carboxylic acid fragment

1–32 17 | Table 2 continued

Enzyme Reaction Lit. Rhizopus arrhizus lipase (RAL) Hydrolyses ester (3-(5-hydroxyhexyloxy)benzaldehyde 74 acetate group and C10 alkyl length) generating C10 alcohol group and carboxylic acid fragment Methyl- Coactivator-associated arginine Methylates R residues in specific proteins such as 105 transferase methyltransferases 1 (CARM1) TARPP and Poly(A)-binding protein 1 Lysine methyltransferase G9a Methylates K when R is -1 position 101 Lysine methyltransferase Methylates K residues in specific proteins such as 103 SETD6 RAB28, HIST1H2B and H3F3B Lysine methyltransferase Methylates K residues in specific proteins such as 103 SETD7 CCNB3, ACOX1 and CSNK2A2 Protein arginine methyltrans- Methylates R likely in the consensus sequence -R-G-G- 105 ferase 1 (PRMT1) or -R-G-R-G- in proteins such as RNA helicase A and hnRNPU Trans- Tissue transglutaminase (TG2), Substitutes amine part of glutamine amide with a donor 70 glutaminases amine Ubiquitin Praja1 Ubiquitinates proteins and can form chains of ubiquitin 85 Ligase (E3 listed) Rsp5 Ubiquitinates proteins with the consensus sequence -P- 203 P-X-Y- where X is preferentially S/P/A SCFSkp2 Ubiquitinates proteins P27Kip1 5

rating 3H-labelled S-adenosyl-L-methionine([3H]AdoMet) on tively speeds up. 30 This suggested that the enzyme activity surfaces. 101,103,105 The arginine methyltransferases coactiva- increases, if it can bind to phosphates via its SH2 adaptor do- tor associated arginine methyltransferase (CARM1) and pro- main. The Mrksich group designed a model system to rep- tein arginine N-methyltransferase (PRMT1) have been stud- resent this where substrates were immobilised to SAMs and ied on an array displaying 37000 proteins from the human the course of phosphorylation followed directly by MALDI- brain and new substrates were found for the two methyltrans- ToF MS. 30 They too found that initially there was a lag where ferases. 105 Rathert et al. studied the lysine methyltransferase initial phosphorylation occurred. Subsequently the rate of the G9a that methylates histone H3K9 and is essential for early reaction increased when the SH2 domain was able to bind to embryogenesis, and also observed methylation of novel non- the previously phosphorylated peptide. 101 histone proteins. Furthermore 9500 human proteins were Smith et al. report the SPOT synthesis of a library of pep- interrogated by the protein lysine methyltransferase (PKMT) tides on nitrocellulose and their analysis by protein kinaseA 103 SETD6. This lead to the discovery of six new potential (PKA) using radio labelled ATP (Fig. 3). 26 All these pep- SETD6 substrates using independent fluorescence and radi- tides had the -R-X-X-S/T- motif, required for phosphoryla- ation strategies. tion by PKA. The phosphorylation of a peptide found pro- 4.1.1.1 Kinases. Protein kinases are the most studied en- tein kinase D (PKD), MARK3 [MAP (microtubule-associated zyme on peptide and protein arrays. 204 We subdivided there- protein)-regulating kinase 3] and RIL by PKA was observed. fore the following Section into single kinase tests, inhibitor Phosphorylation and the specific site of phosphorylation were screening, kinase screen and peptide screen. confirmed by mutation of potential phosphorylated residues with alanine. The biological significance of these phosphory- 4.1.1.2 Single kinase tests. Parker et al. incorporated lation was then ascertained. It was found that when the ala- peptides on photocleavable linkers into a hydrogel surface via nine mutant of RIL was expressed in prostate carcinoma cells copolymerisation. 56 The peptides were then subjected to the (PC-3), the rate of growth of the carcinoma increased mas- kinases v-Abl and c-Abl and the linker cleaved under UV sively. In comparison phosphorylation of PKD and MARK3 exposure. The cleaved peptides were subsequently analysed was found to be purely structural. Further PKA studies were by MALDI-ToF MS directly on the microarray and the phos- performed on peptides immobilised on SPR chips. 109 The de- phorylation detected. Kinetic studies of Abl by MALDI-ToF tection of phosphorylation was done with a novel dinuclear MS revealed that the reactions starts slowly and consecu- zinc(II) complex (Phos-tag), which carried a biotin tag and

18 1–32 | binds specifically to phosphate groups. Addition of strep- Radioactivity was used as a detection method to screen ki- tavidin to this enhanced SPR detection allowing the phos- nases. 25,102Ptacek et al. demonstrated 4000 phosphorylation phorylation event to be detected. Unlike antibodies, the events by studying 82 different kinases. 102 The activity was zinc(II) complex detects any phosphorylation independent of monitored by [γ-33P] ATP on a protein chip from Invitro- the amino acid. The same Zinc (II) complex was also em- gen. The 82 employed kinases represented most of the 125 ployed in a different method involving detection by fluores- kinases in yeast (2011). Min et al. showed in 2004 the fea- cently labelled streptavidin, which also studied the purified sibility of MALDI-ToF MS analysis for the semi-quantitative kinases PKA and PKC. 76 Then PKA activity was shown in and high-throughput detection of kinase activity by immobil- a cell lysate of MCF-7 cancer cells native, after stimulation ising 7 peptide substrates for 7 different kinases (Src, PKA, and after inhibition. It was shown that stimulated cells (by PKG, CaMKII, CK I, Abl and Erk) onto a SAM on gold. 58 forskolin) showed higher phosphorylation on the immobilised Using this methodology they were also able to monitor inhi- peptides than cells which were incubated with a PKA inhibitor bition and determine IC50 values, showing the applicability as would be expected. This was the first example of direct of this method not only in the high-throughput study of ki- monitoring of the activity of an intracellular enzyme that had nases and substrates, but also inhibitors, which could be im- been perturbed by drug-stimuli. portant pharmacologically. Su et al. further developed this array by addition of a microfluidic system so the 7 kinases 4.1.1.3 Inhibitors for kinases. By using SPR and phos- were flowed over the 7 peptide substrates and catalysis anal- phoimaging, the inhibition of tyrosine kinase c-Src was stud- ysed by MALDI-ToF MS (inhibition was also performed with 59,87 ied. Immobilised peptides were subjected to the kinase this new microfluidic method). 61 They tested their system to- and product conversion was shown independently by radioac- wards the feasibility for multi-enzyme screens. They fur- γ 32 tive ATP ([ - P]ATP) or SPR enhanced by binding of an anti- ther expanded this method though to study 3 different phos- phosphotyrosine antibody. An IC50 value of 31 nM was deter- phatases (phosphatase 2A,2C, tyrosine phosphatase and alka- mined for the kinase inhibitor PP1. Further proof-of-principle line phosphatase) within a cell lysate on a phosphopeptide ar- studies for the analysis of kinase inhibitorswere performedus- ray. These microfluidic devices are favourable as multiple pro- 52,89,165 ing nanoparticles. Sun et al. used biontinylated-ATP cessing steps can be performed on a single platform. A range as a substrate for PKA and analysed the binding of avidin of kinases and a cell lysate (A431 human epithelial carcinoma coated gold nanoparticles to the modified peptides by res- cell) have also been shown to be able to be directly detected 165 onance light scattering (RLS). Kerman et al. used the by SPRi (following antibody binding) in a higher throughput same approach to study inhibitors for PKA and tyrosine kinase manner than conventional SPR, proving the suitability of this c Src 52 p60 − by differential pulse voltammetry (DPV). Both system for future analysis 111 techniques enabled the study of inhibitors and to obtain IC50 values. A different approach was done by Kim et al., who immobilised the peptides onto AuNPs, which were then re- 4.1.1.5 Screen of peptides for kinases. A proof-of- acted with Abl tyrosine kinase. 89 The reaction detection was principle study for the screen immobilised peptides was per- achieved by fluorescence, SPR and QCM. This method was formed by Han et al. who showed an approach for the 72 used to study the kinase inhibition by staurosporine. Inhibi- study of the non-receptor kinase c-Src. The attached pep- tion of PKA by staurosporine was also shown by Akita et al. tides were treated with purified and a cell lysates of c-Src, re- who used β-elimination of the phosphate group followed by spectively. Using a fluorescence labelled antibody they were Michael addition of a thiol-containing fluorescent group. 86 able to detect the formed product and showed that there technique is quantitative, high-throughput and applicable to com- c Src 4.1.1.4 High-throughput screening of kinases on peptide plex mixtures. The p60 − kinase was screened towards its and protein arrays. The high-throughput analysis of kinase peptide specificity by varying the putative peptide substrate substrates, kinases and kinase inhibitors is important to im- YIYGSFK in three different ways: i) alanine-scanning, which prove knowledge of cell signalling pathways and pharmacy. means the stepwise substitution of every amino acid in the se- Typically analytical techniques are limited to fluorescence and quence apart from the underlined tyrosine for alanine ii) short- radiation, which can be quickly analysed by fluorescence and ening the peptide sequence left and right of the central ty- radiation readers respectively. However, these arrays can suf- rosine and iii) scanning the effect of a range of amino acids c Src fer from false positives due to intense fluorescence of an ad- directly next to the tyrosine. The highest p60 − activity jacent spot. Also MALDI-ToF MS is also used as mass spec- was observed when the first tyrosine or last lysine were re- tra can be obtained rapidly, although data storage could be a moved. 80 Schutkowski et al. immobilised 710 peptide 13mers problem for very large screens. Finally the recently devel- which were known to be phosphorylation sites. 100 They inter- oped SPRi could allow the direct monitoring of enzymatic rogated their array with the kinase PKA and were able to find transformations without a label in a high-throughput manner. the repeating sequence -R-R-X-S- as crucial for high phos-

1–32 19 | phorylation. Also the priming effect was investigated by pre- quencher and a fluorophor was attached to the linker. This incubating a slide with the kinase CK2 and the phosphoryla- way, a protease would cleave off the quencher allowing the tion was monitored using radioactively labelled ATP. The re- fluorophor to fluoresce, which could then be subsequently de- sults showed a novel selectivity towards the immobilised pep- tected. Two different proteases, chymopapain and subtilisin, tides. In a third set of experiments, Schutkowski et al. im- were analysed and its specificity screened on multiple pep- mobilised monophosphorylated peptides to study the priming tides. They were able to show a consensus sequence of -D-F- effect more in detail. They were able to show a new selectivity X-A/L/V- for subtilisin. for phosphorylated +3 tyrosine amino acids for two peptides. Kim et al. developed the successful SPOT synthesis on This effect was subsequently confirmed and quantified in so- their glass slides and optimised it for the application of bioas- lution phase experiments. MPK3 and MPK6 are known to be says. 16 To screen the protease α-chymotrypsin, the sequences activated by stress factors and have been studied by Feilner et Gly-Ala-P -Gly were synthesised where P was any amino 99 1 1 al. on a 1690 protein presenting surface. The obtained re- acid. The N-terminal glycine carried a biotin which was results revealed 48 and 39 new potential substrates for MPK3 moved if the α-chymotrypsin cleaved the peptide from the and MPK6, respectively, where 26 were substrates for both surface. Interrogation with fluorescence labelled streptavidin kinases. Since the prokaryotic serine/threonine kinase PknB revealed whether the biotinylated peptides were still attached. from was shown to be released from a pathogen (Staphylococ- This way, Kim et al. were able to show the specificity of 51 cus aureus) into the external milieu. As this kinase could be α-chymotrypsin towards the amino acids with aromatic sys- released by the pathogen in the human body, Miler et al. stud- tems such as phenylalanine and tryptophan. Using their SPOT ied the kinase PknB on different human peptides. They were synthesis system with an incorporated terminal biotin moi- able to show that 68 peptides are potential substrates for this ety could be applicable for the screen of consensus sequences kinase. for any protease active on surfaces. The same approach was 4.1.2 Hydrolases. applied to SPR studying three different caspases (caspase- 3, caspase-6 and caspase-8) in cell lysates of Chinese ham- 4.1.2.1 Proteases for hydrolysis. Proteases are critical en- ster ovarian (CHO) cells. 112 The cells were incubated for dif- zymes that hydrolyse proteins and peptides (with varying lev- ferent times with staurosporine because this is known to in- els of specificities) to shorter peptides. They are linked to crease caspase activity. 205 Then, the cell lysates were anal- many important biological processes such as apoptosis, em- ysed after different staurosporine incubation times and dif- bryogenesis, blood clotting to name a few. Trypsin, Lys-C and ferent caspase activities were detected. Deere et al. stud- chymotrypsin are often used in protein digestion. 107,108,197 ied the protease α-chymotrypsin on peptides immobilised to The protease factor Xa is crucial for the blood clotting in the aminopropylated controlled pore glass (CPG) and the poly- 108 human body by cleaving prothrombin. Wegner et al. stud- mer PEGA1900 (polyethylene glycol copolymer cross-linked ied this protein on two different peptides where one contained with polyacrylamide; 1900 refers to the average pore size) the prothrombinsequence and the other was the mutant of this on varying length oligoglycine spacers and peptide loadings peptide where arginine was substituted by alanine. The re- to determine an optimal platform for proteolysis. 11 They de- action was monitored by real-time SPR where the loss of the termined that there was an optimum enzymatic cleavage rate peptide 8mer was observed after successful proteolytic cleav- when the oligoglycine spacers were 4 glycines in length for age. Wegner et al. were able to show that the single arginine CPG and 2< for PEGA1900. Furthermore, optimal loading replacement yields into a ten times loss in enzyme activity. A for the CPG support was determined as 80% with reactivities system to study protease action with the fairly new technique especially low when 40> % or 100% were used. Foose et SPR imaging was employed by Park et al. to study caspase-3 al. reported in 2007 the formation of uniform protein films which is crucial in apoptosis. (Fig. 8). 107 A peptide substrate on an amine functionalised silicon wafer where the proteins containing the caspase cleaving sequence DEVD and a termi- were cross coupled using glutaraldehyde. 116 The formed pro- nal protein was engineered. The caspase cleaved the peptide tein film of ovalbumin was stable with detergents and on-chip thus releasing the terminal protein which causes the necessary digestion with subtilisin Carlsberg, which are used in laundry weight difference for SPR detection. detergents, could be analysed by ellipsometry and AFM. The Diaz-Mochon et al. developed a new immobilisation and observed decrease of surface thickness appeared to be linear detection methodology for monitoring protease action. 67 For over time for a constant subtilisin concentration. As gold is peptide immobilisation, they tagged peptides with protein nu- known to quench fluorescence, attachment of fluorophors to cleic acids (PNA) which were subsequently applied to DNA nanoparticles results in no fluorescence. Wang et al. there- microarrays. It was shown that the PNA tag interacted specif- fore used a substrate peptide with cleavage sites for four dif- ically with certain DNA sequences. In order to monitor en- ferent proteases and a flurophor and attached it to a AuNP. 3 zyme action on the peptide it was labelled with a fluorescence The proteases trypsin, chymotrypsin, proteinase K and ther-

20 1–32 | molysin were added to a solution of these AuNPs. The release of the fluorophor by the proteolytic cleavage resulted in detectable fluorescence. With this system, they showed the inhibition of proteinase K by Hg2+ and phenylmethyl-sulfonyl fluoride (PMSF). This AuNP system has the advantage that it can be used directly in cells and is potentially applicable to any hydrolase providing the correctly engineered peptide is available.

4.1.2.2 Proteases for Synthesis. Proteases have been shown to be able to perform peptide synthesis as well as hydrolysis which has been reviewed previously. 206,207 The Flitsch group was able two show this effect on different surfaces using the proteases thermolysin, and α-chymotrypsin. 36,38,42 The first time this phenomenon was explored in 2002 by using thermolysin for the formation of dipeptides on polymer beads. 42 To drive the reaction towards the synthesis an excess of substrate was used. The reaction Fig. 18 Employing the protease thermolysin for on chip peptide 36 was monitored by cleavage of the resin and subsequent HPLC synthesis. and LCMS analysis. Phenylalanine was immobilised via the C-terminus and treated with thermolysin and a range of acyl donors. It was observed that especially hydrophobic 4.1.2.3 De/Acetylase. Like phosphorylation, acetylation donors helped to drive the reaction to completion with yields is another very common post translational modification oc- of 99% for Fmoc protected phenylalanine and only 77% curring mainly on lysine residues, but can also occur at the 208 for Cbz protected phenylalanine. Also, they were able to N-terminus. As there are few non-histone substrates for show the enantioselective synthesis of phenylalanine out ofa de/acetylase enzymes and the biological importance of acety- DL-mixture. Moreover, they showed the convenient coupling lation of histones, studies of these types of enzymes have been of amino acids carrying functional side groups without using conducted. any protecting group chemistry. In 2004, PEGA beads were MALDI-ToF MS has been used to analyse deacetylation of functionalised via the C-terminus with either asparagine lysine residues on peptide arrays (NB: sometimes MALDI- or phenylalanine and treated with α-chymotrypsin and ToF MS on SAMs is referred to a SAMDI-ToF MS; Fig. thermolysin in presence of Fmoc-asparagine and Fmoc- 19). 35,57 Acetylation is a fairly common and important PTM phenylalanine, respectively. 38 The reaction was monitored that is often involved in regulating cell function. In the work by fluorescence detection of the Fmoc protecting group. by the Mrksich group 361 hexamer peptides with the sequence The results confirmed the specificity of α-chymotrypsin AcGXKAcZGC-NH2, where X and Z represent any amino for an aromatic amino acid on P1 and thermolysin for acid except cysteine, were immobilised to a SAM monolayer a large hydrophobic group in P’1. With these results in on a gold surface. These were screened against 5 deacetylase hand thermolysin was screened by applying it to all twenty isoforms lysine deacetylase 1 (KDAC1), KDAC2, KDAC3, immobilised amino acids. Furthermore, they used three KDAC8 and sirtuin 1 (SIRT1). Peaks corresponding to a loss different Fmoc protected P1 amino acids: phenylalanine, of 42 m/z were interrogated which corresponds to a loss of asparagine and glycine. The results showed the expected the acetyl group and formation of a primary amine on ly- specificity of thermolysin for hydrophobic amino acids in P’1. sine. KDAC1 was found to have no activity for any of the (For more details on preferential cleavage sites and further substrates 57 and each enzyme had slightly different peptide references, see http://www.chem.qmul.ac.uk/iubmb/enzyme/- specificities with KDAC8 being the most specific and only EC3/4/21/1.html and http://www.chem.qmul.ac.uk/- deacetylating peptides with the trimer RKAcF and any case iubmb/enzyme/EC3/4/24/27.html). In 2008, thermolysin was where X is F. Peptides that were found to be deacetylated shown to catalyse the dipeptide synthesis on self-assembling were then screened against nuclear extracts from HeLa, Ju- monolayers (SAMs) on gold (Fig. 18). 36 Phenylalanine was rkat and smooth muscle cells. Endogenous enzyme activities immobilised and thermolysin applied in presence of Fmo- from HeLa cell lines were found to be particularly active and cLeu. The MALDI-ToF MS analysis revealed full conversion substrate specificities suggested that HeLa cell lines contain to the corresponding dipeptide. However, the same reaction KDAC2, KDAC3 and SIRT1. Finally these arrays were used in presence FmocGly did not show any dipeptide formation. to determine KDAC activity through a HeLa cell cycle by ar-

1–32 21 | resting cell cycle at certain stages using selective and non- min. 79 They used six replicates of each 144 phosphopeptide selective reagents. 35 Finally Lin et al. used a yeast proteome chip and incubated it with the different enzymes for six differ- array to determine substrates of the acetlytransferase NuA4. ent times. The obtained data for the five most active peptides Acetylation was monitored using 14C-AcCoA. It was found were further verified by in solution studies and showed simi- that NuA4 acetylated mainly enzymes or effectors involved lar numbers for most peptides. Therefore, Gao et al. showed in the nutrient availability and energy signalling pathway sug- that their microarray is suitable for semi-quantitative studies gesting NuA4 is a regulator of this pathway. 17 on throughput of 144 peptides. 4.1.3 Ligases. 4.1.3.1 Ubiquitation and ubiquitin like proteins (SUMO and NEDD). Merbl et al. reported in 2009 an extensive study on polyubiquitination of protein arrays by ubiquitin ligase. 84 For the study of mitotic checkpoint systems they used checkpoint extracts of HeLa S3 cells. The cells were arrested with nocodazole in order to prevent mitotic spindle formation (al- Fig. 19 Deacetylation of an array of acetylated peptides with the lowed on/off control). Then, the HeLa cell extracts were di- Ac vided into three different aliquots and treated in three differ- amino acid sequence Ac-GXK ZGC-NH2 where X and Z are any amino acid except cysteine using a series of lysine deacetylases and ent ways: i) not modified ii) treatment with UbcH10 to acti- analysis by MALDI-ToF MS. 57 vate ubiquitin ligase activity (checkpoint (CP) released) and iii) treatment with UbcH10 together with an anaphase pro- moting complex (APC) inhibitor (emi1). Before the applica- 4.1.2.4 Phosphatases. Gao et al. used a microarray for tion of these cell extracts to the microarray, radioactively la- the fabrication of phosphopeptides followed by the analysisof 79 belled securin, a ubiquitin ligase substrate which is degraded phosphatases (Fig. 20). Since phosphatases are very similar by APC was added. The results confirmed that securin was in substrate specificity, they attempted to find a specific sub- stable in i) and iii) but degraded by ii) as expected. Then strate patterns for seven phosphatases (see table xy). They im- the aliquots ii) and iii) were applied to the 8000 protein pre- mobilised 144 potential phosphopeptide substrates on a glass senting microarray. The polyubiquitination was monitored by slide and interrogated it with the phosphatases. Remaining an anti-polyubiquitin antibody followed by a fluorescence la- phosphate groups were detected with a phosphate specific belled secondary antibody. Eleven of the sixteen known APC fluorophor (Pro-Q). Absence or decrease of fluorescent sig- substrates were found to be polyubiquitinated. Overall, the nals indicated phosphatase action. The system to study phos- CP-released and APC inhibited cell extracts showed 132 dif- phatases in cell lysates by microfluidics was already shown 61 ferentially modified proteins. Loch et al. used a protein ar- (see above). ray, displaying over 2000 human proteins, to interrogate the E3 ubiquitin ligase Praja1 for new potential substrates. 85 Af- ter initial studies towards the optimal E2 binding partner of Praja1, UBE2D3 was chosen for on-chip enzymatic analysis. Praja1 was spotted together with UBE2D3 and ubiquitin to the protein chip and the reaction was monitored with Fig. 20 Concept of the phosphatase screen performed by Gao et two different antibodies. One antibody detects both mono- al. 79 and polyubiquitination (rabbit anti-ubiquitin) and the other only binds to polyubiquitinated proteins (TUBE2). The rabbit The researchers were able to detect distinguished enzyme anti-ubiquitin antibody detected 26 and TUBE2 25 ubiquity- activity on the different phosphopeptides and also to reveal lation, respectively, where 14 were the same proteins. Loch new substrates. It was observed that PTP1B and TCPTP et al. analysed the ubiquitinylated substrates in a database re- favour acidic amino acids and show less activity towards basic search and found that most of the substrates detected by the residues. It was possible to find a different specificity for the rabbit anti-ubiquitin antibody fit in the categories: kinase ac- phosphatases ∆SHP1 and ∆SHP2. Both showed higher activ- tivity, anti-apoptosis, brain development, RNA polIII and oth- ity on acidic amino acids but in different positions relative to ers. The substrates detected by TUBE2 fitted in the categories: the central tyrosine. ATP binding, RNA processing and others. Even though one Another way to distinguish between different phosphatases has to be careful with this sort of analysis and they first have with similar specificity is via kinetics. Therefore, Gao et al. to be confirmed by in vivo analysis, it showed that microarrays analysed the time dependency over the time course of 120 can help to locate enzyme activity in subcellular functions.

22 1–32 | Del Rincon et al. purified the E3 ubiquitin ligase SCFSkp2 4.2 Glycans and applied it together with the necessary purified E1 and E2, ubiquitin and the system for ATP regeneration on the array A list of the thus far employed enzymes on immobilised gly- 5 cans can be found in table 3. Different glycosylations on sur- containing more than 8000 recombinant proteins (Fig. 21). 21 Since some of the ubiquitin was biotinylated, the product for- faces has been previously reviewed by Voglmeir et al. . mation could be monitored by fluorescence labelled strepta- 4.2.1 Glycosyltransferases. vidin. The kinase inhibitor p27Kip1 was found to be ubiquiti- nated which is a known SCFSkp2 substrate. But also new sub- 4.2.1.1 POMGnT and Transsialidase. Sardzik et al. β strates were identified in this proof-of-principle study. Then, were the first to successfully employ the enzyme -1,2-N- Del Rincon et al. used two cell systems—rabbit reticulocyte acetyl-glucosaminyltransferase 1 (POMGnT1) on an immo- lysate and the S-100 fraction of HeLa cells and applied them bilised mannopeptide and further elongated the formed dis- α β β to protein array. They found 66 proteins which are putative accharide to the O-glycan NeuNAc 2-3Gal 1-4GlcNAc 1- α α substrates for both cell systems. Many of these proteins have 2Man which is naturally found on -dystroglycan (Fig. 10 α been previously associated with ubiquitin. 6). -Dystroglycan is a cell surface glycoprotein that acts as an important receptor for cell-cell adhesion interaction. In fact defects in the O-mannosyl glycans has been linked to congenital muscular dystrophy. 213 A natural peptide sequence with α- Man bound to a threonine residue was chemically synthesised and coupled to a SAM on gold. The immobilised glycopeptide was a substrate of the peptide specific enzyme POMGnT1 214 which transferred β-1,2-GlcNAc to the mannose. A β1-4 galactosyltransferase and an α2-3 trans-sialidase were subsequently employed to form the tetrasaccharide. Since sialic acid groups are not stable during MALDI ionisation, Sardzik et al. (2011) methylated the sialic acids after the enzymatic reactions, which improved their stability. 27 4.2.1.2 Galactosyltransferase. Galactosyltransferases are very reliable enzymes, so are often employed for proof-of- principle studies. They are capable of forming terminal galactoses on surfaces which has been used for the synthesis of glycans and screening the specificity of enzymes and Fig. 21 Workflow of the detection of ubiquitin ligase action on a lectins. 48,53,77). 36 Sanchez-Ruiz et al. non-covalently immo- 5 purchased protein microarray. bilised a range of multiantennary N-glycans that had been derivatised with hydrophobic alkyl groups to a hydrophobic SAM on gold. 62 The glycans were treated with β1,4-GalT Then, del Rincon et al. studied human breast tumour spec- and then incubated with galactose specific lectin ECA and this imens. They observed a clear difference in ubiquitination for lectin was detected by MALDI-ToF MS. The array could then high- and low-grade tumours. One protein which was ubiqui- be treated with a galactosidase, removing the binding site for tinylated by high-grade tumour cells RAD23A is performing the lectin which was also confirmed by MALDI-ToF MS. This DNA repair in conjunction with an E3 ligase. Ubiquitination showed the applicability of these arrays to follow glycosyl- and degradationof this protein mightbe the reason for the lack transferase reactions and screen for lectin binding partners. of DNA repair. The protein array system was further explored Similarly Park et al. employed microarrays to study the β β by the analysis of the post translational modifications (PTMs) acceptor specificity of 1,4-GalT. They showed that the - α with SUMO1 (small ubiquitin-like modifier 1) and NEDD8 GlcNAc is 11-times more galactosylated than -GlcNAc 215 (neural precursor cell expressed and developmentally down- which agreed with solution studies. An alteration of ac- regulated 8). They used HeLa cell extracts with the corre- ceptor specificity was shown by Ban et al. by complexing β α β sponding biotin-labelled donor and applied it to the protein the -1,4-GalT with the protein -lactalbumin. The -1,4- α chip. This way, they were able to show SUMOylation and GalT/ -lactalbumin complex preferably galactosylated glu- 37 NEDDylation for many substrates. For SUMO1 a consensus cose not GlcNAc. sequence (yKxE/D) was found. In comparison, Bovin and co-workers screened the substrate specificity of α-1,3-galactosyltransferase (α-1,3-GalT), an enzyme that is known to be involved in the synthesis of

1–32 23 | Table 3 Enzymatic reactions on Glycoarrays

Enzyme Reaction Lit.

Glycosyl- α-1,3-Fucosyltransferase Transfer of Fuc α1-3 to reducing GlcNAc moeity 10,39,48,77,92 transferases α-1,3-GalT Transfer of Gal α1-3 to Gal in the disaccharide Galβ1-4GlcNAc 91 α-1,6-Fucosyltransferase Transfer of Fuc α1-6 to reducing GlcNAc moeity 39 α-2,3 transsialidase (TcTS) Transfer of Neu5Ac α2-3 to Gal or the disaccharide Galβ1- 27,27 4GlcNAc/Glc in the presence of fetuin Acetyltransferase Transfers an azidoacetyl group from azidoacetal coenzyme A (AzAc- 82,98 CoA) to various aminoglycans (e.g. 2-deoxystreptamine) Alternansucrase Transfer of Glc from sucrose to a Glc acceptor with alternating α1-6 114 and α1-3 linkages AtFUT1 Transfer of Fuc α1-2 to tamarind xyloglucan 106 bovine β-1,4-GalT Transfer of Gal β1-4 to GlcNAc. In presence of lactalbumin the speci- 10,36,37,48,53,62,77,78,209 ﬁcity changes to Glc instead Glucosyltransferase (glyco- Transfer of Glc α1-6 to Glc or primary alcohols 90 syltransferase R (GFTR)) Glycogen phosphorylase Phospholysis of the α1,4-glucosidic linkage from the non-reducing end 33,96 of glycans such as amyopectin to produce glucose-1-phosphate Human ST6GalNAc-1 Transfer of Neu5Ac α2-6 to the 6-hydroxyl group of the reducing 71 GalNAc moeity for the O-glycans GalNAcα- (Tn antigen), Galb1- 3GalNAcα- (T antigen) and Neu5Acα2-3Galβ1-3GalNAcα- (Sialyl T antigen) POMGnT Transfer of GlcNAc β1-2 to O-mannosyl-peptides of certain sequences 10 (exact sequence requirements are unknown) ppGalNAcT Transfers αGalNAc onto serine or threonine residues ideally with pro- 36,49 line or alanine 1 and not bulky amino acids 1 ± ± Rat ST3Gal-III Transfer of Neu5Ac α2-3 to the disaccharide Galβ1-3/4GlcNAc or 53,71,77,210 Galβ1-3GalNAcα/β- O-glycans. Also activity for Galβ1-3[Fucα1- 4]GalNAc- (Lea) Glycosidases Abg (E358G/S) Glycosyn- Transfers Gal β1-4 to βGlc glycosides/peptides using an α-Galactosyl 211 thase Fluoride donor β-1,4-Galactosidases Removes β1-4 Gal from the non-reducing terminus of a Galβ1- 62,209,210 4GlcNAc moeity endo-arabinase Cleaves off mono-, di- and tri-α1,5arabinose units from carbohydrate 63 structure endo-β-glucanase (cellu- Cleaves off mono-, di- and tri-β1,4Glc units from carbohydrate struc- 45,63 lase) ture endo-xylanase Cleaves off Xyl and Xylβ1-4Xyl units from carbohydrate structure 63 endo-xyloglucanase Cleaves off Xylα1,6-Glcβ- units from carbohydrate structure 63 exo-β-glucanase Cleaves cellobiose units from the terminus of β1-4Glc polysaccharides 45 Methylesterases Catalyse the removal of methyl esters from the homogalacturonan (HG) 83 backbone domain of pectin, a ubiquitous polysaccharide in plant cell walls Pancreatic α-Amylase Endo-hydrolyses α1-4 Glc 212 Sialidase (Neuraminidase) Cleaves α2-3 Neu5Ac from the disaccharide Galβ1-3/4GlcNAc or 53 Galβ1-3GalNAcα/β-O-glycans γ-Amylase Removes α1-4/6 Glc from the non-reducing terminus 54,63,97 Oxido- Galactose Oxidase Oxidises the C6-OH group to an aldehyde in either Gal or Tal In prep. reductases

24 1–32 | the carbohydrate xenoantigen (Galα1-3Galβ1-4GlcNAc), by and analysed more closely by MALDI-ToF MS. They were immobilising many natural glycans to a microtiter plates and able to show the different specificities of the three isoforms incubating them with the α-1,3-GalT (ELLA assay). This re- by their reaction with the peptides OSM, GCSF, hCGβ and action was monitored by addition of the α1-3 specific lectin CD59. Viscum album agglutinin (VAA) tagged with biotin, used to 4.2.1.4 Sialyltransferase and Fucosyltransferase. Sialyl- couple streptavidin tagged horseradish peroxidise (HRPO). transferases have been employed on many different microar- HRPO oxidised o-phenylenediamine resulting in a colour ray based systems using different analytical techniques such change when α1-3 galactosylation occurred. 91 They found as SPR 53 mass spectrometry 210 and lectin binding. 77 . 48 They Galβ1-4GlcNAc was a preferable substrate for α-1,3-GalT in are of great interest especially in the synthesis of sialyl lex, an accordance to the synthesis of xenoantigen. important receptor molecule involved in may cell recognition 92 4.2.1.3 ppGalNAcT. Polypeptide-N-acetyl-D- events such as inflammation. Park et al. described the on x galactosaminyl transferase (ppGalNAcT2) is an enzyme chip synthesis of sialyl le using three enzymes including α- that transfers GalNAc onto serine/threonine residues, forming 2,3-Siallyltransferase (SiaT) and a final α-(1,3)-fucosyltation GalNAcα-OPep, when they are in certain peptide motifs. and showed the product formation by binding of the fluores- x 77 The peptide sequence specificity of ppGalNAcT2 was studied cently tagged anti-sialyl Le antibody. Also inhibition of the by Laurent et al. 36 They synthesised 32 different peptides α-(1,3)-fucosyltransferase (FucT) has been studied as it could 92 of the consensus sequence AHG-X-T-X-APA which only disrupt the inflammatory response pathway. differed in the +1 or -1 position relative to threonine in the Blixt and co-workers screened the acceptor specificity Muc1 peptide AHGVTSAPA. Enzymatic transformation was of a range of sialyltransferases by incorporation an biotin- followed by MALDI-ToF MS. Results showed enhanced labelled Neu5Ac label succeeded by addition of fluo- 71 peptide glycosylation in presence of proline in either/both the rescently tagged streptavidin (Fig. 23). They discov- +1 or -1 position. ered four new enzyme specificities such as the sialylation of chitobiose (Galβ1-4GlcNAc)n by human α-2,6- sialyltransferase-I (hST6Gal-I), the transfer of sialic acid on Lewisa by rat α-2,3-sialyltransferase (rST3Gal-III) and that porcine α-2,3-sialyltransferase-I (pST3Gal-I) accepts ganglio-oligosaccharides. The sialylation of Lewisa was especially surprising as it is fucosylated, which normally prevents further enzymatic modification of glycans.

Fig. 23 Screen for acceptor speciﬁcity of different sialyltransferases using biotin labelled glycosyldonors. 71

Fig. 22 Screening of the transferase ppGalNAcT2 towards its A combination of sialyl and fucosyltransferases were used specificity for the amino acids before and after the threonine. 36 by Reichardt and co-workers to produce complex glycan arrays that could be screened against for example glycan bind- Blixt et al. studied the specificity of three different isoforms ing proteins. This complexity was generated by incubat- of ppGalNAcT—T1-T3—on a series of peptides immobilised ing an α-2,6-SialT and an α-1,3-fucosyltransferase, with a to NHS activated hydrogel coated glass slides using the fluo- range of synthetic N-glycan core structures. Product forma- rescently tagged lectin VVA. 49 Potential substrate candidates tion was elucidated using fluorescence tagged lectins spe- were then immobilised to beads, incubated with one of the pp- cific for the glycosylation being studied. 48 This was later GalNAcT isoforms and subsequently cleaved from the resin developed where they screened two recombinant fucosyl-

1–32 25 | transferases, α-1,6-fucosyltransferase from Caenorhabditis although have been fairly poorly studied on array platforms. elegans (CeFUT8) and α-1,3-fucosyltransferase Arabidopsis However Seibel et al. studied glycosyltransferase R (GFTR) thaliana (AtFucTA), on 18 different covalently immobilised on the immobilised acceptors maltose, lactose and a primary glycans (Fig. 24). 39 They were able to obtain a complex alcohols (Fig. 25). 90 GFTR transfers glucose from sucrose fucosylated glycan and used it to study the protein bind- in an α-1,6-linkage. Product formation was visualised using ing specificities of the lectins AAL, AOL and LCA and the the fluorescently tagged lectin ConA. Maltose and the primary antibody anti-HRP. With this expanded methodology, they alcohol were glycosylation whereas lactose showed no prod- demonstrated the rapid nanoscale production of fucosylated uct conversion, suggesting that only terminal glucose residues N-glycans and found new lectin specificities. are acceptors for GFTR. Since maltose is a glucose disaccharide and hence binds to the lectin ConA as well, there were comparison studies between the starting material and the product undertaken which showed the literature known increase of ConA binding to the α-1,6-linked Glc over the α-1,4-linked Glc in maltose. citeGoldstein1965

Fig. 24 Screen for the acceptor speciﬁcity of the fucosyltransferases CeFUT8 and AtFucTA using ﬂuorescence tagged lectins. 39

4.2.1.5 Glycogen Phosphorylase. Glycogen phosphorylase is an enzyme that is found in Potato capable of phospho- rlysing α1,4-glucosidic bonds from the non-reducing termi- Fig. 25 Scheme of the enzymatic transfer of glucose by 90 nus releasing glucose-1-phosphate in the presence of phos- glycosyltransferase R on an alcohol and maltose. phoric acid. Conversely the reaction can be driven towards polymerisation if an excess of glucose-1-phosphate is added. 4.2.1.8 Acetyltransferase. Acetyltransferases catalyse the Okahata and co-workers directly studied the glycogen phos- transfer of an acetyl group from acetyl coenzyme A (AcCoA) α phorylase acting on amylopectin, a polymer of 1,4-glucose to an amine residue forming an amide. 217 They are pharma- α with some 1,6-glucose branching, in both the forwards and cologically of great importance as they are produced by bacte- reverse reaction by QCM (Fig. 10). Using this methodology ria as a defence mechanism against antibacterials. Acetylating they were able to determine all kinetic parameters for the re- antibacterials reduces or stops the ability of those molecules to 96 actions. bind to their therapeutic targets (bacterial immunity). Due to 4.2.1.6 Alternansucrase. Cle et al. used SPR to observe this Barrett et al. studied the acetyltransferases AAC(2’) and the enzyme alternansucrase NRRL B-1355 from Leuconostoc AAC(3), from the pathogens Mycobacterium tuberculosis and 114 Escherichia coli respectively, on immobilised aminoglycoside mesenteroides directly in real time. Alternansucrase catal- 98 yses the transfer of glucose from a sucrose donor, to an immo- antibacterials. They used radioactively labelled AcCoA to- bilised substrate generating alternan, a polymer with an α-1,6 gether with the transferases (from a cell lysate) and analysed and α-1,3 glycosidic structure. 216. Further analysis was con- acetylation with a phosphoimager. Then, in a different assay, ducted by AFM which showed directly the formation of an they incubated the acetylated aminoglycosides with the ﬂu- amorphous surface by a 4.6 fold increase in peak-to-trough orescence labelled bacterial rRNA aminoacyl-tRNA site (A- height. site) which is the target of the antibacterials. The ﬂuorescence readout showed that reaction with AAC(2’) does not affect the 4.2.1.7 Glucosyltransferase. Glucosyltransferases cataly- binding ability. AAC(3), however, showed a decrease of the ses the addition of glucose to molecules such as carbohydrates binding to the rRNA showing the applicability of these sys- or proteins and thus are a very important class of enzymes, tems in studying antibacterial resistance.

26 1–32 | AAC(3)-IV also accepts azidoacetyl coenzyme A (AzAc- mined all kinetic parameters of the reaction by QCM (Fig. CoA), so is able to incorporate an azido group into amino- 27). 33. Glucoamylase catalysis was also monitored electro- glycosides. Tsitovich et al. immobilised aminoglycoside chemically on a novel starch coated single-walled carbon nan- apramycin on an agarose surface via “click-chemistry” us- otubes (SWNT) as degradation of the starch exposed areas of ing an azide group chemically attached to the molecule (Fig. the SWNT (confirmed by AFM). 54. Glycosidases have also 26). 82 The immobilised glycan was subsequently treated with been mutated to change their specificity from hydrolysis to the AAC(3) in presence of the azido coenzyme A (AzAcCoA). glycosylation. These are termed glycosynthases. Withers and The incorporated azide was then used as the ’handle’ in an- co-workers were the first to employ glycosynthases on solid- other “click-chemistry” reaction with an alkyne derivatised support. 211 They replaced the carboxylate carrying glutamic fluorescent dye allowing the acetylation reaction product to acid in the active site of Agrobacterium sp. β-glucosidase be detected (Fig. 26). for serine (Abg E358S) and glycine (Abg E358G), respectively. These glycosynthases were used on glycopeptides on a PEGA resin which carries an O-GlcNAc site to transfer α-D- galactosyl fluoride in 1,4 linkage to the GlcNAc. The glycine mutant showed higher yields over the serine mutant which was analysed by cleavage from the resin and HPLC analysis.

Fig. 26 Enzymatic incorporation of an azide tag followed by a readout with an alkyne carrying ﬂuorophor. 82

4.2.2 Glycosidases. Glycosidases hydrolyses glucosidic bonds resulting in the degradation of carbohydrates into mono-, di- or tri-saccharides, which can then be used in an- abolic reactions. They are also one of the fastest acting enzymes classes found in nature where rate constants can be 218 Fig. 27 QCM detection of γ-Amylase action on immobilised starch up to 1000 s−1. Most biological fluids possess glycosi- 33 dases like milk, blood and saliva. This is exemplified by Re- and direct determination of kinetics. ichardt and co-workers (2012) who applied saliva and soil to their glycan arrays on SAMs and studied their degrada- 63 tion by MALDI-ToF MS (Fig. 14). Glycosidases are also 4.3 Small Molecules frequently employed to regenerate glycoarrays after enzymatic reactions. Plath et al. used a sialidase to recover the A summary of all employed enzymes on immobilised small surface of Galβ-1,4-GlcNAc that had been previously sialy- molecules are shown in table 4. lated with rST3Gal-III allowing for further repetitions on the same platform . 53 Similarly Sanchez-Ruiz et al. used β-1,4- 4.3.1 Lipases. Lipid arrays and lipases are the least stud- galactosidase to remove a previously enzymatically attached ied out of the main biomolecule groups (DNA, protein/peptide galactose. 62 This galactosidase was also used to confirm the and carbohydrates). However, a high throughput analysis of galactosylation linkage as it was known to only hydrolyse β- lipases on C2-C12 monoesters was developed by Grognux et 1,4 glucosidic bonds. Glycosidases have also been studied al. (Fig. 28). 74 Different lipases were applied on the sur- from cell lysates on polysaccharides. 33,210 Northen et al. im- face and cleaved the esters (with an alkyl side had an adjacent mobilised lactose non-covalently via a perfluorinated linker alcohol group in either R or S enantiomer) resulting in the and analysed the enzymatic reaction from an Escherichia formation of a diol on the surface. This diol was chemose- coli cell lysate by nanostructure-initiator mass spectrometry lectively oxidised to the aldehyde and treated with the sulfo- (NIMS). 210 They used the same method to study the inhibitors hydrazine carrying dye Rhodamine B. The hydrazine reacted phenylethyl-D-thiogalactopyranoside and deoxygalactonojir- with the aldehyde and rhodamine was released enabling the imycin and obtained results which were in agreement with fluorescence visualisation of this reaction (see table 4). Sur- previous reports. 219 In a similar study, Chang et al. were face based systems were also shown to be feasible for stud- able to monitor endo- and exo- cellulases by MALDI-ToF ies of lipase enantioselectivities. The transfer of R/S-vinyl-3- MS and found their method could discriminate between the phenylbutanoate to bound diketones catalysed by Chromobac- two classes. 45 Glucoamylase hydrolyses α-1,4/6 glucose from terium Viscosum lipase (CVL) was shown by Humphrey et the non-reducing terminus of carbohydrates. Nishino studied al. 43. They showed that CVL transferred the R enantiomer glucoamylase kinetics (γ-amylase) on amylopectin and deter- in enantioselectivity (ee>99%) and acceptable yields (38%),

1–32 27 | Table 4 Enzymatic reactions of small molecules on arrays

Enzyme Reaction Lit

Esterase Cutinase Serine esterase that hydrolyses 4- 118 hydroxyphenyl valerate to corresponding alcohol

Epoxide Hy- Epoxide hydrolases (Rhodococcus Transforms an epoxide into an 1,2 diol reveling 29 drolase rhodochrous (45299)) coumarin ﬂuorescent moeity

Protease Trypsin (bovine pancreas (T-8003) Cleaves C-terminal to lysine revealing 29 coumarin ﬂuorescent moiety

Phosphatase Alkaline phosphatases (bovine intesti- Dephosphorylates coumarin allowing it to ﬂuo- 29 nal mucosa (P-7640) resce but the enantioselectivity of CVL was challenged by using S- valerate. 118 The substrate was immobilised to SAMs on gold. vinyl-3-phenylbutanoate (ee=92%) as a substrate (yield 7%). Enzymatic action cleaves the ester and produces redox active hydroquinone, allowing the analysis by CV. The further showed the suitability of their system for the real-time analysis of cutinase activity, as they were able to obtain comparable kinetic data to previous solution studies.

5 Future

6 Conclusions

This review has tried to compile the most common enzymatic reactions performed on surfaces and address issues of surface material and analytical readout. It is clear from the numerous examples cited that the technology is very useful for high- throughput enzymatic studies . In some areas such as DNA polymerisation and assay involving protein kinases, the technology is already very mature and has been applied to important biological problems. Physicochemical methods for the Fig. 28 Reaction pathway for monitoring lipase action. 74 detailed structural analysis of enzyme substrates with label or label-free have been developed. With the increasing interest in high-throughputand miniaturisation in biology, biotechnol- 4.3.2 Hydrolases. Zhu et al. proposed a small-molecule ogy and medicine, the surface-based methods describe here microarray for the screening of the enzyme hydrolases: epox- should become important tools. ide hydrolase, acetylcholine esterase, trypsin, and alkaline phosphatase. 29 Different substrates were linked to coumarin which was subsequently attached to the amine functionalised References glass slide. In this configuration, coumarin has very low fluo- 1 W. Chan, Biol. Blood Marrow Transplant., 2006, 12, 87–91. rescence emission due to the alcohol being masked. Once the 2 Y.-C. Yeh, B. Creran and V. M. Rotello, Nanoscale, 2012, 4, 1871–1880. hydrolases released the unmasked coumarin, the fluorescent 3 X. H. Wang, J. Geng, D. Miyoshi, J. S. Ren, N. Sugimoto and X. G. Qu, emission dramatically increases. This way Zhu et al. were Biosensors & Bioelectronics, 2010, 26, 743–747. 4 O. Blixt, S. Head, T. Mondala, C. Scanlan, M. E. Huflejt, R. Alvarez, able to show a fast screen for hydrolases of small molecules. M. C. Bryan, F. Fazio, D. Calarese, J. Stevens, N. Razi, D. J. Stevens, J. J. Skehel, I. van Die, D. R. Burton, I. A. Wilson, R. Cummings, 4.3.2.1 Esterase. Nayak et al. analysed the serine es- N. Bovin, C. H. Wong and J. C. Paulson, Proc. Natl. Acad. Sci. U. S. terase cutinase on the immobilised ester 4-hydroxyphenyl A., 2004, 101, 17033–17038.

28 1–32 | 5 S. V. del Rincon, J. Rogers, M. Widschwendter, D. Sun, H. B. Sieburg 36 N. Laurent, R. Haddoub, J. Voglmeir, S. C. C. Wong, S. J. Gaskell and and C. Spruck, Plos One, 2010, 5, e11332. S. L. Flitsch, ChemBioChem, 2008, 9, 2592–2596. 6 M. von Nickisch-Rosenegk, X. Marschan, D. Andresen and F. F. Bier, 37 L. Ban and M. Mrksich, Angewandte Chemie International Edition, Anal. Bioanal. Chem., 2008, 391, 1671–1678. 2008, 47, 3396–3399. 7 Y. F. Zhao, Y. Liu, I. Lee, Y. Song, X. D. Qin, F. Zaera and J. Y. Liao, 38 R. H. P. Doeze, B. A. Maltman, C. L. Egan, R. V. Ulijn and S. L. Flitsch, Journal of Biomedical Materials Research Part A, 2012, 100A, 103– Angew. Chem., Int. Ed., 2004, 43, 3138–3141. 110. 39 S. Serna, S. Yan, M. Martin-Lomas, I. B. H. Wilson and N. C. Reichardt, 8 N. Laurent, J. Voglmeir, A. Wright, J. Blackburn, N. T. Pham, S. C. C. J. Am. Chem. Soc., 2011, 133, 16495–16502. Wong, S. J. Gaskell and S. L. Flitsch, ChemBioChem, 2008, 9, 883–887. 40 Y. Chevolot, J. Martins, N. Milosevic, D. Leonard, S. Zeng, M. Malis- 9 N. Laurent, R. Haddoub, J. Voglmeir and S. L. Flitsch, Methods in sard, E. G. Berger, P. Maier, H. J. Mathieu, D. H. G. Crout and H. Sigrist, molecular biology (Clifton, N.J.), 2012, 808, 269–84. Bioorganic & Medicinal Chemistry, 2001, 9, 2943–2953. 10 R. Sardzik, A. P. Green, N. Laurent, P. Both, C. Fontana, J. Voglmeir, 41 O. A. Gutierrez, M. Chavez and E. Lissi, Anal. Chem., 2004, 76, 2664– M. J. Weissenborn, R. Haddoub, P. Grassi, S. M. Haslam, G. Widmalm 2668. and S. L. Flitsch, J. Am. Chem. Soc., 2012, 134, 4521–4. 42 R. V. Ulijn, B. Baragana, P. J. Halling and S. L. Flitsch, J. Am. Chem. 11 J. Deere, R. F. De Oliveira, B. Tomaszewski, S. Millar, A. Lalaouni, Soc., 2002, 124, 10988–10989. L. F. Solares, S. L. Flitsch and P. J. Halling, Langmuir, 2008, 24, 11762– 43 C. E. Humphrey, N. J. Turner, M. A. M. Easson, S. L. Flitsch and R. V. 11769. Ulijn, J. Am. Chem. Soc., 2003, 125, 13952–13953. 12 M. J. Weissenborn, J. W. Wehner, C. J. Gray, R. Sardzik, C. E. Eyers, 44 E. Ostuni, R. Chapman, R. Holmlin, S. Takayama and G. Whitesides, T. K. Lindhorst and S. L. Flitsch, Beilstein J. Org. Chem., 2012, 8, 753– Langmuir, 2001, 17, 5605–5620. 762. 45 S. H. Chang, J. L. Han, S. Y. Tseng, H. Y. Lee, C. W. Lin, Y. C. Lin, 13 O. J. Plante, E. R. Palmacci and P. H. Seeberger, Science, 2001, 291, W. Y. Jeng, A. H. J. Wang, C. Y. Wu and C. H. Wong, J. Am. Chem. 1523–1527. Soc., 2010, 132, 13371–13380. 14 A. Y. Bosma, R. V. Ulijn, G. McConnell, J. Girkin, P. J. Halling and S. L. 46 N. Laurent, J. Voglmeir and S. L. Flitsch, Chemical Communications, Flitsch, Chemical Communications, 2003, 2790–2791. 2008, 4400–4412. 15 C. M. Ruan, W. Wang and B. H. Gu, Anal. Chem., 2006, 78, 3379–3384. 47 M. J. Weissenborn, R. Castangia, J. W. Wehner, R. Sardzik, T. K. Lind- 16 D. H. Kim, D. S. Shin and Y.S. Lee, J. Pept. Sci., 2007, 13, 625–633. horst and S. L. Flitsch, Chem. Commun., 2012, 48, 4444–4446. 17 Y. Y. Lin, J. Y. Lu, J. M. Zhang, W. Walter, W. W. Dang, J. Wan, S. C. 48 S. Serna, J. Etxebarria, N. Ruiz, M. Martin-Lomas and N. C. Reichardt, Tao, J. Qian, Y. M. Zhao, J. D. Boeke, S. L. Berger and H. Zhu, Cell, Chemistry-a European Journal, 2010, 16, 13163–13175. 2009, 136, 1073–1084. 49 O. Blixt, E. Clo, A. S. Nudelman, K. K. Sorensen, T. Clausen, H. H. 18 R. B. Merriﬁeld, J. Am. Chem. Soc., 1963, 85, 2149–&. Wandall, P. O. Livingston, H. Clausen and K. J. Jensen, J. Proteome 19 P. J. Halling, R. V. Ulijn and S. L. Flitsch, Curr. Opin. Biotechnol., 2005, Res., 2010, 9, 5250–5261. 16, 385–392. 50 M. H. Kwon, S. H. Jung, Y. M. Kim and K. S. Ha, Anal. Chem., 2011, 20 M. Fais, R. Karamanska, D. A. Russell and R. A. Field, J. Cereal Sci., 83, 8718–8724. 2009, 50, 306–311. 51 M. Miller, S. Donat, S. Rakette, T. Stehle, T. Kouwen, S. H. Diks, 21 J. Voglmeir, R. Sardzik, M. J. Weissenborn and S. L. Flitsch, OMICS, A. Dreisbach, E. Reilman, K. Gronau, D. Becher, M. P. Peppelenbosch, 2010, 14, 437–444. J. M. van Dijl and K. Ohlsen, Plos One, 2010, 5, e9057. 22 H. Chandra, P. J. Reddy and S. Srivastava, Expert Review of Proteomics, 52 K. Kerman, M. Chikae, S. Yamamura and E. Tamiya, Anal. Chim. Acta, 2011, 8, 61–79. 2007, 588, 26–33. 23 A. Thiele, G. I. Stangl and M. Schutkowski, Mol. Biotechnol., 2011, 49, 53 C. Plath, T. Weimar, H. Peters and T. Peters, ChemBioChem, 2006, 7, 283–305. 1226–1230. 24 T. Liu, R. Xue, L. Dong, H. Wu, D. Zhang and X. Shen, Acta Biochimica 54 A. Star, V. Joshi, T. R. Han, M. V. P. Altoe, G. Gruner and J. F. Stoddart, et Biophysica Sinica, 2011, 43, 45–51. Org. Lett., 2004, 6, 2089–2092. 25 L. Meng, G. A. Michaud, J. S. Merkel, F. Zhou, J. Huang, D. R. Mattoon 55 M. Castronovo, S. Radovic, C. Grunwald, L. Casalis, M. Morgante and and B. Schweitzer, Bmc Biotechnology, 2008, 8, 22. G. Scoles, Nano Lett., 2008, 8, 4140–4145. 26 F. D. Smith, B. K. Samelson and J. D. Scott, Biochem. J., 2011, 438, 56 L. L. Parker, S. B. Brueggemeier, W. J. Rhee, D. Wu, S. B. H. Kent, S. J. 103–110. Kron and S. P. Palecek, Analyst, 2006, 131, 1097–1104. 27 R. Sardzik, R. Sharma, S. Kaloo, J. Voglmeir, P. R. Crocker and S. L. 57 Z. A. Gurard-Levin, J. Kim and M. Mrksich, ChemBioChem, 2009, 10, Flitsch, Chemical Communications, 2011, 47, 5425–5427. 2159–2161. 28 H. J. Lee, A. W. Wark and R. M. Corn, Langmuir, 2006, 22, 5241–5250. 58 D. H. Min, W. S. Yeo and M. Mrksich, Anal. Chem., 2004, 76, 3923– 29 Q. Zhu, M. Uttamchandani, D. B. Li, M. L. Lesaicherre and S. Q. Yao, 3929. Org. Lett., 2003, 5, 1257–1260. 59 B. T. Houseman, E. S. Gawalt and M. Mrksich, Langmuir, 2003, 19, 30 X. L. Liao, J. Su and M. Mrksich, Chemistry-a European Journal, 2009, 1522–1531. 15, 12303–12309. 60 J. Kim, S. Kim and W.-S. Yeo, BioChip Journal, 2011, 5, 199–205. 31 J. H. Kim, S. H. Han and B. H. Chung, Biosensors & Bioelectronics, 61 J. Su, M. R. Bringer, R. F. Ismagilov and M. Mrksich, J. Am. Chem. Soc., 2011, 26, 2125–2129. 2005, 127, 7280–7281. 32 O. Kosik, R. P. Auburn, S. Russell, E. Stratilova, S. Garajova, 62 A. Sanchez-Ruiz, S. Serna, N. Ruiz, M. Martin-Lomas and N. C. Re- M. Hrmova and V. Farkas, Glycoconjugate J., 2010, 27, 79–87. ichardt, Angewandte Chemie-International Edition, 2011, 50, 1801– 33 H. Nishino, A. Murakawa, T. Mori and Y. Okahata, J. Am. Chem. Soc., 1804. 2004, 126, 14752–14757. 63 A. Beloqui, A. Sanchez-Ruiz, M. Martin-Lomas and N. C. Reichardt, 34 D.-H. Kim, D.-S. Shin and Y.-S. Lee, J. Pept. Sci., 2012, 18, 394–399. Chemical Communications, 2012, 48, 1701–1703. 35 Z. A. Gurard-Levin, K. A. Kilian, J. Kim, K. Bahr and M. Mrksich, Acs 64 J. Kim and M. Mrksich, Nucleic Acids Res., 2010, 38, e2–e2. Chemical Biology, 2010, 5, 863–873. 65 W. S. Yeo and M. Mrksich, Angewandte Chemie-International Edition,

1–32 29 | 2003, 42, 3121–3124. 98 O. J. Barrett, A. Pushechnikov, M. Wu and M. D. Disney, Carbohydr. 66 C. M. Salisbury, D. J. Maly and J. A. Ellman, J. Am. Chem. Soc., 2002, Res., 2008, 343, 2924–2931. 124, 14868–14870. 99 T. Feilner, C. Hultschig, J. Lee, S. Meyer, R. G. H. Immink, A. Koenig, 67 J. J. Diaz-Mochon, L. Bialy and M. Bradley, Chemical Communications, A. Possling, H. Seitz, A. Beveridge, D. Scheel, D. J. Cahill, H. Lehrach, 2006, 3984–3986. J. Kreutzberger and B. Kersten, Molecular & Cellular Proteomics, 2005, 68 M. L. Bulyk, E. Gentalen, D. J. Lockhart and G. M. Church, Nat Biotech, 4, 1558–1568. 1999, 17, 573–577. 100 M. Schutkowski, U. Reimer, S. Panse, L. Dong, J. M. Lizcano, D. R. 69 M. Huber, D. Losert, R. Hiller, C. Harwanegg, M. W. Mueller and W. M. Alessi and J. Schneider-Mergener, Angewandte Chemie International Schmidt, Anal. Biochem., 2001, 299, 24–30. Edition, 2004, 43, 2671–2674. 70 M.-H. Kwon, J.-W. Jung, S.-H. Jung, J.-Y. Park, Y.-M. Kim and K.-S. 101 P. Rathert, A. Dhayalan, M. Murakami, X. Zhang, R. Tamas, R. Ju- Ha, Mol. Cells, 2009, 27, 337–343. rkowska, Y. Komatsu, Y. Shinkai, X. D. Cheng and A. Jeltsch, Nat. 71 O. Blixt, K. Allin, O. Bohorov, X. F. Liu, H. Andersson-Sand, J. Hoff- Chem. Biol., 2008, 4, 344–346. mann and N. Razi, Glycoconjugate J., 2008, 25, 59–68. 102 J. Ptacek, G. Devgan, G. Michaud, H. Zhu, X. W. Zhu, J. Fasolo, H. Guo, 72 X. Han, S. Shigaki, T. Yamaji, G. Yarnanouchi, T. Mori, T. Niidome and G. Jona, A. Breitkreutz, R. Sopko, R. R. McCartney, M. C. Schmidt, Y. Katayama, Anal. Biochem., 2008, 372, 106–115. N. Rachidi, S. J. Lee, A. S. Mah, L. Meng, M. J. R. Stark, D. F. Stern, 73 Y. H. Oh, M. Y. Hong, Z. Jin, T. Lee, M. K. Han, S. Park and H. S. Kim, C. De Virgilio, M. Tyers, B. Andrews, M. Gerstein, B. Schweitzer, P. F. Biosensors & Bioelectronics, 2007, 22, 1260–1267. Predki and M. Snyder, Nature, 2005, 438, 679–684. 74 J. Grognux and J. L. Reymond, Mol. BioSyst., 2006, 2, 492–498. 103 D. Levy, C. L. Liu, Z. Yang, A. M. Newman, A. A. Alizadeh, P. J. Utz and O. Gozani, Epigenetics & Chromatin, 2011, 4, 19–31. 75 N. Ehrlich, K. Anhalt, C. Hubner and S. Brakmann, Anal. Biochem., 2010, 399, 251–256. 104 K. Inamori, M. Kyo, K. Matsukawa, Y. Inoue, T. Sonoda, K. Tatematsu, K. Tanizawa, T. Mori and Y. Katayama, Anal. Chem., 2008, 80, 643– 76 S. Shigaki, T. Yamaji, X. M. Han, G. Yamanouchi, T. Sonoda, O. Okitsu, 650. T. Mori, T. Niidome and Y. Katayama, Anal. Sci., 2007, 23, 271–275. 105 J. Lee and M. T. Bedford, EMBO Rep., 2002, 3, 268–273. 77 S. Park, M. R. Lee, S. J. Pyo and I. Shin, J. Am. Chem. Soc., 2004, 126, 4812–4819. 106 M. Shipp, R. Nadella, H. Gao, V. Farkas, H. Sigrist and A. Faik, Glyco- conjugate J., 2008, 25, 49–58. 78 S. Park and I. Shin, Org. Lett., 2007, 9, 1675–1678. 107 K. Park, J. Ahn, S. Yeon, M. Kim and B. H. Chung, Biochem. Biophys. 79 L. Q. Gao, H. Y. Sun and S. Q. Yao, Biopolymers, 2010, 94, 810–819. Res. Commun., 2008, 368, 684–689. 80 M. Uttamchandani, E. W. S. Chan, G. Y. J. Chen and S. Q. Yao, Bioor- 108 G. J. Wegner, A. W. Wark, H. J. Lee, E. Codner, T. Saeki, S. P. Fang and ganic & Medicinal Chemistry Letters, 2003, 13, 2997–3000. R. M. Corn, Anal. Chem., 2004, 76, 5677–5684. 81 S. Kiyonaka, K. Sada, I. Yoshimura, S. Shinkai, N. Kato and I. Hamachi, 109 K. Inamori, M. Kyo, Y. Nishiya, Y. Inoue, T. Sonoda, E. Kinoshita, Nat. Mater., 2004, 3, 58–64. T. Koike and Y. Katayama, Anal. Chem., 2005, 77, 3979–3985. 82 P. B. Tsitovich, A. Pushechnikov, J. M. French and M. D. Disney, Chem- 110 Z. L. Zhi, N. Laurent, A. K. Powel, R. Karamanska, M. Fais, J. Voglmeir, BioChem, 2010, 11, 1656–1660. A. Wright, J. M. Blackburn, P. R. Crocker, D. A. Russell, S. Flitsch, 83 J. Obro, T. Sorensen, P. Derkx, C. T. Madsen, M. Drews, M. Willer, J. D. R. A. Field and J. E. Turnbull, ChemBioChem, 2008, 9, 1568–1575. Mikkelsen and W. G. T. Willats, Proteomics, 2009, 9, 1861–1868. 111 T. Mori, K. Inamori, Y. Inoue, X. Han, G. Yamanouchi, T. Niidome and 84 Y. Merbl and M. W. Kirschner, Proc. Natl. Acad. Sci. U. S. A., 2009, Y. Katayama, Anal. Biochem., 2008, 375, 223–231. 106, 2543–2548. 112 Y. Inoue, T. Mori, G. Yamanouchi, X. Han, T. Sonoda, T. Niidome and 85 C. M. Loch, C. L. Cuccherini, C. A. Leach and J. E. Strickler, Molecular Y. Katayama, Anal. Biochem., 2008, 375, 147–149. & Cellular Proteomics, 2011, 10, 1–11. 113 C. Corne, J. B. Fiche, D. Gasparutto, V. Cunin, E. Suraniti, A. Buhot, 86 S. Akita, N. Umezawa, N. Kato and T. Higuchi, Bioorganic & Medicinal J. Fuchs, R. Calemczuk, T. Livache and A. Favier, Analyst, 2008, 133, Chemistry, 2008, 16, 7788–7794. 1036–1045. 87 B. T. Houseman, J. H. Huh, S. J. Kron and M. Mrksich, Nat. Biotechnol., 114 C. Cle, A. P. Gunning, K. Syson, L. Bowater, R. A. Field and S. Borne- 2002, 20, 270–274. mann, J. Am. Chem. Soc., 2008, 130, 15234–+. 88 S. Pal, M. J. Kim and J. M. Song, Lab Chip, 2008, 8, 1332–1341. 115 J. Hyun, J. Kim, S. L. Craig and A. Chilkoti, J. Am. Chem. Soc., 2004, 89 Y. P. Kim, Y. H. Oh and H. S. Kim, Biosensors & Bioelectronics, 2008, 126, 4770–4771. 23, 980–986. 116 L. L. Foose, H. W. Blanch and C. J. Radke, J. Biotechnol., 2007, 132, 90 J. Seibel, H. Hellmuth, B. Hofer, A. M. Kicinska and B. Schmalbruch, 32–37. ChemBioChem, 2006, 7, 310–320. 117 Y. Cao, J. Wang, Y. Y. Xu and G. X. Li, Biosensors & Bioelectronics, 91 L. S. Khraltsova, M. A. Sablina, T. D. Melikhova, D. H. Joziasse, 2010, 26, 87–91. H. Kaltner, H. J. Gabius and N. V. Bovin, Anal. Biochem., 2000, 280, 118 S. Nayak, W. S. Yeo and M. Mrksich, Langmuir, 2007, 23, 5578–5583. 250–257. 119 H. F. Song, K. Kerman and H. B. Kraatz, Chemical Communications, 92 M. C. Bryan, L. V. Lee and C. H. Wong, Bioorganic & Medicinal Chem- 2008, 502–504. istry Letters, 2004, 14, 3185–3188. 120 P. Miao, L. M. Ning, X. X. Li, P. F. Li and G. X. Li, Bioconjugate Chem., 93 M. Meldal, Biopolymers, 2002, 66, 93–100. 2012, 23, 141–145. 94 M. Kohn, M. Gutierrez-Rodriguez, P. Jonkheijm, S. Wetzel, R. Wacker, 121 M.-M. Be, N. Coursol, B. Duchmein, E. Browne, V. Chechev, R. Helmer H. Schroeder, H. Prinz, C. M. Niemeyer, R. Breinbauer, S. E. Szedlacsek and E. Sch¨onfeld, LNHB report, 2008, 1–850. and H. Waldmann, Angewandte Chemie-International Edition, 2007, 46, 122 M. Schena, Protein Microarray, Jones and Bartlett, 2005. 7700–7703. 123 J. Albala and I. Humphery-Smith, Protein Arrays, Biochips and Pro- 95 D. Dressman, H. Yan, G. Traverso, K. W. Kinzler and B. Vogelstein, teomics: The Next Phase of Genomic Discovery, Marcel Dekker, Incor- Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 8817–8822. porated, 2003. 96 A. Murakawa, T. Mori and Y. Okahata, Chem. Lett., 2007, 36, 312–313. 124 O. Kosik and V. Farkas, Anal. Biochem., 2008, 375, 232–236. 97 H. Nishino, T. Nihira, T. Mori and Y. Okahata, J. Am. Chem. Soc., 2004, 125 P. T. Charles, V. R. Stubbs, C. M. Soto, B. D. Martin, B. J. White and 126, 2264–2265.

30 1–32 | C. R. Taitt, Sensors, 2009, 9, 645–655. S. S. Shevkoplyas and G. M. Whitesides, J. Am. Chem. Soc., 2012, 134, 126 R. Chapman, E. Ostuni, L. Yan and G. Whitesides, Langmuir, 2000, 16, 5637–5646. 6927–6936. 161 E. Blalock, A Beginner’s Guide to Microarrays, Kluwer Academic Pub- 127 D. Kambhampati, Protein Microarray Technology, John Wiley & Sons, lishers, 2003. 2006. 162 Bio-Rad, A Demonstration of the Accuracy and Precision of BioOdyssey 128 O. Melnyk, X. Duburcq, C. Olivier, F. Urbs, C. Auriault and H. Gras- MCP Pins Used With the BioOdyssey Calligrapher MiniArrayer; Bul- Masse, Bioconjugate Chem., 2002, 13, 713–720. letin, 2012, tech note 5559, 1–6. 129 C. W. Tornoe, C. Christensen and M. Meldal, J. Org. Chem., 2002, 67, 163 H. Zhu, M. Bilgin, R. Bangham, D. Hall, A. Casamayor, P. Bertone, 3057–3064. N. Lan, R. Jansen, S. Bidlingmaier, T. Houfek, T. Mitchell, P. Miller, 130 M. Mazan, E. Ragni, L. Popolo and V. Farkas, Biochem. J., 2011, 438, R. A. Dean, M. Gerstein and M. Snyder, Science, 2001, 293, 2101–2105. 275–282. 164 J. Y. Lu, Y. Y. Lin, J. Oian, S. C. Tao, J. Zhu, C. Pickart and H. Zhu, 131 J. P. Bressler and G. W. Goldstein, Biochem. Pharmacol., 1991, 41, 479– Molecular & Cellular Proteomics, 2008, 7, 35–45. 484. 165 L. L. Sun, D. J. Liu and Z. X. Wang, Anal. Chem., 2007, 79, 773–777. 132 K. E. Sapsford, L. Berti and I. L. Medintz, Angewandte Chemie- 166 R. Pasternack and P. Collings, Science, 1995, 269, 935–939. International Edition, 2006, 45, 4562–4588. 167 B. A. Stillman and J. L. Tonkinson, BioTechniques, 2000, 29, 630–+. 133 G. T. Noble, F. L. Craven, J. Voglmeir, R. Sardzk, S. L. Flitsch and S. J. 168 L. T. Yin, C. Y. Hu and C. H. Chang, Sensors and Actuators B-Chemical, Webb, J. Am. Chem. Soc., 2012, 134, 13010–13017. 2008, 130, 374–378. 134 B. T. Houseman and M. Mrksich, Angewandte Chemie-International 169 R. Frank, Tetrahedron, 1992, 48, 9217–9232. Edition, 1999, 38, 782–785. 170 R. Volkmer, ChemBioChem, 2009, 10, 1431–1442. 135 T. Shimomura, X. M. Han, A. Hata, T. Niidome, T. Mori and 171 X. Espanel and R. H. van Huijsduijnen, Methods, 2005, 35, 64–72. Y. Katayama, Anal. Sci., 2011, 27, 13–17. 172 A. Y. Rubina, A. Kolchinsky, A. A. Makarov and A. S. Zasedatelev, 136 E. Kinoshita, A. Yamada, H. Takeda, E. Kinoshita-Kikuta and T. Koike, Proteomics, 2008, 8, 817–831. J. Sep. Sci., 2005, 28, 155–162. 173 S. M. Patrie and M. Mrksich, Anal. Chem., 2007, 79, 5878–5887. 137 R. Compton and C. Banks, Understanding Voltammetry, Imperial Col- 174 B. T. Houseman and M. Mrksich, Chemistry & Biology, 2002, 9, 443– lege Press, 2011. 454. 138 E. Ahrne, M. Muller and F. Lisacek, Proteomics, 2010, 10, 671–686. 175 J. I. Kim, S. Kim and W. S. Yeo, Biochip Journal, 2011, 5, 199–205. 139 M. Mann, S.-E. Ong, M. Grnborg, H. Steen, O. N. Jensen and A. Pandey, 176 J. Turkevich, P. C. Stevenson and J. Hillier, Discussions of the Faraday Trends Biotechnol., 2002, 20, 261–268. Society, 1951, 55–&. 140 H. Johnson and C. E. Eyers, Methods Mol Biol., 2010, 658, 93–108. 177 J. Kimling, M. Maier, B. Okenve, V. Kotaidis, H. Ballot and A. Plech, J. 141 M. Davies and M. Bradley, Tetrahedron Lett., 1997, 38, 8565–8568. Phys. Chem. B, 2006, 110, 15700–15707. 142 M. Karas, M. Glckmann and J. Schfer, J Mass Spectrom., 2000, 35, 1– 178 K. Kneipp, Y. Wang, H. Kneipp, L. T. Perelman, I. Itzkan, R. Dasari and 12. M. S. Feld, Phys. Rev. Lett., 1997, 78, 1667–1670. 143 J. R. Yates, J. Mass Spectrom., 1998, 33, 1–19. 179 F. Schindler, P. P. Lajolo, H. Pinczowski, F. L. A. Fonseca, A. Barbieri, 144 J. Watson and O. Sparkman, Introduction to mass spectrometry: instru- L. H. Massonetto, F. T. Katto and A. Del Giglio, European Journal of mentation, applications and strategies for data interpretation, John Wi- Cancer Care, 2008, 17, 152–156. ley & Sons, 2007. 180 J. Kress, R. Zanaletti, A. Amour, M. Ladlow, J. G. Frey and M. Bradley, 145 M. Mann, R. C. Hendrickson and A. Pandey, Annu. Rev. Biochem., 2001, Chemistry-a European Journal, 2002, 8, 3769–3772. 70, 437–473. 181 A. Basso, R. V. Ulijn, S. L. Flitsch, G. Margetts, I. Brazendale, C. Ebert, 146 T. Kinumi, H. Niwa and H. Matsumoto, Anal Biochem., 2000, 277, 177– L. De Martin, P. Linda, S. Verdelli and L. Gardossi, Tetrahedron, 2004, 186. 60, 589–594. 147 A. J. Ibanez, A. Muck and A. Svatos, J. Mass Spectrom., 2007, 42, 634– 182 B. J. Egner, S. Rana, H. Smith, N. Bouloc, J. G. Frey, W. S. Brocklesby 640. and M. Bradley, Chemical Communications, 1997, 735–736. 148 A. Chipman, Nature, 2007, 449, 131–131. 183 L. Zou, H. L. Pang, P. H. Chan, Z. S. Huang, L. Q. Gu and K. Y. Wong, 149 D. H. Min, J. Su and M. Mrksich, Angewandte Chemie-International Analyst, 2008, 133, 1195–1200. Edition, 2004, 43, 5973–5977. 184 M. Meldal, F. I. Auzanneau, O. Hindsgaul and M. M. Palcic, Journal of 150 J. C. Love, L. A. Estroff, J. K. Kriebel, R. G. Nuzzo and G. M. White- the Chemical Society-Chemical Communications, 1994, 1849–1850. sides, Chemical Reviews, 2005, 105, 1103–1169. 185 G. MacBeath, Nat. Genet., 2002, 32, 526–532. 151 N. Thornberry and Y. Lazebnik, Science, 1998, 281, 1312–1316. 186 M. Salim, S. L. McArthur, S. Vaidyanathan and P. C. Wright, Mol. 152 G. Steiner, Anal. Bioanal. Chem., 2004, 379, 328–331. BioSyst., 2011, 7, 101–115. 153 S. Scarano, M. Mascini, A. P. F. Turner and M. Minunni, Biosensors & 187 N. Patel, M. C. Davies, M. Hartshorne, R. J. Heaton, C. J. Roberts, Bioelectronics, 2010, 25, 957–966. S. J. B. Tendler and P. M. Williams, Langmuir, 1997, 13, 6485–6490. 154 P. Braga and D. Ricci, Atomic force microscopy: biomedical methods 188 T. Viitala, I. Vikholm and J. Peltonen, Langmuir, 2000, 16, 4953–4961. and applications, Humana Press, 2004. 189 C. Mateo, R. Torres, G. Fernandez-Lorente, C. Ortiz, M. Fuentes, A. Hi- 155 T. Ando, Nanotechnology, 2012, 23, 062001. dalgo, F. Lopez-Gallego, O. Abian, J. M. Palomo, L. Betancor, B. C. C. 156 G. Sauerbrey, Zeitschrift Fur Physik, 1959, 155, 206–222. Pessela, J. M. Guisan and R. Fernandez-Lafuente, Biomacromolecules, 157 E. Casero, L. Vazquez, A. M. Parra-Alfambra and E. Lorenzo, Analyst, 2003, 4, 772–777. 2010, 135, 1878–1903. 190 F. Rusmini, Z. Y. Zhong and J. Feijen, Biomacromolecules, 2007, 8, 158 D. Barcelo and P. Hansen, Biosensors; Indicator Assays and Chem- 1775–1789. ical Methods for Endocrine Disrupting Compounds in Wastewaters, 191 R. Fernandez-Lafuente, C. M. Rosell, V. Rodriguez, C. Santana, Springer, 2009. G. Soler, A. Bastida and J. M. Guisan, Enzyme Microb. Technol., 1993, 159 B. Becker and M. A. Cooper, J. Mol. Recognit., 2011, 24, 754–787. 15, 546–550. 160 N. D. Shapiro, K. A. Mirica, S. Soh, S. T. Phillips, O. Taran, C. R. Mace, 192 A. D. de Araujo, J. M. Palomo, J. Cramer, M. Kohn, H. Schroder,

1–32 31 | R. Wacker, C. Niemeyer, K. Alexandrov and H. Waldmann, Angewandte Chemie-International Edition, 2006, 45, 296–301. 193 B. P. Duckworth, J. H. Xu, T. A. Taton, A. Guo and M. D. Distefano, Bioconjugate Chem., 2006, 17, 967–974. 194 N. Nath, R. Hurst, B. Hook, P. Meisenheimer, K. Q. Zhao, N. Nassif, R. F. Bulleit and D. R. Storts, J. Proteome Res., 2008, 7, 4475–4482. 195 D. Gosalia, C. Salisbury, J. Ellman and S. Diamond, Mol. Cell. Pro- teomics, 2005, 4, 626–636. 196 D. H. Altreuter, J. S. Dordick and D. S. Clark, Biotechnol. Bioeng., 2003, 81, 809–817. 197 M.-A. Alouini, E.-F. Moustoifa, S. A. Rubio, A. Bartegi, T. Berthelot and G. Deleris, Anal. Bioanal. Chem., 2012, 403, 185–194. 198 D.-H. Kong, S.-H. Jung, S.-T. Lee, Y.-M. Kim and K.-S. Ha, Biosens Bioelectron, 2012, 36, 147–153. 199 S. J. Lee and S. Y. Lee, Anal. Biochem., 2004, 330, 311–316. 200 E. Snir, J. Joore, P. Timmerman and S. Yitzchaik, Langmuir, 2011, 27, 11212–11221. 201 R. Amanchy, J. Zhong, H. Molina, R. Chaerkady, A. Iwahori, D. E. Kalume, M. Gronborg, J. Joore, L. Cope and A. Pandey, J. Proteome Res., 2008, 7, 3900–3910. 202 J. M. Lizcano, M. Deak, N. Morrice, A. Kieloch, C. J. Hastie, L. Y. Dong, M. Schutkowski, U. Reimer and D. R. Alessi, J. Biol. Chem., 2002, 277, 27839–27849. 203 R. Gupta, B. Kus, C. Fladd, J. Wasmuth, R. Tonikian, S. Sidhu, N. J. Krogan, J. Parkinson and D. Rotin, Mol. Syst. Biol., 2007, 3, 1–12. 204 J. Rush, A. Moritz, K. A. Lee, A. Guo, V. L. Goss, E. J. Spek, H. Zhang, X. M. Zha, R. D. Polakiewicz and M. J. Comb, Nat. Biotechnol., 2005, 23, 94–101. 205 G. Tesco, Y. H. Koh and Rudolph E. Tanzi, J. Biol. Chem., 2003, 278, 46074–46080. 206 C. Lombard, J. Saulnier and J. M. Wallach, Protein Pept. Lett., 2005, 12, 621–9. 207 F. Guzman, S. Barberis and A. A. Illanes, Electronic Journal of Biotech- nology, 2007, 10, 279–314. 208 C. Walsh, Posttranslational Modification Of Proteins: Expanding Na- ture’s Inventory, Roberts and Company Publishers, 2006. 209 J. Su and M. Mrksich, Angewandte Chemie-International Edition, 2002, 41, 4715–4718. 210 T. R. Northen, J. C. Lee, L. Hoang, J. Raymond, D. R. Hwang, S. M. Yannone, C. H. Wong and G. Siuzdak, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 3678–3683. 211 J. F. Tolborg, L. Petersen, K. J. Jensen, C. Mayer, D. L. Jakeman, R. A. J. Warren and S. G. Withers, J. Org. Chem., 2002, 67, 4143–4149. 212 T. Sasaki, T. R. Noel and S. G. Rino, J. Agric. Food Chem., 2008, 56, 1091–1096. 213 T. Yoshida-Moriguchi, L. P. Yu, S. H. Stalnaker, S. Davis, S. Kunz, M. Madson, M. B. A. Oldstone, H. Schachter, L. Wells and K. P. Camp- bell, Science, 2010, 327, 88–92. 214 J. Voglmeir, S. Kaloo, N. Laurent, M. M. Meloni, L. Bohlmann, I. B. H. Wilson and S. L. Flitsch, Biochem. J., 2011, 436, 447–455. 215 C. H. Wong, Y. Ichikawa, T. Krach, C. Gautheron-Le Narvor, D. P. Du- mas and G. C. Look, J. Am. Chem. Soc., 1991, 113, 8137–8145. 216 G. L. Côtéand J. F. Robyt, Carbohydr. Res., 1982, 101, 57–74. 217 M. L. B. Magalhaes and J. S. Blanchard, Biochemistry, 2005, 44, 16275– 16283. 218 D. Zechel and S. Withers, Accounts Chem. Res., 2000, 33, 11–18. 219 Miyakem Yukio and Ebata Mitsuo, Agricultural and biological chemistry, 1988, 52, 1649–1654.

32 1–32 | CHAPTER THREE

ENZYMATIC GLYCOSYLATIONS ON ARRAYS

Printed with Permission by Mary Ann Liebert, Inc.,

J. Voglmeir, R. Sardzik, M. J. Weissenborn, S. L. Flitsch, Enzymatic Glycosylations on Arrays, Omics 2010, 14, 4, 437–444.

J. Voglmeir, S. L. Flitsch and M. J. Weissenborn wrote the review. R. Sardzik provided help with the graphics and proof read the manuscript.

12 OMICS A Journal of Integrative Biology Volume 14, Number 4, 2010 ª Mary Ann Liebert, Inc. DOI: 10.1089/omi.2010.0035

Enzymatic Glycosylations on Arrays

Josef Voglmeir, Robert Sˇ ardzı´k, Martin J. Weissenborn, and Sabine L. Flitsch

Abstract

The enzymatic glycosylation of microarrays is a relatively young ﬁeld in glycoscience. Platforms developed from other array technologies (e.g., proteins and nucleic acids) were successfully adopted in several proof-of-principle studies as a high-throughput tool for the generation of more complex carbohydrate structures using carbohydrate-processing enzymes. These arrays and the developed on-chip enzymatic glycosylation methodologies are reviewed in this article.

Introduction measurements, but they have also the advantage of using less quantities of precious carbohydrate material due to minia- nlike nucleic acids and protein sequences, the se- turization that is not possible in solution studies. Although Uquences of carbohydrates are not directly encoded in the many advances have been made in recent years to isolate and genome, but are determined by the expression and specificity synthesize oligosaccharides and glycoconjugates chemically of a set of ‘‘glycoenzymes’’ such as glycosyltransferases and or chemoenzymatically (Blixt and Razi, 2006; Davis, 2000; glycosidases, which are responsible for the biosynthesis and Nicolaou and Mitchell, 2001; Seeberger and Haase, 2000; processing of carbohydrate structures. It has been estimated Zhang et al., 1999), the quantities obtained are mostly in the that around 2% of the genome is dedicated to such gly- low milligram scale, and therefore too small for extensive tests coenzymes. Given that such enzymes can have many in conventional biochemical studies. The use of carbohydrate (glyco)protein and glycolipid substrates, the impact of gly- arrays to study carbohydrate–lectin/antibody binding has cosylation is much larger; for example, it has been estimated allowed to reduce the amount of immobilized carbohydrates that over 70% of all proteins in humans are glycosylated. to a femtogram scale, compared to the picogram scale Because each glycoenzyme has many different lipid and required for ELISA-based screening or nanogram scale in protein substrates, many of the ‘‘genetic’’ tools currently used immunodot assays (Willats et al., 2002). in proteomics such as knock-out strains, iRNA, or two-hybrid The present review will focus on the application of gly- systems are not as informative for glycomics as in the other coarrays to study enzymatic reactions involving Carbohy- ‘‘omic’’ disciplines. Most successful in the glycosciences have drate Active Enzymes, in particular, glycosyltransferases and been biochemical and biophysical tools such as mass spec- glycosidases. These enzymes form a significant part of the trometry for the determination of carbohydrate structures, proteome of higher organisms and their sequenced and bio- enzymology of the glycoenzymes to determine their substrate chemical data are well recorded on the Carbohydrate Active specificity, and hence, predict glycan structures, and the ap- Enzymes database (http://www.cazy.org/) (Cantarel et al., plications of the lectins and glycoenzymes themselves as tools 2009). to generate and characterize glycan structures. Such biochemical studies have been slow and have given us very Glycosylation Reactions on Array Platforms valuable yet patchy information of protein–carbohydrate interactions. In recent times, glycoscientists have therefore The specificity of enzymes involved in formation or identified carbohydrate microarrays (glycoarrays) as the key hydrolysis of glycosidic linkages (Fig. 1) relates to three tool for the high-throughput studies that are necessary to components of the glycosidic bond: specificity for the carbo- understand this complex area of biology. hydrate glycoside (glycosyl donor), stereospecificity for alpha The first generation of glycoarrays have been used to or beta linkages and specificity for the aglycon component identify carbohydrate–protein binding interactions, and this (glycosyl acceptor in the synthesis direction). Although genes application of glycoarrays has been reviewed elsewhere corresponding to glycoenzymes can be identified in genomic (Laurent et al., 2008b; Park et al., 2008). These studies have databases based on consensus sequences and homology to shown that carbohydrate arrays are particularly useful tools known enzymes, the bioinformatic tools for predictions of for glycomics: not only do they shorten the time of biological selectivity of a given enzyme are currently not accurate, and

Manchester Interdisciplinary Biocentre, University of Manchester, Manchester M1 7ND, UK.

437 438 VOGLMEIR ET AL.

human a2,3-sialyltransferase IV (hST3Gal-IV) had a broad acceptor specificity for Galb1,3/4GlcNAc and Galb1,3GalNAc structures, and porcine a2,3-sialyltransferase I (pST3Gal-I) also accepted ganglio-oligosaccharides and core-2 structures. Faik and coworkers (Shipp et al., 2008) have generated a plant cell wall glycoarray platform for analyzing both enzymes as well as acceptors involved in cell wall biosynthesis. Their work has focused on fucosyltransferase 1 (AtFUT1). Chemically derivatized cello-oligosaccharides were attached 14 FIG. 1. Substrate specificity of glycoenzymes for acceptor, onto photoactivatable glass slides. Using [ C]GDP-Fuc as a donor, and glycosidic linkage using galactosylation as an donor substrate, fucosylation of the immobilized acceptor example. was detected with a phosphoimager scanner. Interestingly, the fucosylation of some of the smaller acceptors, such of the xyloglucan trimer, strongly depends on the anchoring on the biochemical data for substrate specificity are required. Given array surface. Experiments showed that no fucosylation was the complexity of substrates for such enzymatic reactions and detected when the trimer was randomly immobilized on the the number of enzymes that need to be investigated, high- chip without covalent attachment. throughput array platforms are considered key tools for the Chevolot et al. (2001) investigated the application of ga- determination of substrate specificity. lactose and lactose-modified polystyrene (PS) surfaces. Their Array experiments involve the preparation of defined car- immobilization technology is based on a photochemical carbohydrate arrays that can be interrogated by a glycosyl- bene formation and a subsequent reaction with the polymeric transferase for acceptor specificity in the presence of defined solid support (Fig. 4). Lactose modified with a linker con- glycosyl donors or be examined as a substrate for glycosi- taining a light sensitive aryl diazirine group was prepared dases. Table 1 presents a list of enzymatic reactions that have chemically and attached by photoactivation onto the PS sur- been studied so far. This list shows that a diversity of enzymes face. The surfaces were characterized by X-ray photoelectron and array platforms have been used. spectroscopy (XPS) and time of flight secondary ion mass Except for the glycosyltransferase R (GTFR) used in the spectrometry (ToF-SIMS). In addition to hybridization ex- work of Seibel et al. (2006), which is a non-Leloir-type glu- periments using rat hepatocytes and Allo A lectin, an enzy- cosyltransferase from Streptococcus oralis, all listed glycosyl- matic solid-phase glycosylation reaction of the lactose transferases have used nucleotide sugars as activated donor modified PS surface was investigated. Using recombinant

substrates. Array technologies described are diverse and have a2,6-sialyltransferase expressed in Pichia pastoris allowed the generally been adopted from other array platforms, including incorporation of radioactive Neu5Ac from [14C]-labeled CMP- glass slides, microtiter plates, and self-assembled monolayers Neu5Ac at linear conversion rates for up to 4 h reaction time. (SAMs) on gold. The choice of readout technologies is also After 6 h incubation time, enzymatic sialylations gave up to diverse and presents a challenge in glycosylation reactions, 40% product formation. because carbohydrates do not generally contain chromo- phores. In the following we will discuss glycan arrays ac- Detection of Enzymatic Transformations Using cording to three general detection methods used: detection of Lectins and Antibodies glycosylation by using labeled glycosyl donors, by using lectins and antibodies for product detection, and label-free A limitation of using labeled substrate donors as described methods using mass spectrometry. in the last section is the limited availability of such compounds. Biotinylation is tolerated for sialyltransferases but might not be tolerated by other enzymes. An alternative way Detection of Glycosylation Reactions of detecting glycan structures is through well-established Using Labeled Glycosyl Donors characterization using lectins and carbohydrate-binding an- Figure 2 outlines the general strategy of using labeled tibodies (Fig. 5). Such detection methods are again limited by glycosyl donors to monitor glycosylating enzymes. Blixt et al. the availability of suitable lectins, but have the advantage that (2008b) have utilized array technology to evaluate the sub- lectins are often linkage specific, and thus acceptor and donor strate specificity of various sialyltransferases (Fig. 3). A glycan specificity as well as regio- and stereoselectivity of the gly- platform based on glass slides containing more than 200 po- cosidic bond can be examined. tential sialyltransferase acceptor substrates was incubated The binding of lectins and antibodies to glycan arrays can with recombinant mammalian sialyltransferases and CMP- be monitored in different ways. An increasingly popular gly- Neu5Ac as donor substrate, which was biotinylated at the coarray platform displays carbohydrates on self-assembled position 9 of the Neu5Ac (CMP-9Biot-Neu5Ac). The transfer monolayers on gold allowing for the detection of protein of the biotinylated sialic acids onto the microarray could be binding by surface plasmon resonance (SPR). Houseman and detected using fluorescein labeled streptavidin. Each enzyme Mrksich (1999) have applied SPR to characterize the binding showed a specific glycosylation profile, and in several of two lectins specific for GlcNAc and LacNAc (B. simplicifolia cases new specificities could be revealed. For example, lectin II, E. cristagalli lectin) to determine glycosylation effi- it was demonstrated that the human a2,6-sialyltransferase I ciencies of bovine b1,4-galactosyltransferase I. SPR has al- (hST6Gal-I) has the ability to accept structures terminating lowed them to obtain quantitative binding data, and they with GlcNAcb1,4GlcNAc. The rat a2,3-sialyltransferase have described experiments using different enzyme concen- (rST3Gal-III) also allows fucosylated acceptors such as Lewisa, trations, ligand densities on the array, and kinetic studies. Cle´

Table 1. Applications of Enzymatic Glycosylation Reactions Performed on Various Array Platforms

Enzyme Substrate Acceptor Product Platform Detection Ref.

Streptococcus oralis Sucrose Maltose, aliphatic primary Panose, glucoside Amin coated 96-well plate FITC-ConA (Seibel et al., 2006) Glycosyltransferase R (GTFR) alcohol L. mesenteroides 6-a- Sucrose Carboxymethylated alternan Carboxymethylated SPR (Cle et al., 2008) glucosyltransferase a-1,6 dextran dextran on gold surface (Alternansucrase) Bovine b1,4-galactosyl- UDP-Gal/ GlcNAc LacNAc SAMs-coated gold surface Radioactive Gal, SPR (Houseman and Mrksich, transferase (GalT1) [14 C]UDP-Gal 1999) UDP-Gal GlcNAc LacNAc SAMs-coated gold surface Rhodamine-labeled (Houseman and Mrksich, E. cristagalli or 2002) B. simplicifolia I UDP-Gal GlcNAc LacNAc Thiol-coated glass slides Cy5-EC (Park et al., 2004) UDP-Gal GlcNAc LacNAc SAMs-coated gold surface MALDI-ToF-MS (Su and Mrksich, 2002) UDP-Gal GlcNAc-peptide LacNAc-peptide SAMs-coated gold surface MALDI-ToF-MS (Laurent et al., 2008a) UDP-Gal 9 different mono- and 9 different galactosylated mono SAMs-coated gold surface MALDI-ToF-MS (Zhi et al., 2008) disaccharides and disaccharides UDP-Gal b-1,2;1,3;1,4;1,6-Disacchar- b-1,4-galactosylated SAMs-coated gold surface MALDI-ToF-MS (Ban and Mrksich, 2008) ides (Glc-Gal, Glc-Glc, b-1,2;1,3;1,4;1,6-Disaccharides GlcNAc-Gal, GlcNAc- (Glc-Gal, Glc-Glc, GlcNAc-Gal, Glc GlcNAc-Glc) UDP-Gala b-1,2;1,3;1,4;1,6-Disacchar- b-1,4-galactosylated SAMs-coated gold surface MALDI-ToF-MS (Ban and Mrksich, 2008) ides (Glc-Gal, Glc-Glc, b-1,2;1,3;1,4;1,6-Disaccharides GlcNAc-Gal, GlcNAc- (Glc-Gal, Glc-Glc, GlcNAc-Gal, Glc GlcNAc-Glc) Bovine b1,4-galactosyltransfer- UDP-Gal 20 different mono-, Galactosylated mono-, Epoxide coated glass slides Cy5-PA-1L, Cy5-BS-I, (Park and Shin, 2007)

439 ase (GalT1) di- and trisaccharides di- and trisaccharides Cy3-RCA120 Porcine a2,3-sialyl CMP-Neu5Ac LacNAc Neu5Aca2,3LacNAc Thiol-coated glass slides Cy5-ECb (Park et al., 2004) transferase (SialT) Human a1,3-fucosyl GDP-Fuc Neu5Aca2,3LacNAc Sialyl LeX Thiol-coated glass slides anti-sialyl LeX (Park et al., 2004) transferase (FucT) Human a1,3-fucosyl GDP-Fuc LacNAc LeX Lipid-alkyne activated Peroxidase conjugated (Bryan 2004, Fazio 2002) transferase IV polystyrene zTPL (FucT-IV) Galactosidase — LacNAc GlcNAc SAMs-coated gold surface MALDI-ToF-MS (Su and Mrksich, 2002) Recombinant a2,6-sialyl- [14C]CMP-Neu5Ac, Lactose Neu5Aca2,6Lac Lipid-alkyne activated Radioactive Neu5Ac (Chevolot et al., 2001) transferase (ST6Gal) CMP-Neu5Ac polystyrene from Pichia pastoris Recombinant Arabidopsis GDP-Fuc / Tamarind xyloglucan a(1,2)fucosylated tamarind Photoactivable glass slides Radioactive Fuc (Shipp et al., 2008) fucosyltransferase 1 [14 C]GDP-Fuc polymer and trimer xyloglucan polymer and trimer (AtFUT1) Human ppGalNAcT2 UDP-GalNAc 32 mucin–type peptides Tn-antigens SAMs-coated gold surface MALDI-ToF-MS (Laurent et al., 2008a) UDP-GalNAc 10 different amino acid 9 glycosylated amino acid SAMs-coated gold surface MALDI-ToF-MS (Laurent et al., 2008c) sequences sequences Human a2,6-sialyltransferase I CMP-9Biot-Neu5Ac >200 glycan structures >200 sialylated glycan structures NHS activated glass slides Alexa488-Streptavidin (Blixt et al., 2008a) (hST6Gal-I) Porcine a2,3-sialyltransferase I CMP-9Biot-Neu5Ac >200 glycan structures >200 sialylated glycan structures NHS activated glass slides Alexa488-Streptavidin (Blixt et al., 2008a) (pST3Gal-I) Human a2,6-sialyltransferase CMP-9Biot-Neu5Ac >200 glycan structures >200 sialylated glycan structures NHS activated glass slides Alexa488-Streptavidin (Blixt et al., 2008a) I (hST6GalNAc I) Human a2,3-sialyltransferase CMP-9Biot-Neu5Ac >200 glycan structures >200 sialylated glycan structures NHS activated glass slides Alexa488-Streptavidin (Blixt et al., 2008a) IV (hST3Gal-IV) Rat a2,3-sialyltransferase CMP-9Biot-Neu5Ac >200 glycan structures >200 sialylated glycan structures NHS activated glass slides Alexa488-Streptavidin (Blixt et al., 2008a) (rST3Gal-III)

aIn the presence of lactalbumin; if not other indicated always in the absence of lactalbumin. bNo signal indicates sialic acid on Gal. 440 VOGLMEIR ET AL.

FIG. 2. Detection of enzyme activity on arrays using labelled glycosyl donors. et al. (2008) have demonstrated the enzymatic transfer of glucose units to carboxymethylated dextran immobilized on gold arrays using real time monitoring by SPR spectroscopy in a label-free manner. Glucose was attached onto the immobilized carbohydrates using an a1,6-glucosyltransferase from Leuconostoc mesenteroides and sucrose as donor substrate. Besides the determination of the kM value of the enzyme for sucrose, a linear relationship between increase of glucose transfer and enzyme concentration was demonstrated. Fur- thermore, the turnover numbers of the enzyme on the surface were estimated by the rate of polymer synthesis. The bovine b1,4-galactosyltransferase I (GalT1) has been a popular enzyme for proof-of-concept studies on different platforms, and there has generally been good agreements between methods with respect to acceptor specificities (Houseman and Mrksich, 2002; Zhi et al., 2008). The product can be detected using fluorescently labeled lectins even on gold surfaces, where fluorescence quenching can reduce sensitivity (Houseman and Mrksich, 2002; Zhi er al., 2006). Park and Shin (2007) have developed a glycoarray platform for the screening of b1,4-galactosyltransferase on epoxide- coated glass slides. They attached 20 different mono-, di-, and trisaccharides to the glass surface, performed an enzymatic reaction and detected the product with a fluorescence marked lectin assay (Park and Shin, 2007). GalT1 showed higher ac- FIG. 4. Enzymatic modification of immobilized diazirine tivity with b-GlcNAc than a-GlcNAc. linked lactose by rat liver a2,6-sialyltransferase using radi- olabelled CMP-Neu5Ac donor substrate (Chevolot, 2001).

The most complex glycan structure enzymatically synthesized on solid support so far is sialyl Lex (Park et al., 2004; Shin et al., 2005). Park et al. synthesized this target by immobilizing maleimide-derivatized b-GlcNAc onto thiol-modified glass slides, followed by a series of consecutive enzymatic reactions using three recombinant glycosyltransferases (Fig. 6). To generate immobilized LacNAc, microspots containing immobilized GlcNAc where incubated with a reaction mixture containing bovine b1,4-galactosyltransferase (GalT1) and UDP-galactose as a donor substrate. After rinsing of the solid support, this structure was further incubated with a2,3-sialyltransferase from porcine liver (SiaT) and CMP-Neu5Ac in an overnight reaction to generate immobilized Neu5Aca2,3- LacNAc. Finally, this spots were incubated with a human a1,3-fucosyltransferase (FucT) in the presence of GDP-Fuc, to generate the final sialyl Lex tetrasaccharide immobilized on FIG. 3. Glycoarrays are used to monitor the substrate the chip. After each step the enzymatic elongation was mon- specificity of a range of alpha-sialyltransferases. The glyco- itored using fluorescence labelled lectins or antibodies specific syldonor is labeled as a biotin derivative (Blixt, 2006). for the generated glycan (b-GlcNAc was probed with T. vul- ENZYMATIC GLYCOSYLATIONS ON ARRAYS 441

FIG. 5. Detection of enzymatic activity using labeled lectins or antibodies.

garis lectin (Cy3-TV), LacNAc with E. cristagalli lectin (Cy5- of fluorescence using the FITC-labeled mannose-/glucose- EC), sialylation by loss of binding of Cy5-EC, and sialyl Lex by specific lectin Concanavalin A (ConA). For the first time, en- using sequentially mouse anti-sialyl Lex- and goat Cy5-anti- zymatic glycosylation of a primary aliphatic alcohol been antibody). A similar approach to generate this structure on reported using microarray technology. The probing of the arrays was demonstrated by Fazio et al. (2002). In this case, enzymatic glycosylation is based on an increase of the fluo- the sialylated trisaccharide was immobilized and only the last rescence signal for the product formed compared to the signal fucosylation step was carried out directly on the array surface. of the immobilized acceptors. When maltose, which is known Seibel et al. (2006) demonstrated new acceptor specificities to bind to ConA, was a starting material, an increase of fluo- of the glycosyltransferase R (GTFR) from Streptococcus oralis rescence after the glycosylation reaction was observed. This is (E.C. 2.4.1.5) using a microarray platform presenting im- in agreement with earlier reports by Goldstein et al. (1965), mobilized carbohydrates and aliphatic alcohols as substrates. who described a twofold increase of ConA binding specificity The covalent attachment of the compounds was performed by of the trisaccharide compared to maltose. In contrast, no 1,3-dipolar cycloaddition of azidoalkyl glycosides with pro- glycosylation was observed on immobilized lactose. To pyonic acid followed by NHS/EDC activation and coupling identify the products further studies were performed in so- to amino-coated 96-well microtiter plates (Fig. 7) (Fazio et al., lution with maltose, various primary alcohols as well as serine 2002). In contrast to other glycosyltransferases tested on mi- and threonine derivatives. croarray platforms so far (see Table 1), GTFR is a non-Leloir- Array technology has also been used as a tool for screening type enzyme, and therefore utilizes nonactivated substrates libraries of potential glycosyltransferase inhibitors. The group like sucrose. This enzyme allows the direct transfer of a glu- of Wong has studied the effect of known (Fazio et al., 2002) cose unit from sucrose onto the immobilized acceptors. The and novel (Bryan et al., 2004) inhibitors of a human a1,3- glycosylation of acceptors was determined by measuring fucosyltransferase IV (FucT-IV). All of the 86 tested compounds contained a guanosine diphosphate (GDP) moiety and a hydrophobic part, which should help to enhance the affinity toward a hydrophobic binding pocket close to the active site of the enzyme. The screening platform was based on lipid- alkyne coated microtiter plates and LacNAc as acceptor substrate was immobilized via Cu(I)-catalyzed 1,3-dipolar cycloaddition on the surface (Fig. 7). The enzymatic transfer of fucose from the donor substrate, GDP-Fuc onto the array platform was tested in the presence of potential inhibitors, and quantified using the fucose specific lectin Tetragonolobus purpureas (TPL) conjugated to a peroxidase. Four of the screened compounds showed inhibitory effects in the nanomolar concentration range.

Label-Free Detection of Enzymatic Activity on Arrays The first studies involving enzymatic modifications on array platforms were reported a decade ago in the group of Mrksich (Houseman and Mrksich, 1999). These approaches were based on self-assembled monolayer-coated gold arrays in combination with SAMDI-ToF MS (self-assembled monolayers for matrix-assisted laser desorption–ionization time-of- flight mass spectrometry) as detection systems (Ban and Mrksich, 2008; Houseman and Mrksich, 1999, 2002; Su and Mrksich, 2002). Combination of this array platform with the detection technique allows the analysis of immobilized glycan FIG. 6. Enzymatic generation of sialyl Lex on solid support structures in a direct, label-free manner without the require- and detection of reaction products using antibodies and ment for a lectin or antibody binding assays. In this meth- lectins (Shin, 2005). odology, arrays of 24 thioalkane derivatized disaccharides 442 VOGLMEIR ET AL.

FIG. 7. Left: glycosylation of covalently attached aliphatic alcohols and maltose with the enzyme GTFR (Glycosyltransferase R) (Seibel, 2006). Right: similar array system has been used to screen a panel of fucosyltransferase inhibitors (Bryan, 2004). where immobilized via covalent attachment of the terminal during development and also in adult tissue (Ten Hagen et al., thio group to a gold surface to screen the acceptor specificity 2003). However, in contrast to N-glycosylation, a consensus of the bovine b1,4-galactosyltransferase I. After addition of sequence for mucin-type glycosylation has not been found matrix (such as THAP, 2,4,6-trihydroxyacetophenone) the (Cai et al., 1997). The analysis by MALDI-ToF mass spec- product can be directly measured by mass spectrometry trometry could reveal different glycosylation efficiencies of without the need for cleavage (Schanbacher and Ebner, 1970). different peptide substrates, and allowed the direct evaluation Approximate reaction yields have been determined by inte- of the number of GalNAc residues added to peptides con- gration of MS peaks and comparing signal of starting material taining more than one potential glycosylation site (Fig. 8). to that of product. The gold platform is sufficiently robust to tolerate multi- A method for screening of glycosyltransferases involved in step chemical synthesis and allowed the combination of solid- protein O-glycosylation of human mucins was developed by supported peptide synthesis with enzymatic glycosylation Laurent et al. (2008c) using the gold/SAM platform technol- methods. Peptide synthesis in array format are useful where a ogy (Houseman et al., 2003; Houseman and Mrksich, 1999; large number of peptide sequences need to be investigated. Lamb et al., 2008; Petty et al., 2007; Qian et al., 2002; Yea et al., Based on the principle of SPOT synthesis (Frank, 1992), all 2008; Zhi et al., 2006). A set of 10 peptides was immobilized chemical steps required for the peptide synthesis, such as re- in array format onto the SAM-coated gold surface, which petitive cycles of amino acid couplings, N-Fmoc deprotection, was interrogated with isoform 2 of the human UDP- acetylation, and TFA-mediated cleavage of protecting groups GalNAc:polypeptide GalNAc-transferase family (ppGal- could be performed in situ (Fig. 8) (Laurent et al., 2008a). A NAcT). This family contains more than 20 human isoforms, all series of peptides derived from the mucin Muc1 fragment with distinct tissue specific and temporary expression pattern AHGVTSAPA (with threonine as glycosyl-acceptor), was ENZYMATIC GLYCOSYLATIONS ON ARRAYS 443

arrays, it will be possible to integrate glycans in multimodal analyses tools in future to study the intricate details of posttranslational modiﬁcation.

Acknowledgments The authors thank EPSRC and BBSRC (PhD Plus scheme for J.V.), the European Commission (Marie Curie Intra- European Fellowship for R.S and M.J.W), and the Royal Society (Wolfson Merit Award to S.L.F.) for ﬁnancial support.

Author Disclosure Statement The authors declare that no conﬂicting ﬁnancial interests exist.

References Ban, L., and Mrksich, M. (2008). On-chip synthesis and label-free assays of oligosaccharide arrays. Angew Chem Int Ed Engl 47, 3396–3399. Blixt, O., and Razi, N. (2006). Chemoenzymatic synthesis of glycan libraries. Methods Enzymol 415, 137–153. Blixt, O., Allin, K., Bohorov, O., Liu, X., Andersson-Sand, H., Hoffmann, J., et al. (2008a). Glycan microarrays for screening sialyltransferase specificities. Glycoconj J 25, 59–68. Blixt, O., Hoffmann, J., Svenson, S., and Norberg, T. (2008b). Pathogen specific carbohydrate antigen microarrays: a chip for detection of Salmonella O-antigen specific antibodies. Glyco- conj J 25, 27–36. Bryan, M.C., Lee, L.V., and Wong, C.H. (2004). High-throughput identification of fucosyltransferase inhibitors using carbohydrate microarrays. Bioorg Med Chem Lett 14, 3185–3188. FIG. 8. Glycosylation of on gold immobilized peptides Cai, Y.D., Yu, H., and Chou, K.C. (1997). Artificial neural with the enzymes ppGalNAcT2 and GalT1. Products were network method for predicting the specificity of GalNAc- detected in a label free manner using in situ MALDI-ToF transferase. J Protein Chem 16, 689–700. mass spectrometry (Laurent, 2008a, 2008c). Cantarel, B.L., Coutinho, P.M., Rancurel, C., Bernard, T., Lom- bard, V., and Henrissat, B. (2009). The Carbohydrate-Active EnZymes database (CAZy): an expert resource for glycoge- prepared directly on the array. The effect of sequence varia- nomics. Nucleic Acids Res 37(Database issue), D233–D238. tions at positions þ1 and 1 in relation to threonine on the Chevolot, Y., Martins, J., Milosevic, N., Leonard, D., Zeng, S., Malissard, M., et al., (2001). Immobilisation on polystyrene of enzyme activity of the glycosyltransferase ppGalNAcT2 diazirine derivatives of mono- and disaccharides: biological could be determined in a quantitative manner by MALDI-ToF activities of modified surfaces. Bioorg Med Chem 9, 2943–2953. MS. It was found that modifications at the 1 position relative Cle, C., Gunning, A.P., Syson, K., Bowater, L., Field, R.A., and to the threonine had a more severe effect on the glycosylation Bornemann, S. (2008). Detection of transglucosidase-catalyzed capability of ppGalNAcT2 than the þ1 position, and that an polysaccharide synthesis on a surface in real time using sur- introduction of proline close to the threonine enhanced the face plasmon resonance spectroscopy. J Am Chem Soc 130, glycosylation efficiency of the enzyme. 15234–15235. Davis, B.G. (2000). Recent developments in oligosaccharide Conclusion synthesis. J Chem Soc Perkin Trans 1, 2137–2160. The application of array technology to the study of en- Fazio, F., Bryan, M.C., Blixt, O., Paulson, J.C., and Wong, C.H. (2002). Synthesis of sugar arrays in microtiter plate. J Am zymes involved in glycosylation is a relatively young field Chem Soc 124, 14397–14402. and so far most reports have concentrated on method devel- Frank, R. (1992). Spot-Synthesis—an easy technique for the po- opment. A range of platforms have been adapted from other sitionally addressable, parallel chemical synthesis on a mem- array technologies (e.g., protein and nucleic acids) and so far brane support. Tetrahedron 48, 9217–9232. have generated activity data that are in agreement with each Goldstein, I.J., Hollerman, C.E., and Smith, E.E. (1965). Protein– other and with solution studies, although the data are still carbohydrate interaction. Ii. Inhibition studies on the interac- very limited. Given the advantages arrays bring to the in- tion of concanavalin a with polysaccharides. Biochemistry 4, vestigation of glycoenzymes in terms of miniaturization and 876–883. throughput, it is expected that glycoarrays will become im- Houseman, B.T., and Mrksich, M. (1999). The role of ligand portant tools for glycoscientists. Because the glycoarray density in the enzymatic glycosylation of carbohydrates pre- technologies use similar immobilization techniques and sented on self-assembled monolayers of alkanethiolates on platforms as used for nucleic acid and peptide/proteins gold. Angew Chem Int Ed 38, 782–785. 444 VOGLMEIR ET AL.

Houseman, B.T., and Mrksich, M. (2002). Carbohydrate arrays Seeberger, P.H., and Haase, W.C. (2000). Solid-phase oligosac- for the evaluation of protein binding and enzymatic modifi- charide synthesis and combinatorial carbohydrate libraries. cation. Chem Biol 9, 443–454. Chem Rev 100, 4349–4394. Houseman, B.T., Gawalt, E.S., and Mrksich, M. (2003). Seibel, J., Hellmuth, H., Hofer, B., Kicinska, A.M., and Schmal- Maleimide-functionalized self-assembled monolayers for the bruch, B. (2006). Identification of new acceptor specificities of preparation of peptide and carbohydrate biochips. Langmuir glycosyltransferase R with the aid of substrate microarrays. 19, 1522–1531. Chembiochem 7, 310–320. Lamb, B.M., Westcott, N.P., and Yousaf, M.N. (2008). Live-cell Shin, I., Park, S., and Lee, M.R. (2005). Carbohydrate micro- fluorescence microscopy of directed cell migration on partially arrays: an advanced technology for functional studies of gly- etched electroactive SAM gold surfaces. Chembiochem 9, cans. Chemistry 11, 2894–2901. 2220–2224. Shipp, M., Nadella, R., Gao, H., Farkas, V., Sigrist, H., and Faik, A. Laurent, N., Haddoub, R., Voglmeir, J., Wong, S.C., Gaskell, S.J., (2008). Glyco-array technology for efficient monitoring of plant and Flitsch, S.L. (2008a). SPOT synthesis of peptide arrays on cell wall glycosyltransferase activities. Glycoconj J 25, 49–58. self-assembled monolayers and their evaluation as enzyme Su, J., and Mrksich, M. (2002). Using mass spectrometry to charac- substrates. Chembiochem 9, 2592–2596. terize self-assembled monolayers presenting peptides, proteins, Laurent, N., Voglmeir, J., and Flitsch, S.L. (2008b). and carbohydrates. Angew Chem Int Ed Engl 41, 4715–4718. Glycoarrays—tools for determining protein–carbohydrate in- Ten Hagen, K.G., Fritz, T.A., and Tabak, L.A. (2003). All in the teractions and glycoenzyme specificity. Chem Commun family: the UDP-GalNAc: polypeptide N-acetylgalactosaminyl- (Camb) 37, 4400–4412. transferases. Glycobiology 13, 1R–16R. Laurent, N., Voglmeir, J., Wright, A., Blackburn, J., Pham, N.T., Willats, W.G., Rasmussen, S.E., Kristensen, T., Mikkelsen, J.D., Wong, S.C., et al. (2008c). Enzymatic glycosylation of peptide and Knox, J.P. (2002). Sugar-coated microarrays: a novel slide arrays on gold surfaces. Chembiochem 9, 883–887. surface for the high-throughput analysis of glycans. Pro- Nicolaou, K.C., and Mitchell, H.J. (2001). Adventures in carbo- teomics 2, 1666–1671. hydrate chemistry: new synthetic technologies, chemical syn- Yea, C.H., Lee, B., Kim, H., Kim, S.U., El-Said, W.A., Min, J., et al. thesis, molecular design, and chemical biology. Angew Chem (2008). The immobilization of animal cells using the cysteine- Int Ed Engl 40, 1576–1624. modified RGD oligopeptide. Ultramicroscopy 108, 1144–1147. Park, S., and Shin, I. (2007). Carbohydrate microarrays for Zhang, Z., Ollmann, I.R., Ye, X.-S., Wischnat, R., Baasov, T., and assaying galactosyltransferase activity. Org Lett 9, 1675– Wong, C.-H. (1999). Programmable one-pot oligosaccharide 1678. synthesis. J Am Chem Soc 121, 734–753. Park, S., Lee, M.R., Pyo, S.J., and Shin, I. (2004). Carbohydrate Zhi, Z.L., Powell, A.K., and Turnbull, J.E. (2006). Fabrication of chips for studying high-throughput carbohydrate-protein in- carbohydrate microarrays on gold surfaces: direct attachment teractions. J Am Chem Soc 126, 4812–4819. of nonderivatized oligosaccharides to hydrazide-modified Park, S., Lee, M.R., and Shin, I. (2008). Carbohydrate micro- self-assembled monolayers. Anal Chem 78, 4786–4793. arrays as powerful tools in studies of carbohydrate-mediated Zhi, Z.L., Laurent, N., Powell, A.K., Karamanska, R., Fais, M., biological processes. Chem Commun (Camb) 37, 4389–4399. Voglmeir, J., et al. (2008). A versatile gold surface approach for Petty, R.T., Li, H.W., Maduram, J.H., Ismagilov, R., and Mrksich, fabrication and interrogation of glycoarrays. Chembiochem 9, M. (2007). Attachment of cells to islands presenting gradients 1568–1575. of adhesion ligands. J Am Chem Soc 129, 29, 8966. Qian, X.P., Metallo, S.J., Choi, I.S., Wu, H.K., Liang, M.N., and Address correspondence to: Whitesides, G.M. (2002). Arrays of self-assembled monolayers Prof. Sabine L. Flitsch for studying inhibition of bacterial adhesion. Anal Chem 74, Manchester Interdisciplinary Biocentre 1805–1810. University of Manchester Schanbacher, F.L., and Ebner, K.E. (1970). Galactosyltransferase Manchester M1 7ND, United Kingdom acceptor specificity of the lactose synthetase A protein. J Biol Chem 245, 5057–5061. E-mail: sabine.fl[email protected] CHAPTER FOUR

GLYCOARRAYS ON GOLD SURFACES

P. Both, R. Sardzik, M. J. Weissenborn, A. P. Green, J. Voglmeir, S. L. Flitsch, Gly- coarrays on gold surfaces, Proceedings of the International Beilstein Symposium 2012, 93–106.

All authors contributed to these symposium proceedings.

13 93 Cracking the Sugar Code by Navigating the Glycospace Beilstein-Institut June 27th – July 1st, 2011, Potsdam, Germany

Glycoarrays on Gold Surfaces

Peter Both, Robert ardzk, Martin Weissenborn, Anthony Green, Josef Voglmeir and Sabine Flitsch* School of Chemistry & Manchester Interdisciplinary Biocentre, The University of Manchester, 131 Princess Street, Manchester M1 7ND, U.K. E-Mail: *[email protected]

Received: 16th December 2011/Published: 11th July 2012

Abstract

Self-assembled monolayers (SAMs) on gold have become widely used as a platform for studying chemical and biochemical reactions, for studying biomolecular interactions and for the development of nanoscale devices. We have used this platform to study the solid-supported synthesis of carbohydrates and glycopeptides using both chemical and enzymatic methods. An attractive feature of the technology is the opportunity for miniaturisation and in situ analysis using mass spectrometry, SPR and fluorescence spectroscopy. Applications for the synthesis of complex oligosaccharides and glycopeptides to generate glycoarrays and their application in biology and medicine are discussed.

Introduction

The sequences of oligo- and polysaccharides in cells and tissues are not directly encoded in their genomes, but are determined by the expression and substrate specificity of a large set of ‘glycoenzymes’, which catalyse the formation or hydrolysis of glycosidic bonds (Figure 1). These enzymes are involved in the biosynthesis of glycan structures by controlling regio- and stereoselectivity of glycosylation, a process which is highly dynamic. The understanding of the activity and substrate specificity of these glycoenzymes is a key to determining and understanding the ‘glycome’, the set of carbohydrate structures in a biological system.

http://www.beilstein-institut.de/glycobioinf2011/Proceedings/Flitsch/Flitsch.pdf 94

Both, P. et al.

Figure 1. The Glycome presents the set of carbohydrate sequences in a cell or organism. The sequences are determined by glycoenzymes such as glycosyltransferases and glycosidases which control formation and hydrolysis of specific glycosidic linkages.

Glycoenzymes can often be identified in genomes for their characteristic conserved polypeptide sequence motifs and over 100,000 members are currently catalogued in the CAZy database (www.cazy.org/). Most of these glycoenzymes are highly specific for generating or cleaving a defined saccharide sequence, including selectivity for monosaccharide units as well as regio- and stereochemistry of linkage. This precise substrate specificity needs to be determined for each enzyme by biochemical studies, since computational methods are currently not able to make accurate predictions on substrate recognition. Thus, there is a major drive in the glycosciences to find biochemical methods that can allow us to go beyond sequence determination to obtain structural and functional information on glycoenzymes.

One of the emerging technologies in the glycosciences involves microarrays containing diverse carbohydrate probes (Glycan arrays) which have been used to study carbohydrate- protein interactions. Such microarrays have been generated from either natural or synthetic carbohydrate samples (Figure 2) and have been interrogated by fluorescently labelled protein for binding. Over the past five years, our laboratory and others have applied this glycan array technology to the study of glycoenzymes [1] 95

Glycoarrays on Gold Surfaces

Figure 2. Carbohydrate arrays have become important tools for the investigation of protein/carbohydtae interactions in a high-throughput manner.

Methods and Results

Biochemical reactions are generally studied in solution, but for microarray studies, such reactions need to be transferred to solid phase. Although immobilised enzymes have been used frequently in biotechnological applications, the study of enzymatic modifications of immobilised substrates (Figure 3) has been less well explored [2]. Extensive fundamental studies of enzyme-catalysed reactions on diverse solid supports have shown that issues of substrate accessibility [3 – 13], reaction equilibria [14 – 17] and stereospecificity [18 – 21] need to be considered when choosing solid supports. Porous supports such as copolymers of polystyrene or polyacrylamide with polyethylene glycol are generally not suitable as supports for efficient enzyme catalysis because of poor yields. Such polymer supports also offer little opportunity for in situ molecular analysis of reactions. 96

Both, P. et al.

Figure 3. Classically, enzymes have been in used solution (upper panel) or immobilised on solid phase (middle panel). Applications on microarrays require the substrate to be immobiled during the reaction (lower panel).

Given these considerations, a solid support consisting of self-assembled monolayers (SAMs) on gold (subsequently referred to as ‘gold platform’) [22, 23] was chosen as a more suitable system to study glycoenzymes on arrays (Figure 4). Such arrays can easily be derivatized with structures containing an amine functionality such as peptides, glycopeptides or glycosyl ethanolamines. The SAM surfaces can be used with both organic solvents and aqueous buffers and have been optimised for biocompatibility [22]. Peptides can be coupled directly through their amino-termini or amino side chains, whereas glycosyl aminoethanols are easily prepared by short synthetic routes [23]. 97

Glycoarrays on Gold Surfaces

Figure 4. The ‘gold platform’ presented in this paper consists of self-assembled monolayers linked via thiols to gold surfaces.

A particular advantage of the gold platform is the possibility of in situ analysis of surfaces through MALDI-TOF mass spectrometry, a technology termed SAMDI [24, 25] (Figure 5). Thus, any chemical or enzymatic transformation can be monitored directly on the surface without the need for cleavage. The surface can also be re-used for further chemical or enzymatic reactions since only small fractions of the analyte are sampled by MALDI-TOF MS analysis. Figure 5 shows as an example the mass spectra taken from array spots before and after glycosylation using a b-1,4-galactosyltransferase. The glycan probe shows several peaks due to thiol dimerization to generate homo- and hetero-disulfides, but all peaks can be assigned unambiguously to either starting material or product. The MALDI-TOF mass spectrum of the product illustrates that the glycosylation can be high-yielding – in this case no starting material was observed in the spectrum, suggestion quantitative conversion. We have investigated a diverse range of glycosyltransferases on the gold platform such as GlucNAc-, GalNAc-, Neu5Ac- and Fuc-transferases (Figure 6, unpublished results) and have found that this platform provides accessible acceptor substrates to all of these enzymes [26 – 28]. A further advantage of the gold platform is compatibility with other spectroscopic methods, In particular surface plasmon resonance spectroscopy [29], which also needs attachment of ligands to conducting surfaces such as gold, in order to measure carbohydrate/protein interactions [29]. 98

Both, P. et al.

Figure 5. The enzymatic galactosylation of a surface-linked GlcNAc can be monitored using MALDI-TOF mass spectrometry with trihydroxy acetophenone as a matrix. The spectrum shows monomers, homo- and heterodimers of the functionalised alkyl thiols.

Figure 6. In collaboration with Prof Iain Wilson(BOKU, Austria), we have investigated a number of fucosyltransferases on the array platform. Varying degrees of glycosylation were observed with recombinant enzymes. 99

Glycoarrays on Gold Surfaces

One of the most intriguing questions in protein glycosylations concerns the choice of glycosylation site on a polypeptide sequence. Whilst the site of N-glycosylation is highly conserved and predictable for tripeptide motifs (Asn-X-Thr/Ser), consensus motifs for O-glycosylation are poorly understood and need to be determined by biochemical studies for each glycosyltransferase. Microarrays of peptide acceptor substrates are very useful for this purpose, because of high-throughput and miniaturisation. Particularly attractive for these studies is spot-synthesis of peptides directly on an array, because it allows for the fast and cheap synthesis of tailor-made peptide libraries to investigate specific peptide substrates around a known lead structure.

A common problem with spot synthesis, however, is quality control. Peptide synthesis is not always reliable and on chip synthesis can be difficult to monitor. Given the in situ analysis capability of the gold platform, we have developed a methodology that allows us to do spot synthesis with concomitant analysis. Figure 7 shows the individual steps involved, which are fully compatible with Fmoc-protected amino acids used in many automated peptide synthesis systems [30, 31]. Hence, all the building blocks needed are commercially available. We have shown that the in situ analysis using MALDI-TOF MS allows for tight quality control after every step. Thus, poor couplings can easily be detected and if necessary be repeated before proceeding with synthesis. MALDI-TOF analysis can also give us information about the purity of the final peptide, and in particular detect any truncated side products.

Figure 7. Solid phase peptide synthesis on ‘gold array’ (SPOT) using Fmoc technology.

This spot synthesis was used to make a glycopeptide array to probe the substrate specificity of ppGalNAc T2, an isoform of a family of over 20 human glycosyltransferases involved in the first steps of mucin glycosylation (Figure 8). A peptide library around a natural polypeptide sequence was generated and it was found that glycosyltransferase activity was 100

Both, P. et al.

highly dependent on the nature of the two residues next to the threonine glycosylation site [30]. Interestingly, placing a proline on either site dramatically increased the activity of the enzyme and made the peptide a much better substrate than the natural sequence studied.

Figure 8. Glycopeptide synthesis appears to proceed to completion on the array surfaces as indicated by MALDI-TOFMS analysis of starting material and product.

Another application of our platform is in the analysis of genetic orders of glycosylation [32, 33]. The glycosyltransferase POMGnT catalyses the glycosylation of an unusual class of O- mannosyl glycans (Figure 9), which have been isolated as a major component of mouse brain, but so far have only been identified on a single protein, a-dystroglycan (a-DG), which has been found in muscle and brain tissue. Defective glycosylation of a-DG leads to congenital muscular dystrophies and a number of patients have been diagnosed with mutations in the POMGnT gene. Our technology has allowed us to study clinical mutants of POMGnT for activity against peptides derived from a-DG. Interestingly, one of these clinical mutations did not seem to affect enzyme kinetics, whereas other clinical point mutation abolished activity. Some correlation between enzyme activity and severity of disease was found, although this requires further investigation. 101

Glycoarrays on Gold Surfaces

Figure 9. POMGnT1 is involved in the biosynthesis of unusual O-mannosyl glycans found in the brain. Mutations in the gene encoding for POMGnT1 can lead to congenital disorders of glycosylations, in particular muscular dystrophies.

It is interesting to note that the glycosylation of peptides using this enzymatic approach appears to proceed with excellent yields. Although quantification of reaction yield using mass spectrometry is difficult, the analysis of the gold surface by MALDI-TOF MS as shown in figures 8 and 9 gives a very clean spectrum in which only the product peak is visible, suggesting near quantitative yields of reaction. Thus, the solid-phase enzymatic synthesis of glycopeptides is possible. Such enzymatic routes are particularly interesting for the generation of linkages that are chemically more difficult to achieve, such as a- GlucNAc and a-GalNAc linkages as shown in figures 8 and 9 respectively.

The results shown so far demonstrate that mass spectrometric readout on gold arrays is very useful for following chemical and enzymatic reactions in a label-free manner. Because of the high resolution of MALDI-TOF MS even at high molecular weights, it should be possible to follow multiple reactions on the surface in parallel (multiplex analysis), provided that there is a difference in molecular weights between individual reaction sets. For proof of principle studies, we have immobilised mixtures of three Muc1 derived peptides on gold arrays (Figure 10, unpublished results). In previous studies we had shown that peptides Muc1fr1 is a good substrate for ppGalNAc T2 and complete glycosylation is observed when incubated with enzyme and UDP-GalNAc. Under the same reaction conditions, Muc1fr2 is only partially glycosylated and for Muc1fr3 we could not observe any glycosylation. The same 102

Both, P. et al.

results were obtained in peptide mixtures, suggesting that the methodology allows us to investigate several potential substrates in single spots. These investigations are currently extended to more complex mixtures.

Figure 10. The resolution of the platform readout by mass spetrometry allows for the investigation of multiple substrates in one spot as shown here for pairs of peptides derived from the Muc1 repeat sequence of mucins, which are interrogated as substrates for the mucin transferase ppGalNAc T2.

Some of the most important terminal monosaccharides on cell surfaces of humans and higher organisms are the sialic acids. A number of glycosyltransferases which generate sialosides are available and it was of interest to study these enzymes on gold surfaces. However, the mass spectrometric analysis of sialic acids can present problems, since sialosides are often not stable under MALDI-TOF conditions, and hydrolysis products are observed. This can be overcome by methylation of sialic acids prior to analysis; in particular formation of the methyl ester significantly increases the signal intensity. For this purpose, we tested a number of methylation protocols for suitability on gold surfaces and found that these significantly improved the signal and allowed us to demonstrate that sialylation had indeed taken place. For example, the transsialidase from Trypanosoma cruzi was able to transfer sialic acids in an a2,3 manner from the glycoprotein fetuin onto the gold surfaces containing lactose acceptors to generate a2,3 sialyllactose. In collaboration with the group of Paul Crocker at the University of Dundee we could show that CHO cells, engineered to display sialoadhesin on their surfaces, selectively recognised gold surfaces displaying a2,3 sialyllactose over lactose (Figure 11 [34]). 103

Glycoarrays on Gold Surfaces

Figure 11: Sialic acids (Neu5Ac) are important components of cell surface oligosaccharides. The transsialidase TcTS from Trypanosoma cruzi can transfer sialic acids from glycoproteins such as fetuin to immobilised lactosides. The resulting sialolacto- sides are recognised by cell surface sialoadhesin receptors (collaboration with Prof Paul Crocker, University of Dundee)

Conclusions

An array platform consisting of gold surfaces covered with self-assembled functionalised monolayers has proven to be chemically and biologically robust and versatile. Many chemical and biochemical reactions can be studied on this surface and can proceed with very high efficiency, as measured by in situ mass spectrometry. In the past few years we have explored the application of this technology to glycosylation reactions and have found that it is suitable for many applications, in particular exploring enzyme specificity and activity. We believe that this technology is ideally suited to address the challenges of glycomics – high- throughput investigation of glycoenzyme activity and investigation of carbohydrate-protein interactions. The platform has the additional advantage that it is suited to a number of analytical techniques such as mass spectrometry and surface plasmon resonance spectroscopy, which makes it suitable for the development of diagnostic devices.

Acknowledgements

We are grateful to the EPSRC, BBSRC and FP7 for funding. SLF is the recipient of a Wolfson Merit Award of the Royal Society. 104

Both, P. et al.

References

[1] Laurent, N., Voglmeir, J., Flitsch S.L. (2008) Chem. Commun. 37:4371 – 4384. doi: 10.1039/B814463J.

[2] Halling, P.J., Ulijn, R.V., Flitsch, S.L. (2005) Current Opinion in Biotechnology 16:385 – 392. doi: http://dx.doi.org/10.1016/j.copbio.2005.06.006.

[3] Ulijn, R.V., Brazendale, I., Margetts, G., Flitsch, S.L., McConnell, G., Girkin, J., Halling, P.J. (2003) J. Combi. Chem. 5:215 – 217. doi: http://dx.doi.org/10.1021/cc030024c.

[4] Ulijn, R.V., Bisek, N., Flitsch, S.L. (2003) Org. Biomol. Chem. 1:621 – 622. doi: http://dx.doi.org/10.1039/b211887d.

[5] Ulijn, R.V., Bisek, N., Halling, P. J., Flitsch, S.L. (2003) Org. Biomol. Chem. 1:1277 – 1281. doi: http://dx.doi.org/10.1039/b211890d.

[6] Basso, A., De Martin, L., Gardossi, L., Margetts, G., Brazendale, I., Bosma, A.Y., Ulijn, R.V., Flitsch, S.L. (2003) Chem. Commun. 11:1296 – 1297. doi: 10.1039/B301680C.

[7] Basso, A., De Martin, L., Ebert, C., Linda, P., Gardossi, L., Ulijn, R.V., Flitsch, S.L. (2003) Tetrahedron Lett. 44:6083 – 6085. doi: http://dx.doi.org/10.1016/S0040-4039(03)01464-3.

[8] Bosma, A.Y., Ulijn, R.V., McConnell, G., Girkin, J., Halling, P.J., Flitsch, S.L. (2003) Chem. Commun. 22:2790 – 2791. doi: http://dx.doi.org/10.1039/b308078a.

[9] Basso, A., Ulijn, R.V., Flitsch, S.L., Margetts, G., Brazendale, I., Ebert, C., De Martin, L., Linda, P., Verdelli, S., Gardossi, L. (2004) Tetrahedron 60(3):589 – 594. doi: http://dx.doi.org/10.1016/j.tet.2003.10.125.

[10] Daines, A.M., Maltman, B.A., Flitsch, S.L. (2004) Current Opinion in Chemical Biology 8:106 – 113. doi: http://dx.doi.org/10.1016/j.cbpa.2004.02.003.

[11] Basso, A., Maltman, B.A., Flitsch, S.L., Margetts, G., Brazendale, I., Ebert, C., Linda, P., Verdellia, S., Gardossia, L. (2004) Tetrahedron 61:971 – 976. doi: http://dx.doi.org/10.1016/j.tet.2004.11.015.

[12] Bejugam, M., Maltman, B.A., Flitsch, S.L. (2005) Tetrahedron: Asymmetry 16:21 – 24. doi: http://dx.doi.org/10.1016/j.tetasy.2004.11.031. 105

Glycoarrays on Gold Surfaces

[13] Maltman, B.A., Bejugam, M., Flitsch, S.L. (2005) Org. Biomol. Chem. 3:2505 – 2507. doi: http://dx.doi.org/10.1039/b506154g.

[14] Ulijn, R.V., Baragana, B., Halling, P.J., Flitsch, S.L. (2002) J. Am. Chem. Soc. 124:10988 – 9. doi: http://dx.doi.org/10.1021/ja026912d.

[15] Basso, A., Maltman, B.A., Flitsch, S.L., Margetts, G., Brazendale, I., Ebert, C., Linda, P., Verdellia, S., Gardossia, L. (2004) Tetrahedron 61:971 – 976. doi: http://dx.doi.org/10.1016/j.tet.2004.11.015.

[16] Deere, J., McConnell, G., Lalaouni, A., Maltman, B.A., Flitsch, S.L., Halling, P.J. (2007) Advanced Synthesis & Catalysis 349:1321 – 1326. doi: http://dx.doi.org/10.1002/adsc.200700044.

[17] Deere, J., Halling, P., Flitsch, S.L., De Oliveira, R., Tomaszewski, B., Millar, S., Lalaouni, A., Solares, L. (2008) Langmuir 24:11762 – 11769. doi: http://dx.doi.org/10.1021/la801932f.

[18] Humphrey, C.E., Turner, N.J., Easson, M.A.M., Flitsch, S.L., Ulijn, R.V. (2003) J. Am. Chem. Soc. 125:13952 – 3. doi: http://dx.doi.org/10.1021/ja037922x.

[19] Doeze, R.H.P., Egan, C.L., Ulijn, R.V., Flitsch, S.L. (2004) Angewandte Chemie 43:3138 – 3141. doi: http://dx.doi.org/10.1002/anie.200353367.

[20] Bejugam, M., Maltman, B.A., Flitsch, S.L. (2005) Tetrahedron: Asymmetry 16:21 – 24. doi: http://dx.doi.org/10.1016/j.tetasy.2004.11.031.

[21] Maltman, B.A., Bejugam, M., Flitsch, S.L. (2005) Org. Biomol. Chem. 3:2505 – 2507. doi: http://dx.doi.org/10.1039/b506154g.

[22] Sˇardzı´k, R., Noble, G.T., Weissenborn, M.J, Martin, A., Webb, S.J., Flitsch, S.L. (2010) Beilstein J. Org. Chem. 6:699 – 703. doi: http://dx.doi.org/10.3762/bjoc.6.81.

[23] Ostuni, E., Chapman, R.G., Liang, M.N., Meluleni, G., Pier, G., Ingber, D., White- sides, G.M. (2001) Langmuir 17:5605 – 5620. doi: http://dx.doi.org/10.1021/la010384m.

[24] Su, J., Mrksich, M. (2002) Angew. Chem. Int. Ed. 41:4715 – 4718. doi: http://dx.doi.org/10.1002/anie.200290026. 106

Both, P. et al.

[25] Gurard-Levin, Z.A., Scholle, M., Eisenberg, A.H., Mrksich, M. (2011) ACS Comb. Sci. 13(4):347 – 50. doi: http://dx.doi.org/10.1021/co2000373.

[26] Laurent, N., Voglmeir, J., Wright, A., Blackburn, J., Pham, N.T., Wong, S.C.C., Gaskell, S.J., Flitsch, S.L. (2008) ChemBioChem. 9:883 – 887. doi: http://dx.doi.org/10.1002/cbic.200700692..

[27] Zhi, Z., Laurent, N., Powell, A.K., Karamanska, R., Fais, M., Voglmeir, J., Wright, A., Blackburn, J.M., Crocker, P.R., Russell, D.A., Flitsch, S.L., Field, R.A., Turnbull, J.E. (2008) ChemBioChem. 9:1568 – 1575. doi: http://dx.doi.org/10.1002/cbic.200700788.

[28] Laurent, N., Haddoub, R., Flitsch, S.L. (2008) Trends in Biotechnology 26:328 – 337. doi: http://dx.doi.org/10.1016/j.tibtech.2008.03.003.

[29] Karamanska, R., Clarke, J., Blixt, O., MacRae, J.I., Zhang, J.Q., Crocker, P.R., Laurent, N., Wright, A., Flitsch, S.L., Russell, D.A., Field, R.A. (2008) Glycoconju- gate Journal 25:69 – 74. doi: http://dx.doi.org/10.1007/s10719-007-9047-y.

[30] Laurent, N., Haddoub, R., Voglmeir, J., Wong, S.C.C., Gaskell, S.J., Flitsch, S.L. (2008) ChemBioChem 9:2592 – 2596. doi: http://dx.doi.org/10.1002/cbic.200800481.

[31] Haddoub, R., Dauner, M., Stefanowicz, F.A., Barattini, V., Laurent, N., Flitsch, S.L. (2009) Org. Biomol. Chem. 7:665 – 670. doi: http://dx.doi.org/10.1039/b816847d.

[32] Voglmeir, J., Sˇardzı´k, R., Weissenborn M.J., Flitsch, S.L. (2010) OMICS J. 14:437 – 444. doi: http://dx.doi.org/10.1089/omi.2010.0035.

[33] Voglmeir, J., Kaloo, S., Laurent, N., Meloni, M.M., Bohlmann, L., Wilson, I.B., Flitsch, S.L. (2011) Biochem. J. 436:447 – 55. doi: http://dx.doi.org/10.1042/BJ20101059.

[34] Sˇardzı´k, R., Sharma, R., Kaloo, S., Voglmeir, J., Crocker, P.R., Flitsch, S.L. (2011) Chem. Commun. 47:5425 – 7. doi: http://dx.doi.org/10.1039/c1cc10745c. CHAPTER FIVE

OBJECTIVES OF THIS THESIS

Protein-carbohydrate interactions are important in diverse intra- and extracellular signalling processes and the mediation of biomolecular interactions.1 For example these specific interactions distinguish between different blood groups and enable bacteria to adhere to host cells in the human body.1,2 Array based systems of immobilised glycans and glycoproteins enable high throughput analyses of proteins, pathogens and cells.3,4 Microarrays have been employed for clinical studies and as a diagnostic tool, for example to identify cancer biomarkers, since they only require minute sample amounts.5 Enzymes have been employed on microarrays to study substrate specificity and to produce complex surface-bound ligands by enzymatic transformations.6 The development of surface based systems for investigation of protein-carbohydrate interactions and enzymatic reactions is central to the expansion of our understanding of these biological processes and has far reaching consequences for the diverse fields of biological and medical sciences.7–9 The goal of our work was to increase the impact and versatility of peptide- and glycoarrays. It was envisioned that such goals can be attained by: (i) developing synthetic routes for preparation of the appropriate ligands and novel methods for array formation, (ii) the MALDI-ToF MS analysis of substrates immobilised on glass and polystyrene surfaces, and (iii) engineering new tools to determine carbohydrate-protein interactions and determine protein glycosylations.

5.1 Array Formation

Availability of the appropriate ligands is often a limiting factor in the production of arrays for studies of biologically relevant events.10,11 The design and optimisation of low-tech synthetic routes that can be performed in non-specialist laboratories could help to address this problem, particularly when simple ligands are required. More complex ligands are still produced by laborious syntheses or difﬁcult isolation and are consequently available only in minute quantities.10 Due to these limitations, the protocols for immobilisation of such ligands on surfaces must be highly efﬁcient. Since such microarrays are often used for studies of enzymatic reactions or protein binding interactions that involve various

14 5.2. ANALYSIS OF ARRAYS biomolecules, the coupling reactions should ideally proceed under physiological conditions and produce stable products reminiscent of or identical to the natural products. To allow for the coupling of target molecules from crude mixtures such as cell lysates these coupling reactions should also be highly chemoselective and biocompatible. Both of these approaches were pursued: a robust synthetic route for nine aminoethyl glycosides was developed (chapter7) and the development of a general coupling method based on native chemical ligation was undertaken as detailed in chapter8.

5.2 Analysis of Arrays

Most array-based systems are produced on functionalised glass surfaces and the analysis of immobilised ligands is typically based on detection of protein-ligand interactions.12,13 Although it offers high resolution and does not require secondary protein interaction, direct MALDI-ToF MS analysis of arrays is rarely utilised because it can only be performed on electrically conductive surfaces with non-covalently attached or cleavable ligands.14,15 Conductivity is required for the charged ligands to dissipate from the array surface and travel to the ToF mass analyser, making analysis on poorly conducting surfaces, such as glass and most polymers, challenging.14 Coating the surface with indium-tin-oxide (ITO) improves the conductivity but these techniques are expensive and must be performed prior to ligand immobilisation.16 A system that can overcome the problems related to the low electrical conductivity of glass- or polymer-based microarrays would enable MS analysis of most non-covalent arrays or surfaces where the ligands are immobilised by cleavable linkers. One could then perform conventional array analysis by protein interactions in combination with MALDI-ToF MS to detect even small changes of the surface ligands. Such a system would largely be of benefit if the MALDI analysis could be performed without application of a matrix, i.e. by ’self-matrix’ properties of the ligands. Addition- ally, these ligands should be suitable for various array surfaces and coupling techniques. This could also enhance the availability of required surface-ligands. The work on the development of a simple, modified array system that fulfils the out- lined criteria was performed by using S-tritylated compounds which were non-covalently immobilised on polystyrene plates. Polystyrene plates were backed with conducting tape to make them compatible with MALDI analysis. The trityl groups served as a ’self- matrix’ (chapter9). Moreover, the S-trityl group was shown to be readily removed and the free thiols obtained were used for array formation on plain and maleimide functionalised gold (described in chapter 10).

15 5.3. APPLICATION OF MICROARRAYS

5.3 Application of Microarrays to Study Carbohydrate Protein Interactions

Protein glycosylation patterns cannot be accurately predicted. They are diverse, vary in each expression system and have significant functional roles.17,18 Since existing techniques — such as X-ray crystallography and antibody binding studies — often fail to provide sufficiently detailed information, many glycopeptides require novel analytical tools for the identification of glycosylation and evaluation of its functional roles.1 To address these issues, a biophotonic scattering technique first developed by the Shaw group was employed and further developed for the analysis of 96 different glycans and glycoproteins. These molecules were immobilised on slides and consecutively interrogated with a range of lectins. The glycosylation pattern was determined by the specific lectin binding. This method was employed for proteins with known (chapter 11) and unknown (chapter 12) glycosylations. One example of diversely glycosylated proteins is α-dystroglycan, a human protein present in muscle and brain tissues.19 The tetrasaccharide NeuNAcα2-3Galβ1-4GlcNAc- β1-2Manα is present in α-dystroglycan but its exact molecular function remains unknown. To study the exact functional role of this compound and its synthesis it needs to be made available for biological studies. The development of a chemoenzymatic route is described for the tetrasaccharide on its natural peptide sequence in solution and on solid-phase (chapter 13).

16 CHAPTER SIX

METHODOLOGIES APPLIED IN THIS THESIS

6.1 Chemical Synthesis

The chemical synthesis and characterisation of aminoethyl galactose and aminoethyl maltose is described in chapter7. The aminoethyl derivatives of galactose, glucose, N- acetylglucosamine and mannose were coupled to microarrays and interrogated with lectins (Chapter 11).

6.2 Arrays on Gold

A majority of arrays used in this thesis were prepared on gold platforms (chapters8, 10, 11, 12, and 13). Gold surfaces were employed as they are compatible with numerous analytical techniques such as MALDI-ToF MS20,21 and surface plasmon resonance (SPR).22,23 The plain gold was functionalised with alkane thiols that carry six ethylene glycol units and carboxylic acid moieties. The alkanethiols form spontaneously into self- assembled monolayers (SAMs) (Figure 6.1).24,25 The carboxylic acids act as point for chemical modiﬁcation and the ethylene glycol prevents a unspeciﬁc protein adsorption.25 Thiolates with only three ethylene glycol units and without carboxylic acid were employed to provide space between the different functional groups and avoid steric hindrance of proteins in e.g. enzymatic reactions.

6.2.1 Coupling into SAMs

After the plain gold 1 was modiﬁed with SAMs, 2 was activated by the formation of the N-hydroxysuccinimide (NHS) ester 3 (Figure 6.2). The activated ester 3 is formed by the application of a mixture of EDC (1-Ethyl-3-(3-dimethylaminopropyl)carbodiimid) and NHS. The mechanism (Figure 6.3) goes via the inital coupling of the EDC to the carboxylic acid to form the compound 2-EDC. The activated compound 2-EDC then reacts with NHS under formation of 3 and 1-Ethyl-3-(3-dimethylaminopropyl)urea. The formed NHS ester can readily react with any primary amine such as amino acids and

17 6.2. ARRAYS ON GOLD

Figure 6.1: Formation of SAMs on gold. Left: The spotting of linker and spacer solution (Picture was taken by Barbara Adamczyk). Right: The functions of each part of the self-assembled monolayers (SAMs) and the gold. aminoethyl glycosides (Section 6.1). Figure 6.2 shows the formation of 4 by coupling aminoethyl galactose to the surface. Coupling efﬁciency was improved by employing pentaﬂuorophenol activated esters and a coupling methodology that relied on the native chemical ligation principals were discussed in chapter8.

Figure 6.2: Formation and functionalisation of self-assembling monolayers on gold.

Maleimide functionalised SAMs 5 were produced in order to be able to anchor thiol- containing molecules to the SAMs 2. The formed maleimide surface 5 can be treated with thiols, thus forming thioethers. The mannoside 6 was deprotected in situ using tri- ﬂuoroacetic acid. As this deprotection step afford a trityl cation triethylsilane was added

18 6.2. ARRAYS ON GOLD

Figure 6.3: The mechanism of the activation of the linker 2 by EDC to form the activated ester 3. to trap the produced cation to avoid undesired side reactions.26 7 was directly applied to the maleimide 5 to form the thioether 8 (ﬁgure 6.4). The described maleimide functionalisation and formation of thioethers is described in detail in chapter 10.

Figure 6.4: Functionalisation of the SAMs 2 with (a) aminoethyl maleimide to form 5. The mannoside 6 was detritylated in presence of triﬂuoroacetic acid and triethylsilane (b). The formed thiol was applied to the maleimide 5 to form 8.

6.2.2 MALDI-ToF MS analysis of SAMs

The different analytical techniques on microarrays are discussed in chapter2 with the focus on gold arrays in chapter4. Matrix-assisted laser desorption/ionisation time-of-ﬂight (MALDI-ToF) mass spectrometry (MS) is a label-free analytical tool and thus enables direct surface analysis.

19 6.3. SURFACE PLASMON RESONANCE (SPR) ON SAMS

Small mass changes of modified substrates and products can be detected which is particularly important when monitoring new chemical and enzymatic reactions. Mass spectrometers consists of three main parts: i) the ion source, ii) the mass analyser and iii) the ion detector. In this thesis, MALDI was employed as ion source and time- of-flight (ToF) as mass analyser. The ionisation and mass analysis has been described elsewhere.27,28 Briefly, the ion generation process in MALDI consists of the irradiation of a conductive plate loaded with the sample — a mixture composed of analytes and a photoactive molecule called the matrix — by a laser pulse of a specific wavelength (usually in the UV range). This pulse is absorbed by the matrix molecules, causing an explosive desorption of the sample from the plate and the formation of an ion plume composed of excited matrix and analyte molecules. In the plume, matrix excitation energy is transferred to the analytes, which are then ionised in a mild process. The charged analytes are then extracted from the ion source through electrostatic lenses towards the mass analyser and are analysed relative to the mass to charge (m/z) ratio in the ToF. A reflector in the ToF mass analyser is employed to compensate for the differences in the kinetic energy of ions with the same m/z by forcing them to reflect in a curve trajectory that minimises the differences of their velocities.28 When used on SAMs, MALDI-ToF MS does not rely on co-crystallisation of the matrix with the analyte. The matrix solution crystallises instead on top of the analytes, which are the gold attached SAMs. Another difference is the requirement to cleave the Au-S bond during the ionisation process to enable the SAMs to travel to the mass analyser (figure 6.5). Analysis of 2 by MALDI produces peaks corresponding to the hetero- (terminal carboxylic group and terminal hydroxy-group type linker) and homo disulphides (solely consisting of the carboxylic terminal SAMs) formed during the ionisation process (figure 6.6) and is a useful method for detection of correct SAM formation. Following activation and coupling of aminoethyl galactose to form 4, MALDI analysis again produces the disulphide peaks, this time shifted to the right to reflect the additional mass from the aminoethyl galactose.

6.3 Surface Plasmon Resonance (SPR) on SAMs

Since protein-ligand interactions are difficult to detect, characterise and quantify via MAL- DI-ToF MS, surface plasmon resonance (SPR) was employed. SPR analysis, in contrast to fluorescence analysis, does not rely on protein tagging and is hence applicable to non- derivatised biological samples.22,29–31 SPR is based on the physical effects i) total internal reflection (TIR), ii) evanescent electric field and iii) surface plasmon resonance (SPR). It is introduced in chapter2 and discussed elsewhere.32–34 Briefly, monochromatic light that passes through two media with different refractive indices shows an absorbance maximum if shone on a thin gold surface. This absorbance maximum occurs at a certain angle of the light beam to the gold

20 6.3. SURFACE PLASMON RESONANCE (SPR) ON SAMS

Figure 6.5: Schematic illustration of direct MALDI-ToF MS analysis of SAMs. SAMs were coated with a matrix (not shown) ionised and directed through the ToF via the re- ﬂector to the detector.

Figure 6.6: The spectra resulting from the MALDI-ToF MS analysis of 2 (left) and 4 (right) in the m/z range of 1040 to 1700. For the spectrum of 2 the peaks for the hetero disulphide (1051 m/z, terminal carboxylic group and terminal hydroxy-group type linker) and the homo disulphide (1241 m/z, two terminal carboxylic-group type linker). The spectrum of 4 showed the hetero and the homo disulphide in a m/z ratio of 1256 and 1651, respectively. surface: the surface plasmon resonance (SPR) angle (ﬁgure 6.7). This SPR angle changes in proportion to the mass change on the gold surface if, e.g., a protein binds to a ligand on gold. In this thesis, plain SPR gold chips were functionalised with SAMs as described in Section 6.2 and were employed to analyse the coupling speciﬁcity of cysteine compounds over amines (described in chapter8). In chapter 10, SPR was used to analyse differ-

21 6.4. ARRAYS ON POLYSTYRENE

Figure 6.7: Illustration of the surface plasmon resonance (SPR) technique. ent binding speciﬁcities of thiol-mannoses with different spacer length towards the lectin concanavalin A (ConA).

6.4 Arrays on Polystyrene

The use of non-covalent polystyrene slides has the advantage that it can be applied directly on microtiter plates and is therefore convenient and cheap. C-H Wong and co-workers showed that arrays of glycan-carrying fatty acids could be formed on polystyrene by hydrophobic interaction. They were able to analyse glycan-protein interactions proteins and study enzyme inhibitors using this method.35,36 K-Y Wong et al. improved this system by using glycans which were linked with triphenylmethyl (trityl) groups.37 These trityl groups carried saccharides and served as ’anchors’ to the polystyrene via hydrophobic interaction. Using this method, it was possible to analyse lectin binding to the immobilised carbohydrate ligands. The system described in chapters9 and 10 used the S-tritylated glycan 6 with different linker length for array formation. Furthermore, the ﬁrst MALDI-ToF MS analysis directly on non-modiﬁed polystyrene was shown by application of conducting tape. Also, as the trityl groups worked as a ’self-matrix’, no additional matrix application was required.

22 CHAPTER SEVEN

PREPARATION OF AMINOETHYL GLYCOSIDES FOR GLYCOCONJUGATION

R. Sardzik, G. T. Noble, M. J Weissenborn, A. Martin, S. L. Flitsch, Preparation of aminoethyl glycosides for glycoconjugation, Beilstein J. Org. Chem. 2010, 6, 699–703.

In this project, M. J. Weissenborn synthesised and characterised the molecules 3 and 8: The other molecules were synthesised by R. Sardzik, G. T. Noble and A. Martin. The article was written by S. L. Flitsch, R. Sardzik and S. J. Webb.

23 Preparation of aminoethyl glycosides for glycoconjugation

Robert Šardzík, Gavin T. Noble, Martin J. Weissenborn, Andrew Martin, Simon J. Webb and Sabine L. Flitsch*

Full Research Paper Open Access

Address: Beilstein J. Org. Chem. 2010, 6, 699–703. Manchester Interdisciplinary Biocentre & School of Chemistry, The doi:10.3762/bjoc.6.81 University of Manchester, 131 Princess Street, Manchester, M1 7DN, UK Received: 26 May 2010 Accepted: 21 July 2010 Email: Published: 29 July 2010 Robert Šardzík - [email protected]; Sabine L. Flitsch* - [email protected] Guest Editor: T. K. Lindhorst

Abstract The synthesis of a number of aminoethyl glycosides of cell-surface carbohydrates, which are important intermediates for glycoarray synthesis, is described. A set of protocols was developed which provide these intermediates, in a short number of steps, from commercially available starting materials.

Introduction The chemical conjugation of carbohydrates through the thesis. Here we describe a systematic study with the aim of anomeric centre to biomolecules such as peptides, proteins, finding such robust and efficient methods for a number of lipids, metabolites and to array surfaces is an important commonly used mono- and disaccharides starting from synthetic challenge [1-5]. A diverse range of linkers and spacers commercially available reagents and with a minimal number of has been described in the literature [2-12], among which steps. In our studies no such general aminoethylation method aminoalkyl glycosides have become the most popular, in par- that was applicable to all targets was found, which rather ticular the aminoethyl linker. This linker has been tested in a suggests that protocols need to be tailored for each sugar. large number of arrays and appears to be biocompatible in array screening [2,3,7]. Given that aminoethyl glycosides are Results and Discussion conveniently conjugated to surfaces containing activated Coupling reactions carboxylates, they have become a useful generic anomeric func- Aminoethyl glycosides have previously been generated in a tional group for glycoconjugation. The importance of this linker number of ways. Free sugars have been glycosylated with merits efforts into finding a robust synthetic method than can be 2-chloroethanol under acid catalysis, followed by peracetyla- used by scientists who are not experienced in carbohydrate syn- tion, nucleophilic substitution with azide and finally, reduction

699 Beilstein J. Org. Chem. 2010, 6, 699–703.

Figure 1: Aminoethyl glycosides (1–9) which were synthesised in this study. of the azido group [13,14]. Alternatively, the carbohydrate was first activated as the trichloroacetimidate or bromide followed by glycosylation with N-Cbz-aminoethanol [15], bromoethanol [16] or azidoethanol [17] and subsequently transformed into the amine. Scheme 1: General reaction scheme for generation of aminoethyl glycosides. X = OAc, Br or Cl. In the interest of finding fast reliable methods, we have investigated two general aminoethylation protocols: First, the direct glycosylation of peracetylated sugars, which can be either with moderate selectivity and from the reaction mixture the purchased or easily prepared from free sugars and can be used pure β-anomer (>95:5) was isolated by column chromato- without purification. Where these proved to be unreactive, the graphy. anomeric acetates were converted to glycosyl bromides, usually in quantitative yields, and products were used without further The glucoside 11, galactoside 12 and mannoside 13 were purification. Where possible, N-Cbz-protected aminoethanol prepared in moderate to good yield directly from the acetate was used as the glycosyl acceptor because it is commercially (X = OAc in Scheme 1) giving rapid access to these monosac- available, crystalline and can be easily deprotected in one step charide derivatives. avoiding use of azides. Figure 1 lists the target aminoethyl glycosides (1–9) generated in this study (q.v. Scheme 1 and The fucoside 14 was generated both from the acetate and bro- Table 1). mide. Higher alpha selectivity was observed with the bromide (α/β ratio of crude product 82:18). The key glycosylation step is shown in Scheme 1 and the results of the different glycosylation reactions are summarised in N-Acetylglucosamine 15 was successfully prepared from the

Table 1. β-acetate using SnCl4 (method D). When starting from the α-acetate, no reaction was observed and only starting material Fully protected xylopyranoside 10 could be prepared both from could be recovered. Under microwave conditions (method B or the corresponding bromide as well as the acetate (the β-anomer using other Lewis acids as Yb(OTf)3 in DCM, 90 °C, 30 min, was prepared from xylose with sodium acetate in acetic anhy- 200 W) the reaction was not reproducible, giving low yields and dride) in similar yields. In each case both anomers were formed leading to decomposition products.

700 Beilstein J. Org. Chem. 2010, 6, 699–703.

Table 1: Results of glycosylation reactions as shown in Scheme 1.

Entry Product X Method α/βc Yieldd

1 Br A 17:83 67 + 23e 10 2 10 Br B 23:77 58 + 19e 3 10 OAc C 22:78 52 + 17e

4 OAc C 15:85 36

5 OAc C 8:92 62

6 OAc C >95:5 57

7 Br A 14:86 75 + 19e

14 8 14 Br B 15:85 75 + 20e 9 14 OAc C 35:65 n.d.

10 OAc D >5:95 61

11 Br A 10:90 56

16 12 16 Br B 10:90 86 13 16 Br Aa 31:68 30 14 16 Br Bb 30:70 73e 15 16 Br E 37:63 59

16 Br E 10:90 88

701 Beilstein J. Org. Chem. 2010, 6, 699–703.

Table 1: Results of glycosylation reactions as shown in Scheme 1. (continued)

17 Br B 14:86 47

18 Cl F 10:90 70

19 a b c 13 d The reaction was performed overnight at r.t. The reaction was performed in CH2Cl2. determined by C NMR from crude reaction mixture. Yield of e pure major anomer (>95:5) after column chromatography. Mixture of both anomers. Method A: Hg(CN)2, CH3CN, 60 °C, 2–4 h. Method B: Hg(CN)2, CH3CN, 90 °C, microwave, 200 W, 15 min. Method C: BF3·Et2O, CH3CN, 0 °C, 1 h, r.t., overnight. Method D: SnCl4, CH3CN, 60 °C, 16 h. Method E: Hg(CN)2, HgBr2, CH3CN, r.t., overnight. Method F: Ag2CO3, CH2Cl2, r.t., overnight.

Lactose is both cheap and readily available. It is an important anomers were formed. The best reaction conditions were component of glycoprotein glycans and also a substrate for combined to give Method A. The success of Method A led to sialyltransferases to generate biologically important sialyllacto- attempts to improve the method further and to use microwave sides. The aminoethyl lactoside 16 was prepared in greatest irradiation as in method B. Method B also works well for mono- yield from the bromide and attempts to prepare 16 directly from saccharides and maltoside. the acetate using BF3·Et2O as the activator only resulted in decomposition. Given the problems with purification, the use of azidoethanol as a glycosyl acceptor was also investigated. This reaction A number of reaction conditions for generating 16 from the bro- (Table 1) was much more successful and produced mainly the mide with N-Cbz-aminoethanol were investigated. With beta anomer 17.

Ag2CO3 in dichloromethane at room temperature low yields of product 16 along with a number of side-products (orthoester, Maltoside 18 was generated by the same microwave-mediated elimination or hydrolysis) were observed and the product was glycosylation as developed for lactoside 16 (Method B) and in difficult to separate from starting materials, in particular the reasonable yield.

N-Cbz-aminoethanol. With Hg(CN)2, or the more reactive Hg(CN)2/HgBr2-mixture, in dichloromethane or acetonitrile, N-Acetyl neuraminic acid (sialic acid) is an important glycosylation was more successful, but both anomers were component of cell surfaces and chemical glycosylation pro- generated. Given the problems previously encountered with cedures involving sialic acid are generally challenging. In our purification, the glycosylation with Hg(CN)2 was further opti- hands activation as the chloride (prepared from Neu5Ac in 3 mised by increasing both the temperature and the amount of steps) using silver carbonate (Method F) gave reasonable yields acceptor. The problem of separation of the alcohol from the of 19. product was solved by acetylation of the crude reaction mixture to lower the polarity of the free alcohol. Attempts to speed up Deprotection reactions the reaction by heating led to the observation that in acetoni- The general deprotection for compounds 10–18 is shown in trile predominantly one (β) anomer is formed, but anomerisa- Scheme 2. Acetates were cleaved using NaOMe followed by tion occurs with longer reaction times. In dichloromethane both hydrogenation to generate 1–8 in good yields.

Scheme 2: Deprotection protocols.

702 Beilstein J. Org. Chem. 2010, 6, 699–703.

Deprotection was also successful when the hydrogenation was 7. Zhi, Z.; Laurent, N.; Powell, A. K.; Karamanska, R.; Fais, M.; performed first, but in some cases migration of acetate to the Voglmeir, J.; Wright, A.; Blackburn, J. M.; Crocker, P. R.; Russell, D. A.; Flitsch, S. L.; Field, R. A.; Turnbull, J. E. ChemBioChem aminoethyl linker was observed. However, this can be avoided 2008, 9, 1568–1575. doi:10.1002/cbic.200700788 by using palladium hydroxide on charcoal as the hydrogenation 8. Karamanska, R.; Clarke, J.; Blixt, O.; MacRae, J. I.; Zhang, J. Q.; catalyst, with short reaction times, followed by the immediate Crocker, P. R.; Laurent, N.; Wright, A.; Flitsch, S. L.; Russell, D. A.; use of the resulting amine in further coupling [18]. Field, R. A. Glycoconjugate J. 2008, 25, 69–74. doi:10.1007/s10719-007-9047-y Sialoside 19 was deprotected by treatment with NaOMe, 9. Laurent, N.; Voglmeir, J.; Wright, A.; Blackburn, J.; Pham, N. T.; Wong, S. C. C.; Gaskell, S. J.; Flitsch, S. L. ChemBioChem 2008, 9, followed by LiOH and subsequent hydrogenation to give 9. 883–887. doi:10.1002/cbic.200700692 10. Laurent, N.; Haddoub, R.; Flitsch, S. L. Trends Biotechnol. 2008, 26, Conclusion 328–337. doi:10.1016/j.tibtech.2008.03.003 We have described rapid and convenient methods for the syn- 11. Laurent, N.; Voglmeir, J.; Flitsch, S. L. Chem. Commun. 2008, thesis of a range of aminoethyl glycosides (1–9) of common 4400–4412. doi:10.1039/b806983m 12. Laurent, N.; Haddoub, R.; Voglmeir, J.; Wong, S. C. C.; Gaskell, S. J.; mono- and disaccharides. Although some of the glycosylation Flitsch, S. L. ChemBioChem 2008, 9, 2592–2596. reactions could be improved by using alternative glycosylation doi:10.1002/cbic.200800481 methods (such as trichloroacetimidates, thiols), these would 13. Ni, J.; Singh, S.; Wang, L.-X. Bioconjugate Chem. 2003, 14, 232–238. require more steps with chromatographic purifications and less doi:10.1021/bc025617f overall yields. These aminoethyl glycosides are now readily 14. Sanki, A. K.; Mahal, L. K. Synlett 2006, 455–459. accessible for incorporation into glycan arrays. doi:10.1055/s-2006-926264 15. Orlandi, S.; Annuziata, R.; Benaglia, M.; Cozzi, F.; Manzoni, L. Tetrahedron 2005, 61, 10048–10060. doi:10.1016/j.tet.2005.08.018 16. Park, S.; Shin, I. Org. Lett. 2007, 9, 1675–1678. doi:10.1021/ol070250l Supporting Information 17. Chernyak, A. Y.; Sharma, G. V. M.; Kononov, L. O.; Krishna, P. R.; Carbohydr. Res. A Supporting Information containing all experimental Levinsky, A. B.; Kochetkov, N. K.; Rama Rao, A. V. 1992, 223, 303–309. doi:10.1016/0008-6215(92)80029-Z details and analytical data of all compounds described in 18. Noble, G. T.; Flitsch, S. L.; Liem, K. P.; Webb, S. J. the article as well as their precursors is available. Org. Biomol. Chem. 2009, 7, 5245–5254. doi:10.1039/b910976e

Supporting Information File 1 Experimental procedures and analytical data [http://www.beilstein-journals.org/bjoc/content/ License and Terms supplementary/1860-5397-6-81-S1.pdf] This is an Open Access article under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which Acknowledgements permits unrestricted use, distribution, and reproduction in We are grateful to the EPSRC, BBSRC, EC (Marie Curie ITN) any medium, provided the original work is properly cited. and Royal Society (Wolfson Merit Award to SLF) for funding. The license is subject to the Beilstein Journal of Organic References Chemistry terms and conditions: 1. Lepenies, B.; Seeberger, P. H. Immunopharmacol. Immunotoxicol. (http://www.beilstein-journals.org/bjoc) 2010, 32, 196–207. doi:10.3109/08923970903292663 2. Blixt, O.; Han, S.; Liao, L.; Zeng, Y.; Hoffmann, J.; Futakawa, S.; Paulson, J. C. J. Am. Chem. Soc. 2008, 130, 6680–6681. The definitive version of this article is the electronic one doi:10.1021/ja801052g which can be found at: 3. Huflejt, M. E.; Vuskovic, M.; Vasiliu, D.; Xu, H.; Obukhova, P.; doi:10.3762/bjoc.6.81 Shilova, N.; Tuzikov, A.; Galanina, O.; Arun, B.; Lu, K.; Bovin, N. Mol. Immunol. 2009, 46, 3037–3049. doi:10.1016/j.molimm.2009.06.010 4. Bejugam, M.; Flitsch, S. L. Org. Lett. 2004, 6, 4001–4004. doi:10.1021/ol048342n 5. Macmillan, D.; Daines, A. M.; Bayrhuber, M.; Flitsch, S. L. Org. Lett. 2002, 4, 1467–1470. doi:10.1021/ol025627w 6. Hartmann, M.; Horst, A. K.; Klemm, P.; Lindhorst, T. K. Chem. Commun. 2010, 46, 330–332. doi:10.1039/b922525k

703 7.1. SUPPORTING INFORMATION

7.1 Supporting Information

The supporting information given below is speciﬁc to those experiments carried out by M. J. Weissenborn. For the complete supporting information please see http://www.beilstein- journals.org/bjoc/ content/supplementary/1860-5397-6-81-S1.pdf.

24 Experimental Procedures and Analytical Data for

Preparation of aminoethyl glycosides for glycoconjugation

Robert Šardzík1, Gavin T. Noble1, Martin J. Weissenborn1, Andrew Martin1, Simon Webb,1 Sabine L. Flitsch*1

Address: 1Manchester Interdisciplinary Biocentre & School of Chemistry, The University of Manchester, 131 Princess Street, Manchester, M1 7DN, UK

Email: Sabine L. Flitsch - [email protected] * Corresponding author

General Methods

Unless stated otherwise, all chemicals were of analytical grade and used as received from Sigma-Aldrich. All solvents used were from commercial suppliers (Sigma-Aldrich, Fisher Scientific or Romil). Microwave mediated reactions were performed on an CEM-Discover SP-D workstation fitted with an automated Explorer 24/48 module in pressurised vessels (10 or 35 mL) with snap-on caps. NMR spectra were recorded on Bruker Avance 300, DPX400 and Avance II+ 500 spectrometers at room temperature and calibrated according to the chemical shift of tetramethysilane or 3-(trimethylsilyl)propionic-2,2,3,3-d4 acid sodium salt 1 13 for samples in D2O (δ = 0 ppm). Compound spectra were assigned with H, C, DEPT, COSY, HSQC, HMQC and HMBC NMR experiments as appropriate. Chemical shifts are given in ppm, coupling constants in Hertz (Hz) and multiplicities indicated with the appropriate abbreviations: singlet (s), doublet (d), triplet (t), double doublet (dd), double double doublet (ddd) and multiplet (m). The determination of diastereomeric ratios are based on comparison of signal intensities of separated signal pairs in 13C NMR spectra. ES+ mass spectra were obtained with Micromass Prospec and Micromass Platform spectrometers. IR spectra were measured and recorded using a PerkinElmer Sprctrum RX I FT-IR Spectrometer. Optical activity was measured using an Optical Activity Ltd AA-1C00 polarimeter. Melting points were measured with a Gallenkamp apparatus and are not corrected.

General procedures

General Procedure 1; Glycosylation Method A

Glycosyl bromide and N-Cbz-aminoethanol (2 eq.) were dissolved in abs. CH3CN under nitrogen. Hg(CN)2 (1.1 eq.) was added and the reaction mixture was heated to 70 °C for 2–4 h. After this time, the solvent was evaporated in vacuo, the residue re-dissolved in CH2Cl2, washed with water, NaHCO3-solution, brine, dried over MgSO4 and concentrated in vacuo. The crude product was purified by column chromatography on silica.

General Procedure 2; Glycosylation Method B

To a solution of glycosyl bromide and N-Cbz-aminoethanol (2 eq.) in abs. CH3CN in a microwave vessel was added Hg(CN)2 (1.1 eq.), the vessel was closed and the reaction mixture heated to 90 °C for 15 min using 200 W power. The solvent was removed under vacuum, the residue re-dissolved in CH2Cl2, washed with water, NaHCO3-solution, brine, dried over MgSO4 and concentrated in vacuo. The crude product was purified by column chromatography on silica.

General Procedure 3; Glycosylation Method C

Peracetylated glycosyl acceptor (1 eq.) and N-Cbz-ethanolamine (1.2 eq.) were dissolved in . dry acetonitrile under nitrogen. The solution was cooled to 0 °C and BF3 Et2O (5 eq.) was added dropwise. The reaction was stirred 30 min at 0 °C and overnight at r.t. The reaction was quenched with Et3N and concentrated in vacuo, the residue re-dissolved in dichloromethane then washed once with sat. NaHCO3, water and brine. The organic layers were dried over Na2SO4, the solvent removed under reduced pressure and the product purified using column chromatography on silica.

General Procedure 4; Deacetylation with sodium methanolate

Peracetylated 2-(benzyloxycarbonyl)aminoethyl glucoside was dissolved in methanol and NaOMe in methanol (0.33 eq.) was added. The reaction was then stirred for 3 h to o/n at r.t. The base was neutralised with activated Amberlite IR-120, the resin was then removed via filtration and the solvent evaporated in vacuo to yield the product.

General Procedure 5; Hydrogenolysis of N-Cbz-protecting groups

2-(Benzyloxycarbonyl)aminoethyl glycoside was dissolved in MeOH and Pd/C (10 %) was added. The reaction was then stirred under a H2 atmosphere for 2 h to o/n. The solution was then filtered through Celite and the solvent removed under reduced pressure to yield the free amine.

Synthesis of aminoethyl galactoside 3

2-(Benzyloxycarbonyl)aminoethyl 2,3,4,6-tetra-O-acetyl--D-galactopyranoside (12)

Pentaacetyl β-D-galactose (1.95 g, 5.00 mmol) was coupled with N-Cbz-ethanolamine in 6 h as described in Method C. The product was purified using column chromatography (ethyl acetate/hexane 50:50) to yield 12 as a clear oil (1.64 g, 62 %).

21 22 20  3  4   D = −1.4 (c 1.0, CHCl3), Lit.  D = +4.4 (c 1.2, CH2Cl2), Lit.  D = +20.7 (c 1, 1 CH2Cl2); H NMR (500 MHz, CDCl3): δ (ppm) = 1.90 (s, 3H, COCH3), 1.93 (s, 3H,

COCH3), 1.95 (s, 3H, COCH3), 2.07 (s, 3H, COCH3), 3.32 (m, 2H, CH2NH), 3.62 (ddd, J =

3.6, 7.1, 10.2 Hz, 1H, CHaHbCH2NH), 3.80–3.82 (m, 2H, 5-H, CHaHbCH2NH), 4.06 (d, J =

6.6 Hz, 2H, 6-H2), 4.38 (d, J = 7.9 Hz, 1H, 1-H), 4.93 (dd, J = 3.4, 10.5 Hz, 1H, 3-H), 5.02 (s,

2H, CH2Ph), 5.10 (dd, J = 8.0, 10.4 Hz, 1H, 2-H), 5.19 (t, J = 5.4 Hz, 1H, NH), 5.31 (dd, J = 13 0.7, 3.4 H, 1H, 4-H), 7.22–7.30 (m, 5H, C6H5); C NMR (126 MHz, CDCl3) δ (ppm) = 20.5–

20.7 (4 q, 4 COCH3), 40.8 (t, CH2NH), 61.3 (t, C-6), 66.7 (t, CH2Ph), 67.0 (d, C-4), 68.8 (d,

C-2), 69.4 (t, OCH2CH2NH), 70.7 (d, C-3, C-5), 101.5 (d, C-1), 128.1, 128.5 (2 d, o-, m-, p-C from C6H5), 136.5 (s, i-C from C6H5), 156.3 (s, NCOO), 169.6, 170.1, 170.2, 170.4 (4 s, 4 ~ −1 COCH3); IR:  (cm ) = 3393, 2947, 1743, 1714, 1524, 1371, 1222, 1048; HRMS (ESI+): + m/z calcd for C24H31NO12 [M+Na] 548.1744, found 548.1747.

2-(Benzyloxycarbonyl)aminoethyl -D-galactopyranoside (23) Tetraacetate 12 (1.62 g, 3.08 mmol) was deacetylated as described above to obtain galactoside 23 in quantitative yield (1.10 g, 3.08 mmol) as clear oil.

21 1  D = +1.8 (c 1.0, MeOH); H NMR (400 MHz, MeOD): δ (ppm) =3.30 (ddd, J = 4.2, 6.8,

14.2 Hz, 1H, CHaHbNH), 3.40 (ddd, J = 4.1, 6.2, 14.2 Hz, 1H, CHaHbNH), 3.46 (dd, J = 3.2, 9.7 Hz, 1H, 3-H), 3.50 (ddd, J = 1.0, 5.3, 6.8 Hz, 1H, 5-H), 3.52 (dd, J = 7.3, 9.8 Hz, 1H, 2-

H), 3.63 (ddd, J = 4.0, 6.8, 10.5 Hz, 1H, CHaHbCH2NH), 3.70 (dd, J = 5.3, 11.4 Hz, 1H, 6-

Ha), 3.75 (dd, J = 6.9, 11.3 Hz, 1H, 6-Hb), 3.82 (dd, J = 1.0, 3.2 Hz, 1H, 4-H), 3.91 (ddd, J =

4.2, 6.2, 10.4 Hz, 1H, CHaHbCH2NH), 4.22 (d, J = 7.3 Hz, 1H, 1-H), 5.06 (s, 2H, CH2Ph), 13 7.24–7.37 (m, 5H, C6H5); C NMR (101 MHz, MeOD) δ (ppm) = 42.0 (t, CH2NH), 62.4 (t,

C-6), 67.4 (t, CH2Ph), 69.9 (t, CH2CH2NH), 70.2 (d, C-4), 72.5 (d, C-2), 74.8 (d, C-3), 76.6

(d, C-5), 105.0 (d, C-1), 128.8, 129.0, 129.5 (3 d, o-, m-, p-C from C6H5), 138.3 (s, i-C from −1 C6H5), 158.9 (s, NCOO); IR: (cm ) = 3403, 1643, 1264, 1077; HRMS (ESI+): m/z calcd + for C16H23NO8 [M+Na] 380.1316, found 380.1308.

2-Aminoethyl -D-galactopyranoside (3) Prepared by hydrogenation (General Procedure 5) from 23 (1.00 g, 3.89 mmol) in MeOH (30 mL) in 6 h. Yield: 806 mg (3.61 mmol, 93 %), colourless oil.

20 5  1  D = −12.9 (c 2.4, MeOH), Lit.  D = −11.3 (c 0.23, MeOH); H NMR (400 MHz,

MeOD): δ (ppm) = 2.80 (ddd, J = 4.2, 6.3, 13.4 Hz, 1H, CHaHbNH2), 2.84 (ddd, J = 4.4, 5.5,

13.4 Hz, 1H, CHaHbNH2), 3.45 (dd, J = 3.3, 9.7 Hz, 1H, 3-H), 3.49 (ddd, J = 1.0, 5.3, 7.0 Hz, 1H, 5-H), 3.52 (dd, J = 7.5, 9.7 Hz, 1H, 2-H), 3.61 (ddd, J = 4.4, 6.3, 10.5 Hz, 1H,

CHaHbCH2NH2), 3.69 (dd, J = 5.3, 11.3 Hz, 1H, 6-Ha), 3.73 (dd, J = 7.0, 11.3 Hz, 1H, 6-Hb),

3.80 (dd, J = 1.0, 3.3 Hz, 1H, 4-H), 3.91 (ddd, J = 4.2, 5.5, 10.3 Hz, 1H, CHaHbCH2NH2), 13 4.21 (d, J = 7.5 Hz, 1H, 1-H); C NMR (101 MHz, MeOD) δ (ppm) = 42.2 (t, CH2NH2),

62.5 (t, C-6), 70.3 (d, C-4), 71.9 (t, CH2CH2NH2), 72.6 (d, C-2), 74.9 (d, C-3), 76.7 (d, C-5), 105.1 (d, C-1); IR: (cm−1) = 3320, 3359, 2929, 2887, 1645, 1598, 1073,1042; HRMS + (ESI+): m/z calcd for C8H17NO6 [M+H] 224.1134, found 224.1133.

Synthesis of aminoethyl maltoside 8

2-(Benzyloxycarbonyl)aminoethyl 2,3,4,6-tetra-O-acetyl-α-D-glucopyranosyl-(1→4)- 2,3,6-tri-O-acetyl-β-D-glucopyranoside (18)

To a solution of β-D-maltose octaacetate (1.00 g, 1.47 mmol) in dry CH2Cl2 (5 mL) was added acetic anhydride (0.2 mL) and the solution was cooled to 0 °C. HBr in acetic acid (2 mL, 33 %) was added dropwise and the reaction mixture was stirred for 0.5 h at 0 °C and 2 h at r.t. The solution was diluted with CH2Cl2 (20 mL) and washed with water (10 mL),

NaHCO3 solution (2x10 mL) and brine (10 mL), dried over MgSO4 and concentrated in vacuo. The obtained crude maltose bromide was glycosylated according to General Procedure 2 to yield maltoside 18 (560 mg, 0.69 mmol, 47%) as a white solid.

20  1 m.p. = 59–62 °C;  D = +57.4 (c 2.4, CHCl3); H NMR (400 MHz, CDCl3): δ (ppm) = 1.91,

1.93, 1.94, 1.96, 1.98, 1.99, 2.03 (6 s, 21H, COCH3), 3.27 - 3.34 (m, 2H, CH2N), 3.56 (ddd, J

= 2.7, 4.2, 9.6 Hz, 1H, 5-H), 3.60 – 3.70 (m, 1H, OCHaHb), 3.72 – 3.81 (m, 1H, OCHaHb), 3.83 (dd, J = 4.2, 10.1 Hz, 1H, 5'-H), 3.9 (t, J = 9.1 Hz, 1H, 4-H), 4.0 (dd, J = 2.1, 12.4 Hz,

1H, 6'-Ha), 4.1 (dd, J = 4.4, 12.2 Hz, 1H, 6-Ha), 4.2 (dd, J = 3.8, 12.5 Hz, 1H, 6'-Hb), 4.4 (d, J

= 8.1 Hz, 2H, 1-H, 6-Hb), 4.7 (dd, J = 8.1, 9.5 Hz, 1H, 2-H), 4.8 (dd, J = 4.0, 10.5 Hz, 1H, 2'-

H), 4.99 (t, J = 9.9 Hz, 1H, 4'-H), 5.02 – 5.06 (m, 2H, CH2Ph), 5.17 (t, J = 9.2 Hz, 1H, 3-H), 13 5.28 (t, J = 10.0 Hz, 1H, 3'-H), 5.34 (d, J = 3.9 Hz, 1H, 1'-H), 7.2 – 7.3 (m, 5H, C6H5); C

NMR (101 MHz, CDCl3) δ (ppm) = 20.75, 20.77, 20.79, 20.87, 20.95, 21.07, 21.13 (7 q, 7

COCH3); 41.1 (t, CH2N), 61.61 (t, C-6'), 62.8 (t, C-6), 66.9 (t, CH2Ph), 68.1 (d, C-5'), 68.7 (d,

C-3'), 69.5 (t, OCH2), 69.8 (d, C-2'), 70.1 (d, C-2), 72.2 (d, C-5), 72.4 (d, C-4), 72.6 (d, C-4'),

75.4 (d, C-3), 95.7 (d, C-1'), 100.8 (d, C-1), 128.4, 128.7 (2 d, o-, m-, p-C from C6H5), 136.6

(s, i-C from C6H5), 156.5 (s, NCOO), 169.6, 169.9, 170.1, 170.4, 170.65, 170.72 (6 s, 7 ~ −1 COCH3); IR:  (cm ) = 3394, 2959, 1752, 1523, 1432, 1369, 1238, 1040; HRMS (ESI+): + m/z calcd for C36H47NO20 [M+H] 814.2770, found 814.2772.

2-Aminoethyl α-D-glucopyranosyl-(1→4)-β-D-glucopyranoside (8) The glycosylated maltose 18 (304 mg, 0.374 mmol) was first deacetylated according to General Procedure 4 followed by hydrogenation as described in General Procedure 5. The free maltoside 8 (144 mg, 0.372, 99 %) was obtained as a white solid.

20 −1 m.p. = 165–167 °C;  D = +69.7 (c 6.4, MeOH); IR: (cm ) = 3346, 1643, 1039; HRMS + (ESI+): m/z calcd for C14H27NO11 [M+H] 386.1662, found 386.1662.

1. Hudson, C. S.; Johnson, J. M., J. Am. Chem. Soc. 1915, 37 (12), 2748-2753. 2. Schroede.Lr; Counts, K. M.; Haigh, F. C., Carbohydr. Res. 1974, 37 (2), 368-372. 3. Orlandi, S.; Annunziata, R.; Benaglia, M.; Cozzia, F.; Manzoni, L., Tetrahedron 2005, 61 (42), 10048-10060. 4. Denoyelle, S.; Polidori, A.; Brunelle, M.; Vuillaume, P. Y.; Laurent, S.; ElAzhary, Y.; Pucci, B., New J. Chem. 2006, 30 (4), 629-646. 5. Susaki, H.; Suzuki, K.; Ikeda, M.; Yamada, H.; Watanabe, H. K., Chem. Pharm. Bull. 1994, 42 (10), 2090-2096. 6. Kamst, E.; Zegelaar-Jaarsveld, K.; van der Marel, G. A.; van Boom, J. H.; Lugtenberg, B. J. J.; Spaink, H. P., Carbohydr. Res. 1999, 321 (3-4), 176-189. 7. King, R. R.; Cooper, F. P.; Bishop, C. T., Carbohydr. Res. 1977, 55 (MAY), 83-93. 8. Ogura, H.; Furuhata, K.; Itoh, M.; Shitori, Y., Carbohydr. Res. 1986, 158, 37-51. 9. Roy, R.; Laferriere, C. A., Can. J. Chem. 1990, 68 (11), 2045-2054. 10. Eschenfelder, V.; Brossmer, R., Carbohydr. Res. 1980, 78 (1), 190-194.

CHAPTER EIGHT

OXO-ESTER MEDIATED NATIVE CHEMICAL LIGATION ON MICROARRAYS: AN EFFICIENT AND CHEMOSELECTIVE COUPLING METHODOLOGY

Reproduced by permission of The Royal Society of Chemistry

M. J. Weissenborn, R. Castangia, J. W. Wehner, R. Sardzik, Th. K. Lindhorst, S. L. Flitsch, Oxo-Ester Mediated Native Chemical Ligation on Microarrays: An Efﬁcient and Chemoselective Coupling Methodology, Chem. Commun. 2012, 48, 4444–4446.

The idea of applying Native Chemical Ligation to SAMs on gold was developed by J. W. Wehner, T. K. Lindhorst and M. J. Weissenborn. The change from thioesters towards oxo-esters was proposed by S. L. Flitsch and the use of pentaﬂuorophenol activated esters by R. Castangia. J. W. Wehner synthesised and characterised the carbohydrate molecules and R. Castangia synthesised the peptide 11. R. Sardzik and M. J. Weissenborn did the SPR analysis. R. Castangia and M. J. Weissenborn designed all coupling experiments and M. J. Weissenborn carried them out. S. L. Flitsch, T. K. Lindhorst, J. W. Wehner and M. J. Weissenborn wrote the article and J. W. Wehner, R. Castangia and M. J. Weissenborn the supporting information.

25 View Online / Journal Homepage / Table of Contents for this issue ChemComm Dynamic Article Links

Cite this: Chem. Commun., 2012, 48, 4444–4446

www.rsc.org/chemcomm COMMUNICATION

Oxo-ester mediated native chemical ligation on microarrays: an eﬃcient and chemoselective coupling methodologyw

Martin J. Weissenborn,za Roberto Castangia,za Johannes W. Wehner,b Robert Sˇ ardzı´k,a Thisbe K. Lindhorst*b and Sabine L. Flitsch*a

Received 6th February 2012, Accepted 19th March 2012 DOI: 10.1039/c2cc30844d

We report a highly eﬃcient and selective method for the coupling of peptides and glycoconjugates bearing N-terminal cysteines to activated surfaces. This chemoselective method generates stable amide linkages without using any thiol additives.

Fig. 1 The native chemical ligation (NCL) performed on thioester- Peptide and protein microarrays are well established in the functionalised surfaces. field of proteomics and provide platform technologies for drug discovery and clinical diagnosis.1 Such arrays are mostly generated by covalent coupling of pre-prepared ligands to solid surfaces.2 The conjugation chemistry needs to be highly chemoselective, site-specific and efficient to ensure correct presentation of the ligand on the surface. Next to rather classical chemical acylation methods, more specific bioorthogonal coupling 3 4 methods have been developed such as Staudinger ligation, ‘‘click Fig. 2 The chemoselective coupling of Cys derivatives on PFP- 5 6,7 chemistry’’, and native chemical ligation (NCL). activated gold plates is a powerful method for oxo-ester mediated NCL is particularly attractive as it can be carried out under NCL on surfaces. physiological conditions.8 NCL generally involves the reaction

Downloaded by John Rylands University Library on 14 April 2012 of a thioester and a cysteine (Cys) derivative, such as an selectivity of cysteines over other amines could be retained in

Published on 20 March 2012 http://pubs.rsc.org | doi:10.1039/C2CC30844D N-terminal Cys-peptide. If the NCL is performed on solid the coupling reaction. phase, either the Cys- or the thioester-peptide needs to be All our experiments were conducted on gold arrays covered immobilised.6,9 Starting with immobilised Cys requires thioester- with self-assembled monolayers (SAMs) formed by alkane 14 peptides, which are incompatible with Fmoc protecting groups in thiols, which are omega-functionalised with hexaethylene 10 solid phase peptide synthesis. The reverse coupling method glycol (OEG6) carboxylic acid, thus providing points for proceeds from thioester-functionalised surfaces 1 by coupling of chemical modification (4, Fig. 3). Any reaction on this surface the Cys derivative via a reversible transthioesterification to give can be monitored directly and label-free using matrix-assisted thioester 2 and the native peptide 3 after irreversible rearrangement laser desorption/ionisation-time-of-flight mass spectrometry 15 (Fig. 1). The formation of 1 has been shown first by Lesaicherre (MALDI-ToF MS). et al.11 albeit in only 1/10 of monoloyer coverage.8 In the first instance, a range of activated oxo-esters such as As an alternative, we have therefore investigated NCL para-nitrophenyl (pNP), N-hydroxysuccinimidoyl (NHS) and coupling protocols starting from readily available activated 2,3,4,5,6-pentafluorophenyl (PFP) esters were generated from oxo-ester 5 on surfaces (Fig. 2) instead of using thioester carboxy-functionalised surfaces and were reacted with Cys. surface 1.12,13 A concern using such oxo-esters would be higher Analysis of the reactions by MALDI-ToF mass spectrometry reactivity and it was therefore important to investigate if suggested that the coupling reactions were most efficient using PFP activation and this activation method was therefore chosen for subsequent studies. a School of Chemistry & Manchester Interdisciplinary Biocentre, The coupling of Cys itself was then further investigated for The University of Manchester, 131 Princess Street, efficiency of coupling and selectivity over other amino acids.16,17 Manchester M17DN, UK. E-mail: sabine.fl[email protected]; Fax: +44-(0)161-275-1311; Tel: +44-(0)161-306-5172 The coupling was tested at different Cys concentrations using b Otto Diels Institute of Organic Chemistry, Christiana Albertina the standard NCL conditions (7.5 eq. tris(2-carboxyethyl)- University of Kiel, Otto-Hahn-Platz 3/4, 24098 Kiel, Germany. phosphine hydrochloride (TCEP), 5 M guanidine buffer, 75 mM E-mail: [email protected]; Fax: +49-431-880-7410 Na HPO ,pH7)18 and it was found that coupling was w Electronic supplementary information (ESI) available. See DOI: 2 4 10.1039/c2cc30844d observed even at a concentration of 2 mM of Cys, which z These authors contributed equally to this work. compared favourably to the 50 mM previously necessary for

4444 Chem. Commun., 2012, 48, 4444–4446 This journal is c The Royal Society of Chemistry 2012 View Online

Table 1 Competitive assay employing N-Cys and N-amino peptides and glycoamino acids

Fig. 3 (a) Coupling experiments with 4-mercaptophenylacetic acid (MPAA). (b) MALDI-ToF MS analysis of 7. (c) MALDI-ToF MS analysis of 7.

peptide coupling. By addition of imidazole12,17 we were able to further lower the Cys concentration necessary to 0.25 mM. Thiophenol and 4-mercaptophenylacetic acid (MPAA) were less eﬀective. With a robust coupling protocol in hand, we investigated the selectivity of Cys coupling over other amino acids. Inter- estingly, when the activated surfaces 5 were reacted with

Downloaded by John Rylands University Library on 14 April 2012 glycine (Gly) or phenylalanine (Phe) in the presence of thiol

Published on 20 March 2012 http://pubs.rsc.org | doi:10.1039/C2CC30844D MPAA as additive, only thioester 8 was formed and no conversion to the respective amide was observed (Fig. 3). The same reaction with Cys and MPAA, however, showed the Cys amide 7 as the sole product. These experiments suggest 2 mM each on PFP activated SAMs. The SPR slide was that on the surface O-toS-transesterification (5 to 8) is fast. subsequently interrogated with the mannose binding lectin The thioester 8 consequently undergoes a reaction with Cys via concanavalin A (ConA) followed by the N-acetylglucosamine transthioesterification which rearranges to the amide 7. This (GlcNAc) binding lectin wheat germ agglutinin (WGA).19 difference in acylation rates was also observed in competition There was strong ConA and no WGA binding observed which experiments. When Gly, Phe and Cys were added to the showed that exclusively the Cys-Man 13 had coupled to the reaction mixture in equimolar amount in the presence of surface, but not 14. MPAA only the Cys-product 7 was formed. After these promising results with small amino acids and Interestingly, similar selectivities were observed in the peptides, the coupling method was tested on more complex presence of imidazole (instead of the thiol MPAA).12 Competitive bioconjugates. Of particular interest was the coupling of studies with mixtures of b-alanine (b-Ala), Phe, Gly and Cys glycoconjugates, which are used in glycoarrays20 and which (1 mM each) also showed exclusively the Cys product even at can often only be obtained in small quantities. Efficient 1 : 25 ratio of Cys : Gly (Fig. S17, ESIw). All subsequent reactions coupling strategies are therefore of great importance to the were therefore performed with imidazole as additive. field. The cysteine derivatives such as 13 are easily obtained Similar selectivities were observed with longer peptides from aminoethyl glycosides 15. In competition with 15, containing either cysteine or glycine at their N-termini cysteine derivative 13 was exclusively coupled to surfaces (Table 1). When mixtures of peptides 9 and 10, 11 and 10, giving the product 26 as shown in MALDI-ToF MS analysis. and 11 and 12, respectively, were applied to the surface, only To test this coupling strategy further, we investigated the products 24 or 25 were observed. multivalent glycoconjugates, which have become important As an alternative and more sensitive method of analysis, tools to probe multivalent binding events on cell surfaces.21 The surface plasmon resonance (SPR) was employed. 13 and 14 required trivalent glycoclusters22 were synthesised as shown in were applied in a competitive assay at a concentration of Scheme 1. Trivalent glycoclusters 21 and 23 were synthesized

This journal is c The Royal Society of Chemistry 2012 Chem. Commun., 2012, 48, 4444–4446 4445 View Online

2 Z. A. Gurard-Levin and M. Mrksich, Annu. Rev. Anal. Chem., 2008, 1, 767–800. 3 E. M. Sletten and C. R. Bertozzi, Acc. Chem. Res., 2011, 44, 666–676; C. P. R. Hackenberger and D. Schwarzer, Angew. Chem., Int. Ed., 2008, 47, 10030–10074. 4 A. Watzke, M. Ko¨hn, M. Gutierrez-Rodriguez, R. Wacker, H. Schroder, R. Breinbauer, J. Kuhlmann, K. Alexandrov, C. M. Niemeyer, R. S. Goody and H. Waldmann, Angew. Chem., Int. Ed., 2006, 45, 1408–1412; S. S. van Berkel, M. B. van Eldijk and J. C. M. van Hest, Angew. Chem., Int. Ed., 2011, 50, 8806–8827. 5 H. C. Kolb, M. G. Finn and K. B. Sharpless, Angew. Chem., Int. Ed., 2001, 40, 2004; M. Meldal and C. W. Tornoe, Chem. Rev., 2008, 108, 2952–3015. 6 J. A. Camarero, Y. Kwon and M. A. Coleman, J. Am. Chem. Soc., 2004, 126, 14730–14731. 7 S. B. H. Kent, Chem. Soc. Rev., 2009, 38, 338–351. Scheme 1 Synthesis of trivalent glycoclusters 21 and 23. 8 S. Anderson, Langmuir, 2008, 24, 13962–13968. 9 B. Helms, I. van Baal, M. Merkx and E. W. Meijer, ChemBio- starting from the known building block 17.23 For the preparation Chem, 2007, 8, 1790–1794. of the trivalent wedge 19 Fmoc-glycine (16) was coupled in 10 S. K. Mahto, C. J. Howard, J. C. Shimko and J. J. Ottesen, ChemBioChem, 2011, 12, 2488–2494. a HBTU/DIPEA-mediated reaction with 17 followed by 11 M. L. Lesaicherre, M. Uttamchandani, G. Y. J. Chen and deprotection of the tBu esters with formic acid yielding S. Q. Yao, Bioorg. Med. Chem. Lett., 2002, 12, 2079–2083. compound 19 in 85% over two steps. Mannosides were 12 G. M. Fang, H. K. Cui, J. S. Zheng and L. Liu, ChemBioChem, 24 2010, 11, 1061–1065. introduced via peptide coupling following published chemistry 13 Q. Wan, J. Chen, Y. Yuan and S. J. Danishefsky, J. Am. Chem. giving 20 in a satisfying yield of 64%. Fmoc deprotection with Soc., 2008, 130, 15814–15816. piperidine in DMF gave the trivalent cluster mannoside 21 in 14 E. Ostuni, R. G. Chapman, R. E. Holmlin, S. Takayama and 89% yield. An amount of 21 was further functionalized with G. M. Whitesides, Langmuir, 2001, 17, 5605–5620. 15 N. Laurent, R. Haddoub, J. Voglmeir, S. C. C. Wong, S. J. Gaskell Boc-L-Cys(Trt)-OH in another peptide coupling reaction yielding and S. L. Flitsch, ChemBioChem, 2008, 9, 2592–2596; 22 in 58%. This step was followed by one step removal of the B. T. Houseman and M. Mrksich, Chem. Biol., 2002, 9, 443–454. Boc and Trityl protecting groups using TFA in DCM under 16 P. E. Dawson, T. W. Muir, I. Clarklewis and S. B. H. Kent, addition of TES followed by a treatment with basic ion exchange Science, 1994, 266, 776–779. 25 17 E. C. B. Johnson and S. B. H. Kent, J. Am. Chem. Soc., 2006, 128, resin Amberlyst A-21 following a literature protocol. After 6640–6646. size exclusion and reversed phase chromatography the Cys- 18 P. E. Dawson and S. B. H. Kent, Annu. Rev. Biochem., 2000, 69, terminated ligand 23 was obtained in 65% yield. 923–960; D. Macmillan, M. De Cecco, N. L. Reynolds, The trivalent cluster mannosides manno-Cys 23 and manno- L. F. A. Santos, P. E. Barran and J. R. Dorin, ChemBioChem, 2011, 12, 2133–2136. Gly 21 were then compared in coupling experiments to the 19 N. Laurent, J. Voglmeir and S. L. Flitsch, Chem. Commun., 2008,

Downloaded by John Rylands University Library on 14 April 2012 gold array surfaces. From reaction mixtures containing 4 mM 4400–4412.

Published on 20 March 2012 http://pubs.rsc.org | doi:10.1039/C2CC30844D each of 21 and 23 the manno-Cys product 27 was formed 20 R. Sardzik, G. T. Noble, M. J. Weissenborn, A. Martin, S. J. Webb and S. L. Flitsch, Beilstein J. Org. Chem., 2010, 6, exclusively (Table 1). 699–703; O. Bohorov, H. Andersson-Sand, J. Hoffmann and Complex bioconjugates containing N-terminal cysteines can O. Blixt, Glycobiology, 2006, 16, 21c–27c; O. Blixt, S. Head, be efficiently coupled to surfaces through amide bond formation T. Mondala, C. Scanlan, M. E. Huflejt, R. Alvarez, via native chemical ligation starting with activated oxo-esters on M. C. Bryan, F. Fazio, D. Calarese, J. Stevens, N. Razi, D. J. Stevens, J. J. Skehel, I. van Die, D. R. Burton, the surface. These surface oxo-esters are easily prepared from I. A. Wilson, R. Cummings, N. Bovin, C. H. Wong and their carboxylic acids. Our studies have shown that the ligation in J. C. Paulson, Proc. Natl. Acad. Sci. U. S. A., 2004, 101, the presence of imidazole is highly selective for cysteine derivatives 17033–17038; M. Hartmann, A. K. Horst, P. Klemm and over competing amines. T. K. Lindhorst, Chem. Commun., 2010, 46, 330–332. 21 M. Hartmann and T. K. Lindhorst, Eur. J. Org. Chem., 2011, This work was supported by the Royal Society (Wolfson 3583–3609. Award to SLF), the European Commission (MJW), the 22 H. A. Shaikh, F. D. So¨nnichsen and T. K. Lindhorst, Carbohydr. EPSRC and the Evonik Foundation (JWW). Res., 2008, 343, 1665–1674. 23 G. R. Newkome, R. K. Behera, C. N. Moorefield and G. R. Baker, Notes and references J. Org. Chem., 1991, 56, 7162–7167. 24 J. W. Wehner and T. K. Lindhorst, Synthesis, 2010, 3070–3082; 1 G. MacBeath and S. L. Schreiber, Science,2000,289, 1760–1763; A. Schierholt, M. Hartmann and T. K. Lindhorst, Carbohydr. D. Weinrich, P. Jonkheijm, C. M. Niemeyer and H. Waldmann, Res., 2011, 346, 1519–1526. Angew. Chem., Int. Ed.,2009,48, 7744–7751; L. Berrade, 25 N. Srinivasan, A. Yurek-George and A. Ganesan, Mol. Diversity, A. E. Garcia and J. A. Camarero, Pharm. Res.,2011,28, 1480–1499. 2005, 9, 291–293.

4446 Chem. Commun., 2012, 48, 4444–4446 This journal is c The Royal Society of Chemistry 2012 8.1. SUPPORTING INFORMATION

8.1 Supporting Information

The supporting information given below is speciﬁc to those experiments carried out by M. J. Weissenborn. For the complete supporting information please see http://www.rsc.org/suppdata/cc/c2/c2cc30844d/c2cc30844d.pdf.

26 Supporting Information Oxo-Ester Mediated Native Chemical Ligation on Microarrays: An Efficient and Chemoselective Coupling Methodology

a a b Martin J. Weissenborn, Roberto Castangia, Johannes W. Wehner, Robert Šardzík, a Thisbe K. Lindhorst*b and Sabine L. Flitsch*a a School of Chemistry & Manchester Interdisciplinary Biocentre, The University of Manchester, 131 Princess Street, Manchester, M17DN (UK). Fax: 44- 161-275-1311; Tel: -44- 161-306-5172; E-mail: [email protected] b Otto Diels Institute of Organic Chemistry, Christiana Albertina University of Kiel, Otto-Hahn-Platz 3/4, 24098 Kiel, Germany. Fax: 49-431-880-7410; E-mail: [email protected]

1. General methods ...... 2

2. Coupling on Self-Assembled Monolayers (SAMs) ...... 3

3. MALDI-ToF MS spectra ...... 6

4. Surface Plasmon Resonance (SPR) Analysis ...... 24

S 1

1. General methods

Unless otherwise noted, all reagents including anhydrous solvents were obtained from commercial suppliers and used as delivered without further purification. Alkanethiol linkers [HS-(CH2)17-(OC2H4)3-OH and HS-(CH2)17-(OC2H4)6-OCH2COOH] were purchased from Prochimia Surfaces (Poland). The peptide CQDSETRTFY 9 was purchased from Bachem (UK). Reverse phase HPLC was performed on Agilent 1100 and Agilent 1200 systems using Phenomenex Luna 5 µm, 250 x 150 mm semi-preparative C18 column with UV detection at 210 nm. Unless otherwise stated all MALDI-ToF MS experiments were carried out on an Ultraflex II instrument (Bruker Daltonics) in positive reflectron mode. A solution of matrix (2,4,6- trihydroxyacetophenone, 10 mg/mL in acetone) was applied on the gold and allowed to dry before analysis. All spectra were analysed with FlexAnalysis software (Bruker, USA) using default integration settings. Calibration was either performed before the analysis at the Ultraflex II instrument or afterwards in FlexAnalysis. Unless otherwise noted, all m/z values refer to the [M+Na]+ ion and the corresponding disulphide which is formed during ionisation.

S 2

2. Coupling on Self-Assembled Monolayers (SAMs)

2.1 Preparation and activation of SAMs A disposable 64-well gold plate (Applied Biosystems) was cleaned with Piranha solution (12 mL, 5:1 conc. H2SO4/ 30% H2O2) for 30 min, rinsed with distilled water, ethanol and dried under a stream of nitrogen. A DMSO solution of carboxylic acid-terminated [HS- (CH2)17-(OC2H4)6-OCH2-COOH] and tri(ethylene glycol) [HS-(CH2)17-(OC2H4)3-OH] alkanethiols (final concentration 0.4 mg/mL, molar ratio 1:4) was applied on the plate (~1 µL per well) and left overnight at RT to form a mixed SAM. The plate was washed with ethanol and dried under nitrogen. The carboxylic group was activated with EDC and PFP (final concentrations 0.180 M and 0.174 M, respectively) in dry DMF for 0.5-1 h, followed by washing with water and ethanol and drying as above. The product formation was analysed by MALDI-ToF MS.

2.2 General coupling method of amino acids Cys, Gly, Phe and -Ala

General Native Chemical Ligation procedure of Cys 7 An aqueous buffer containing 5 M guanidine hydrochloride (Gn·HCl), 75 mM Na2HPO4, 20% MeCN was added to imidazole (2.5 M) and stock solution of amino acid(s) and Tris(2- carboxyethyl)phosphine hydrochloride (TCEP, 7.5 eq. relative to Cys). The pH was adjusted to 7 using 2 M HCl. For Cys concentrations other than 4 mM, the solution was diluted with Gn·HCl (75 mM Na2HPO4) to obtain the different cysteine concentrations. The reactions were carried out overnight and were directly analysed.

S 3

Reactions in absence of imidazole The Cys or Gly solutions without imidazole were prepared as above with the absence of imidazole and MeCN.

Coupling reactions of Gly The coupling reactions of Gly were carried out in the same way as above but in the absence of TCEP.

Reactions in the presence of MPAA According to the general procedure (see above) with the difference that the solution was used without imidazole and a saturated 4-mercaptophenylacetic acid (MPAA) solution was generated. Illustrated in Fig. 3 in the main document with the products 7 and 8.

Cys and Gly experiments in different ratios An aqueous buffer containing 5 M Guanidine (Gn·HCl), 75 mM Na2HPO4, 20% MeCN, and 2.5 M imidazole HCl (for reactions in absence of imidazole, no imidazole HCl and MeCN was used) was premixed and the pH was adjusted to 7 using 5 M NaOH. Cysteine (4 mM) and Gly (4-400 mM) with TCEP (7.5 eq. relative to Cys) were applied in various different ratios (1:1  1:100). The reactions were carried out overnight and were directly analysed.

Figure S1: The chemoselective coupling on PFP activated gold plates at a 25 times excess of Gly over Cys.

S 4

2.3 Method for Native Chemical Ligation (NCL) of N-terminal Cys-peptides (9, 11) and –carbohydrates (13, 23) Unless otherwise stated, an aqueous buffer containing 5 M Guanidine (Gn·HCl), 75 mM Na2HPO4, 20% MeCN, and 2.5 M imidazole HCl (for reactions in absence of imidazole, no imidazole HCl and MeCN was used) was premixed and the pH was adjusted to 7 using 5 M NaOH. Cysteine-derivatives (9, 11, 13, 23) and/or amino-derivatives (10, 12, 14, 15, 21) with TCEP (7.5 eq. relative to Cys-derivative; reactions of only the amino-derivative were carried out without TCEP) were added in various concentrations. Reactions were carried out overnight and were directly analysed.

S 5

3. MALDI-ToF MS spectra

3.1 Amino acids

A = [M+Na]+

A = [M+2 Na-H]+

A = [M+K]+

A = [M+Na+K-H]+

+ A = [M+2 K -H]+ B = [M+Na]

Figure S2 Coupling of Cys at a concentration of 0.25 mM. The different Na and K ions and salts are only here indicated for clarification. A is the disulphide of 4 and B is the disulphide of 7 as it is most commonly detected by MALDI-ToF MS on SAMs.

S 6

Figure S3 Coupling of Cys (2 mM) in presence (a) and absence (b) of imidazole. A is the disulphide of 4 and B is the disulphide of 7.

S 7

a) A

b) A

Figure S4 No coupling of Gly (2 mM) is observed in presence of imidazole (a) but in absence of imidazole (b).

S 8

Figure S5 The competitive coupling reaction of β-Ala, Gly, Phe and Cys (all 1 mM) in presence of imidazole. Only the Cys product was observed. The peak at 1123 m/z was detected in the negative control as well and is therefore not the β-Ala product. A is the disulphide of 4 and D is the disulphide of 7. This reaction is analogous to Fig. 3 in the main document with difference of imidazole instead of MPAA and additional -Ala.

S 9

a) A C

A E

c) E A

d) C

Figure S6 The reaction in presence of MPAA: a) Cys (4 mM) , b) Gly (4 mM), c) Phe (4 mM) and d) competitive with Gly, Phe and Cys (all 4 mM). A is the disulphide of 4, C is the disulphide of 7 and E is the disulphide of 8.

S 10

Figure S7 The chemoselective coupling on PFP activated gold plates at a 25 times excess of Gly over Cys. Only Cys product 7 was detected in MALDI-ToF MS. A is the disulphide of 4 and B is the disulphide of 7.

S 11

3.2 Peptides

A a)

b) A

Figure S8 The coupling of the peptide 9 (0.25 mM) in presence (a) and absence (b) of imidazole. All detected ions are the [M+H]+ ions. A is the disulphide of 4 and B is the disulphide of 24.

S 12

Figure S9 Coupling of the peptide 11 with imidazole at 0.25 mM. A is the disulphide of 4, B the thiol of 25 and C is the disulphide of 25.

S 13

A a) C

A b)

Figure S10 Coupling of the peptide 11 in presence of imidazole in a) 4 mM and b) 0.25 mM concentration. A is the disulphide of 4, B the thiol of 25 and C is the disulphide of 25.

S 14

a) A

C B

A C

Figure S11 Coupling of the Gly-peptide 10 at 4 mM in a) presence and b) absence of imidazole. A is the disulphide of 4, B the thiol of 10 and C is the disulphide of 10 on the surface.

S 15 a) A

A b)

Figure S12 Coupling of the Lysine containing peptide 12 at 2 mM in a) presence and b) absence of imidazole. A is the disulphide of 4, B the thiol of 25 and C is the disulphide of 25 on the surface.

S 16

a) C

b) D

A B

Figure S13 Competitive coupling reactions at 2 mM and in presence of imidazole: a) Cys-peptide 9 with Gly- peptide 10. The detected ion E is corresponding to the [M+H]+ ion. b) Cys-peptide 11 with Gly-peptide 10. A is the disulphide of 4, B the thiol of 25, C is the disulphide of 24 and D is the disulphide of 25.

S 17

Figure S14 Competitive coupling reactions at 2 mM and in presence of imidazole: Cys-peptide 11 with Lysine containing peptide 12. A is the disulphide of 4, B the thiol of 25, C is the disulphide of 25.

S 18

3.3 Coupling of carbohydrates

a) A

Figure S15 The coupling of the Cys-Man 13 in presence of imidazole at 0.1 (a) and 0.025 mM (b). In absence of imidazole at 0.1 mM concentration was no product detected (c). A is the disulphide of 26.

S 19

A B

B= [M+K]+

b) A C

Figure S16 The coupling of aminoethyl-Man 15 showed no product formation at 4 mM in the presence of imidazole (a). In absence of imidazole at 4 mM was coupling of 15 observed. A is the disulphide of 4 and C is the disulphide of 15 on the surface.

S 20

a) B

A= [M+K]+

A= [M+K]+ B

Figure S17 The coupling of trivalent Cys-Man 23 showed product formation in presence of imidazole at 1 mM (a) and 0.1 mM (b). A is the thiol of 27 and B the disulphide of 27.

S 21

B= [M+K]+ a)

b) A

B= [M+K]+

Figure S18 The coupling of trivalent Gly-Man 21 showed coupling in absence of imidazole at 4 mM (a) and 1 mM (b). A is the disulphide of 4, B the disulphide of 21 on the surface

S 22

B A

b) D= [M+K]+

C= [M+K]+

Figure S19 The competitive reactions in presence of imidazole: a) aminoethyl-Man 15 and Cys-Man 13 at 2 mM concentration. b) Trivalent Gly-Man 21 and trivalent Cys-Man 23 at a concentration of 4 mM. A is the disulphide of 4, B the disulphide of 26, C the thiol of 27 and D is the disulphide of 27.

S 23

4. Surface Plasmon Resonance (SPR) Analysis

The experiments were performed on a Biacore 3000 system (GE Healthcare, Sweden) using Sensor Chip Au (GE Healthcare). The gold coated sensor was cleaned with piranha solution, rinsed with water, ethanol and dried in a stream of nitrogen. The formation of self-assembling monolayers was performed in the same way as described (see above). The chip was washed with ethanol, dried under nitrogen and mounted into a chip holder following instructions in the supplier’s manual. After docking in the instrument the sensor was equilibrated with phosphate buffer saline (PBS, degassed and filtered) at a flow rate of 10 L/min. For surface activation, 70 l of a 1:1 mixture of freshly prepared NHS (0.4 M) and EDC (0.1 M) in water were injected (Channels 1-3). Reference spot (channel 1) was blocked by injecting 70 l of amino ethanol HCl (1 M). Aminoethyl GlcNAc 14 (10 mM, Channel 2) and mannosyl- cysteine 13 (2 mM and 15 mM TCEP, Channel 3) were immobilised by injecting 100 l of the substrate solution in PBS followed by blocking with amino ethanol (70 l, 1 M). The SAMs in channel 4 were PFP-activation. In order to perform this activation the chip was undocked, and the gold surface coated with 2 L of EDC and PFP (final concentrations 0.180 M and 0.174 M, respectively) in dry DMF at r.t. After 30 min the chip was washed with ethanol, and dried under nitrogen. After docking of the chip, the SPR chip was equilibrated with PBS buffer at a flow rate of 5 L/min and 250 L of the reaction mixture of

2 mM GlcNAc-NH2 14 and 2 mM Cys-Man 13 under NCL conditions (see section 4) were injected within an hour (channel 4).

Binding studies were carried out using HEPES buffer (10 mM HEPES, 0.15 M NaCl, 1 mM

CaCl2, 1 mM MnCl2, pH 7.4) at a flow rate of 25 l/min. 250 l of lectin solution (50 g/mL) were injected followed by 600 s dissociation sequence. The chip surface was regenerated by injecting 200 l of 100 mM glycine-HCl solution (pH 2).

Figure S20 The cysteine compound Cys-Man 13 and the amino-compound GlcNAc 14 were applied in a competitive assay at 2 mM on PFP activated SAMs.

S 24 a)

Figure S21 SPR analysis of the reaction in Figure S20 shown in Channel 4. Black: immobilised GlcNAc 14, Red: immobilised Cys-Man 13. a) Readout with the mannose binding lectin concanavalin A (ConA) (50 g/mL). ConA showed strong binding to the spot (blue) were the competitive reaction was carried out. This shows the presence of the Cys-Man 13. b) Readout with the GlcNAc binding lectin wheat germ agglutinin (WGA) (50 g/mL) showed no binding to the spot (blue) were the competitive reaction was carried out. This shows that no GlcNAc 14 has bound to the surface.

S 25

References for supporting information

1. J. W. Wehner and T. K. Lindhorst, Synthesis, 2010, 3070-3082. 2. C. Akpo, E. Weber and U. Reiche, New J. Chem., 2006, 30, 1820-1833; G. R. Newkome, R. K. Behera, C. N. Moorefield and G. R. Baker, J. Org. Chem., 1991, 56, 7162-7167. 3. M. Kleinert, N. Röckendorf and T. K. Lindhorst, Eur. J. Org. Chem., 2004, 3931-3940; T. K. Lindhorst, S. Kötter, U. Krallmann-Wenzel and S. Ehlers, J. Chem. Soc. Perkin Trans. 1, 2001, 823-831; J. Dahmen, T. Frejd, G. Gronberg, T. Lave, G. Magnusson and G. Noori, Carbohydr. Res., 1983, 116, 303-307. 4. R. Sardzik, G. T. Noble, M. J. Weissenborn, A. Martin, S. J. Webb and S. L. Flitsch, Beilstein J. Org. Chem., 2010, 6, 699-703. 5. N. Srinivasan, A. Yurek-George and A. Ganesan, Mol. Divers., 2005, 9, 291-293.

S 26

CHAPTER NINE

FORMATION OF CARBOHYDRATE-FUNCTIONALISED POLYSTYRENE AND GLASS SLIDES AND THEIR ANALYSIS BY MALDI-TOF MS

M. J. Weissenborn, J. W. Wehner, C. J. Gray, R. Sardzik, Th. K. Lindhorst, S. L. Flitsch, Formation of carbohydrate-functionalised polystyrene and glass slides and their analysis by MALDI-TOF MS, Beilstein J. Org. Chem. 2012, 8, 753–762.

M. J. Weissenborn had the idea and also carried out the initial experiments for gaining electrical conductivity on low conductive surfaces by applying aluminium tape to the back of the surface. J. W. Wehner and M. J. Weissenborn further developed the idea. J. W. Wehner synthesised and characterised the molecules used in this manuscript. R. Sardzik provided the GlcNAc ligand for the synthesis of 7. C. J. Gray and M. J. Weissenborn designed, performed and analysed all MALDI-ToF MS experiments with valuable input of C. E. Eyers. C. J. Gray, J. W. Wehner and M. J. Weissenborn wrote the experimental part and supporting information. All authors contributed to the writing of the article.

27 Formation of carbohydrate-functionalised polystyrene and glass slides and their analysis by MALDI-TOF MS

Martin J. Weissenborn1, Johannes W. Wehner2, Christopher J. Gray1, Robert Šardzík1, Claire E. Eyers*1, Thisbe K. Lindhorst*2 and Sabine L. Flitsch*1

Full Research Paper Open Access

Address: Beilstein J. Org. Chem. 2012, 8, 753–762. 1School of Chemistry & Manchester Interdisciplinary Biocentre, The doi:10.3762/bjoc.8.86 University of Manchester, 131 Princess Street, Manchester, M1 7DN, United Kingdom and 2Otto Diels Institute of Organic Chemistry, Received: 10 February 2012 Christiana Albertina University of Kiel, Otto-Hahn-Platz 3/4, 24098 Accepted: 02 May 2012 Kiel, Germany Published: 21 May 2012

Email: This article is part of the Thematic Series "Synthesis in the Claire E. Eyers* - [email protected]; glycosciences II". Thisbe K. Lindhorst* - [email protected]; Sabine L. Flitsch* - [email protected] Associate Editor: A. Kirschning

* Corresponding author © 2012 Weissenborn et al; licensee Beilstein-Institut. License and terms: see end of document. Keywords: carbohydrate array; conductive tape; MALDI-TOF MS; nonconductive surface; trityl-mediated adhesion

Abstract Glycans functionalised with hydrophobic trityl groups were synthesised and adsorbed onto polystyrene and glass slides in an array format. The adsorbed glycans could be analysed directly on these minimally conducting surfaces by MALDI-TOF mass spectrometry analysis after aluminium tape was attached to the underside of the slides. Furthermore, the trityl group appeared to act as an internal matrix and no additional matrix was necessary for the MS analysis. Thus, trityl groups can be used as simple hydrophobic, noncovalently linked anchors for ligands on surfaces and at the same time facilitate the in situ mass spectrometric analysis of such ligands.

Introduction Microarrays have become valuable tools in the high-throughput environments [1]. The initial success with DNA microarrays analysis of biological interactions and have promising applica- has prompted investigations into other biomolecular ligands, tions for the development of diagnostic devices in clinical such as protein, peptide and carbohydrate arrays [2,3]. An

753 Beilstein J. Org. Chem. 2012, 8, 753–762.

important aspect of this field is the immobilisation of such desorption/ionisation time-of-flight (MALDI-TOF) mass spec- ligands on solid array surfaces, which can be polymers, such as trometry (MS) analysis (Figure 1B), which has been highly polystyrene, or glass or gold, i.a. [4-7]. The challenge for successful on ligands immobilised on gold plates [10]. MALDI- immobilisations is in the efficiency of coupling and analysis of TOF MS requires an electrically conducting surface and a the attached ligands to ensure quality control. matrix for analysis. The matrix is typically cocrystallised with the sample, which can lead to irregular surfaces, which can be a Noncovalent attachment of biomolecules to hydrophobic problem for reproducible analysis, especially when used in surfaces has been used for a long time in ELISA assays and is array format as a high-throughput tool. To avoid the use of such attractive because no coupling reagents are required. However, a matrix, we were interested in investigating trityl functionalisa- it requires inherent hydrophobicity in the biomolecule or attach- tion, which has the potential for self-formation of a matrix (self- ment of a hydrophobic tether. The latter has been used highly matrix) for MS analysis, at the same time as acting as a hydro- successfully by the Feizi group as part of the neoglycolipid phobic tether [11-13]. However, polystyrene has only minimal array technology [8]. More recently, two groups [7,9] have innate electrical conductivity and to our knowledge has never reported the application of hydrophobic tethers for binding to been used successfully, unmodified, as a target for MALDI- polystyrene slides for glycan analysis. Initially, simple alkyl TOF MS analysis. Based on previous work one predicts that chains were used as tethers [9] but more recently, Wong and photoelectrons generated by UV laser irradiation are not dissip- co-workers improved on this technology by using trityl-derived ated by the polymeric surface. These photoelectrons distort the glycans, which are easily attached to glycans and were reported local electric field causing a significant loss in resolution of the to bind strongly to polystyrene (Figure 1A) [7]. analyte ions and a nonlinear shift in the mass-to-charge ratio [14]. The attachment of glycans to the surface was generally confirmed indirectly by lectin binding, which severely limits the Previous attempts to get around this issue have involved coating ligands that can be interrogated to those that can be detected by polymer or glass surfaces with a thin membrane of conductive carbohydrate-binding proteins. material, such as gold, carbon or indium-tin oxide [15-17], or the addition of electron-accepting additives, such as methyl To overcome this limitation, we were interested in developing viologen dichloride hydrate [14]. The addition of electron- label-free methods for ligand detection on these polystyrene accepting additives, however, did not completely suppress the surfaces, and have investigated the use of matrix-assisted laser mass shifts observed during MALDI-TOF MS on low-conduc-

Figure 1: Carbohydrate arrays on polystyrene slides can be obtained by noncovalent immobilisation of tritylated saccharide derivatives. (A) Lectin- mediated analysis of carbohydrate arrays [7]. (B) The new concept of label-free MALDI-TOF MS analysis by aluminium-backing of polystyrene or glass slides.

754 Beilstein J. Org. Chem. 2012, 8, 753–762.

tivity supports. Additionally, glass slides coated with conduc- acid 3 was coupled with either of the aminoethyl glycosides 4 tive material are expensive and limited in their utility [14]. [24-26] or the GlcNAc derivative 6 [27], which were prepared according to literature procedures. For the coupling a combina- In order to address these issues, we have investigated a simple tion of HBTU/DIPEA or HATU/DIPEA in dry DMF was and cheap method to enable MALDI-TOF MS analysis on applied yielding 74% of 5 and 71% of 7. minimally conductive supports by applying a commercially available aluminium tape to the underside of glass and poly- The analysis of molecules on materials such as polystyrene by styrene slides. This has allowed us to make standard micro- MALDI-TOF MS is difficult and irreproducible, largely due to scope slides suitable for MALDI-TOF MS analysis. their minimal electrical conductivity. To our knowledge, successful MS analysis on such surfaces has not been reported. Results and Discussion In order to circumvent this issue, commercially available The hydrophobic trityl tethers chosen for our studies consisted aluminium tape was applied to the underside of a polystyrene of S-tritylated instead of N-tritylated groups, which were used microscope slide. previously [7]. This followed the idea of “orthogonal” surface functionalisation. Thus, a thiol group would make the tethers The tape significantly enhanced both the signal intensity and generally compatible with other platforms in our laboratories, resolution of MALDI-TOF MS analysis of the Man–Trt com- such as through the formation of SAMs on gold [18] or by pound 5 (Figure 2). Furthermore, we observed that analysis of coupling into maleimide-functionalised surfaces in a chemose- the Man–Trt compound 5 could not only be performed in the lective fashion [19]. absence of any additional matrix [11,12], but under these conditions a modest increase in mass-spectrometric resolution was For the initial studies two carbohydrate derivatives, 5 and 7, also observed. Such self-matrix properties are very convenient, were synthesised. The α-D-mannoside 5 would be useful in a yielding more robust and reproducible analyses, and negating bacterial adhesion inhibition assay against the bacterial lectin the need to search for “sweet spots”, as no crystal formation is FimH [20,21]. The second glycoside 7 has been used previ- required, in contrast to conventional MALDI-TOF MS analysis. ously for well-established enzymatic surface modifications [22]. Both these compounds can be synthesised by starting from Interestingly, both Na and K cation adducts of Man–Trt 5 and commercially available 11-mercaptoundecanoic acid (1), which its disulfide 8 were observed (Scheme 2). The relative ratios of is tritylated with 2 in a straightforward synthesis, in 98% yield, the monomer 5 and the disulfide 8 were found to be concentra- following the procedure of Kovács et al. [23] (Scheme 1). The tion-dependent in the analysis on stainless steel (Supporting

Scheme 1: Synthesis of the tritylated compounds 3 [28], 5 and 7: (a) dichloromethane, 2.5 h, rt, 98%, (b) HBTU/DIPEA, dry DMF, overnight, rt, 74%, (c) HATU/DIPEA, dry DMF, 3 h, rt, 71%.

755 Beilstein J. Org. Chem. 2012, 8, 753–762.

Figure 2: Comparison of the polystyrene and glass surfaces with and without aluminium backing by matrix-free MALDI-TOF MS analysis. Each spot contains 15 nmol Man–Trt (5): (A) polystyrene without aluminium; (B) glass without aluminium; (C) polystyrene with aluminium and (D) glass with aluminium. The peaks at m/z of 688.3, 704.3, 867.5 and 883.4 correspond to [5 + Na]+, [5 + K]+, [8 + Na]+, and [8 + K]+, respectively.

Scheme 2: The Man–Trt compound 5 forms the disulfide 8 during UV ionisation in MALDI-TOF MS analysis.

756 Beilstein J. Org. Chem. 2012, 8, 753–762.

Information File 1). It was observed that at 15 nmol, almost observed with the polystyrene slides, the resolution and signal exclusively the disulfide 8 was detected. Conversely at 50 pmol intensity of MALDI-TOF MS analysis (Figure 2) was dramati- only K+ and Na+ adducts of Man–Trt 5 were found. cally improved following application of the aluminium tape to the back of the glass slides. We were interested to see whether the addition of the aluminium tape would also enhance signals on surfaces other Next, the limit of detection of MS analysis of aluminium- than polystyrene, and we therefore assessed the influence of the backed polystyrene and glass slides was compared (Figure 3), conductive tape on analysis using glass slides, which are widely using dilutions of the analyte Man–Trt 5 from 15 nmol to used in protein-, peptide- and glycoarrays [2,29,30]. As 50 pmol. The resolution between the two supports was compar-

Figure 3: Limit-of-detection analysis of 5 on aluminium-backed glass and polystyrene slides. Both systems showed the MALDI-TOF MS detection limit at 0.5 nmol. The peaks at m/z of 688.3, 704.3, 867.4 and 883.4 correspond to [5 + Na]+, [5 + K]+, [8 + Na]+, and [8 + K]+, respectively.

757 Beilstein J. Org. Chem. 2012, 8, 753–762.

able and analysis showed that Man–Trt 5 could still be detected We were intrigued by the role of the aluminium backing and at a concentration of 0.5 nmol for aluminium-backed poly- decided to investigate it in more detail. The aluminium backing styrene and glass slides. was applied in three different ways: First, the entire underside of the slide was covered as in the previous experiments. Second, After application of trityl samples, the polystyrene and glass only a narrow strip of tape was applied to the back of the slide, slides were gently washed with 1 μL of water (washing proce- making sure that the strip was in contact with the frame of the dure 1). Subsequent analysis by MALDI-TOF MS showed no slide adapter (Figure 5). This configuration should still allow noticeable change to the prewashed samples, confirming the for efficient dissipation of any produced photoelectrons. Spots trityl-group-mediated noncovalent adhesion of the ligands to the were analysed directly over and also next to the strip. The surfaces. On the other hand, much more rigorous washing under results showed similar intensity and resolution as for the fully running distilled water (washing procedure 2) caused a signifi- aluminium-backed polystyrene slide (Supporting Information cant reduction in signal on the polystyrene slide (Figure 4). File 1, Figures S2 and S3). Third, only a small aluminium rect- Analysis of the glass slide, however, showed only a slight angle was attached to the back of the polystyrene slide, this time decrease in signal intensity even after three rigorous washes. making sure that there was no electrical contact to the slide

Figure 4: Comparison of MALDI-TOF MS spectra on the aluminium-backed polystyrene and glass slides after washing. (A) At 7.5 nmol following washing procedure 1 and (B) at 15 nmol following washing procedure 2.

758 Beilstein J. Org. Chem. 2012, 8, 753–762.

frame (Supporting Information File 1, Figures S4 and S5). In 2-Aminoethyl α-D-mannopyranoside (4) [24-26] and 2-amino- this last case very poor signals, both in resolution and intensity, ethyl 2-acetamido-2-deoxy-β-D-glucopyranoside (6) [27] were were observed, which were analogous to the non-aluminium- prepared according to the literature. Reactions were monitored backed polystyrene slide. Thus, contact of the tape to the slide by thin-layer chromatography using silica gel 60 GF254 on adapter frame appears to be essential for good signal intensity, aluminium foil (Merck) with detection by UV light and char- but it is not necessary to cover the slide fully. ring with sulfuric acid in EtOH (10%). Preparative MPLC was performed on a Büchi apparatus using a LiChroprep Si 60 (40–60 mm, Merck) column for normal-phase silica-gel chromatography. Analytical HPLC was performed on a Merck Hitachi LaChrom L-7000 series apparatus with a LiChrospher 100 RP-8 (5 μm, Merck) column. 1H and 13C NMR spectra were recorded on a Bruker DRX-500 spectrometer. NMR spectra were calibrated with respect to the solvent peak. 2D NMR techniques (COSY, HSQC, HMBC) were used for full Figure 5: Photo of the aluminium strip on the back of the polystyrene assignment of the spectra. ESI-MS measurements were support. performed on a Mariner ESI-TOF 5280 instrument (Applied Biosystems). High-resolution mass spectra (HRMS) were Given that the MS analysis on the polystyrene was successful, obtained with the Waters Micromass LCT-TOF mass spectro- we attempted an enzymatic galactosylation of GlcNAc–Trt 7 meter. MALDI-TOF mass spectra were recorded on a Bruker using bovine β-(1→4)-galactosyltransferase (β-(1→4)-GalT, Biflex-III 19 kV instrument with Cl-CCA (4-chloro-α-cyano- EC 2.4.1.38) on the polystyrene slides. This enzymatic trans- cinnamic acid) or DHB (2,5-dihydroxybenzoic acid) as matrix. formation has proven to be a very reproducible and robust reac- Optical rotation was measured on a Perkin-Elmer polarimeter tion, which appears to proceed to completion on gold arrays 341 (Na-D-line: 589 nm, length of cell 1 dm). IR spectra were [31] and is routinely performed in our laboratory. After treat- recorded on a Perkin-Elmer Paragon 1000 FTIR spectrometer. ment of slides containing GlcNAc–Trt 7 with the enzyme by For sample preparation a Golden Gate diamond ATR unit with using previously reported procedures [31], followed by washing a sapphire stamp was used. procedure 1, neither product nor starting material could be observed by MALDI-TOF MS analysis. In fluorescence- 11-Tritylsulfanylundecanoic acid (3) [28] assisted studies this problem is mostly overcome by blocking Chlorotriphenylmethane (2, 2.53 g, 9.07 mmol) was dissolved with nonfluorescent milk proteins or BSA; however, in in dichloromethane. 11-Mercaptoundecanoic acid (1, 2.00 g, MALDI-TOF MS analysis the blocking proteins would also be 9.15 mmol) dissolved in dichloromethane (60 mL) was added ionised and consequently quench the signal [32]. dropwise over 1 h. The reaction mixture was stirred for 1.5 h at ambient temperature until TLC (cyclohexane/ethyl acetate 3:1) Conclusion indicated no further conversion. The reaction mixture was Successful MALDI-TOF MS analysis on minimally conductive washed with H2O (50 mL), the organic layer was dried over surfaces was achieved by application of aluminium tape. Poly- MgSO4, and the solvent was removed under reduced pressure. styrene and glass surfaces were spotted with the analyte The crude product was purified by MPLC (100 g silica column, Man–Trt 5 and were analysed over a range of concentrations A: cyclohexane, B: ethyl acetate, A: 90% → 40%, 120 min) and after washing. This new technique enables the direct yielding 3 (4.10 g, 8.90 mmol, 98%) as a colourless solid. analysis of any noncovalent glycoarray on glass and polystyrene. So far, our attempts to study enzymatic reactions on the Rf 0.61 (methanol/dichloromethane 3:18); mp 80–82 °C; HPLC modified polystyrene surface have been unsuccessful and will tR 7.31 min (A = water, B = methanol, A: 20%, 10 min, 1 require further investigation. 1.2 mL/min); H NMR (500 MHz, CD3OD, 300 K) δ 7.38 (mc, 6H, Haryl,Trt), 7.25 (mc, 6H, Haryl,Trt), 7.19 (mc, 3H, Haryl,Trt), Experimental 3 3 2.18 (t, J = 7.6 Hz, 2H, HO(O)CCH2CH2), 2.11 (t, J = 7.4 Hz, 3 3 General experimental section for the 2H, CH2STrt), 1.59 (q, J = 7.9 Hz, J = 7.0 Hz, 2H, saccharide synthesis HO(O)CCH2CH2), 1.37–1.09 (m, 14H, CH2CH2CH2) ppm; 13 Commercially available starting materials and reagents were C NMR (125 MHz, CD3OD, 300 K) δ 179.3 (C(O)OH), used without further purification. Reactions requiring dry condi- 146.5 (3 Caryl,Trt), 130.8 (6 CHaryl,Trt), 128.8 (6 CHaryl,Trt), tions were performed under an atmosphere of nitrogen. Anhy- 127.6 (3 CHaryl,Trt), 67.6 (Cquart,Trt), 36.5 (HO(O)CCH2CH2), drous DMF was purchased. 32.9 (CH2STrt), 30.5, 30.4, 30.4, 30.3, 30.1, 30.0, 29.7 (7

759 Beilstein J. Org. Chem. 2012, 8, 753–762.

CH2CH2CH2), 26.9 (HN(O)CCH2CH2) ppm; MALDI-TOF MS (HN(O)CCH2CH2) ppm; MALDI-TOF MS (DHB) m/z: 688.11 (DHB) m/z: 483.13 [M + Na]+, 499.10 [M + K]+; HRMS–ESI [M + Na]+, 704.08 [M + K]+; HRMS–ESI (m/z): [M + Na]+ + (m/z): [M + Na] calcd for C30H36NNaO2S, 483.2328; found, calcd for C40H54N2NaO7S, 729.3544; found, 729.3506; IR 483.2343; IR (ATR–IR) : 3384, 3189, 3056, 2923, 2850, (ATR–IR) : 3293, 2923, 2852, 1645, 1548, 1488, 1443, 1253, 1651, 1594, 1489, 1444, 1419, 1032, 770, 740, 695, 674, 1132, 1057, 1031, 975, 810, 741, 697, 676, 616 cm−1. 621 cm−1. 2-((11-Tritylsulfanylundecanoyl)amino)ethyl 2-acet- 2-((11-Tritylsulfanylundecanoyl)amino)ethyl α-D- amido-2-deoxy-β-D-glucopyranoside (7) mannopyranoside (5) 11-Tritylsulfanylundecanoic acid (3, 18.2 mg, 39.5 μmol) and 11-Tritylsulfanylundecanoic acid (3, 750 mg, 1.63 mmol) and HATU (30.0 mg, 79.0 μmol) were dried for 1 h under vacuum. HBTU (743 mg, 1.96 mmol) were dried for 2 h under vacuum, Then, dry DMF (2 mL) and DIPEA (7.00 μL, 40.9 μmol) were and then dry DMF (5 mL) and DIPEA (400 μL, 2.33 mmol) added, and the mixture was stirred for 20 min under a nitrogen were added, and the mixture was stirred for 20 min under a atmosphere at ambient temperature. Simultaneously, in a nitrogen atmosphere at ambient temperature. Simultaneously, in different reaction vessel 2-aminoethyl 2-acetamido-2-deoxy-β- a different reaction vessel aminoethyl mannoside 4 (438 mg, D-glucopyranoside (6, 11.5 mg, 43.5 μmol) was dried for 1 h 1.96 mmol) was dried for 2 h under vacuum, and then dissolved under vacuum and dissolved in dry DMF (1 mL), and then in dry DMF (5 mL), and DIPEA (160 μL, 931 μmol) was DIPEA (7.00 μL, 40.9 μmol) was added. The mixture was added. The mixture was stirred for 20 min under a nitrogen stirred for 20 min under a nitrogen atmosphere at ambient atmosphere at ambient temperature. The reaction mixture with temperature. The solution of 6 in dry DMF was added to the the preactivated 11-tritylsulfanylundecanoic acid (3) was cooled preactivated 11-tritylsulfanylundecanoic acid (3) and it was to 0 °C, the solution of mannoside 4 was added and the stirred under a nitrogen atmosphere at ambient temperature for resulting mixture was stirred under a nitrogen atmosphere at 3 h. All volatile compounds were removed under reduced pres- ambient temperature overnight. All volatile compounds were sure and the crude product was subjected to MPLC (50 g silica removed under reduced pressure and the crude product was column, A: dichloromethane, B: methanol, A: 99% → 85%, subjected to MPLC (150 g silica column, A: dichloromethane, 180 min) yielding 7 (19.7 mg, 27.9 μmol, 71%) as a colourless B: methanol, A: 99% → 90%, 120 min) and another round of lyophylisate. MPLC (125 g silica column, A: ethyl acetate, B: methanol, A:

99% → 90%, 120 min) yielding 5 (808 mg, 1.21 mmol, 74%) as Rf 0.21 (methanol/dichloromethane, 3:18); HPLC tR 5.44 min a colourless foam. (A = water, B = methanol, A: 20%, 10 min, 1.2 mL/min); 1 −1.6 (c 0.1, methanol); H NMR (500 MHz, CD3OD, Rf 0.16 (methanol/dichloromethane, 1:9); HPLC tR = 5.49 min 300 K) δ 7.39 (mc, 6H, Haryl,Trt), 7.28 (mc, 6H, Haryl,Trt), 7.21 3 (A = water, B = methanol, A: 20%, 10 min, 1.2 mL/min); (mc, 3H, Haryl,Trt), 4.39 (d, J = 8.4 Hz, 1H, H1GlcNAc), 3.88 1 2 3 2 +23.7 (c 0.5, MeOH); H NMR (500 MHz, CD3OD, (dd, J = 11.8 Hz, J = 2.2 Hz, 1H, H6aGlcNAc), 3.82 (ddd, J = 3 3 300 K) δ 7.38 (mc, 6H, Haryl,Trt), 7.27 (mc, 6H, Haryl,Trt), 7.20 10.6 Hz, J = 6.7 Hz, J = 4.5 Hz, 1H, OCHHCH2NH), 3.67 3 4 3 2 3 (dt, J = 7.3 Hz, J = 1.3 Hz, 3H, Haryl,Trt), 4.76 (d, J = 1.7 Hz, (dd, J = 11.8 Hz, J = 5.8 Hz, 1H, H6bGlcNAc), 3.67–3.59 (m, 2 3 3 3 1H, H1Man), 3.83 (dd, J = 11.6 Hz, J = 2.3 Hz, 1H, H6aMan), 2H, H2GlcNAc, OCHHCH2NH), 3.43 (dd, J = 10.4 Hz, J = 3 3 3.80 (dd, J = 1.7 Hz, J = 3.3 Hz, 1H, H2Man), 3.77–3.67 (m, 8.3 Hz, 1H, H3GlcNAc), 3.40–3.36 (m, 1H, OCH2CHHNH), 3 3H, OCHHCH2NH, H3Man, H6bMan), 3.60 (dd~t, J = 9.5 Hz, 3.40–3.26 (m, 3H, OCH2CHHNH, H4GlcNAc, H5GlcNAc), 2.18 3 3 1H, H4Man), 3.56–3.50 (m, 2H, H5Man, OCHHCH2NH), 3.41 (t, J = 7.6 Hz, 2H, HN(O)CCH2CH2), 2.12 (t, J = 7.4 Hz, 2H, 2 3 3 (ddd, J = 14.0 Hz, J = 6.3 Hz, J = 4.6 Hz, 1H, CH2STrt), 1.98 (s, 3H, NHAc), 1.59 (m, 2H, 2 3 3 OCH2CHHNH), 3.35 (ddd, J = 14.0 Hz, J = 6.7 Hz, J = 4.7 HN(O)CCH2CH2), 1.38–1.10 (m, 14H, 7 CH2CH2CH2) ppm; 3 13 Hz, 1H, OCH2CHHNH), 2.19 (t, J = 7.6 Hz, 2H, C NMR (125 MHz, CD3OD, 300 K) δ 176.4 (HNC(O)CH2), 3 HN(O)CCH2CH2), 2.12 (t, J = 7.4 Hz, 2H, CH2STrt), 1.59 (q, 173.9 (HNC(O)CH3), 146.5 (3 Caryl,Trt), 130.8 (6 CHaryl,Trt), 3 3 J = 7.6 Hz, J = 7.3 Hz, 2H, HN(O)CCH2CH2), 1.38–1.10 (m, 128.8 (6 CHaryl,Trt), 127.7 (3 CHaryl,Trt), 102.9 (C1GlcNAc), 78.0 13 14H, 7 CH2CH2CH2) ppm; C NMR (125 MHz, CD3OD, (C5GlcNAc), 76.1 (C3GlcNAc), 72.1 (C4GlcNAc), 69.2 300 K) δ 176.5 (C(O)NH), 146.5 (3 Caryl,Trt), 130.8 (6 (OCH2CH2NH), 67.3 (Cquart,Trt), 62.8 (C6GlcNAc), 57.3 CHaryl,Trt), 128.8 (6 CHaryl,Trt), 127.7 (3 CHaryl,Trt), 101.7 (C2GlcNAc), 40.6 (OCH2CH2NH), 37.1 (HN(O)CCH2CH2), (C1Man), 74.8 (C5Man), 72.6 (C3Man), 72.1 (C2Man), 68.6 32.9 (CH2STrt), 30.5, 30.4, 30.4, 30.3, 30.1, 30.0, 29.7 (7 (C4Man), 67.3 (Cquart,Trt), 67.3 (OCH2CH2NH), 62.9 (C6Man), CH2CH2CH2), 27.0 (HN(O)CCH2CH2), 23.0 (HNC(O)CH3) + 40.2 (OCH2CH2NH), 37.1 (HN(O)CCH2CH2), 32.9 (CH2STrt), ppm; MALDI-TOF MS (DHB) m/z: 729.48 [M + Na] , 745.45 + + 30.5, 30.4, 30.4, 30.3, 30.1, 30.0, 29.7 (7 CH2CH2CH2), 27.0 [M + K] ; HRMS–ESI (m/z): [M + Na] calcd for

760 Beilstein J. Org. Chem. 2012, 8, 753–762.

C40H54N2NaO7S, 729.3544; found, 729.350; IR (ATR–IR) : Acknowledgements 3270, 2924, 2852, 1640, 1549, 1488, 1443, 1373, 1156, 1109, We thank Dr Stephan Mohr and Dr Jeremy Hawkes for 1080, 1033, 896, 742, 698, 616 cm−1. providing us with the polystyrene slides; and Dr Niels-Chris- tian Reichardt, Javier Calvo Martinez, Dr Sonia Serna and Dr Array washing Antonio Sanchez-Ruiz for help with the initial MALDI-TOF Washing procedure 1 MS analysis. This work was supported by the Royal Society Distilled water (1 μL) was spotted over the dried analyte spot (Wolfson Award to SLF) and the European Commission’s and was subsequently drawn back up with the pipette. This was Seventh Framework Programme (FP7), which funded the repeated three times. The slides were then allowed to dry under EuroGlycoArrays ITN (MJW) and GlycoBioM (RS, SLF, atmospheric conditions. CEE). JWW was supported by the Evonik Foundation and CJG by a BBSRC-funded doctoral training award. Washing procedure 2 The MALDI target slide was washed with cool distilled water References (making sure the spot was not directly under the tap) for 6 s at a 1. Cao, B.; Li, R.; Xiong, S.; Yao, F.; Liu, X.; Wang, M.; Feng, L.; Wang, L. flow rate of 3 L/min and then dried under a stream of nitrogen. Appl. Environ. Microbiol. 2011, 77, 8219–8225. doi:10.1128/AEM.05914-11 MALDI-TOF MS analysis of tritylated com- 2. Berrade, L.; Garcia, A. E.; Camarero, J. A. Pharm. Res. 2011, 28, 1480–1499. doi:10.1007/s11095-010-0325-1 pounds 3. Blixt, O.; Head, S.; Mondala, T.; Scanlan, C.; Huflejt, M. E.; Alvarez, R.; Unless otherwise stated all MALDI-TOF MS experiments were Bryan, M. C.; Fazio, F.; Calarese, D.; Stevens, J.; Razi, N.; carried out on an Ultraflex II instrument (Bruker Daltonics, Stevens, D. J.; Skehel, J. J.; van Die, I.; Burton, D. R.; Wilson, I. A.; USA) in positive reflectron mode in the absence of a matrix. Cummings, R.; Bovin, N.; Wong, C.-H.; Paulson, J. C. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 17033–17038. Spectra were acquired over the mass range 600–2500 m/z with doi:10.1073/pnas.0407902101 500 shots (57% laser energy) per spectrum and a laser firing 4. Weissenborn, M. J.; Castangia, R.; Wehner, J. W.; Šardzík, R.; rate of 200 Hz. Data were processed and analysed with Flex- Lindhorst, T. K.; Flitsch, S. L. Chem. Commun. 2012, 48, 4444–4446. Analysis software (Bruker Daltonics, USA) by using the default doi:10.1039/c2cc30844d integration settings. Smoothing and baseline subtraction was 5. Serna, S.; Yan, S.; Martin-Lomas, M.; Wilson, I. B. H.; Reichardt, N.-C. performed on each spectrum by using the default settings in J. Am. Chem. Soc. 2011, 133, 16495–16502. doi:10.1021/ja205392z 6. Voglmeir, J.; Šardzík, R.; Weissenborn, M. J.; Flitsch, S. L. OMICS FlexAnalysis. Calibration was either performed before the 2010, 14, 437–444. doi:10.1089/omi.2010.0035 analysis on the Ultraflex II instrument or afterwards in Flex- 7. Zou, L.; Pang, H.-L.; Chan, P.-H.; Huang, Z.-S.; Gu, L.-Q.; Wong, K.-Y. Analysis by using Man–Trt 5 as an internal calibrant for poly- Carbohydr. Res. 2008, 343, 2932–2938. styrene and glass slides and a tryptic digest of Qcal protein as doi:10.1016/j.carres.2008.08.021 the calibrant for the steel target [33]. 8. Fukui, S.; Feizi, T.; Galustian, C.; Lawson, A. M.; Chai, W. Nat. Biotechnol. 2002, 20, 1011–1017. doi:10.1038/nbt735 9. Bryan, M. C.; Plettenburg, O.; Sears, P.; Wong, C.-H. Polystyrene slides were manufactured by Goodfellows, U.K., Abstr. Pap. - Am. Chem. Soc. 2002, 224, U181. and standard glass microscope slides, purchased from Yancheng 10. Šardzík, R.; Green, A. P.; Laurent, N.; Both, P.; Fontana, C.; Huida Medical Instruments Co., China, were used. Conductive Voglmeir, J.; Weissenborn, M. J.; Haddoub, R.; Grassi, P.; aluminium tape was purchased from Farnell, U.K., and at- Haslam, S. M.; Widmalm, G.; Flitsch, S. L. J. Am. Chem. Soc. 2012, tached to the back of the nonconductive polystyrene and glass 134, 4521–4524. doi:10.1021/ja211861m 11. Thiery, G.; Shchepinov, M. S.; Southern, E. M.; Audebourg, A.; slides. The slides were mounted on to MTP Slide-Adapter II Audard, V.; Terris, B.; Gut, I. G. Rapid Commun. Mass Spectrom. (Bruker) for analysis. Tritylated sugar in methanol (0.5 µL) was 2007, 21, 823–829. doi:10.1002/rcm.2895 applied to the surface and the solvent was allowed to evaporate 12. Ustinov, A. V.; Shmanai, V. V.; Patel, K.; Stepanova, I. A.; under atmospheric conditions. Unless otherwise stated, the Prokhorenko, I. A.; Astakhova, I. V.; Malakhov, A. D.; spots were washed by following procedure 1. Skorobogatyi, M. V.; Bernad, P. L.; Khan, S.; Shahgholi, M.; Southern, E. M.; Korshun, V. A.; Shchepinov, M. S. Org. Biomol. Chem. 2008, 6, 4593–4608. doi:10.1039/b810600b Supporting Information 13. Aumüller, I.; Lindhorst, T. K. Eur. J. Org. Chem. 2006, 1103–1108. doi:10.1002/ejoc.200500900 Supporting Information File 1 14. Ibáñez, A. J.; Muck, A.; Svatoš, A. J. Mass Spectrom. 2007, 42, Enzyme expression and MALDI MS spectra. 634–640. doi:10.1002/jms.1192 15. Altelaar, A. F. M.; van Minnen, J.; Jiménez, C. R.; Heeren, R. M. A.; [http://www.beilstein-journals.org/bjoc/content/ Piersma, S. R. Anal. Chem. 2005, 77, 735–741. supplementary/1860-5397-8-86-S1.pdf] doi:10.1021/ac048329g

761 Beilstein J. Org. Chem. 2012, 8, 753–762.

16. Scherl, A.; Zimmermann-Ivol, C. G.; Di Dio, J.; Vaezzadeh, A. R.; Binz, P.-A.; Amez-Droz, M.; Cochard, R.; Sanchez, J.-C.; License and Terms Glückmann, M.; Hochstrasser, D. F. Rapid Commun. Mass Spectrom. 2005, 19, 605–610. doi:10.1002/rcm.1831 This is an Open Access article under the terms of the 17. Li, X.; Wilm, M.; Franz, T. Proteomics 2005, 5, 1460–1471. Creative Commons Attribution License doi:10.1002/pmic.200401023 (http://creativecommons.org/licenses/by/2.0), which 18. Laurent, N.; Haddoub, R.; Voglmeir, J.; Wong, S. C. C.; Gaskell, S. J.; permits unrestricted use, distribution, and reproduction in Flitsch, S. L. ChemBioChem 2008, 9, 2592–2596. doi:10.1002/cbic.200800481 any medium, provided the original work is properly cited. 19. Houseman, B. T.; Gawalt, E. S.; Mrksich, M. Langmuir 2003, 19, 1522–1531. doi:10.1021/la0262304 The license is subject to the Beilstein Journal of Organic 20. Hartmann, M.; Lindhorst, T. K. Eur. J. Org. Chem. 2011, 3583–3609. Chemistry terms and conditions: doi:10.1002/ejoc.201100407 (http://www.beilstein-journals.org/bjoc) 21. Hartmann, M.; Horst, A. K.; Klemm, P.; Lindhorst, T. K. Chem. Commun. 2010, 46, 330–332. doi:10.1039/b922525k 22. Laurent, N.; Voglmeir, J.; Flitsch, S. L. Chem. Commun. 2008, The definitive version of this article is the electronic one 4400–4412. doi:10.1039/b806983m which can be found at: 23. Kupihár, Z.; Schmél, Z.; Kovács, L. Molecules 2000, 5, M144–U3. doi:10.3762/bjoc.8.86 doi:10.3390/M144 24. Kleinert, M.; Röckendorf, N.; Lindhorst, T. K. Eur. J. Org. Chem. 2004, 3931–3940. doi:10.1002/ejoc.200400239 25. Lindhorst, T. K.; Kötter, S.; Krallmann-Wenzel, U.; Ehlers, S. J. Chem. Soc., Perkin Trans. 1 2001, 823–831. doi:10.1039/B009786L 26. Dahmén, J.; Frejd, T.; Grönberg, G.; Lave, T.; Magnusson, G.; Noori, G. Carbohydr. Res. 1983, 116, 303–307. doi:10.1016/0008-6215(83)88120-8 27. Šardzík, R.; Noble, G. T.; Weissenborn, M. J.; Martin, A.; Webb, S. J.; Flitsch, S. L. Beilstein J. Org. Chem. 2010, 6, 699–703. doi:10.3762/bjoc.6.81 28. Ryan, D.; Parviz, B. A.; Linder, V.; Semetey, V.; Sia, S. K.; Su, J.; Mrksich, M.; Whitesides, G. M. Langmuir 2004, 20, 9080–9088. doi:10.1021/la048443u 29. MacBeath, G.; Schreiber, S. L. Science 2000, 289, 1760–1763. 30. Weinrich, D.; Jonkheijm, P.; Niemeyer, C. M.; Waldmann, H. Angew. Chem., Int. Ed. 2009, 48, 7744–7751. doi:10.1002/anie.200901480 31. Šardzík, R.; Sharma, R.; Kaloo, S.; Voglmeir, J.; Crocker, P. R.; Flitsch, S. L. Chem. Commun. 2011, 47, 5425–5427. doi:10.1039/c1cc10745c 32. Deere, J.; McConnell, G.; Lalaouni, A.; Maltman, B. A.; Flitsch, S. L.; Halling, P. J. Adv. Synth. Catal. 2007, 349, 1321–1326. doi:10.1002/adsc.200700044 33. Eyers, C. E.; Simpson, D. M.; Wong, S. C. C.; Beynon, R. J.; Gaskell, S. J. J. Am. Soc. Mass Spectrom. 2008, 19, 1275–1280. doi:10.1016/j.jasms.2008.05.019

762 9.1. SUPPORTING INFORMATION

9.1 Supporting Information

The supporting information given below is speciﬁc to those experiments carried out by M. J. Weissenborn. For the original supporting information please see http://www.beilstein-journals.org/bjoc/content/supplementary/1860-5397-8-86-S1.pdf.

28 Supporting Information

Formation of carbohydrate-functionalised polystyrene and glass slides and their analysis by

MALDI-ToF MS

Martin J. Weissenborn1, Johannes W. Wehner2, Christopher J. Gray1, Robert

Šardzík1, Claire E. Eyers*1, Thisbe K. Lindhorst*2 and Sabine L. Flitsch*1

Address: 1 School of Chemistry, University of Manchester, Manchester

Interdisciplinary Biocentre, 131 Princess Street, Manchester, M17DN,

United Kingdom;

2 Otto Diels Institute of Organic Chemistry, Christiana Albertina University of Kiel,

Otto-Hahn-Platz 3/4, 24098 Kiel, Germany.

E-mail: Sabine L. Flitsch* [email protected];

Thisbe K. Lindhorst* [email protected];

Claire E. Eyers* [email protected]

* Corresponding author

1. GalT1 enzyme ...... 2

2. MALDI-ToF mass spectra ...... 4

S 1

1. GalT1 enzyme

Subcloning of wild-type bovine GalT1 from Pichia expression vector into pET30a for expression in E. coli

The P. pastoris expression vector (pGAPz-GalT1) containing the GalT1 gene which was used as template DNA, was kindly provided by Dr. Dubravko Rendić from the University of

Natural Resources and Applied Life Sciences in Vienna.

Amplification of the plasmid was performed after transformation of E. coli TOP10F competent cells. The purified plasmid was used as a template to amplify the soluble region of the GalT1 gene by PCR using the primer pair EcGalT1-Fw/EcGalT1-Rv (EcGalT1-Fw:

AAG AAT TCC TGC GAG GGG TCG CAC CGC CGC CGC CTT TGC AGA ACT CTT CC;

EcGalT1-Rv: AAA AAA AGC GGC CGC CTA GCT CGG CGT CCC GAT GTC CAC TGT

GAT TTT GG). The PCR product and pET30a vector were subjected to endonuclease treatment using EcoRI and NotI followed by ligation using T4 DNA ligase and subsequently dephosphorylated (by Calf Intestinal Alkaline Phosphatase). BL21(DE3) cells were transformed with the resulting plasmid construct for protein expression.

Expression, purification and refolding of the bovine GalT1

After transformation of E. coli (BL21 (DE3) with pET30a-GalT1, a single colony was used to inoculate 5 ml of LB medium containing kanamycin (50 μg/mL) and the culture was incubated at 37°C/250 rpm. The overnight culture was used to inoculate 400 mL of fresh LB medium containing kanamycin (50 μg/mL). Cells were grown at 37°C/250 rpm to an OD600 of 0.6 and protein expression was subsequently induced by the addition of IPTG (1 mM).

The culture was further incubated at 37°C/250 rpm for 6 hours. Cells were harvested

(5000 g, 10 min, 4°C) and resuspended in 10 mL lysis-buffer (50 mM Tris/HCl, 500 mM

NaCl, 1% Triton X-100, 0.5 mM PMSF, pH 8.5). Lysis was performed by sonication (30 min,

10s on/10s off-cycles) followed by centrifugation (30000 g, 30 min, 4°C). The pellet was resuspended in binding buffer (50 mM Tris/HCl, 500 mM NaCl, 6 M urea, 10 mM imidazole,

S 2 pH 8.5) and incubated overnight on a rocker at 4°C in order to solubilize the enzyme from inclusion bodies. After centrifugation (30000 g, 30 min, 4°C) the supernatant was filtered through a 0.2 μm filter and loaded onto a 5 mL HisTrap column pre-equilibrated with binding buffer. After washing the column with 50 mL of loading buffer containing 6 M urea the enzyme was eluted from the column using 50 mM Tris/HCl, 500 mM NaCl, 6 M urea, 500 mM Imidazole, pH 8.5. Eluted fractions containing protein were pooled and added dropwise at 4°C under constant stirring to an excess of 100 mL refolding buffer (50 mM Tris/HCl, 500 mM NaCl, 1% Triton X-100, 8 mM GSH, 1 mM GSSG, 1% EDTA-free Protease Inhibitor

Cocktail, pH 8.5). This solution was applied again to a HisTrap column with binding buffer

(50mM Tris/HCl, 150 mM NaCl, 7 mM imidazole, pH 8.5) and elution buffer (50mM Tris/HCl,

150 mM NaCl,0.5 M imidazole, pH8.5) did not contain urea and contained a lower concentration of NaCl (150 mM). Eluted fractions containing enzyme were pooled, concentrated and rebuffered (25 mM Tris/HCl, 150 mM NaCl, pH 7) using Vivaspin concentrators from Sartorius (MWCO 10,000).

S 3

2. MALDI-ToF mass spectra

2.1. GlcNAc-Trt 7

949.453

5000[a.u.] Intens.

4000

3000

2000

729.228

1000

0 750 1000 1250 1500 1750 2000 2250 m/z

Figure S1 MALDI-ToF mass spectrum of GlcNAc-Trt 7 on aluminium backed polystyrene support.

S 4

2.2. Polystyrene with aluminium strip

Figure S2 MALDI-ToF mass spectrum of Man-Trt 5 spotted away from where the aluminium strip is located.

S 5

Figure S3 MALDI-ToF mass spectrum of Man-Trt 5 spotted over where the aluminium strip is located.

2.3. Polystyrene with aluminium box

Figure S4 Photo of the aluminium box backing. The aluminium box is not in contact with the metal frame.

S 6

Figure S5 MALDI-ToF mass spectrum of Man-Trt 5 when only a box of aluminium backing is applied, which does not make contact with the target support.

S 7

2.4. Limit of detection experiments

Serial dilutions of Man-Trt 5 linker were spotted onto steel, aluminium backed polystyrene and aluminium backed glass plates to determine the limit of detection of the altered non-conductive plate in comparison to conventional steel target plates. In this case, the spots were not washed with water so that the number of moles of linker in each spot would be more consistent.

Figure S6 m/z of the observed ions in the mass spectrum of Man-Trt 5.

S 8

2.4.1. Steel

Figure S7 Serial dilution from 15 to 7.5 nmol of Man-Trt 5 analysed on a steel target plate.

S 9

Figure S8 Serial dilution from 5 to 0.005 nmol of Man-Trt 5 analysed on a steel target plate.

S 10

2.4.2. Glass aluminium backed

Figure S9 Serial dilution from 15 to 7.5 nmol of Man-Trt 5 analysed on an aluminium backed glass slide.

S 11

Figure S10 Serial dilution from 5 to 0.05 nmol of Man-Trt 5 analysed on an aluminium backed glass slide.

S 12

2.4.3. Glass without aluminium backing

Figure S11 Serial dilution from 15 to 12.5 nmol of Man-Trt 5 analysed on a glass slide with no aluminium backing.

S 13

Figure S12 Serial dilution from 10 to 2.5 nmol of Man-Trt 5 analysed on a glass slide with no aluminium backing.

S 14

2.4.4. Polystyrene aluminium backed

Figure S13 Serial dilution from 15 to 7.5 nmol of Man-Trt 5 analysed on aluminium backed polystyrene plate.

S 15

Figure S14 Serial dilution from 5 to 0.05 nmol of Man-Trt linker 5 analysed on aluminium backed polystyrene plate.

As shown in figure 2 of the main report, MALDI-ToF analysis of Man-Trt 5 on polystyrene showed no signal and therefore there is no dilution series for the non- aluminium backed support as there was in glass.

S 16

CHAPTER TEN

DUAL PURPOSE S-TRITYL-LINKERS FOR GLYCOARRAY FABRICATION ON BOTH POLYSTYRENE AND GOLD

Reproduced by permission of The Royal Society of Chemistry

J. W. Wehner, M. J. Weissenborn, M. Hartmann, C. J. Gray, R. Sardzik, Th. K. Lindhorst, S. L. Flitsch, Dual purpose S-trityl-linkers for glycoarray fabrication on both polystyrene and gold, Org. Biomol. Chem. 2012, 10, 8919. DOI: 10.1039/c2ob26118a

This project was initiated by J. W. Wehner and T. K. Lindhorst. J. W. Wehner did the synthesis and characterisation of the molecules and performed the experiments in the microtiter plates — including the bacterial adhesion test where M. Hartmann helped with the interpretation. C. J. Gray, J. W. Wehner and M. J. Weissenborn designed the in situ deprotection and coupling on SAMs and directly on plain gold. C. J. Gray and M. J. Weissenborn performed the experiments and analysed by MALDI-ToF MS where C. E. Eyers gave valuable input. R. Sardzik and M. J. Weissenborn did the SPR analysis. J. W. Wehner, C. J. Gray, T. K. Lindhorst, S. L. Flitsch and M. J. Weissenborn wrote the article. J. W. Wehner, C. J. Gray and M. J. Weissenborn wrote the supporting information.

29 View Online / Journal Homepage / Table of Contents for this issue Organic & Dynamic Article Links Biomolecular Chemistry Cite this: Org. Biomol. Chem., 2012, 10, 8919 www.rsc.org/obc PAPER

Dual purpose S-trityl-linkers for glycoarray fabrication on both polystyrene and gold†

Johannes W. Wehner,a Martin J. Weissenborn,b Mirja Hartmann,a Christopher J. Gray,b Robert Šardzík,b Claire E. Eyers,b Sabine L. Flitsch*b and Thisbe K. Lindhorst*a Received 11th June 2012, Accepted 21st September 2012 DOI: 10.1039/c2ob26118a

There is a wide range of immobilisation reactions to tether substrates to a variety of surfaces for array-based analysis. Most of these immobilisation strategies are speciﬁc for a particular surface and require an additional linker to be attached to the substrate or the surface. Furthermore, the analysis of functionalised surfaces is often restricted to certain analytical techniques and therefore, different immobilisation strategies for different surfaces are desirable. Here we have tested an S-tritylated linker for non-covalent or covalent immobilisation of mannosides to polystyrene or gold surfaces. S-Tritylated mannosides with varying linkers were readily synthesised and used to add to biorepulsive maleimide- terminated preformed SAMs after in situ deprotection of the S-trityl group. In addition, S-tritylated mannosides themselves formed stable glycoarrays on polystyrene microtiter plates. The glycoarrays were successfully analysed by MALDI-ToF mass spectrometry, SPR spectroscopy, and interrogated with GFP-transfected Escherichia coli cells. This work has shown that a dual purpose linker can be used on multiple surfaces to form arrays allowing for different testing as well as analytical approaches.

Introduction via native chemical ligation (NCL),17 1,3-dipolar cycloaddition,18 or Diels–Alder cycloaddition.19,20 Each surface type, Microarrays are valuable tools in the analysis of biological inter- however, requires speciﬁc methods for functionalisation. Downloaded by The University of Manchester Library on 20 November 2012

Published on 21 September 2012 http://pubs.rsc.org | doi:10.1039/C2OB26118A actions in fundamental research and in high-throughput screen- For example, for the formation of self-assembled monolayers ing and have promising applications as diagnostic devices in the (SAMs) on gold, thiol-functionalised derivatives are needed, in 1,2 clinic. Among many different microarrays, glycoarrays are order to form Au–S bonds on the surface. Hence, immersion of carbohydrate-functionalised surfaces, which have received much gold wafers in a solution of thiol-functionalised glycosides has attention in the glycosciences for the investigation of the mole- been developed into a common method for the preparation of 3–7 cular details of carbohydrate–protein interactions, and to glyco-SAMs.18,19,21 Moreover, carbohydrate thiols have been study cellular adhesion such as in the context of carbohydrate- utilised in the formation of glycoarrays through Michael-type 8 specific bacterial colonisation of surfaces. addition to maleimide-terminated surfaces.22 For the formation A key step in the preparation of glycoarrays, which can of glycoarrays on polystyrene, prefunctionalised microtiter plates consist of a variety of materials such as gold, glass or poly- have been employed for covalent immobilisation.8,23 However, styrene, is the immobilisation of the glycoconjugates on the direct non-covalent array formation on polystyrene is especially 9–14 respective surface. A variety of methods have been utilised appealing as it requires no additional immobilisation agents. As for this step, involving covalent or non-covalent attachments. polystyrene is inherently hydrophobic, hydrophobic interactions Common covalent immobilisation techniques include the amide or π–π interactions between the glycoconjugate and the polymer 15,16 formation via direct amine coupling into activated esters or surface can be used in this case to produce robust glycoarrays.3,9,13,24–26 In the course of our work on the preparation and biological aOtto Diels Institute of Organic Chemistry, Christiana Albertina testing of glycoamino acids, we have found that S-trityl- University of Kiel, Otto-Hahn-Platz 3/4, 24098 Kiel, Germany. protected low molecular weight glycoconjugates are readily E-mail: [email protected]; Fax: +49-431-880-7410 27,28 b fi School of Chemistry & Manchester Institute of Biotechnology, The made and puri ed. This has prompted us to test the direct University of Manchester, 131 Princess Street, Manchester, M17DN, application of S-tritylated carbohydrate derivatives for the UK. E-mail: sabine.fl[email protected]; preparation of glycoarrays on different surfaces. De-tritylation Fax: +44 (0)161-275-1311; Tel: +44 (0)161-306-5172 would lead to thiol-modified glycoconjugates to allow immobil- †Electronic supplementary information (ESI) available: Experimental procedures and analytical details (NMR, mass spectrometry, characteris- isation on plain gold or on preformed maleimide-terminated ation and formation of glycoarrays). See DOI: 10.1039/c2ob26118a biorepulsive SAMs. On the other hand, the trityl protecting

This journal is © The Royal Society of Chemistry 2012 Org. Biomol. Chem., 2012, 10, 8919–8926 | 8919 View Online

2-aminoethyl mannoside 129 with the commercially available cysteine derivatives Fmoc–Cys(Trt)–OH (2) or 11-tritylsulphanyl- undecanoic acid (3),9,30 leading to the peptide-coupled mannosides 427 and 6,9 respectively. Removal of the Fmoc protecting group and N-acetylation can be effected in one pot yielding the known S-tritylated glycoamino acid 5.27 Then, it was important to add S-tritylated mannoside derivatives to the collection having a longer spacer, because this usually facilitates immobilisation of the respective compound on a surface. Therefore, 6-amino- 4-thiahexyl mannoside 78,31 was made and subjected to peptide coupling with the S-tritylated thiols 2 and 3, using HATU and DIPEA. In analogy to the preparation of 4 and 6, this reaction led to the trityl-functionalised mannosides 8 and 11, having con- Fig. 1 Dual linkers: thio-functionalised bioprobes (e.g. carbohydrates siderably longer spacers than their analogues 4 and 6. Removal as shown for α-D-mannosides) can be directly attached to polystyrene of the Fmoc protecting group in 8 led to 9 and then acetylation surfaces in their S-tritylated form (left), or added as thiols to gold or to the N-acetylated target molecule 10 (Scheme 1). maleimide-terminated surfaces (right).

Fabrication and interrogation of glycoarrays on gold group should also allow preparation of glycoarrays on simple polystyrene microtiter plates, by hydrophobic interactions Initially we tested, if the prepared S-tritylated glycosides can be between the molecule’s trityl fragment and the hydrophobic immobilised on gold with concomitant removal of the trityl pro- surface (Fig. 1). In this account we report the synthesis of S-trity- tecting group. Thus, in situ de-tritylation of 5, 6, 10, and 11 was lated glycosides, their utilisation in glycoarray fabrication on effected overnight by treatment with trifluoroacetic acid (TFA) gold as well as on polystyrene and the interrogation of the pre- and triethylsilane (TES) in dichloromethane.28 Then, the solvent pared surfaces with lectins as well as live bacterial cells. was removed and the crude free thiol dissolved in PBS buffer, centrifuged and the solution applied to the gold surface according to the standard protocol for preparation of SAMs.17 To test if Results and discussion the immobilisation of in situ deprotected mannosides was Synthesis of S-tritylated glycoconjugates successful, the prepared glycoarrays were analysed by MALDI- ToF mass spectrometry.32 This mass spectrometric protocol is a As we have long-standing interest in the investigation of reliable method for the analysis of SAMs on gold, in which typi- mannose-specific lectins, in particular mannose-specific bacterial cally the masses of the disulphides of the respective thiols are adhesion, we have made a selection of four S-tritylated manno- detected.17 Also here, the detected peaks correspond to the disul-

Downloaded by The University of Manchester Library on 20 November 2012 side derivatives for this study, 5, 6, 10, and 11 (Scheme 1). phides of the thiols derived from 5, 6, 10, and 11 (cf. ESI, Published on 21 September 2012 http://pubs.rsc.org | doi:10.1039/C2OB26118A We have shown earlier that preparation of mannosides 5 and 6 Fig. S16–S19†). Thus, the MS analysis showed the success of is readily accomplished by coupling of the well-known glycoarray formation after in situ de-tritylation. In addition,

Scheme 1 Synthesis of S-tritylated α-D-mannosides 5, 6, 10, and 11. Reaction conditions: (a) HATU, DIPEA, DMF, 0 °C → room temp., overnight, 81% (4,from1 and 2), 74% (6,from1 and 3), 84% (8,from7 and 2), 78% (11,from7 and 3); (b) (i) morpholine, DMF, room temp., quant. (ii) Ac2O, DIPEA, room temp., 4 h, quant.; (c) morpholine, DMF, room temp., 67%; (d) pyridine, Ac2O, room temp., overnight, 96%.

8920 | Org. Biomol. Chem., 2012, 10, 8919–8926 This journal is © The Royal Society of Chemistry 2012 View Online

tritylated 6 was deprotected and purified to deliver the pure thiol Fabrication and interrogation of glycoarrays on polystyrene 6-SH. When pure 6-SH was employed for glycoarray fabrication, mass spectrometric analysis gave very similar results as The next step was to employ the S-trityl group in mannosides 5, when the in situ deprotection–immobilisation approach was 6, 10, and 11 (Scheme 1) to anchor these molecules to a hydro- † fi phobic polystyrene surface, as it has been shown earlier for employed (cf. ESI, Fig. S15 ). The ef ciency of immobilisation 13 of in situ deprotected thiols greatly simplifies fabrication of another type of tritylated molecules. For the non-covalent func- glycoarrays on gold. S-Tritylated derivatives are much easier to tionalisation of polystyrene microplates hydrophobic molecules purify than free thiols, owing to their greater hydrophobicity. have been used regularly. From these studies it is known that the π–π In addition, free thiols are prone to oxidation, forming the interactions established between the polystyrene surface and respective disulphides, a problem which is circumvented in the the aromatic trityl fragment are strong enough to guarantee a in situ deprotection approach. robust direct immobilisation on polystyrene microtiter plates. The same in situ de-tritylation protocol was employed on pre- First, the reaction conditions for glycoarray fabrication on fi formed maleimide-terminal self-assembled monolayers (SAMs) polystyrene were optimised and methanol was identi ed as on gold (cf. Fig. 1). These SAMs include biorepulsive oligoethyl- the most suitable solvent for immobilisation of the prepared eneglycol units,33 which are important in biological studies to tritylated glycosides. Then, in order to determine the stability avoid nonspecific protein adsorption. The S-tritylated bioprobes of produced glycoarrays against different washing conditions, – 5, 6, 10, and 11 were treated as above and the non-purified mix- a colorimetric phenol sulphuric acid assay was performed 34–36 fi tures directly applied to the maleimide-functionalised surface. (Fig. 3). This assay allows quanti cation of glycoconjugates After 1.5 h reaction time, the surface was rinsed with ethanol immobilised on surfaces. Washing with ethanol removed the and again analysed by MALDI-ToF MS. The MALDI MS immobilised glycosides completely, as expected. In contrast, spectra showed the corresponding masses of the coupled ligands washing with twice distilled water and/or PBST buffer led to (ESI, Fig. S24–S27†). Comparison with the coupling results negligible reduction of the carbohydrate content on the surface obtained with previously deprotected and purified thiols revealed in the case of the glycoarrays formed from the tritylated glyco- that glycoarray formation after in situ deprotection is similarly sides 6, 10, and 11. However, the mannoside with the shortest effective. spacer, compound 5, formed the least stable glycoarray on poly- As an additional method to test glycoarray formation on gold, styrene which was washed out by water or buffer to over 50% – SPR spectroscopy was used. Here, the mannose-specific lectin according to the phenol sulphuric acid assay. concanavalin A (ConA) was employed for interrogation of In the next step, the prepared glycoarrays were tested with live glycoarrays prepared after in situ de-tritylation of a pair of com- bacterial cells in a GFP-assisted adhesion assay, which was established earlier.8 Here, the genetically engineered E. coli parable mannosides, 10 having the shorter spacer, and 11 having 8,37 a longer spacer incorporated. The crude thiols were added to a strain PKL1162 was used. Protocols for cellular adhesion maleimide-terminated biorepulsive SAM on gold and ConA was assays on polystyrene microplates usually involve a blocking allowed to interact with the formed glycoarray. In both cases, the step with BSA or skimmed milk for example, to prevent fi expected carbohydrate–lectin interactions were detected, unspeci c binding of the cells to the microtiter plate surface. Downloaded by The University of Manchester Library on 20 November 2012

Published on 21 September 2012 http://pubs.rsc.org | doi:10.1039/C2OB26118A suggesting glycoarray formation (Fig. 2). For the array formed However, we could show that when the tritylated mannosides from thiol derived from 11, a much stronger interaction with ConA was measured than for the analogous case using 10. This suggests that (tritylated) thiols having the bioprobe attached to a rather long spacer are better suited for surface immobilisation.

Fig. 3 Removal of compounds 5, 6, 10 and 11 from polystyrene surfaces using different washing steps. Microtiter plate wells were functionalised with 25 mM methanolic solutions of tritylated mannosides 5, 6, 10, and 11, then it was washed with water and/or buffer and the remaining glycoside content on the surface determined by the phenol–sulphuric Fig. 2 SPR spectroscopy on surfaces functionalised with 10 and 11 colorimetric acid assay. The glycoside content without washing was using the lectin ConA was used to prove the formation of glycoarrays deﬁned as 100%. Six-fold washing with ethanol removed the glycoarray after in situ deprotection of tritylated mannosides 10 and 11. completely (not shown).

This journal is © The Royal Society of Chemistry 2012 Org. Biomol. Chem., 2012, 10, 8919–8926 | 8921 View Online

Fig. 5 Inhibition curves of competitive bacterial adhesion inhibition Fig. 4 Bacterial adhesion curves (GFP-tagged E. coli PKL1162) assays using polystyrene glycoarrays prepared from 6 (top) and 11 obtained by application of glycoarrays consisting of compounds 6 (top) (bottom). Methyl α-D-mannoside was used as an inhibitor and type 1 and 11 (bottom) after 1 h incubation and ﬂuorescence readout. ﬁmbriated E. coli cells (PKL1162) were used to adhere to the surface.

were used for modification of polystyrene plates, no blocking step was necessary (ESI, Fig. S28†). glycoarray formation. This finding is in line with earlier results, Glycoarrays on polystyrene were prepared using tritylated which have indicated that mannosides having a thiahexyl mannosides 6 and 11 at different concentrations. Concentration aglycon moiety show a higher affinity to the mannose-specific dependency of bacterial adhesion to these two glycoarrays was lectin of E. coli than mannosides having an ethyl aglycon.38 tested and was found as expected in both cases, with the inten- Downloaded by The University of Manchester Library on 20 November 2012

Published on 21 September 2012 http://pubs.rsc.org | doi:10.1039/C2OB26118A sity of the GFP fluorescence increasing with higher concentrations of the applied mannoside solutions (Fig. 4). A Conclusions plateau was reached at concentrations between 20 mM and S-Tritylated mannosides were synthesised and shown to be suit- 25 mM. Mannosides 5 and 10 were less suited in this assay. able for the fabrication of glycoarrays on different surfaces such Only little adhesion could be detected and no consistent concen- as gold and polystyrene. An in situ deprotection protocol has tration dependency of bacterial adhesion could be observed in allowed us to apply tritylated carbohydrate derivatives on plain the case of these mannosides linked via short spacers (ESI, gold as well as on maleimide-terminal preformed biorepulsive Fig. S31†). SAMs. As SAMs on gold on the one hand and polystyrene After having shown that tritylated mannosides such as 6 and microtiter plates on the other can be used in quite different appli- 11 form stable glycoarrays on polystyrene microtiter plates upon cations, S-tritylated glycoconjugates can be regarded as facile direct treatment, testing of inhibition of bacterial adhesion to derivatives for orthogonal immobilisation on surfaces of opposite these surfaces could be done next. From the results obtained character. in the adhesion experiments, 25 mM concentrations appeared The prepared glycoarrays were analysed by MALDI-ToF optimal to form microarrays for competitive bacterial adhesion MS and were shown to be robust and suited for interrogation inhibition assays. As described earlier,38 serial dilutions with lectins and live bacterial cells. Next, we will further employ of methyl α-D-mannoside (MeMan), a standard inhibitor this methodology in a 384 well polystyrene microtiter plate of mannose-specific bacterial adhesion, were applied to inhibit format to facilitate inhibitor screening in bacterial adhesion bacterial adhesion to the two different glycoarrays formed with 6 assays. and 11, respectively. The obtained inhibition curves are depicted in Fig. 5. After sigmoidal fitting of the testing results, IC50 ∼ values could be deduced, with IC50 (MeMan) 2.9 mM for the Experimental inhibition of bacterial adhesion to the surface, modified with mannoside 6 and IC50 (MeMan) ∼5.3 mM for 11. Thus, the Commercially available starting materials and reagents were surface prepared from mannosides 11 appears to be slightly used without further purification. Reactions requiring dry con- more adhesive in this testing system than when 6 was used for ditions were performed under an atmosphere of nitrogen using

8922 | Org. Biomol. Chem., 2012, 10, 8919–8926 This journal is © The Royal Society of Chemistry 2012 View Online

oven-dried glassware. Anhydrous DMF was purchased from General procedure for in situ de-tritylation of S-protected Acros. All other used solvents were purified by distillation. mannosides ConA was purchased from Vector labs. Bovine serum albumin The S-tritylated mannosides (5, 6, 10,or11,3μmol) were dis- (BSA), methyl α-D-mannopyranoside (MeMan) and polyethylene glycol sorbitan monolaurate (Tween® 20) were obtained solved in dichloromethane (1 mL). Then, triethylsilane (5 equiv.) fl from Sigma-Aldrich. Microtiter plates with a hydrophobic and tri uoroacetic acid (5 equiv.) were added and the reaction surface (Corning, no. 3540, low volume 384 wells, flat clear mixture was left for 1.5 h at room temperature and was sub- fl bottom, black polystyrene, nontreated and Corning, no. 3631, 96 sequently treated with further tri uoroacetic acid (5 equiv.) and wells, flat clear bottom, black polystyrene, nontreated) were left overnight. Thereafter, the solvent was removed in vacuo and the residue dissolved in 10 mM PBS to obtain a final concen- obtained from Corning. 2-Aminoethyl α-D-mannopyranoside 29 8,31 tration of the corresponding free thiol of 10 mM. This solution (1), 6-amino-4-thiahexyl α-D-mannopyranoside (7), was centrifuged and the supernatant directly applied to the differ- N-(fluoren-9-yl-methoxycarbonyl)-S-(triphenylmethyl)-L-cysteine- 27 ently modified gold surfaces. [2-(α-D-mannopyranosyloxy)ethyl]amide (4), 11-tritylsulphanyl- undecanoic acid, and 2-(11-tritylsulphanyl-undecanoyl)- 9 aminoethyl α-D-mannopyranoside (6) were prepared according Fabrication of glycoarrays on gold and maleimide-terminated to the literature. SAMs and their MALDI-ToF MS analysis Reactions were monitored by thin-layer chromatography using either silica gel 60 GF254 on aluminium foil (Merck) or RP-18 A disposable 64-well gold plate (Applied Biosystems) was F254s on aluminium foil (Merck) with detection by UV light cleaned with a Piranha solution (12 mL, 3 : 1 conc. H2SO4/30% and charring with sulphuric acid in EtOH (10%). Merck silica H2O2) for 30 min, rinsed with distilled water and ethanol gel 60 (0.040–0.063 mm) was used for flash chromatography. and dried under a stream of nitrogen. A solution of carboxylic – – – – Analytical HPLC was performed on a Merck Hitachi LaChrom acid-terminated linkers [HS (CH2)17 (OC2H4)6 OCH2 COOH] L-7000 series apparatus with a LiChrospher 100 RP-8 (5 μm, and tri(ethylene glycol)-terminated alkanethiol spacers – – – fi Merck) column (for HPLC chromatograms see the ESI†). Pre- [HS (CH2)17 (OC2H4)3 OH] in dry DMSO ( nal concentration −1 parative MPLC was performed on a Büchi apparatus using a 0.4 mg mL , molar ratio 1 : 4) was applied on the plate (∼1 μL LiChroprep RP-18 column (40–60 μm, Merck) for reversed- per well) and left overnight at room temperature to form a mixed phase and a LiChroprep Si 60 column (40–60 μm, Merck) for SAM. The plate was washed with ethanol and dried under nitro- normal-phase silica gel chromatography. 1H and 13C NMR gen. The carboxylic acid groups were activated by spotwise spectra were recorded on a Bruker DRX-500 or a Bruker AV-600 treatment with a solution of EDC, NHS and N-(2-aminoethyl)- instrument. NMR spectra were calibrated with respect to the maleimide (all Sigma-Aldrich, 0.180 M, 0.174 M and 0.174 M, – solvent peak (in the case of CDCl3 the reference was tetra- respectively) in dry DMF for 1 2 h, followed by washing with methylsilane (TMS)). 2D NMR techniques (COSY, HSQC, water and ethanol and drying as above. The product formation HMBC) were used for full assignment of the spectra. ESI MS was analysed by MALDI-ToF MS. Unless otherwise stated all measurements were performed on a Mariner ESI-ToF 5280 MALDI-ToF MS experiments on gold surfaces were carried out Downloaded by The University of Manchester Library on 20 November 2012

Published on 21 September 2012 http://pubs.rsc.org | doi:10.1039/C2OB26118A instrument (Applied Biosystems). MALDI-ToF mass spectra on an Ultraflex II instrument (Bruker Daltonics) in positive were recorded on a Bruker Biflex-III 19 kV instrument with Cl- reflection mode. A solution of matrix (2,4,6-trihydroxyacetophe- −1 CCA (4-chloro-α-cyanocinnamic acid) or DHB (2,5-dihydroxy- none, 10 mg mL in acetone) was applied on the gold and benzoic acid) as matrix. Optical rotation was measured on a allowed to dry before analysis. Perkin-Elmer polarimeter 341 (Na-D-line: 589 nm, length of cell 1 dm). IR spectra were recorded on a Perkin-Elmer Paragon Surface plasmon resonance (SPR) analysis using 10 and 11 1000 FT-IR instrument. For sample preparation a Golden Gate diamond ATR unit with a sapphire stamp was used. The SPR The gold-coated sensor was cleaned with a Piranha solution, experiments were performed on a Biacore 3000 system (GE rinsed with water and ethanol and dried in a stream of nitrogen. Healthcare, Sweden) using a gold sensor chip (GE Healthcare). The formation of self-assembling monolayers was performed in For bacterial adhesion studies and phenol–sulphuric acid assays, the same way as described above. The chip was washed with a TECAN infinite 200 multifunction microplate reader was ethanol, dried under nitrogen and mounted onto a chip holder employed. The wavelengths of the band pass filters for excitation following the instructions in the supplier’s manual. After and emission were 485 and 535 nm, respectively. For the docking in the instrument the sensor was equilibrated with PBS phenol–sulphuric acid assay absorbance at 492 nm was buffer (10 mM, degassed and filtered) at a flow rate of 10 μL measured. min−1. For surface activation, 70 μL of a 1 : 1 mixture of E. coli bacteria (PKL1162)8,37 were used and grown in freshly prepared solutions of NHS (0.4 M) and EDC (0.1 M) in LB-media + AMP + CAM (100 mg ampicillin, 50 mg water were injected. The reference spot (channel 1) was blocked chloramphenicol L−1) at 37 °C under slight agitation. Buffers by injecting 70 μL of aminoethanol-hydrochloride (1 M). were used as follows: PBS buffer solution (pH 7.2): sodium Additional channels were modified with N-(2-aminoethyl)- chloride (8.00 g), potassium chloride (200 mg), sodium hydro- maleimide (10 mM, flow rate 10 μL min−1) for 10 min and then gen phosphate-dihydrate (1.44 g), and potassium dihydrogen treated with in situ deprotected mannosides 10 and 11 for about phosphate (200 mg) were dissolved in bidist. water (1.00 L). an hour at a flow rate of 3 μL min−1. PBST buffer solution (pH 7.2): PBS buffer + 0.05% v/v Binding studies were carried out using the lectin ConA (10 μg −1 Tween® 20. mL , 250 μL) in buffer (0.15 M NaCl, 1 mM CaCl2,1mM

This journal is © The Royal Society of Chemistry 2012 Org. Biomol. Chem., 2012, 10, 8919–8926 | 8923 View Online

−1 MnCl2, pH 7.0) at a ﬂow rate of 25 μL min . After the injection and the wells were then washed with PBS buffer (3 × 20 μL per a 600 s dissociation sequence was followed. well). Fluorescence was read out at 485/535 nm.

Fabrication of glycoarrays on polystyrene 2-(11-Sulphhydryl-undecanoyl)aminoethyl α-D-mannopyranoside (6-SH) A series of 5 to 50 mM stock solutions of tritylated mannosides 5, 6, 10, and 11 (in MeOH) were prepared. A 12 μL sample of The tritylated mannoside 6 (50 mg, 75.1 μmol) was dissolved in each solution was pipetted into a 384-well polystyrene microtiter dichloromethane (1 mL), triethylsilane (60 μL, 376 μmol) and plate, which was dried by standing overnight at ambient temp- trifluoroacetic acid (58 μL, 751 μmol) were added and the reac- erature. Each well was then washed three times with deionised tion mixture was stirred for 2 h at ambient temperature. The water and three times with PBST buffer (20 μL per well each). solvent was removed under reduced pressure and the crude product was purified by RP-MPLC (120 g RP-18, A: methanol, B: water, A: 60% → 95%, 90 min) yielding the deprotected title Phenol–sulphuric acid assay compound 6-SH (31.3 mg, 73.6 μmol, 98%) after lyophilisation. 26 Rf 0.33 (methanol–water, 3 : 1); [α]D = +40.4 (c = 0.5, metha- To test the stability of glycoarrays formed by immobilisation of 1 3 nol); H NMR (500 MHz, CD3OD, 300 K): δ = 4.76 (d, J = 5, 6, 10 or 11 against different washing conditions, these com- 2 3 1.7 Hz, 1H, H-1Man), 3.83 (dd, J = 11.7 Hz, J = 2.3 Hz, 1H, pounds were immobilised as described above followed by 3 3 H-6aMan), 3.81 (dd, J = 1.7 Hz, J = 3.4 Hz, 1H, H-2Man), 3.75 6 washing cycles with ethanol (20 μL per well each) followed (mc, 1H, OCHHCH2NH), 3.72–3.67 (m, 2H, H-3Man, H-6bMan), by phenol–sulphuric acid assay. 12 μL of solutions of trityl- 3 3.60 (dd ∼ t, J = 9.5 Hz, 1H, H-4Man), 3.56–3.51 (m, 2H, protected carbohydrate (50 mM, 25 mM, 12.5 mM solutions in H-5Man, OCHHCH2NH), 3.45–3.32 (m, 2H, OCH2CH2NH), MeOH) were pipetted into a 384-well microtiter plate and the 3 3 2.49 (t, J = 7.1 Hz, 2H, CH2CH2SH), 2.19 (t, J = 7.5 Hz, plate was allowed to dry by standing overnight at ambient temp- 2H, HN(O)CCH2CH2), 1.59 (mc, 4H, HN(O)CCH2CH2, erature. The wells were then washed three times with deionised OCH2CH2CH2S), 1.40 (mc, 2H, CH2CH2CH2), 1.35–1.25 (m, water and three times with PBST (20 μL per well each). The 13 12H, CH2CH2CH2) ppm; C NMR (125 MHz, CD3OD, phenol–sulphuric acid assay was performed according to a litera- 300 K): δ = 176.5 (C(O)NH), 101.7 (C-1Man), 74.8 (C-5Man), ture-known method.35 A 5% phenol solution (4.2 μL per well) 72.6 (C-3Man), 72.1 (C-2Man), 68.6 (C-4Man), 67.3 was pipetted to the wells, followed by the addition of concen- (OCH2CH2NH), 62.9 (C-6Man), 40.2 (OCH2CH2NH), 37.1 (HN trated H SO (21 μL per well). The mixture was incubated for 2 4 (O)CCH2CH2), 35.2 (OCH2CH2CH2S), 30.6, 30.5, 30.4, 30.3, 30 min at room temperature, and the absorbance measured at 30.2, 29.4 (6 CH2CH2CH2), 27.0 (HN(O)CCH2CH2), 25.0 492 nm (A492) to determine the amount of carbohydrate (CH2CH2SH) ppm; HR-ESI MS: calcd for C38H72N2NaO14S2 immobilised on the microtiter plate. The amount of immobilised (disulphide): m/z 867.4317 [M + Na]+; found: m/z 867.4309 5, 6, 10 and 11 was estimated from the ratio of the absorption at [M + Na]+; IR (ATR): ν˜ = 3308, 2918, 2850, 1637, 1554, 1463, −1 Downloaded by The University of Manchester Library on 20 November 2012 492 nm of immobilised compounds (subjected to 3 washing 1132, 1057, 1031, 975 cm . Published on 21 September 2012 http://pubs.rsc.org | doi:10.1039/C2OB26118A cycles with deionised water and/or PBST) to the A492 of the corresponding control (unwashed). Washings with ethanol (6 washing cycles using 20 μL ethanol per well) removed the N-(Fluoren-9-yl-methoxycarbonyl)-S-(triphenylmethyl)-L- glycoarrays completely according to the phenol–sulphuric acid cysteine-[6-(α-D-mannopyranosyloxy)-3-thiahexyl]amide (8) assay. A mixture of Fmoc-L-Cys(Trt)-OH (2, 2.46 g, 4.18 mmol), mannoside 1 (1.37 g, 4.60 mmol), and HATU (1.91 g, GFP-based bacterial adhesion assay 5.02 mmol) was dried for 1 h under vacuum and then dissolved in dry DMF (40 mL). It was cooled to 0 °C, DIPEA (853 μL, Determination of bacterial adhesion. Trityl-protected carbo- 5.02 mmol) was added and the reaction mixture stirred overnight hydrates (5, 6, 10, and 11) were immobilised on 384-well micro- at ambient temperature under a nitrogen atmosphere. The solvent titer plates as described above. The wells were incubated with was removed under reduced pressure and the crude product was − E. coli PKL1162 (2 mg mL 1 PBS buffer) for 1 h (37 °C, purified by column chromatography (methanol–ethyl acetate, 120 rpm), and subsequently washed with the same buffer 1:12→ 1 : 9) yielding the title compound 8 (3.01 g, 3.48 mmol, μ (3 × 20 L per well). E. coli binding to the mannoside- 84%) as a colourless foam. Rf 0.38 (methanol–ethyl acetate, fl 22 1 functionalised surface was monitored by uorescence measure- 1 : 9); [α]D = +21.8 (c = 0.5, methanol); H NMR (500 MHz, 3 ments at 485/535 nm using a microplate reader. CD3OD, 300 K): δ = 7.78 (mc, 2H, H-arylFmoc), 7.66 (d, J = 6.8 Hz, 2H, H-arylFmoc), 7.40–7.34 (m, 8H, H-arylTrt, Inhibition of bacterial adhesion with methyl α-D-mannoside H-arylFmoc), 7.30–7.19 (m, 14H, H-arylTrt, H-arylFmoc), 4.72 (d, 3 2 3 (MeMan). Compounds 6 and 11 (12 μL per well, 25 mM) were J = 1.6 Hz, 1H, H-1Man), 4.41 (dd, J = 10.6 Hz, J = 7.1 Hz, 2 3 immobilised on 384-well microtiter plates as described above. 1H, CHHFmoc), 4.30 (dd, J = 10.6 Hz, J = 6.8 Hz, 1H, 3 3 Then, 5 μL of a serial dilution of the standard inhibitor MeMan CHHFmoc), 4.23 (dd ∼ t, J = 6.8 Hz, 1H, CHFmoc), 3.93 (dd, J 3 2 3 (1 μM–1000 mM) were pipetted to the plate followed by = 8.4 Hz, J = 5.5 Hz, 1H, H-αCys), 3.82 (dd, J = 11.7 Hz, J = −1 3 3 addition of 5 μLofE. coli (PKL1162) solution (4 mg mL 2.4 Hz, 1H, H-6aMan), 3.78 (dd, J = 3.3 Hz, J = 1.6 Hz, 1H, PBS buffer). The plate was incubated for 1 h (37 °C, 120 rpm) H-2Man), 3.77–3.74 (m, 1 H, OCHHCH2CH2S), 3.71 (dd,

8924 | Org. Biomol. Chem., 2012, 10, 8919–8926 This journal is © The Royal Society of Chemistry 2012 View Online

2 3 3 N S α J = 11.7 Hz, J = 5.8 Hz, 1H, H-6bMan), 3.67 (dd, J = 9.4 Hz, -(Acetyl)- -(triphenylmethyl)-L-cysteine-[6-( -D- 3 3 J = 3.3 Hz, 1H, H-3Man), 3.61 (dd ∼ t, J = 9.6 Hz, 1H, mannopyranosyloxy)-3-thiahexyl]amide (10) H-4Man), 3.54–3.50 (m, 1H, H-5Man), 3.49–3.43 (m, 1H, OCHHCH CH S), 3.34–3.23 (m, 2H, SCH CH NH), 2.62–2.49 The glycoamino acid 9 was dissolved in pyridine (2 mL) and 2 2 2 2 acetic anhydride (110 μL, 1.17 mmol) was added. The reaction (m, 6H, OCH2CH2CH2S, SCH2CH2NH, H-βCys), 1.79 (mc, 2H, 13 mixture was stirred overnight at room temperature. Then solvents OCH2CH2CH2S) ppm; C NMR (125 MHz, CD3OD, 300 K): δ = 172.7 (C(O)NH), 158.0 (OC(O)NH), 146.0 (C-aryl ), were removed under reduced pressure, it was codistilled with Trt toluene three times (10 mL each) and the crude product was sub- 145.1, 142.6 (C-arylFmoc), 130.8, 129.0 (CH-arylTrt), 128.8, 128.2 (CH-aryl ), 127.9 (CH-aryl ), 126.3, 120.9 jected to RP-MPLC (60 g RP-18, A: methanol, B: water, A: Fmoc Trt → (CH-aryl ), 101.6 (C-1 ), 74.7 (C-5 ), 72.7 (C-3 ), 40% 95%, 120 min) yielding the title compound 10 (154 mg, Fmoc Man Man Man μ 72.2 (C-2 ), 72.0 (C ), 68.6 (C-4 ), 68.1 (CH ), 225 mol, 96%) after lyophilisation. Rf 0.31 (ethyl acetate); Man q,Trt Man 2,Fmoc α 22 66.9 (OCH CH CH S), 62.9 (C-6 ), 55.7 (C-α ), 48.4 [ ]D = +29.4 (c = 0.5, methanol); HPLCtR = 2.64 min (A = 2 2 2 Man ys water, B = methanol, A: 20%, 10 min, 1.2 mL min−1); 1H NMR (CHFmoc), 40.1 (SCH2CH2NH), 35.2 (C-βCys), 31.9 (500 MHz, CD3OD, 300 K): δ = 7.38 (mc, 6H, H-arylTrt), 7.30 (SCH2CH2NH), 30.6 (OCH2CH2CH2S), 29.3 (OCH2CH2CH2S) 3 (mc, 6H, H-arylTrt), 7.23 (mc, 3H, H-arylTrt), 4.73 (d, J = ppm; MALDI-ToF MS (DHB): calcd for C48H52N2NaO9S2: m/z 2 + + 1.6 Hz, 1H, H-1Man), 4.19 (mc, 1H, H-αCys), 3.83 (dd, J = 887.30 [M + Na] ; found: m/z 887.50 [M + Na] ; calcd 3 + 11.7 Hz, J = 2.4 Hz, 1H, H-6aMan), 3.81–3.77 (m, 2H, for C48H52KN2O9S2: m/z 903.27 [M + K] ; found: m/z 903.48 2 3 [M + K]+; IR (ATR): ν˜ = 3316, 3055, 2924, 1705, 1660, 1521, OCHHCH2CH2S, H-2Man), 3.71 (dd, J = 11.7 Hz, J = 5.6 Hz, – ∼ 3 1490, 1445, 1318, 1230, 1130, 1084, 1029, 974, 739 cm−1. 1H, H-6bMan), 3.70 3.67 (m, 1H, H-3Man), 3.61 (dd t, J = 9.6 Hz, 1H, H-4Man), 3.55–3.50 (m, 1H, H-5Man), 3.50–3.45 (m, 1H, OCHHCH2CH2S), 3.36–3.24 (m, 2H, SCH2CH2NH), S-(Triphenylmethyl)-L-cysteine-[6-(α-D-mannopyranosyloxy)- 3.62–2.56 (m, 4H, OCH2CH2CH2S, SCH2CH2NH), 2.56 (dd, 3-thiahexyl]amide (9) 2 3 2 J = 12.4 Hz, J = 6.3 Hz, 1H, H-βaCys), 2.49 (dd, J = 12.4 Hz, 3 β μ J = 7.7 Hz, 1H, H- bCys), 1.91 (s, 3H, NH(O)CCH3), 1.82 (mc, The Fmoc-protected cysteinyl mannoside 8 (406 mg, 467 mol) 13 was dissolved in dry DMF (5 mL) under a nitrogen atmosphere 2H, OCH2CH2CH2S) ppm; C NMR (125 MHz, CD3OD, δ and morpholine (250 μL, 2.87 mmol) was added. The reaction 300 K): = 173.0 (CH3C(O)NH), 172.4 (C(O)NH), 145.9 mixture was stirred for 1 h at ambient temperature and then (C-arylTrt), 130.7, 129.0, 128.0 (CH-arylTrt), 101.6 (C-1Man), another portion of morpholine (250 μL, 2.87 mmol) was added. 74.7 (C-5Man), 72.7 (C-3Man), 72.2 (C-2Man), 68.6 (C-4Man), 68.0 α This was repeated after 3 h and 4 h and stirred overnight. After (Cq,Trt), 66.9 (OCH2CH2CH2S), 62.9 (C-6Man), 54.0 (C- Cys), β 18 h the volatile compounds were removed under reduced 40.1 (SCH2CH2NH), 34.9 (C- Cys), 31.9 (SCH2CH2NH), 30.7 pressure and the crude product was puriﬁed by column chrom- (OCH2CH2CH2S), 29.3 (OCH2CH2CH2S), 22.5 (NHCOCH3) – – ppm; HR-ESI MS: calcd for C35H44N2NaO8S2: m/z 707.2431 atography (methanol ethyl acetate TEA, 100 : 20 : 1) yielding + + μ [M + Na] ; found: m/z 707.2426 [M + Na] ; MALDI-ToF MS the title compound 9 (200 mg, 311 mol, 67%) as a colourless + 22 (Cl-CCA): calcd for C35H44N2NaO8S2: m/z 707.24 [M + Na] ; syrup. Rf 0.28 (RP-18, methanol–water, 4 : 1); [α]D = +22.8 (c =

Downloaded by The University of Manchester Library on 20 November 2012 + 1 Published on 21 September 2012 http://pubs.rsc.org | doi:10.1039/C2OB26118A 0.26, methanol); H NMR (500 MHz, CD OD, 300 K): δ = 7.41 found: m/z 707.33 [M + Na] , calcd for C35H44KN2O8S2: m/z 3 723.22 [M + K]+; found: m/z 723.31 [M + K]+;IR(ATR):ν˜ = (mc, 6H, H-arylTrt), 7.29 (mc, 6H, H-arylTrt), 7.22 (mc, 3H, 3 – 3284, 2926, 1645, 1535, 1489, 1443, 1372, 1202, 1131, 1085, H-arylTrt), 4.74 (d, J = 1.6 Hz, 1H, H-1Man), 3.85 3.77 (m, 3 H, −1 2 1055, 1031, 975, 742, 698, 675 cm . OCHHCH2CH2S, H-6aMan, H-2Man), 3.71 (dd, J = 11.7 Hz, 3 3 3 J = 5.7 Hz, 1H, H-6bMan), 3.69 (dd, J = 9.6 Hz, J = 3.4 Hz, 1H, H-3 ), 3.61 (dd ∼ t, 3J = 9.6 Hz, 1H, H-4 ), Man Man 6-(10-Tritylsulphanyl-undecanoyl)amino-4-thiahexyl-α-D- 3.55–3.46 (m, 2H, H-5Man, OCHHCH2CH2S), 3.41–3.30 (m, 3 mannopyranoside (11) 2H, SCH2CH2NH), 3.11 (dd ∼ t, J = 6.5 Hz, 1H, H-αCys), 2 2.65–2.59 (m, 4H, OCH2CH2CH2S, SCH2CH2NH), 2.55 (dd, J 11-Tritylsulphanyl-undecanoic acid (3, 194 mg, 420 μmol) and 3 2 3 = 12.1 Hz, J = 6.1 Hz, 1H, H-βaCys), 2.40 (dd, J = 12.1 Hz, J HATU (319 mg, 840 μmol) were dried for 1 h under vacuum, = 7.0 Hz, 1H, H-βbCys), 1.90 (s, 3H, NHCOCH3), 1.83 (mc, 2H, dry DMF (2.50 mL) and DIPEA (288 μL, 1.68 mmol) were 13 OCH2CH2CH2S) ppm; C NMR (125 MHz, CD3OD, 300 K): δ added and the mixture was stirred for 20 min under a nitrogen = 175.5 (C(O)NH), 146.1 (C-arylTrt), 130.8, 129.0, 127.9 (CH- atmosphere at ambient temperature. Simultaneously in a different arylTrt), 101.6 (C-1Man), 74.7 (C-5Man), 72.7 (C-3Man), 72.2 reaction vessel 7 (150 mg, 504 μmol) was dried for 1 h under (C-2Man), 68.6 (C-4Man), 67.8 (Cq,Trt), 66.9 (OCH2CH2CH2S), vacuum, dissolved in dry DMF (2.50 mL) and stirred for 20 min 62.9 (C-6Man), 55.3 (C-αCys), 40.0 (SCH2CH2NH), 38.3 under a nitrogen atmosphere at ambient temperature. The reac- (C-βCys), 32.0 (SCH2CH2NH), 30.7 (OCH2CH2CH2S), 29.3 tion mixture with the preactivated 11-tritylsulphanyl-undecanoic + (OCH2CH2CH2S) ppm; HR-ESI MS: m/z = [Trt] 243.1242; acid (3) and HATU was cooled to 0 °C, the solution of manno- + calcd for C33H43N2O7S2: m/z 643.2506 [M + H] ; found: m/z side 7 was added and it was stirred under a nitrogen atmosphere + 643.2534 [M + H] ; calcd for C33H42NaN2O7S2: m/z 665.2326 at ambient temperature overnight. Then, all volatile compounds [M + Na]+; found: m/z 665.2323 [M + Na]+; MALDI-ToF MS were removed under reduced pressure and the crude product was + (DHB): calcd for C33H42NaN2O7S2: m/z 665.23 [M + Na] ; subjected to MPLC (150 g silica column, A: ethyl acetate, B: found: m/z 665.47 [M + Na]+; IR (ATR): ν˜ = 3303, 3062, 2921, methanol, A: 99% → 85%, 120 min) and RP-MPLC (60 g 1658, 1519, 1488, 1443, 1318, 1261, 1129, 1084, 1054, 1024, RP-18, A: methanol, B: water, A: 50% → 95%, 120 min) yield- 975, 803, 743 cm−1. ing the title compound 11 (242 mg, 327 μmol, 78%) as a

This journal is © The Royal Society of Chemistry 2012 Org. Biomol. Chem., 2012, 10, 8919–8926 | 8925 View Online

22 – colourless foam. Rf 0.29 (ethyl acetate), [α]D = +25.3 (c = 0.5, 5(a) K. R. Love and P. H. Seeberger, Angew. Chem., 2002, 114, 3733 3736; K. R. Love and P. H. Seeberger, Angew. Chem., Int. Ed., 2002, 41, methanol); HPLCtR = 6.11 min (A = water, B = methanol, A: – −1 1 3583 3586; (b) M. D. Disney and P. H. Seeberger, Chem. Biol., 2004, 11, 20%, 10 min, 1.2 mL min ); H NMR (500 MHz, CD3OD, 1701–1707; (c) T. Horlacher and P. H. Seeberger, Chem. Soc. Rev., 2008, 300 K): δ = 7.38 (mc, 6H, H-arylTrt), 7.26 (mc, 6H, H-arylTrt), 37, 1414–1422. 3 6 T. Feizi, F. Fazio, W. Chai and C.-H. Wong, Curr. Opin. Struct. Biol., 7.19 (mc, 3H, H-arylTrt), 4.75 (d, J = 1.7 Hz, 1H, H-1Man), – 3 2003, 13, 637–645. 3.85 3.80 (m, 2H, H-6aMan,OCHHCH2CH2S), 3.79 (dd, J = – 3 2 3 7 T. Yue and B. B. Haab, Clin. Lab. Med., 2009, 29,15 29. 3.3 Hz, J = 1.7 Hz, 1H, H-2Man), 3.72 (dd, J = 11.7 Hz, J = 8 M. Hartmann, A. K. Horst, P. Klemm and T. K. Lindhorst, Chem. 3 3 – 5.6 Hz, 1H, H-6bMan), 3.70 (dd, J = 9.3 Hz, J = 3.3 Hz, 1H, Commun., 2010, 46, 330 332. 3 9 M. J. Weissenborn, J. W. Wehner, C. J. Gray, R. Šardzík, C. E. Eyers, H-3Man), 3.62 (dd ∼ t, J = 9.5 Hz, 1H, H-4Man), 3.56–3.48 (m, – T. K. Lindhorst and S. L. Flitsch, Beilstein J. Org. Chem., 2012, 8, 753– 2H, H-5Man, OCHHCH2CH2S), 3.37 3.32 (m, 2H, 762. SCH2CH2NH), 2.63 (mc, 4H, OCH2CH2CH2S, SCH2CH2NH), 10 R. Šardzík, R. Sharma, S. Kaloo, J. Voglmeir, P. R. Crocker and 3 3 – 2.17 (t, J = 7.5 Hz, 2H, HN(O)CCH2CH2), 2.11 (t, J = 7.4 Hz, S. L. Flitsch, Chem. Commun., 2011, 47, 5425 5427. – 11 S. Serna, S. Yan, M. Martin-Lomas, I. B. H. Wilson and N.-C. Reichardt, 2H, CH2STrt), 1.89 1.81 (m, 2H, OCH2CH2CH2S), 1.59 – 3 – J. Am. Chem. Soc., 2011, 133, 16495 16502. (quint., J = 7.6 Hz, 2H, HN(O)CCH2CH2), 1.38 1.09 (m, 14H, 12 J. Voglmeir, R. Šardzík, M. J. Weissenborn and S. L. Flitsch, OMICS, 13 CH2CH2CH2) ppm; C NMR (125 MHz, CD3OD, 300 K): δ = 2010, 14, 437–444. 13 L. Zou, H. L. Pang, P. H. Chan, Z. S. Huang, L. Q. Gu and K. Y. Wong, 176.4 (C(O)NH), 146.5 (C-arylTrt), 130.8, 128.8, 127.7 (CH- Carbohydr. Res., 2008, 343, 2932–2938. arylTrt), 101.6 (C-1Man), 74.7 (C-5Man), 72.7 (C-3Man), 72.2 14 M. C. Bryan, F. Fazio, H.-K. Lee, C.-Y. Huang, A. Chang, M. D. Best, (C-2Man), 68.6 (C-4Man), 67.6 (Cq,Trt), 66.9 (OCH2CH2CH2S), D. A. Calarese, O. Blixt, J. C. Paulson, D. Burton, I. A. Wilson and – 62.9 (C-6Man), 40.1 (SCH2CH2NH), 37.1 (HN(O)CCH2CH2), C.-H. Wong, J. Am. Chem. Soc., 2004, 126, 8640 8641. 32.9 (CH STrt), 32.1 (SCH CH NH), 30.7 (OCH CH CH S), 15 N. Laurent, R. Haddoub, J. Voglmeir, S. C. C. Wong, S. J. Gaskell and 2 2 2 2 2 2 S. L. Flitsch, ChemBioChem, 2008, 9, 2592–2596. 30.5, 30.4, 30.4, 30.3, 30.1, 30.0, 29.7 (7 CH2CH2CH2), 29.3 16 N. Laurent, J. Voglmeir and S. L. Flitsch, Chem. Commun., 2008, 44, (OCH2CH2CH2S), 27.0 (HN(O)CCH2CH2) ppm; HR-ESI MS: 4400–4412. calcd for C H NNaO S : m/z 762.3469 [M + Na]+; found: m/z 17 M. J. Weissenborn, R. Castagnia, J. W. Wehner, R. Šardzík, 41 57 7 2 – 762.3453 [M + Na]+; MALDI-ToF MS (DHB): calcd for T. K. Lindhorst and S. L. Flitsch, Chem. Commun., 2012, 48, 4444 4446. + 18 M. Kleinert, T. Winkler, A. Terfort and T. K. Lindhorst, Org. Biomol. C41H57NNaO7S2: m/z 762.35 [M + Na] ; found: m/z 762.52 Chem., 2008, 6,2118–2132. + + – [M + Na] ; calcd for C41H57KNO7S2: m/z 778.32 [M + K] ; 19 B. T. Houseman and M. Mrksich, Chem. Biol., 2002, 9, 443 454. found: m/z 778.52 [M + K]+; IR (ATR): ν˜ = 3296, 2922, 2852, 20 H. S. G. Beckmann, A. Niederwieser, M. Wiessler and V. Wittmann, Chem.–Eur. J., 2012, 18, 6548–6554. 1643, 1544, 1488, 1443, 1130, 1084, 1054, 1030, 975, 811, 741, 21 R. Šardzík, A. P. Green, N. Laurent, P. Both, C. Fontana, J. Voglmeir, −1 697, 676, 616 cm . M. J. Weissenborn, R. Haddoub, P. Grassi, S. M. Haslam, G. Widmalm and S. L. Flitsch, J. Am. Chem. Soc., 2012, 134, 4521–4524. 22 B. T. Houseman, E. S. Gawalt and M. Mrksich, Langmuir, 2003, 19, 1522–1531. Abbreviations 23 C. Maierhofer, K. Rohmer and V. Wittmann, Bioorg. Med. Chem., 2007, 15, 7661–7676. DIPEA N,N-diisopropylethylamine, 24 M. C. Bryan, O. Plettenburg, P. Sears, D. Rabuka, S. Wacowich-Sgarbi

Downloaded by The University of Manchester Library on 20 November 2012 EDC 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide, and C.-H. Wong, Chem. Biol., 2002, 9, 713–720. Published on 21 September 2012 http://pubs.rsc.org | doi:10.1039/C2OB26118A HATU O-(7-azabenzotriazol-1-yl)-N,N,N′,N′-tetramethyl- 25 F. Fazio, M. C. Bryan, O. Blixt, J. C. Paulson and C.-H. Wong, J. Am. – uronium hexaﬂuorophosphate, Chem. Soc., 2002, 124, 14397 14402. 26 C. Grabosch, K. Kolbe and T. K. Lindhorst, ChemBioChem, 2012, 13, NHS N-hydroxysuccinimide, 1874–1879. PBS phosphate buffered saline, 27 A. Schierholt, M. Hartmann, K. Schwekendiek and T. K. Lindhorst, PBST phosphate buffered saline + 0.05% v/v Tween® 20, Eur. J. Org. Chem., 2010, 3120–3128. 28 J. W. Wehner and T. K. Lindhorst, Synthesis, 2010, 3070–3082. TEA triethylamine. 29 (a) M. Kleinert, N. Röckendorf and T. K. Lindhorst, Eur. J. Org. Chem., 2004, 3931–3940; (b) T. K. Lindhorst, S. Kötter, U. Krallmann-Wenzel Acknowledgements and S. Ehlers, J. Chem. Soc., Perkin Trans. 1, 2001, 823–831; (c) J. Dahmen, T. Frejd, G. Gronberg, T. Lave, G. Magnusson and – This work was supported by the Evonik Foundation (JWW), the G. Noori, Carbohydr. Res., 1983, 116, 303 307. 30 D. Ryan, B. A. Parviz, V. Linder, V. Semetey, S. K. Sia, J. Su, Royal Society (Wolfson Award to SLF), the European Commis- M. Mrksich and G. M. Whitesides, Langmuir, 2004, 20, 9080–9088. sion (MJW), the EPSRC and SFB 677 (DFG collaborative 31 D. Pagé and R. Roy, Bioorg. Med. Chem. Lett., 1996, 6, 1765–1770. network). A further thanks goes to Corning for the kind support 32 N. Laurent, J. Voglmeir, A. Wright, J. Blackburn, N. T. Pham, S. C. Wong, S. J. Gaskell and S. L. Flitsch, ChemBioChem, 2008, 9, with free samples. 883–887. 33 E. Ostuni, R. G. Chapman, R. E. Holmlin, S. Takayama and G. M. Whitesides, Langmuir, 2001, 17, 5605–5620. Notes and references 34 M. Dubois, K. A. Gilles, J. K. Hamilton, P. A. Rebers and F. Smith, Anal. Chem., 1956, 28, 350–356. 1 B. Cao, R. Li, S. Xiong, F. Yao, X. Liu, M. Wang, L. Feng and L. Wang, 35 S. K. Saha and C. F. Brewer, Carbohydr. Res., 1994, 254, 157–167. Appl. Environ. Microbiol., 2011, 77, 8219–8225. 36 T. Masuko, A. Minami, N. Iwasaki, T. Majima, S.-I. Nishimura and 2 M. Fais, R. Karamanska, D. A. Russell and R. A. Field, J. Cereal Sci., Y. C. Lee, Anal. Biochem., 2005, 339,69–72. 2009, 50, 306–311. 37 A. Reisner, J. A. J. Haagensen, M. A. Schembri, E. L. Zechner and 3 S. Fukui, T. Feizi, C. Galustian, A. M. Lawson and W. G. Chai, Nat. S. Molin, Mol. Microbiol., 2003, 48, 933–946. Biotechnol., 2002, 20, 1011–1017. 38 For a review see: M. Hartmann and T. K. Lindhorst, Eur. J. Org. Chem., 4 R. Jelinek and S. Kolusheva, Chem. Rev., 2004, 104, 5987–6015. 2011, 3583–3609.

8926 | Org. Biomol. Chem., 2012, 10, 8919–8926 This journal is © The Royal Society of Chemistry 2012 10.1. SUPPORTING INFORMATION

10.1 Supporting Information

The supporting information given below is speciﬁc to those experiments carried out by M. J. Weissenborn. The complete supporting information can be found under http://www.rsc.org/suppdata/ob/c2/c2ob26118a/c2ob26118a.pdf.

30 Supporting Information

Dual purpose S-trityl-linkers for glycoarray fabrication on both polystyrene and gold

Johannes W. Wehner,a Martin J. Weissenborn,b Mirja Hartmann,a Christopher J. Gray,b Robert Šardzík, b Claire E. Eyers,b Sabine L. Flitsch*b and Thisbe K. Lindhorst*a

Address: 1 Otto Diels Institute of Organic Chemistry, Christiana Albertina University of Kiel, Otto-Hahn-Platz 3/4, 24098 Kiel, Germany. 2 School of Chemistry, University of Manchester, Manchester Interdisciplinary Biocentre, 131 Princess Street, Manchester, M17DN, United Kingdom.

E-mail: Prof. Thisbe K. Lindhorst*: [email protected] Prof. Sabine L. Flitsch*: [email protected] * Corresponding authors

S 1

Table of contents page

1 Glycoarrays on gold S12 1.1 Gold chip functionalisation S12 1.1.1 Coupling of in situ deprotected thiols to unmodified gold surface S12 1.1.2 Coupling of in situ deprotected thiols to maleimide-terminated gold surface S12 1.2 MALDI-ToF MS analysis of SAMs on gold S12 1.2.1 SAM formation on gold surface using in situ deprotected thiols S13 1.2.2 SAM formation using maleimide-terminated thiols SAMs S16 1.2.3 Thiol coupling to maleimide-terminated SAMs S18

2 References S28

S 2

1 Glycoarrays on gold

1.1 Gold chip functionalisation

1.1.1 Coupling of in situ deprotected thiols to unmodified gold surface Thiols were directly applied to the cleaned gold chip in PBS buffer (10 mM) and left for 1.5-12 h. Afterwards, the plate was thoroughly washed with ethanol, water and dichloromethane.

1.1.2 Coupling of in situ deprotected thiols to maleimide-terminated gold surface The respective thiol solution (10 mM in 10 mM PBS, 1 L) was applied spotwise to the maleimide- functionalised SAMs, left for 1-3 h and analysed by MALDI-ToF MS.

1.2 MALDI-ToF MS analysis of SAMs on gold All spectra were analysed with FlexAnalysis software (Bruker, USA) using default integration settings. Calibration was either performed before the analysis at the Ultraflex II instrument or afterwards in FlexAnalysis. Unless otherwise noted, all m/z values refer to the [M+Na]+ ion and the corresponding disulphide which is formed during ionisation.

S 3

1.2.1 SAM formation on gold surface using in situ deprotected thiols

Figure S1: MALDI-ToF mass spectrum of gold surface modified with the purified thiol mannoside 6-SH. The respective disulphide (m/z 866.940 [M+Na]+ and m/z 882.913 [M+K]+) was detected.

Figure S2: MALDI-ToF mass spectrum of gold surface modified with in situ deprotected mannoside 6-SH. The obtained spectrum is in analogy to the result obtained with the purified 6-SH. The respective disulphide (m/z 866.989 [M+Na]+ and m/z 882.970 [M+K]+) was detected.

S 4

Figure S3: MALDI-ToF mass spectrum of gold surface modified with in situ deprotected mannoside 5-SH. The respective disulphide (m/z 756.941 [M+Na]+) was detected.

S 5

Figure S4: MALDI-ToF mass spectrum of gold surface modified with in situ deprotected mannoside 11-SH. The respective disulphide (m/z 1015.040 [M+Na]+ and m/z 1031.022 [M+K]+) was detected.

Figure S5: MALDI-ToF mass spectrum of gold surface modified with in situ deprotected mannoside 10-SH. Thiol 10-SH (m/z 464.746 [M+Na]+ and m/z 480.732 [M+K]+) and the respective disulphide (m/z 904.996 [M+Na]+) were detected.

S 6

2.2.2 SAM formation using maleimide-terminated thiols SAMs

Figure S6: MALDI-ToF mass spectrum of gold surface modified with hydroxyl- and carboxy- terminal thiols. Analysis of the formed SAM on gold showed: A the mixed disulphide formed in situ from the hydroxyl- and carboxyl-terminal thiols (m/z 1051.220) and B the disulphide formed in situ from the carboxyl-terminal spacer (m/z 1241.240).

S 7

Figure S7: MALDI-ToF mass spectrum of gold surface modified with hydroxyl- and carboxy- terminal thiols (cf. Figure S20) after coupling of N-(2-aminoethyl)maleimide (Mal). Analysis of the formed SAM on gold showed: A the mixed disulphide formed in situ from the hydroxyl- and carboxyl-terminal thiols (m/z 1051.205) and B the mixed disulphide formed in situ from the maleimide and the hydroxyl-terminal molecules (m/z 1173.214).

S 8

2.2.3 Thiol coupling to maleimide-terminated SAMs

Figure S8: MALDI-ToF mass spectrum of gold surface modified with N-(2-aminoethyl)maleimide (Mal) (cf. Figure S21) after coupling of purified thiol 6-SH (positive control). Analysis of the formed SAM on gold showed: A the mixed disulphide formed in situ from the hydroxyl- and carboxyl-terminal thiols (m/z 1051), B the thiol 6-SH coupled to the maleimide (m/z 1178.127) and C the mixed disulphide formed in situ from the hydroxyl-terminal linker and thiol 6-SH coupled to the maleimide (m/z 1596.332).

Figure S9: MALDI-ToF mass spectrum of gold surface modified with N-(2-aminoethyl)maleimide (Mal) (cf. Figure S21) after coupling of Man-Trt 6 (negative control). Analysis of the formed SAM on gold showed no coupling of 6 and therefore only starting material was observed (cf. Figure S21).

S 9

A B

Figure S10: MALDI-ToF mass spectrum of gold surface modified with N-(2-aminoethyl)maleimide (Mal) (cf. Figure S21) after coupling of in situ deprotected thiol 6-SH. Analysis of the formed SAM on gold showed: A the mixed disulphide formed in situ from the hydroxyl- and carboxyl-terminal thiols (m/z 1051.545), B the thiol 6-SH coupled to the maleimide (m/z 1178) and C the mixed disulphide formed in situ from the hydroxyl-terminal linker and thiol 6-SH coupled to the maleimide (m/z 1596.470).

S 10

A C

Figure S11: MALDI-ToF mass spectrum of gold surface modified with N-(2-aminoethyl)maleimide (Mal) (cf. Figure S21) after coupling of in situ deprotected thiol 5-SH. Analysis of the formed SAM on gold showed: A the mixed disulphide formed in situ from the hydroxyl- and carboxyl-terminal thiols (m/z 1051.336), B the mixed disulphide formed in situ from the maleimide and the hydroxyl- terminal molecules (m/z 1173.350) and C the mixed disulphide formed in situ from the hydroxyl- terminal linker and thiol 5-SH coupled to the maleimide (m/z 1541.416).

S 11

A C

Figure S12: MALDI-ToF mass spectrum of gold surface modified with N-(2-aminoethyl)maleimide (Mal) (cf. Figure S21) after coupling of in situ deprotected thiol 11-SH. Analysis of the formed SAM on gold showed: A the mixed disulphide formed in situ from the hydroxyl- and carboxyl- terminal thiols (m/z 1051.534), B the mixed disulphide formed in situ from the maleimide and the hydroxyl-terminal molecules (m/z 1173.586) and C the mixed disulphide formed in situ from the hydroxyl-terminal linker and thiol 11-SH coupled to the maleimide (m/z 1670.823).

Figure S13: MALDI-ToF mass spectrum of gold surface modified with N-(2-aminoethyl)maleimide (Mal) (cf. Figure S21) after coupling of in situ deprotected thiol 10-SH. Analysis of the formed SAM on gold showed: A the mixed disulphide formed in situ from the hydroxyl- and carboxyl- terminal thiols (m/z 1051.401), B the thiol 10-SH coupled to the maleimide (m/z 1197) and C the mixed disulphide formed in situ from the hydroxyl-terminal linker and thiol 10-SH coupled to the maleimide (m/z 1615.521).

S 12

2 References [1] A. Schierholt, M. Hartmann, K. Schwekendiek and T. K. Lindhorst, Eur. J. Org. Chem., 2010, 3120-3128.

[2] M. J. Weissenborn, J. W. Wehner, C. J. Gray, R. Šardzík, C. E. Eyers, T. K. Lindhorst and S. L. Flitsch, Beilstein J. Org. Chem., 2012, 8, 753-762.

[3] J. W. Wehner and T. K. Lindhorst, Synthesis, 2010, 3070-3082.

[4] S. Kötter, U. Krallmann-Wenzel, S. Ehlers and T. K. Lindhorst, J. Chem. Soc. Perkin Trans. 1, 1998, 2193-2200.

S 13

CHAPTER ELEVEN

HIGH-THROUGHPUT SCREENING OF PROTEIN GLYCOSYLATION USING LECTIN-BINDING BIOPHOTONIC MICROARRAY IMAGING

M. J. Weissenborn, R. V. Olkhov, S. L. Flitsch, A. M. Shaw, High-throughput screening of protein glycosylation using lectin-binding biophotonic microarray imaging, Anal. Biochem., submitted.

The idea for this project came from Andrew M. Shaw and S. L. Flitsch to study lectin binding to monosaccharides by biophotonic microarray imaging. R. V. Olkhov and A. M. Shaw developed the biophotonic microarray imaging. M. J. Weissenborn developed the use of SAMs on gold nanoparticles, including the formation and functionalisation of SAMs, the synthesis and characterisation of the carbohydrates. The aminoethyl galactose ligand was kindly provided by R. Sardzik, for which he was acknowledged. R. V.Olkhov, A. M. Shaw and M. J. Weissenborn saw the feasibility of this system towards glycoproteins and S. L. Flitsch stressed the importance of its ﬁnding. R. V. Olkhov and A. M. Shaw developed the biophotonic scattering technique. R. V. Olkhov carried out the biophotonic experiments and A. M. Shaw, R. V. Olkhov and M. J. Weissenborn interpreted the results. All authors were involved in the writing of the manuscript.

31 High-Throughput Screening of Protein Glycosylation

Using Lectin-Binding Biophotonic Microarray Imaging

Martin J. Weissenborn,a Rouslan V. Olkhov,b Sabine L. Flitsch,a Andrew M. Shaw b*

a School of Chemistry & MIB, The University of Manchester, Manchester, M1 7DN, United Kingdom b College of Life and Environmental Sciences, University of Exeter, Exeter, EX4 4QD, United

Kingdom

* corresponding author [email protected]

Short Title: Lectin Screening

Abstract

Lectin binding with tethered monosaccharides has been studied using the particle plasmon light-scattering properties of gold nanoparticles printed into an array format. Aminoethyl- functionalised monosaccharides were printed onto self-assembled monolayers of hydroxyl- and carboxyl-groups. A detailed analysis of the binding of concanavalin A (ConA) and wheat germ agglutinin (WGA) to their target sugars indicate affinity constants in the order of KD ~10 nM for the presented monosaccharides. The detection limits for the lectins following a 200 seconds injection time were determined as 10 ng/ml or 0.23 nM and 100 ng/ml or 0.93 nM, respectively. An eight-lectin screen was performed on the glycoprotein pig fibrinogen (FBR). Four lectins showed specific binding with a spectrum of KD values similar in magnitude to the tethered-sugar-lectin interactions. The array technology has potential to perform a multi-lectin screen of large numbers of proteins in 200 seconds providing information on protein glycosylation and their micro-heterogeneity.

Keywords: protein glycosylation, posttranslational modification (PTM), lectin, microarray imaging

Introduction

The analysis of protein glycosylation has become an important facet of biopharmaceutical production as glycoproteins make up more than one third of approved biopharmaceuticals [1-3]. For glycoprotein hormones, such as human erythropoietin, glycosylation determines pharmacokinetic and pharmacodynamic profiles [4,5]. The plethora of therapeutic antibodies currently in the clinic and development need to be expressed with precise glycosylation patterns for correct therapeutic function

[6]. Given the dynamic character of glycosylation during and after protein translation in the cell, tight quality control of biopharmaceutical production is needed with fast, efficient and precise analytical techniques to monitor protein glycosylation.

Glycosylation is the most abundant post-translational modification (PTM) in nature, where glycans are attached to proteins or lipids [7]. The glycosylation is highly diverse and is therefore used as a unique pattern for cell-cell recognition [8]. Protein glycosylation is important for the refolding, solubility and stability of the proteins [9], and defects in glycosylation are observed in diseases such as allergies or muscular dystrophies [10,11]. The determination and detection of different glycosylation is non-trivial, laborious and requires a complex set of analytical techniques [12]. Amongst them are nuclear magnetic resonance [13], mass spectrometry [14] in combination with gas and liquid chromatography [15], and eastern or lectin blotting [16,17]. An important tool for characterising glycans is the specific binding of lectins, which are commercially available and can be highly selective for mostly terminal saccharides [18-21]. In combination with surface plasmon resonance (SPR) [22], lectins can provide useful quantitative data on protein glycosylation without the need for the more conventional fluorescent labelling of probes [21,23]. Using the SPR label-free technology in an array format allows the study of a large number of carbohydrate-protein interactions including determination of binding kinetics, affinity and avidity constants and relative energies of interactions.

We have developed an SPR related label free biophotonic sensor array technique [24] for the analysis of proteins from solution including antibodies from complex fluids [25]. The technique uses the sensitivity of the light scattering properties of the gold nanoparticle localised surface plasmons to provide intensity changes observed in real time with a video camera. The gold nanoparticle array spots

3 can be functionalised by self-assembling monolayers (SAMs) which carry a terminal carboxylic group

(linker) for chemical modification and a terminal hydroxyl group (spacer), respectively [26,27]. The carboxylic group can be used to couple various different chemically modified sugars on the surface

[28]. The hydroxyl group carrying molecule spacer allow spacer dilutions, which is important to understand the different affinities of lectins to high and low concentrated sugars on the surface [29,30].

Also larger molecules, such as glycoproteins, can be coupled into the SAMs and their glycosylation can be studied.

Here, we report the first analysis for the label-free biophotonic imaging detection of glycans on the glycoprotein fibrinogen. For proof-of-principle studies, the lectin binding specificities and its kinetics were studied using monosaccharide standards galactose (Gal), mannose (Man), glucose (Glc) and N-acetylglucosamine (GlcNAc), which were analysed with eight different lectins. The surface concentration of each sugar was varied to assess the effect on the association and dissociation kinetics and hence differential lectin affinity for the sugars. The optimised protocols were subsequently applied to the protein fibrinogen using bovine serum albumin as a negative control, where four different lectins showed binding affinity towards fibrinogen. This new technique enables the fast screen of protein glycosylation and could therefore become valuable in the quality control of glycosylated biopharmaceuticals.

Experimental Methods

Reagents

Self-assembling monolayer (SAM) components: HS-(CH2)17-(OC2H4)3-OH (used as a ‘spacer’) and HS-(CH2)17-(OC2H4)6-OCH2COOH (used as a ‘linker’), were obtained from ProChimia Surfaces

(Poland). N-(3-dimethylaminopropyl)-N'-ethylcarbodiimide hydrochloride (EDC), N- hydroxysuccinimide (NHS), bovine serum albumin (BSA), fibrinogen fraction I from pig plasma

(FBR), were obtained from Sigma-Aldrich. The lectins erythrina cristagalli (ECA), lens culinaris

(LcH), galanthus nivalis (GNA), arachis hypogaea (peanut) (PNA), sambucus nigra (SNA), and aleuria

4 aurantia (AAL) were kindly provided by Galab (Germany) as 1 mg/ml solutions in Tris buffer. The concanavalin A (ConA), and wheat germ (WGA) lectins were acquired from Vector labs as 5 mg/mL solution in HEPES buffer. All lectins were supplied FITC tagged. The standard running and dilution buffer was phosphate buffered saline (0.01 M phosphate buffer, 0.0027 M potassium chloride and 0.137

M sodium chloride, pH 7.4), containing 5×10-5 w/w Tween 20 surfactant, PBST. Aqueous 100 mM phosphoric acid solution, pH 1.9, was used as the regeneration buffer.

Aminoethyl glycoside derivatives of the monosaccharides -D-mannose (Man), -D-glucose

(Glc), -D-galactose (Gal), and N-acetyl--D-glucosamine (GlcNAc) were prepared by glycosylation methods reported elsewhere [28].

Sensor array preparation

The manufacture of the gold nanoparticle biophotonic arrays used in these experiments are described in detail elsewhere [24,31]. Briefly, the rectangular 12  8 sensor arrays are inkjet printed with a 300 µm pitch between the spots and a spot diameter of 200 µm, Fig. 1. Each spot is printed with seed nanoparticles 4 nm in diameter, removed from the printer and placed in growth solution to grow truncated polyhedral gold nanoparticles with an approximate diameter of 120 nm. These particles are present with a surface density of approximately 25% and are the plasmon light-scattering centres for assays. The arrays are then returned to the inkjet printer to prepare the assays. Analyte binding specificity is introduced by functionalisation of the gold surface with a particular monosaccharide or a protein. Before the functionalisation the sensor surface was activated by coating the gold nanoparticles with a mixed SAM of linker and spacer molecules, Fig. 1A, followed by activation of the linker carboxylic groups with the common EDC/NHS chemistry to produce succinimide esters reactive towards primary amino groups. Five mixtures with different ratios of linker to spacer SAM components were used: 1:4, 1:10, 1:100, 1:1000, and 1:2500.

Four monosaccharides, mannose (Man), galactose (Gal), glucose (Glc), and N- acetylglucosamine (GlcNAc), and two proteins, bovine serum albumin (BSA), and human fibrinogen

(FBR) were inkjet printed onto activated sensors according to the legend shown in Fig. 1B, incubated overnight at room temperature, rinse washed with deionised (DI) water, and stored at 4 °C until used.

Lectin binding kinetic assay

The sensor arrays are re-hydrated in PBST buffer for 15 minutes, washed with regeneration buffer, blocked with 2 mg/ml solution of BSA in PBST for 10 minutes, washed again with regeneration buffer and finally stabilised in PBST running buffer flow. An injection of doubly concentrated PBS buffer was performed to establish the sensitivity of the array spots to the change of an analyte bulk refractive index, these data were used to convert relative brightness response into a scale equivalent to refractive index change, conventionally this is given in response units, RU (RU≡10-6 refractive index units [32]). The sensitivity of each array spot is typically 8×10-5 RIU and the assays are averaged over

8 or 16 spots as required for multiplexed results.

A typical binding experiment procedure is shown in Fig. 2A: (i) the baseline plasmon scattering intensity is recorded in the flow of running PBST buffer; (ii) the lectin sample solution was injected over the surface for approximately 200 s to record the association binding phase with sufficient accuracy to determine the association rate constant, ka ; (iii) the flow was switched to running buffer and dissociation phase kinetics were recorded for approximately 10 minutes; and (iv) the sensor was regenerated with 100 mM phosphoric acid solution.

The regeneration step (iv) removes the adsorbed analyte material from the assay spot and allows the array to be re-used two or three times without loss of sensitivity or specificity. This allows assays to be repeated several times on the same sensor array without noticeable degradation in performance. The assay is repeated with several concentrations of the target binding lectin and the kinetic data subjected to a global fit for all concentrations, both association and dissociation simultaneously for all concentration, from which the rate constants, ka and kd can then be determined.

The variation of the rates with time for a simple 1:1 binding model is performed and deviations from this model may provide information on more complex interaction such as co-operative binding and

6 avidity. Affinity and avidity constants can then be determined for all of the interactions. An example of the procedure is shown in Fig. 2B with 2 µg/ml of ConA which has a strong binding of the lectin to mannose and FBR functionalised sensor surfaces. The rest of the monosaccharide surfaces—Gal, Glc, and GlcNAc—do not show any affinity towards the ConA analyte. Qualitative comparison of the responses in Man and FBR channels revealed that although the rate of association with fibrinogen is slower the dissociation is also markedly slower than in Man the channel.

Results

The Man and GlcNAc saccharides are known to bind selectively to ConA and WGA [33], respectively. The specific binding interactions between the modified sugars printed onto the array with the lectins ConA and WGA were studied over a range of lectin concentrations. The concentration- dependence series of kinetic binding responses were globally fitted (simultaneously for all concentrations in association and dissociation) analysed with a 1:1 interaction model resulting in association and dissociation rate constants listed in Fig. 3 and Table 1. The thermodynamic equilibrium dissociation constants KD may then be derived from the fitted rate constants: KD = kd/ka. The sensograms with lectin concentrations above 10 µg/ml significantly deviate from the exponential functional form expected for a simple 1:1 binding model and were excluded from the global fit.

The strongest specific responses were observed for lectin-monosaccharide pairs of ConA-Man and WGA-GlcNAc shown in Fig. 3 with their corresponding global fit analyses. For the lower concentrations of both lectins, the 1:1 binding model shows a good fit. At higher ConA concentrations,

Fig. 3A, the fit to the exponential function remains reasonable whereas there are significant deviations from the single exponential functional form for the WGA-GlcNAc pair, Fig. 3B, indicating a departure from the 1:1 binding model. The kinetic response for 48 µg/ml WGA appears to consist of faster and slower processes, the latter can be readily characterised if double exponential function was used to fit the data, resulting in the kinetic and thermodynamic parameters of the slower process ka = (1.1±0.2) ×

4 -1 -1 -3 -1 -1 10 s M , kd = (3.7±0.1) × 10 s , KD = 346 ± 55 nM, G = 36.4 ± 0.5 kJ M .

The sensitivity of the assays may be assessed for the 200 s injection time indicating a detection limit of 10 ng/ml or 0.23 nM for WGA and 100 ng/ml or 0.93 nM for ConA. These values compare favourably with other techniques, for example the bulk phase mannose glycol-conjugated nanoparticle based absorption and scattering measurements were shown to have ConA limits of detection 2-3 nM

[34].

The deviations from the 1:1 model indicated a more complex avidity interaction at high concentrations of lectin so we have performed a series of experiments to determine the variation of the kinetic and thermodynamic parameters for the interactions between the lectins and different densities of sugars on the surface. The sugar surface density was varied by varying the SAM composition by changing the linker:spacer ratio, Fig. 4. This reduced the surface sugar density and the potential for multiple lectin-sugar interactions. The variation of the association and dissociation rate constants for the interactions is shown in Fig. 4 and summarised in Table 1.

We observed significant binding to FBR compared to the BSA control channel during the experiments since FBR is N-glycosylated [35]. The data showed interactions of FBR with the lectins

ConA and WGA, respectively. FBR was coupled to the SAMs in a 1000:1 linker:spacer dilution and was subsequently interrogated with six different lectins, Fig. 5. The monosaccharides Glc, GlcNAc,

Man and Gal were used as controls. The strong specific binding to FBR was observed for WGA and

ConA, in addition specific binding was observed for SNA and LcH—which interact with sialic acid and a fucosylated core regions, respectively, Fig. 5 and Fig. 6.

Discussion

The aim of the current work was to assess the potential of an array-based, label-free particle plasmon screening technology for the rapid assessment of the posttranslational modification of proteins, specifically glycosylation. The monosaccharide standards Glc, GlcNAc, Gal and Man aminoethyl glycosides were prepared as published elsewhere [28]. The aminoethyl glycosides were coupled to the carboxyl groups in SAMs allowing the surface sugar concentration to be varied by varying the linker-spacer ratio. Lower sugar surface densities are expected to result in a smaller number

8 of binding sites on the surface reflecting simple 1:1 lectin-sugar binding kinetics. The kinetic parameters in Table 1 indicate a trend to a consistent determination and reproducibility especially in the dissociation rate constant, kd and hence the determination in KD and the energy of interaction, Fig.

4A. The number of lectins bound to the surface, however, derived from the fitted surface coverage,m in Fig. 4B, showed variations of only 20%. The theoretical separation between any linker-spacer in the

SAM is 0.45 nm [36,37] with the separations in a rectangular matrix changing with increasing dilution:

0.9, 1.4, 4.5, 14.2, and 22.5 nm for 1:4, 1:10, 1:100, 1:1000, 1:2500 respectively. The distance between saccharide binding pockets of ConA is about 7.0±0.5 nm [38], which is two-three fold less than ideal inter-linker distance. Although this may suggest that ConA would bind only one site for the linker:spacer ratios 1:1000 and 1:2500, the realistic SAM with randomly distributed linker molecules shall offer a sample amount of closer spaced sugars and two-site binding is reasonable. Moreover there is the fluidity within the SAM and a lectin may recruit additional binding sugars once bound to the initial site on the surface. The trend with monovalent binding is consistent with the trends observed in

Fig. 4 indicating binding affinity constants and energies for the sugar-lectin interactions as shown in

Table 2.

The kinetic parameters derived in this study reflect the binding of the lectin specifically to the tethered sugar moieties and non-specifically to the glycol spacer molecule which dominates the SAM composition. It has been observed previously that the measured interactions between lectins and sugars depends on whether the sugar or the lectin is immobilised on the sensor surface and on whether the sugars are monomers or oligomers [39-43]. The derived ConA-Man interaction parameters listed in

Table 1 indicate KD in low nanomolar range consistent with the expected values for the sugar surface

[39]. In addition, aggregation of the lectins has been observed at higher concentration observing multi- lectin binding to the tethered-sugar surfaces [39] which is consistent with the Fig. 3B for WGA binding to GlcNAc. The relevant dissociation constants KD for tethered-lectin-(mannose containing saccharide) binding reported in thermodynamic studies are in the range of 50-770 M for mono and disaccharides

[44-46], and up to 750 nM for more complex longer and branched oligosaccharides [45].

The lectins specificity, Fig. 5, indicates binding to the expected tethered-sugars with some notable exceptions and binding of a series of lectins to glycoprotein, pig FBR, Table 2 and Table 3. All interactions, whether for the tethered sugar presented in the SAM surface or the sugar presented by the printed FBR protein, showed KD values of ~10 nM, Fig. 6. Similarly, large concentration effects were observed for WGA-FBR binding, Fig. 6B, consistent with lectin clustering. The lectin binding screening for all six lectins may be summarised in the KD matrix shown in Table 3. Only three of the lectins bound to their expected tethered-sugar moieties: ConA-Man, WGA-GlcNAc and ECA-GlcNAc but four lectins showed specific binding to FBR. LcH binds to FBR indicating the presence of a fucose core structure on the surface of the protein in the correct presentation for the sugar to interact with the binding site of the lectin[47]. Neither AAL nor LcH binding to the Gal tethered to the surface nor does

LcH bind to the Man tethered to the surface. The nature of the interactions between the lectins and the presentation of the sugars both on the surface and as presented on the glycoprotein FBR is not the same and indicative of the glycosylation pattern on FBR in some way [35]. The lectins WGA and ConA require terminal GlcNAc and Man monosaccharides, respectively, for binding but binding may also occur to specific patterns of oligosaccharides. The terminal monosaccharides can always occur due to the microheterogeneity in the glycosylation [48]. Furthermore, the lectin SNA showed specific binding to sialic acid moieties on surface proteins including FBR [49,50].

Conclusions

Our biophotonic scattering array screening technique for interactions with tethered sugars with a series of eight lectins producing a set of KD with values ~10 nM with the expected specificity. The presentation of tethered sugars within the SAMs may in part explain the lack of binding of lectins to known sugar targets. A lectin screen of the tethered protein pig FBR shows a pattern of lectin binding with KD values showing significant differences from the single sugar binding. The arrays have potential to be a new high-throughput screening technology to determine pattern of lectin binding for protein glycosylation and potentially from a detailed analysis of the KD values, a pattern of the protein and potentially glycosylation micro heterogeneity. The new array-based high-throughput technology has

10 the potential to screen glycosylation of proteins rapidly and could therefore become a valuable tool in pharmaceutical research and industry.

Acknowledgements

The authors like to thank Dr Robert Sardzik for providing aminoethyl galactose. This work was supported by the Royal Society (Wolfson Award to SLF), the European Commission’s Marie Curie program which funded the EuroGlycoArrays ITN (MJW) and the EPSRC.

References

[1] J.O.B. Hourihane, Peanut Allergy. Pediatr. Clin. North Am. 58 (2011) 445-458.

[2] R. Jefferis, Recombinant antibody therapeutics: the impact of glycosylation on mechanisms of action. Trends Pharmacol. Sci. 30 (2009) 356-362.

[3] V. Kayser, N. Chennamsetty, V. Voynov, K. Forrer, B. Helk, and B.L. Trout, Glycosylation influences on the aggregation propensity of therapeutic monoclonal antibodies. Biotechnol. J 6 (2011)

38-44.

[4] L.O. Narhi, T. Arakawa, K.H. Aoki, R. Elmore, M.F. Rohde, T. Boone, and T.W. Strickland, The

Effect of Carbohydrate on the Structure and Stability of Erythropoietin. J. Biol. Chem. 266 (1991)

23022-23026.

[5] H.J. Li, and M. d'Anjou, Pharmacological significance of glycosylation in therapeutic proteins.

Curr. Opin. Biotechnol. 20 (2009) 678-684.

[6] R. Jefferis, and M.P. Lefranc, Human immunoglobulin allotypes Possible implications for immunogenicity. Mabs 1 (2009) 332-338.

[7] I. Shin, S. Park, and M.R. Lee, Carbohydrate microarrays: An advanced technology for functional studies of glycans. Chem. Eur. J. 11 (2005) 2894-2901.

[8] M. Hartmann, and T.K. Lindhorst, The Bacterial Lectin FimH, a Target for Drug Discovery -

Carbohydrate Inhibitors of Type 1 Fimbriae-Mediated Bacterial Adhesion. Eur. J. Org. Chem. (2011)

3583-3609.

[9] C.T. Walsh, S. Garneau-Tsodikova, and G.J. Gatto, Protein posttranslational modifications: The chemistry of proteome diversifications. Angew. Chem. Int. Ed. 44 (2005) 7342-7372.

[10] T. Yoshida-Moriguchi, L.P. Yu, S.H. Stalnaker, S. Davis, S. Kunz, M. Madson, M.B.A. Oldstone,

H. Schachter, L. Wells, and K.P. Campbell, O-Mannosyl Phosphorylation of Alpha-Dystroglycan Is

Required for Laminin Binding. Science 327 (2010) 88-92.

[11] J.N. Arnold, M.R. Wormald, R.B. Sim, P.M. Rudd, and R.A. Dwek, The impact of glycosylation on the biological function and structure of human immunoglobulins, Annu. Rev. Immunol., Annual

Reviews, Palo Alto, 2007, pp. 21-50.

[12] Y. Wada, P. Azadi, C.E. Costello, A. Dell, R.A. Dwek, H. Geyer, R. Geyer, K. Kakehi, N.G.

Karlsson, K. Kato, N. Kawasaki, K.H. Khoo, S. Kim, A. Kondo, E. Lattova, Y. Mechref, E. Miyoshi,

K. Nakamura, H. Narimatsu, M.V. Novotny, N.H. Packer, H. Perreault, J. Peter-Katalinic, G. Pohlentz,

V.N. Reinhold, P.M. Rudd, A. Suzuki, and N. Taniguchi, Comparison of the methods for profiling glycoprotein glycans - HUPO Human Disease Glycomics/Proteome Initiative multi-institutional study.

Glycobiology 17 (2007) 411-422.

[13] V. Slynko, M. Schubert, S. Numao, M. Kowarik, M. Aebi, and F.H.T. Allain, NMR Structure

Determination of a Segmentally Labeled Glycoprotein Using in Vitro Glycosylation. J. Am. Chem.

Soc. 131 (2009) 1274-1281.

[14] W. Morelle, Analysis of Glycosylation and Other Post-Translational Modifications by Mass

Spectrometry. Curr. Anal. Chem. 5 (2009) 144-165.

[15] A. Antonopoulos, S.J. North, S.M. Haslam, and A. Dell, Glycosylation of mouse and human immune cells: insights emerging from N-glycomics analyses. Biochem Soc T 39 (2011) 1334-1340.

[16] H. Tanaka, N. Fukuda, and Y. Shoyama, Eastern blotting and immunoaffinity concentration using monoclonal antibody for ginseng saponins in the field of traditional Chinese medicines. J. Agric. Food

Chem. 55 (2007) 3783-3787.

[17] S. Oguri, Analysis of sugar chain-binding specificity of tomato lectin using lectin blot: recognition of high mannose-type N-glycans produced by plants and yeast. Glycoconjugate J. 22 (2005) 453-461.

[18] M.J. Weissenborn, R. Castangia, J.W. Wehner, R. Sardzik, T.K. Lindhorst, and S.L. Flitsch, Oxo- ester mediated native chemical ligation on microarrays: an efficient and chemoselective coupling methodology. Chem. Commun. 48 (2012) 4444-4446.

[19] H. Lis, and N. Sharon, Lectins: Carbohydrate-specific proteins that mediate cellular recognition.

Chem. Rev. 98 (1998) 637-674.

[20] R. Šardzík, A.P. Green, N. Laurent, P. Both, C. Fontana, J. Voglmeir, M.J. Weissenborn, R.

Haddoub, P. Grassi, S.M. Haslam, G. Widmalm, and S.L. Flitsch, Chemoenzymatic Synthesis of O-

Mannosylpeptides in Solution and on Solid Phase. J. Am. Chem. Soc. 134 (2012) 4521-4524.

[21] O. Blixt, S. Head, T. Mondala, C. Scanlan, M.E. Huflejt, R. Alvarez, M.C. Bryan, F. Fazio, D.

Calarese, J. Stevens, N. Razi, D.J. Stevens, J.J. Skehel, I. van Die, D.R. Burton, I.A. Wilson, R.

Cummings, N. Bovin, C.H. Wong, and J.C. Paulson, Printed covalent glycan array for ligand profiling of diverse glycan binding proteins. Proc. Natl. Acad. Sci. U. S. A. 101 (2004) 17033-17038.

[22] R. Karamanska, J. Clarke, O. Blixt, J.I. MacRae, J.Q.Q. Zhang, P.R. Crocker, N. Laurent, A.

Wright, S.L. Flitsch, D.A. Russell, and R.A. Field, Surface plasmon resonance imaging for real-time, label-free analysis of protein interactions with carbohydrate microarrays. Glycoconjugate J. 25 (2008)

69-74.

[23] J. Voglmeir, R. Sardzik, M.J. Weissenborn, and S.L. Flitsch, Enzymatic Glycosylations on Arrays.

Omics 14 (2010) 437-444.

[24] R.V. Olkhov, and A.M. Shaw, Label-free antibody–antigen binding detection by optical sensor array based on surface-synthesized gold nanoparticles. Biosens. Bioelectron. 23 (2008) 1298-1302.

[25] B. Jansen van Vuuren, T. Read, R.V. Olkhov, and A.M. Shaw, Human serum albumin interference on plasmon-based immunokinetic assay for antibody screening in model blood sera. Anal. Biochem.

405 (2010) 114-120.

[26] E. Ostuni, R.G. Chapman, R.E. Holmlin, S. Takayama, and G.M. Whitesides, A survey of structure-property relationships of surfaces that resist the adsorption of protein. Langmuir 17 (2001)

5605-5620.

[27] N. Laurent, R. Haddoub, J. Voglmeir, S.C. Wong, S.J. Gaskell, and S.L. Flitsch, SPOT synthesis of peptide arrays on self-assembled monolayers and their evaluation as enzyme substrates.

ChemBioChem 9 (2008) 2592-2596.

[28] R. Sardzik, G.T. Noble, M.J. Weissenborn, A. Martin, S.J. Webb, and S.L. Flitsch, Preparation of aminoethyl glycosides for glycoconjugation. Beilstein journal of organic chemistry 6 (2010) 699-703.

[29] L.L. Kiessling, and R.A. Splain, Chemical approaches to glycobiology. Annu. Rev. Biochem 79

(2010) 619-653.

[30] J.E. Gestwicki, C.W. Cairo, L.E. Strong, K.A. Oetjen, and L.L. Kiessling, Influencing receptor- ligand binding mechanisms with multivalent ligand architecture. J. Am. Chem. Soc. 124 (2002) 14922-

14933.

[31] R.V. Olkhov, J.D. Fowke, and A.M. Shaw, Whole serum BSA antibody screening using a label- free biophotonic nanoparticle array. Anal. Biochem. 385 (2009) 234-241.

[32] Biacore User Training Documentation, GE Healthcare.

[33] G. Bains, R.T. Lee, Y.C. Lee, and E. Freire, Microcalorimetric study of wheat germ agglutinin binding to N-acetylglucosamine and its oligomers. Biochemistry (Mosc). 31 (1992) 12624-12628.

[34] S. Watanabe, S. Yamamoto, K. Yoshida, K. Shinkawa, D. Kumagawa, and H. Seguchi, Surface plasmon resonance scattering and absorption sensing of Concanavalin A using glycoconjugated gold nanoparticles. Supramol. Chem. 23 (2011) 297-303.

[35] J.M. Kollman, L. Pandi, M.R. Sawaya, M. Riley, and R.F. Doolittle, Crystal Structure of Human

Fibrinogen. Biochemistry 48 (2009) 3877-3886.

[36] J.P. Bareman, and M.L. Klein, Collective Tilt Behavior in Dense, Substrate-Supported Monolayers of Long-Chain Molecules: a Molecular-Dynamics Study. J. Phys. Chem. 94 (1990) 5202-5205.

[37] J.C. Love, L.A. Estroff, J.K. Kriebel, R.G. Nuzzo, and G.M. Whitesides, Self-assembled monolayers of thiolates on metals as a form of nanotechnology. Chem. Rev. 105 (2005) 1103-1169.

[38] D.A.R. Sanders, D.N. Moothoo, J. Raftery, A.J. Howard, J.R. Helliwell, and J.H. Naismith, The

1.2 Г… resolution structure of the con A-dimannose complex. J. Mol. Biol. 310 (2001) 875-884.

[39] E. Duverger, N. Frison, A.-C. Roche, and M. Monsigny, Carbohydrate-lectin interactions assessed by surface plasmon resonance. Biochimie 85 (2003) 167-179.

[40] Y. Shinohara, H. Sota, F. Kim, M. Shimizu, M. Gotoh, M. Tosu, and Y. Hasegawa, Use of a

Biosensor Based on Surface Plasmon Resonance and Biotinyl Glycans for Analysis of Sugar Binding

Specificities of Lectins. J. Biochem. 117 (1995) 1076-1082.

[41] M.I. Khan, M.V. Sastry, and A. Surolia, Thermodynamic and kinetic analysis of carbohydrate binding to the basic lectin from winged bean (Psophocarpus tetragonolobus). J. Biol. Chem. 261 (1986)

3013-3019.

[42] Y. Shinohara, F. Kim, M. Shimizu, M. Goto, M. Tosu, and Y. Hasegawa, Kinetic measurement of the interaction between an oligosaccharide and lectins by a biosensor based on surface plasmon resonance. Eur. J. Biochem. 223 (1994) 189-194.

[43] Z. Dai, A.-N. Kawde, Y. Xiang, J.T. La Belle, J. Gerlach, V.P. Bhavanandan, L. Joshi, and J.

Wang, Nanoparticle-Based Sensing of Glycan-Lectin Interactions. J. Am. Chem. Soc. 128 (2006)

10018-10019.

[44] B.T. Houseman, and M. Mrksich, Carbohydrate Arrays for the Evaluation of Protein Binding and

Enzymatic Modification. Chemistry & Biology 9 (2002) 443-454.

[45] D.K. Mandal, N. Kishore, and C.F. Brewer, Thermodynamics of Lectin-Carbohydrate Interactions.

Titration Microcalorimetry Measurements of the Binding of N-Linked Carbohydrates and Ovalbumin to Concanavalin A. Biochemistry (Mosc). 33 (1994) 1149-1156.

[46] F.P. Schwarz, K.D. Puri, R.G. Bhat, and A. Surolia, Thermodynamics of monosaccharide binding to concanavalin A, pea (Pisum sativum) lectin, and lentil (Lens culinaris) lectin. J. Biol. Chem. 268

(1993) 7668-7677.

[47] S. Serna, S. Yan, M. Martin-Lomas, I.B.H. Wilson, and N.C. Reichardt, Fucosyltransferases as

Synthetic Tools: Glycan Array Based Substrate Selection and Core Fucosylation of Synthetic N-

Glycans. J. Am. Chem. Soc. 133 (2011) 16495-16502.

[48] B.J. Harmon, X. Gu, and D.I.C. Wang, Rapid Monitoring of Site-Specific Glycosylation

Microheterogeneity of Recombinant Human Interferon-γ. Anal. Chem. 68 (1996) 1465-1473.

[49] T. Pacchiarotta, P.J. Hensbergen, M. Wuhrer, C. van Nieuwkoop, E. Nevedomskaya, R.J.E. Derks,

B. Schoenmaker, C.A.M. Koeleman, J. van Dissel, A.M. Deelder, and O.A. Mayboroda, Fibrinogen alpha chain O-glycopeptides as possible markers of urinary tract infection. J. Proteomics 75 (2012)

1067-1073.

[50] P.G. Wu, K.B. Lee, Y.C. Lee, and L. Brand, Solution conformations of a biantennary glycopeptide and a series of its exoglycosidase products from sequential trimming of sugar residues. J. Biol. Chem.

271 (1996) 1470-1474.

Figures

A B

Fig. 1. Sensor preparation details: (A) monosaccharides used to functionalise sensor arrays by EDC/NHS coupling to carboxylic groups present in self-assembling monolayers (SAMs) (left). The hydroxyl and carboxy terminal molecules form the SAMs on gold nanoparticles shown in SEM image (right); (B) Array print legend.

i ii iii iv

A a B a 0.9 1.8 b

brightness)

 b c

0.0 0.9

sensor (kRU)sensor response c sensor ( sensor response -0.9 0.0

0 10 20 0 5 10 time (min) time (min)

Fig. 2. Example of the experimental sensogram data recorded in an assay of 2 µg/mL ConA lectin sample: (A) averaged scattering intensity changes recorded from 8 or 16 arrays spots functionalised with the same material showing the experimental phases (i) baseline, (ii) association, (iii) dissociation, (iv) regeneration; (B) Data from panel A is referenced against the BSA control channels to compensate for the differences in refractive indices of running buffer and lectin samples. The control channel further corrects for temperature variations and light source intensity instabilities. The signal scale is converted into response units and injection time lag removed. The marked traces on both panels are: a – Man, b – FBR, and overlapping c – Gal, Glc, GlcNAc, and BSA (panel A only).

A B a

1.6 2

0.8 1 b

response (kRU)response c response (kRU)response b d c d e 0.0 e 0

0 5 10 0 5 10 time (min) time (min)

Fig. 3. Global fit for all concentrations for the specific binding of the lectins to the surface-tethered sugars assuming a

1:1 interaction model: (A) ConA-Man binding with ConA concentrations (a) 24, (b) 2.3, (c) 1.2, (d) 0.57, (e) 0.25 µg/ml;

(B) WGA-GlcNAc, with WGA concentrations (a) 48, (b) 1.2, (c) 0.62, (d) 0.30, (e) 0.15 µg/ml; kinetic trace (a) has a weighting of 10-5 in the global fit, solid curve, while the dotted curve corresponds to the individual fit of this trace to double exponential functions.

A B2.5 10 10 2.0

)

-1

-1 × 1.5

(kRU)

a d 

-1

× 5 5

) -4 1.0

fittted

0.5

0 0 0.0 1E-3 0.01 0.1 1E-3 0.01 0.1 linker:spacer ratio linker:spacer ratio

Fig. 4. Dependence of kinetic parameters derived from the series of ConA-Man sensograms on the linker:spacer ratio:

(A) association, filled squares, and dissociation, open circles, rate constants; (B) fitted values of maximum analyte surface coverage. It appears that smaller numbers of Man binding sites on the sensor surface leads to stronger specific binding of lectin molecules from the analyte solution to the sensor surface.

1.0 1.0 1.0 A B C Man 0.8 * 0.8 0.8 ** FBR 0.6 0.6 0.6 Gal Glc 0.4 0.4 0.4 GlcNAc 0.2 0.2 0.2

response(kRU)

0.0 0.0 0.0

-0.2 -0.2 -0.2

0 5 10 15 0 5 10 15 0 5 10 15

1.0 D * 1.0 E 1.0 F 0.8 0.8 0.8

0.6 0.6 0.6

0.4 0.4 0.4

0.2 0.2 0.2

response(kRU)

0.0 0.0 0.0

-0.2 -0.2 -0.2

0 5 10 15 0 5 10 15 0 5 10 15 time (min) time (min) time (min)

Fig. 5. Binding assays of SNA (A), GNA (B), ECA (C), LcH (D), PNA (E), and AIL (F) lectins performed on sensors coated with 1:1000 linker:spacer SAMs. Each lectin sample was injected for ca. 10 minutes at 9.6 µg/ml concentration.

The SNA and LcH lectins showed a clear affinity towards the FBR surface (traces marked with * on panels A and D).

ECA shows some affinity towards GlcNAc functionalised surfaces (marked ** on panel C).

A a B 1.0 1.0

0.5 0.5

response (kRU)response

response (kRU)response b c b

0.0 d 0.0 c

0 5 10 0 5 10 time (min) time (min)

Fig. 6. Kinetic analysis of lectin-FBR interaction. Response data were obtained on sensor arrays functionalised with

SAMs of 1:1000 linker:spacer ratio. (A) ConA-FBR, lectin analyte concentrations were (a) 24, (b) 2.3, (c) 1.2, (d) 0.57

g/ml. (B) WGA-FBR, lectin analyte concentrations were (a) 48, (b) 1.2, (c) 0.62 g/ml.

Table 1. ConA-Man interaction rate constants observed on sensor surfaces with varying SAM linker:spacer ratios.

-4 -1 -1 4 -1 -1 linker :spacer 10 × ka (M s ) 10 × kd (s ) KD (nM) -G (kJ M )

1:4 4±3 10±0.3 24±20 42.7±4.6

1:10 7.3±0.2 7.4±0.3 10±0.4 45.2±0.5

1:100 7.0±0.4 9.9±0.8 14±1.5 44.4±0.5

1:1000 9.7±0.1 3.2±0.1 3.3±0.1 47.7±0.5

1:2500 11.8±0.2 3.3±0.1 2.8±0.1 48.1±0.5

Table 2. Lectin-ligand interaction rate constants observed on sensor surface with linker:spacer ratio 1:1000.

-4 -1 -1 4 -1 -1 lectin-ligand 10 × ka (M s ) 10 × kd (s ) KD (nM) -G (kJ M )

ConA-FBR 4.9±0.1 0.7±0.1 1.4±0.2 49.8±0.5

WGA-FBR 2.9±0.1 8.2±0.2 28±1 42.7±0.5

SNA-FBR 1.7±0.1 0.91±0.10 5.4±0.7 46.8±0.5

LcH-FBR 1.4±0.1 1.55±0.1 11±1 45.1±0.5

WGA-GlcNAc 78.2±0.5 3.2±0.1 4.1±0.4 47.3±0.5

ECA-GlcNAc 1.6±0.2 13.4±0.5 82.7±11 40.1±0.6

Table 3. KD matrix of sugar-lectin interactions.

AAL SNA PNA LcH WGA ConA GNA ECA

PFBR 5.4±0.7 11±1 28±1 1.4±0.2

BSA

Glc

GlcNAc ? 4.1±0.4 82.7±11

Gal ? ?

Man ? ? 3.3±0.1 ?

? indicates expected binding but not observed

CHAPTER TWELVE

CRYSTAL STRUCTURE OF A SOLUBLE FORM OF HUMAN CD73 WITH ECTO-5’-NUCLEOTIDASE ACTIVITY

Reproduced by permission of John Wiley and Sons

D. P. H. M. Heuts, M. J. Weissenborn, R. V. Olkhov, A. M. Shaw, C. Levy, N. S. Scrutton, Crystal Structure of a Soluble Form of Human CD73 with Ecto-5’-Nucleotidase Activity, Anal. Biochem., 2012, 13, 2384. DOI: 10.1002/cbic.201200426

The protein expression and analysis was done by D. P. H. M. Heuts. The crystal structure was solved by C. Levy. M. J. Weissenborn had the idea to analyse the protein glycosylation by biophotonic scattering and the experiments were carried out by R. V. Olkhov and A. M. Shaw. All authors contributed to the writing of the manuscript.

32 DOI: 10.1002/cbic.201200426 Crystal Structure of a Soluble Form of Human CD73 with Ecto-5’-Nucleotidase Activity Dominic P. H. M. Heuts,[a] Martin J. Weissenborn,[a] Rouslan V. Olkhov,[b] Andrew M. Shaw,[b] Jennet Gummadova,[a] Colin Levy,[a] and Nigel S. Scrutton*[a]

CD73 is a dimeric ecto-5’-nucleotidase that is expressed on the tein. The crystal structure reveals a conserved loop that is di- exterior side of the plasma membrane. CD73 has important rectly involved in the dimer-dimer interaction showing that the regulatory functions in the extracellular metabolism of certain two subunits of the dimer are not linked by disulfide bridges. nucleoside monophosphates, in particular adenosine mono- Using biophotonic microarray imaging we were able to con- phosphate, and has been linked to a number of pathological firm glycosylation of the enzyme and show that the enzyme is conditions such as cancer and myocardial ischaemia. Here, we decorated with a variety of oligosaccharide structures. The present the crystal structure of a soluble form of human solu- crystal structure of sCD73 will aid the design of inhibitors or ble CD73 (sCD73) at 2.2 resolution, a truncated form of CD73 activator molecules for the treatment of several diseases and that retains ecto-5’-nucleotidase activity. With this structure we prove useful in explaining the possible roles of single nucleo- obtained insight into the dimerisation of CD73, active site ar- tide polymorphisms in physiology and disease. chitecture, and a sense of secondary modifications of the pro-

Introduction

Human ecto-5’-nucleotidase (CD73), encoded by the NT5E which would be assisted by the availability of a detailed crystal gene, is a 71 kD dimeric enzyme that is expressed on the exte- structure of the CD73 dimer.[6] rior face of the plasma membrane. CD73 acts by hydrolysing The mature form of CD73 is glycosylated and attached to nucleotide monophosphates, thereby releasing inorganic phos- the plasma membrane by a C-terminal glycosyl phosphatidyl- phate and the corresponding nucleoside. Previous enzymatic inositol (GPI) anchor. CD73 belongs to a family of 5’-nucleoti- studies indicate that purines, and in particular 5’-adenosine dases, which includes 2’,3’-cyclic phosphodiesterases and apyr- monophosphate (AMP), are the preferred substrates of CD73.[1] ases. This family is part of a distantly related superfamily of CD73 is involved in two important extracellular metabolic metallophosphoesterases.[7] Sequence-related 5’-nucleotidases pathways, the adenosine nucleotide and nicotinamide adenine all share a similar structural fold and consist of two distinct do- dinucleotide (NAD +) metabolism. Extracellular NAD + is degrad- mains: an N-terminal domain, which binds two divalent metal ed by the action of enzymes such as NAD +-glycohydrolases, ions important for catalysis, and a C-terminal domain, responsi- ADP-ribosyltransferases and pyrophosphatases.[2] The pyro- ble for binding the nucleotide substrate. The active site is cre- phosphatases cleave NAD+ to yield AMP and b-nicotinamide ated at the interface between the two domains. Substrate- mononucleotide (NMN), thereby providing substrates for bound and substrate-free structures of 5’-nucleotidases indi- CD73. The adenosine released by CD73-catalysed AMP hydroly- cate that a large domain motion, whereby the two domains sis can be used as a substrate for intracellular purine nucleo- rotate relative to each other, is required for catalysis.[8] Based tide production. Alternatively, adenosine can interact with ade- on the presence of conserved catalytic residues, sequence nosine receptors (P1 receptors) through which it is involved in analysis has suggested that the active-site structure of human a plethora of regulatory functions. As a signalling molecule ad- CD73 should resemble that of its bacterial counterparts.[7] Until enosine has a central role in modulating homeostasis in condi- now, no detailed structural information of a CD73 from a verte- tions of ischaemia, hypoxia, inflammation and trauma.[3] Mouse brate has been available. Here, we present the crystal structure models show that myocardial infarction due to ischaemia is exacerbated when ecto-5’-nucleotidase activity is chemically in- [a] Dr. D. P. H. M. Heuts, M. J. Weissenborn, Dr. J. Gummadova, Dr. C. Levy, hibited or genetically deleted. Owing to the multitude of roles Prof. Dr. N. S. Scrutton of adenosine, CD73 is a potential target in several pathological Manchester Institute of Biotechnology, University of Manchester conditions such as myocardial ischaemia and cancer.[4] Recent- 131 Princess Street, Manchester M1 7DN (UK) ly, it has been shown that CD73-catalysed adenosine produc- E-mail: [email protected] tion promotes tumour growth and that CD73 inhibition may [b] Dr. R. V. Olkhov, Dr. A. M. Shaw [5] College of Life and Environmental Sciences, University of Exeter be a route to cancer treatment. The design of tissue-selective Exeter EX4 4QD (UK) CD73 inhibitors is a major focus of current research efforts, Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/cbic.201200426.

2384 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2012, 13, 2384 – 2391 Crystal Structure of a Soluble Human CD73 of a soluble form (sCD73) of human CD73, which lacks the GPI anchor, at 2.2 resolution. The structure reveals how the CD73 dimer is formed and provides insight into active-site architecture that will ultimately assist structure-based design of small- molecule inhibitors (or activators) of human CD73.

Results and Discussion Biochemical analyses The truncated soluble form of CD73 (sCD73; residues 27–549), which lacks the N-terminal signal sequence and the C-terminal

GPI anchor signal sequence, was extended with a His6 tag and was codon-optimised for expression in human cells. From 1 L of culture medium, typically 2–5 mg sCD73 (n=3 purifications) was purified by successively using NiII-affinity chromatography Figure 1. Relative activity of sCD73 before and after reconstitution with vari- and size-exclusion chromatography. SDS-PAGE analysis showed ous divalent metal ions. Native sCD73 was extensively dialyzed against EDTA (2 mm) to remove active-site metal ions. After a second round of dialysis that sCD73 runs at ~65 kD which corresponds approximately against buffer, apo-sCD73 was incubated with various divalent metals À1 to the calculated mass of 58 932 gmol . CD73 contains four (0.1 mm). potential N-glycosylation sites (Asn53, Asn311, Asn333 and Asn403) and is known to be glycosylated, which is the most likely cause of this discrepancy in apparent mass.[9] The enzyme is homodimeric in vivo, covalently bound to the external face of the plasma membrane by a GPI anchor and is also present in a soluble form.[10] Using multiangle light-scattering analysis we were able to confirm that our preparations of sCD73 were also homodimeric in solution. We determined a weight-averaged molecular mass of 121.2 kD from two independent measurements, which is higher than the calculated mass of the dimer (117.9 kD), but again this is most likely due to secondary modifications of the enzyme. To investigate the metal cofactor preference, the binuclear metal centre of sCD73 was stripped of metal ions by extensive dialysis against EDTA (2 mm). The apo-enzyme generated was subsequently reconstituted with various divalent metal ions Figure 2. The binding assays of the lectins ConA, GNA, LcH, WGA, ECL, SNA, and nucleotidase activity was measured. We found that nucle- MAL, AAL and PNA were performed on sensors coated with linker/spacer di- otidase activity was almost completely abolished following dilution of 1:1000. One lectin was injected at a time and the surface was re- alysis against EDTA (ca. 5% residual activity). Incubation of generated after each incubation step. The lectins ConA, GNA, WGA and LcH apo-sCD73 with FeII and CuII caused inhibition of this residual showed specific binding to the immobilised sCD73 protein. activity, whereas MgII had no effect and MnII and NiII resulted in a slight increase in activity (Figure 1). Significant recovery of suggests a significant level of mannose residues on sCD73 and nucleotidase activity could only be achieved with CoII and ZnII this is consistent with what has been found in an earlier study (up to 30 and 60% of the activity of the holo-enzyme, respec- with ecto-5’-nucleotidase from cultured human chorionic tively). Similar to nucleotidases from other sources this is cells.[13] Additionally, we confirmed the proposed complex oli- consistent with CD73 being a zinc-dependent 5’-nucleoti- gosaccharide structure on sCD73 through the binding of other dase.[7,8b,11] lectins.[13] Binding of Lens culinaris lectin (LcH) and Aleuria aur- As CD73 is known to be glycosylated, we set out to verify antia lectin (AAL) suggests a fucosylated core structure.[14] the presence of this secondary modification on sCD73 using Wheat germ agglutinin (WGA) binding and Sambucus nigra biophotonic microarray imaging.[12] Nine lectins were sequen- lectin (SNA) binding suggests the presence of terminal N- tially tested against the sensor array. A total of two series of acetyl-glucosamine (GlcNAc) and sialic acid groups, respective- measurements were performed with a different order of lectin ly.[15] injections, and the association rates of lectin binding were determined (Figure 2). This pattern of association rates reflects Steady-state kinetic analysis the glycosylation of sCD73. The binding of the different lectins varies significantly in terms of the association and dissociation Steady-state kinetic experiments with various monophosphate rates as well as maximum achievable loads. The strong binding nucleosides as substrate confirmed that our recombinant of concanavalin A (ConA) and Galanthus nivalis lectin (GNA) sCD73 displays similar kinetic parameters as reported previous-

ChemBioChem 2012, 13, 2384 – 2391 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2385 N. S. Scrutton et al. ly for ecto-5’-nucleotidase from human seminal plasma, a six C-terminal histidine residues). For clarity, the amino acid human colon adenocarcinoma cell line and human airway epi- numbering used throughout is based on the CD73 primary thelia.[1a,16] The enzyme clearly prefers purine nucleoside mono- structure, including the signal sequence. In the crystal struc- phosphates (AMP, GMP and IMP) over pyrimidine nucleoside ture, the peptide chain could be modelled from Trp27 to monophosphates (UMP, CMP and TMP; Table 1). Recently, it Ser549 in subunits A and B. In the refined model, both subunits are in a closed conformation when compared to the Es- cherichia coli 5’-nucleotidase (5’-NT) structures. The enzyme Table 1. Apparent steady-state kinetic parameters of sCD73 determined occurs as a homodimer and both subunits contain two ZnII with nucleoside monophosphates. Initial rates of hydrolysis were mea- ions and one molecule of inorganic phosphate in the active sured by coupling inorganic phosphate production to a purine nucleo- site. Despite attempts at both co-crystallisation and crystal side phosphorylase-catalysed chromogenic assay. The assays were performed at 378C using 3–6 nm sCD73. soaks with several ligands we were unable to obtain a structure in complex with a ligand or in a significantly different protein À1 m mÀ1 À1 Substrate kcat [s ] KM [m ] kcat/KM [m s ] conformation. AMP 187Æ 8 3.8 Æ0.8 48.8 Æ10.8 The structure of sCD73 is typical for members of this family GMP 132Æ 5 6.5 Æ1.0 20.3 Æ3.3 of zinc-dependent 5’-nucleotidases, consisting of two distinct IMP 104Æ 4 7.1 Æ1.0 14.7 Æ2.1 domains connected by an -helix (residues Pro318–Asn333). TMP 40Æ 1 8.0 Æ0.9 5.0 Æ0.6 a UMP 289Æ 13 31.6 Æ4.5 9.1 Æ1.4 The larger N-terminal domain (residues 27–317) has a four-lay- CMP 202Æ 7 56.4 Æ5.6 3.6 Æ0.4 ered a/b-b-b-a architecture containing two sandwiched mixed 2’-dAMP 174Æ 9114Æ16 1.5 Æ0.2 b-sheets. The C-terminal domain (residues 337–549) has a four- NMN 102Æ 4 880Æ113 0.12 Æ0.02 layered a/b-b-a-b architecture with a five-stranded b-sheet forming the central core. The active site is formed at the interface of the two domains, with the ZnII ions bound to the N-ter- was shown that CD73 is also able to use NMN as a substrate.[8b] minal domain and a pocket for substrate binding in the C-ter-

We show that the catalytic efficiency (kcat/KM) with this sub- minal domain. As discussed above, CD73 contains four puta- strate is ~400 times lower than that with the best substrate tive N-glycosylation sites. In the crystal structure we detected

AMP; this is largely due to the ~230-fold increase in KM. The 2’- the presence of a b-N-acetyl-glucosamine residue attached at hydroxyl group on the ribose moiety of AMP is important for Asn311 thereby unequivocally indicating that this site is N-gly- substrate recognition, as removal of this hydroxyl group in- cosylated (see Figure S1 in the Supporting Information). creases the KM value 30-fold. In light of this, it is surprising that Compared with the crystal structures of 5’-NT and Haemophi- the enzyme exhibits a low KM with TMP which also lacks the 2’- lus influenza NadN, sCD73 displays some marked differences in hydroxyl group. However, the low KM with TMP is accompanied loop structures and extensions. The loop structure in sCD73 by a sharp decrease in kcat, indicating that TMP likely binds in formed by residues Thr44–Gly59 and a disulfide bridge (Cys51– a different conformation compared to AMP, allowing tighter Cys57) is replaced by a b-hairpin (Pro51–Gly68) in NadN that binding but poorer positioning for turnover. A similar observa- interacts with a larger a +b extension (Ser320–Ser355). The b- tion has been made with bull seminal plasma 5’-nucleoti- hairpin is also present in E. coli and Thermus thermophilus 5’- dase.[17] The reactivity with NMN was also linked to NAD ho- nucleotidases. The crystal structure also shows that the eight meostasis because CD73 is supposedly able to degrade NAD cysteines present in sCD73 are all involved in intramolecular to adenosine and nicotinamide riboside (NR), with NMN gener- disulfide bridge formation. The homodimer is therefore not ated as an intermediate.[8b] We tested NAD as a possible sub- linked by intersubunit disulfide bridges. Three disulfide bridges strate for sCD73, but following overnight incubation at 378C (Cys353–Cys358, Cys365–Cys387 and Cys476–Cys479) are locat- and correcting for chemical hydrolysis we did not detect any ed in the C-terminal domain and one is located in the N-termi- inorganic phosphate (using the phosphate assay described in nal domain (Cys51–Cys57; Figure 3). Knowledge of the location the Experimental section) and adenosine, NR, AMP or NMN of these disulfide bridges is useful as a Cys358Ala mutation were not detected by HPLC analysis that could be attributed has been implicated in ectopic tissue calcification.[18] The to sCD73-catalysed hydrolysis of NAD. However, we found that Cys353–Cys358 bond is located in the C-terminal domain and commercial supplies of NAD contained trace amounts of AMP is probably crucial in providing the correct orientation of the and NMN, which were turned over by sCD73 and subsequently loop that is directly involved in dimerisation. The mutation of gave rise to trace amounts of adenosine and NR in HPLC analy- this cysteine residue could result in a loss of structural integrity sis. Our observation that sCD73 does not have detectable causing loss of CD73 functionality. pyrophosphatase activity with NAD is not surprising, as other The crystal structure of sCD73 reveals a homodimer formed pyrophosphate nucleosides such as ADP are not substrates. On by a network of hydrogen bonds and hydrophobic interactions the contrary, they are inhibitors of enzyme activity.[16a] between the C-terminal domains. Analysis of the homodimer by using the protein interfaces, surfaces and assemblies (PISA) service at the European Bioinformatics Institute (Hinxton, UK) General protein structure suggests that the dimer observed in the crystals would be a The purified enzyme used for crystallisation comprised 530 res- stable assembly in solution. The average surface area of a idues (Trp27–Ser549, excluding one N-terminal methionine and sCD73 monomer is 20908 2 and the interface surface area of

2386 www.chembiochem.org 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2012, 13, 2384 – 2391 Crystal Structure of a Soluble Human CD73

Figure 3. Ribbon representation of the three-dimensional structure of dimeric sCD73. The two subunits are shown in black and grey, respectively. The disulfide bridges are displayed as yellow sticks, the potential N-glycosylation sites (Asn) as magenta spheres, GlcNac is indicated with red sticks and the ZnII ions are indicated by cyan spheres. The N-glycosylation sites are clus- tered together at the top of the dimer facing away from the C-terminal site of GPI-modification. the dimer is 1253 2. From these values it was calculated that approximately 6% of surface area is involved in dimer interac- Figure 4. Ribbon representation of the three-dimensional structure of dimer- tions. An important part of the network comprises three short ic sCD73. Subunit A is shown in orange, with the N-terminal domain (lighter stretches of amino acids on one subunit (Ile343–Tyr345, shade) and the C-terminal domain (darker shade) connected with an a-helix at the top. Subunit B is shown in magenta, with the N-terminal domain Asp399–Arg402 and Val537–Val542) that interact with a disul- (lighter shade) and the C-terminal domain (darker shade) also connected fide-bonded loop (Cys476–Pro482) on the opposite subunit with an a-helix at the top. The dimer interface is located between the two (Figure 4A). This latter loop, which is absent in bacterial, fungal C-terminal domains, the ZnII ions are indicated by cyan spheres, and the C- and yeast enzymes but conserved in animal enzymes, is likely terminal serines are indicated by green sticks at the bottom of the dimer interface. A) Enlarged view of the dimer interface showing Arg480 from sub- crucial for dimer formation. A BLAST search revealed that the unit A (loop Cys476–Pro482) that is involved in a cation–p interaction with conserved disulfide bridge Cys476–Cys479 found in this loop Tyr345 from the opposite subunit. Distances are indicated in . The distance in sCD73 is conserved in animal ecto-5’-nucleotidases (Fig- between the Arg480 CZ carbon and Tyr345 CG carbon is 4.48 . B) Enlarged ure 5A). Furthermore, the conserved loop in sCD73 contains view of the active site of subunit B showing the amino acid residues and water molecule (small red sphere) that coordinate the ZnII ions and one in- an arginine residue (Arg480) that is involved in a typical organic phosphate anion in orange that is bound to the metal ions. cation–p interaction with Tyr345 from the opposite subunit[19] and hydrogen bonding with Asp399–Arg402. We also inspect- ed a surface representation of the sCD73 dimer with residues between the metal ions in the active site (3.4 ) also corre- conserved within animal sequences. The highly conserved sponds to the distance typically found with ZnII ions in this ge- nature of the dimer interface suggests that the interface as ob- ometry. Related 5’-nucleotidases originating, for example, from served in the crystals is indeed the physiologically relevant bull seminal plasma, chicken gizzard, snake venom, H. influenza dimer (Figure 5B). and E. coli also require ZnII ions for catalysis.[8b, 11,20] In both sub- For 5’-NT and NadN it has been shown that they exhibit units the ZnII ions are coordinated in a slightly distorted octa- a large domain motion during catalysis to facilitate substrate hedral geometry (Figure 4B). The residues Asp85, Asn117, binding and product release.[8] It has been suggested that in His220 and His243 coordinate Zn1 and Zn3, whereas Asp85, the E. coli 5’-nucleotidase, the C-terminal domain rotates rela- Asp36, His38, and a hydroxyl moiety coordinate Zn2 and Zn4. tive to the N-terminal domain. An overlay of the open and For both ZnII ions in the subunits the fifth and sixth coordinate closed structures of 5’-NT with that of sCD73 shows that with positions of the octahedral coordination sphere are occupied CD73 the N-terminal domain would rotate upwards, away from by the oxygen of a single inorganic phosphate anion and a the plasma membrane (Figure 6). This suggests that in mono- hydroxyl moiety (Figure 4B). meric 5’-nucleotidases C-terminal domain rotation relative to The C-terminal domain holds the substrate-binding pocket, the N-terminal domain can be equally seen as N-terminal which in the structure reported here is unoccupied. This is con- domain rotation relative to the C-terminal domain. In the case trary to an observation made with the NAD nucleotidase from of sCD73, however, only the C-terminal domain is in a position H. influenza, in which an adenosine molecule was found to be to rotate, as the N-terminal domain is in a fixed position be- bound to the enzyme even though the enzyme had not been cause of dimer formation and anchorage to the plasma mem- incubated with the nucleotide.[8b] Comparison of the active site brane. of sCD73 to that of 5’-NT revealed a high degree of structural overlap (Figure 7). Based on active site similarities with 5’-NT, in sCD73 the aromatic purine moiety of the substrate is stacked Active site architecture between Phe417 and Phe500 (Figure 7B). The structure of As shown in Figure 1, apo-sCD73 can be reconstituted with sCD73 in complex with phosphate highly resembles the pro- ZnII indicating that ZnII is essential for catalysis. The distance posed product complex of 5’-NT.[21] Equally, for CD73 we pro-

ChemBioChem 2012, 13, 2384 – 2391 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2387 N. S. Scrutton et al.

Figure 5. A) Partial multiple sequence alignment of (putative) 5’-nucleotidases. The conserved loop structure that is involved in dimerisation of 5’-nucleotidases is indicated by the box outline, with the conserved cysteines highlighted in black. From this alignment it appears that the loop structure is only present in animals, and that dimerisation does not occur in lower organisms such as bacteria and yeast. The full names of the organisms in the alignment and the Gen- Bank, NCBI or PDB accession numbers of the respective protein sequences are as follows: Homo sapiens (AAH65937.1), Bos taurus (AAI14094.1), Mus musculus (AAC13542.1), Oreochromis niloticus (XP_003446719.1), Triatoma infestans (CAE46445.1), Drosophila melanogaster (AAF57855.1), T. thermophilus (YP_144594.1), E. coli (NP_415013.1), Acinetobacter lwoffii (ZP_06070098.1), H. influenza (YP_247923.1) and Candida albicans (PDB ID: 3C9F). B) A surface representation of the sCD73 dimer with residues conserved within animal sequences highlighted in white. The bottom panel shows the dimer interface having broken apart the two monomers and rotated them 608 relative to the axis shown. The highly conserved nature of the dimer interface suggests that the interface as observed in the crystals is indeed the physiologically relevant dimer.

that the soluble form of the enzyme was active with nucleoside monophosphates and contained ZnII as cofactor. Using biophotonic microarray imaging we were able to confirm glycosylation of the enzyme and show that the enzyme has high mannose content in addition to a fucosylated core structure and terminal GlcNAc and sialic acid residues. The crystal structure of human ecto-5’-nucleotidase was determined at 2.2 resolution. This structure gives insight into the mode of dimerisation of CD73, active site geometry and post-translation modifications (N-glycosylation) and disulfide bond formation. A significant feature of this structure is the identification of the disulfide bridge between Cys353 and Cys358, which makes up part of a loop region critical to dimer- Figure 6. Structural overlay in ribbon representation of the sCD73 dimer isation. It has previously been shown that the mutation (red), E. coli 5’-nucleotidase in the closed from (blue, PDB ID:1HPU) and E. coli 5’-nucleotidase in the open conformation (orange, PDB ID: 1HP1). The Cys358Ala in patients that suffer from symptomatic arterial large N-terminal domain motion from a closed to open conformation is in- and joint calcification, renders the enzyme inactive. Compara- dicated with the arrow. The C terminus of sCD73 is located at the bottom tive analysis of the active site and proposed catalytic mecha- of the figure, indicating that the N-terminal domain moves away from the nism of E. coli 5’-nucleotidase with the sCD73 structure allowed plasma membrane. us to propose a catalytic mechanism for CD73. This detailed structural information can now be used in the structure-based pose a catalytic mechanism that involves an in-line nucleophil- design of inhibitors or activator molecules for the treatment of ic attack by a hydroxyl moiety that is coordinated by ZnII (Zn4 several pathological diseases involving CD73. Additionally, this and Zn2 in subunits A and B, respectively) on the substrate crystal structure may prove useful in providing explanations as phosphorus with the nucleoside acting as a leaving group to the possible roles of single-nucleotide polymorphisms in (Scheme 1). disease.

Conclusions Experimental Section We expressed sCD73, an ecto-5’-nucleotidase in HEK293 EBNA1 Materials: 4-Morpholineethanesulfonic acid sodium salt (MES), ad- cells and purified the enzyme in a form suitable for biochemi- enosine 2’,5’-diphosphate sodium salt (ADP), adenosine mono- cal and structural analyses. Biochemical analyses confirmed phosphate disodium salt (AMP), guanosine 5’-monophosphate

2388 www.chembiochem.org 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2012, 13, 2384 – 2391 Crystal Structure of a Soluble Human CD73

Figure 7. Active-site structure of A) E. coli 5’-nucleotidase with the inhibitor adenosine 5’-methylene diphosphonate bound (PBD ID: 1HPU) and B) sCD73 with phosphate bound. Zinc ions are indicated with grey spheres, and waters with red spheres. The high degree of structural similarity between the two active sites indicates how substrate may bind to CD73.

Scheme 1. Proposed mechanism for the CD73-catalysed hydrolysis of a monophosphate nucleoside substrate. A) The proposed Michaelis complex analogous to the proposed catalytic mechanism of E. coli 5’-nucleotidase. B) The proposed product complex that resembles the sCD73 structure with phosphate bound. R represents the nucleoside moiety of the substrate/product.

disodium salt hydrate (GMP), cytidine 5’-monophosphate disodium Expression and purification of soluble CD73: Cloning and expres- salt (CMP), uridine 5’-monophosphate disodium salt (UMP), inosine sion of sCD73 was performed as previously described for human 5’-monophosphate disodium salt (IMP), thymidine 5’-monophos- vascular adhesion protein-1.[22] In short, an intronless gene encod- phate disodium salt hydrate (TMP), thiamine monophosphate chlo- ing the soluble form of CD73 (sCD73; residues 27–549) and a His6 ride dihydrate, NMN, pyridoxal 5’-phosphate and metal chloride tag at the C terminus was codon-optimised for expression in salts were obtained from Sigma. The Molecular Probes EnzCheck human cells and synthesised by Qiagen. Residues 1–26 of CD73, phosphate assay kit was bought from Invitrogen, and HisTrap FF which form the signal sequence for protein secretion, were re- columns were from GE Healthcare. Self-assembled monolayer com- placed with a methionine residue in sCD73. The gene was sub- ponents were alkanethiolate anchors coupled to short oligo(ethy- cloned into a modified pCep-Pu vector using the flanking NheI and lene glycol) chains terminated with either a hydroxyl or carboxylic BamHI restriction sites. This vector contains an N-terminal signal sequence that is derived from a human extracellular glycoprotein acid group, termed spacer and linker respectively: HS-(CH2)17- (OC H ) -OH and HS-(CH ) -(OC H ) -OCH COOH. Spacer and linker (osteonectin), residues 1–19 followed by the sequence Leu-Ala-Ser, 2 4 3 2 17 2 4 6 2 [23] were both obtained from ProChimia Surfaces (Sopot, Poland). N-(3- which allows extracellular expression of sCD73. HEK293 EBNA1 dimethylaminopropyl)-N’-ethylcarbodiimide hydrochloride (EDC), cells were transfected with pCep-Pu:sCD73 (2 mg) using the trans- N-hydroxysuccinimide (NHS), bovine serum albumin (BSA; 98%) fection reagent Lipofectamine (Invitrogen) according to the manu- were obtained from Sigma–Aldrich. The lectins from Erythrina cris- facturer’s instructions. The collected culture medium (typically 600 mL) was centrifuged for 10 min at 3000g and 48C and steri- tagalli (ECL-biotin), Lens culinaris (LcH-biotin), Galanthus nivalis lised by passage through a 0.45 mm filter (Millipore). The medium (GNA-biotin), Arachis hypogaea (peanut; PNA-biotin), Sambucus containing sCD73 was concentrated to 50 mL using an Amicon nigra (SNA-biotin), Maackia amurensis (MAL-biotin) and Aleuria aur- stirred ultrafiltration cell fitted with a 30 kD cut-off filter and subse- antia (AAL-biotin) were kindly provided by Galab (Geesthacht, Ger- quently concentrated further to 5 mL using Vivaspin 20 centrifugal many) as 1 mgmLÀ1 solutions in Tris buffer. The concanavalin A units (Sartorius-Stedim). The concentrated sCD73 sample was (ConA-FITC), and WGA lectins were acquired from Vector Laborato- loaded onto a 2 mL HisTrap FF column that was pre-equilibrated ries as 5 mgmLÀ1 solutions in HEPES buffer. with MES buffer (50 mm, pH 7.0) containing NaCl (100 mm; buf-

ChemBioChem 2012, 13, 2384 – 2391 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2389 N. S. Scrutton et al. fer A). The column was washed with buffer A containing imidazole ised with self-assembling monolayers, formed by alkanethiols, (20 mm) and sCD73 was eluted with buffer A containing imidazole which are carboxyl- (linker) and hydroxyl- (spacer) functionalised, (300 mm). For further purification using size-exclusion chromatog- respectively.[26] A linker/spacer ratio of 1:1000 was chosen for the raphy, the eluted fractions were pooled, concentrated and loaded experiments described here. The carboxyl group was activated onto a HiLoad 26/60 Superdex 200 preparative grade column pre- with EDC and NHS to produce a succinimidyl ester group which equilibrated with buffer A containing MgCl2 (1 mm), at 48C. Frac- reacts with the primary amine group of protein surface lysines to tions containing sCD73 were pooled, concentrated and stored at form peptide bonds. The target proteins printed onto the sensor À808C until further use. Purified sCD73 was loaded on a 10% array were sCD73 (2 mgmLÀ1) and BSA (2 mgmLÀ1). BSA served as polyacrylamide gel for Coomassie Brilliant Blue staining. Protein nonspecific control. After 2 h incubation at RT, to allow protein concentration was determined by measuring the absorbance at tethering, the sensors were washed and stored dry at 48C before

280 nm using a calculated extinction coefficient e280 of use. The standard running and dilution buffer was phosphate-buf- 56310mÀ1 cmÀ1.[24] fered saline containing Tween 20 surfactant (PBST; 510À5 w/w), m m CaCl2 (1 m ) and MgCl2 (1 m ), to stabilise the lectins. Aqueous Multiangle light scattering: To determine the oligomeric state of phosphoric acid solution (100 mm) was used as the regeneration sCD73, samples (0.2 mL) containing sCD73 (approximately À1 buffer. Before use, the dry arrays were rehydrated in PBST buffer 0.5 mgmL ) were gel-filtered using a Superdex 200 10/300 GL for 15 min, washed with regeneration buffer, blocked with a solu- m m column in MES buffer (5 m , pH 7.5) containing NaCl (100 m ) tion of BSA (2 gLÀ1) in PBST for 10 min, washed with regeneration m and MgCl2 (1 m ). The elution was monitored using a Wyatt buffer, and finally stabilised in PBST flow. An injection of doubly DAWN HELEOS 18-angle laser photometer and a Wyatt Optilab rEX concentrated PBS was carried out to establish the sensitivity of the refractive index detector (Wyatt Technology, Santa Barbara, CA). array spots to the change in analyte bulk refractive index. These The molecular mass moments were analysed by using Astra data were used to convert relative brightness response into a scale 5.3.4.13 (Wyatt Technology). equivalent to refractive index change, and conventionally this is À6 Steady-state kinetic analysis: Steady-state kinetic studies were given in response units, RU10 refractive index units (RIU). The À5 performed with sCD73 (3–6 nm)at378C using the EnzCheck phos- sensitivity of each array spot is typically 810 RIU and the assays phate assay kit, by coupling the sCD73-driven release of inorganic are averaged over 16 spots as required for multiplexed results. The À1 phosphate to the purine nucleoside phosphorylase-catalysed con- following analytes were used in the assay: ConA (27 mgL ), WGA À1 À1 À1 À1 version of 2-amino-6-mercapto-7-methylpurine riboside (MESG) to (27 mgL ), GNA (18 mgL ), ECL (18 mgL ), MAL (18 mgL ), PNA À1 À1 À1 À1 ribose 1-phosphate and 2-amino-6-mercapto-7-methylpurine. The (18 mgL ), SNA (18 mgL ), LcH (18 mgL ) and AAL (18 mgL ). conversion of MESG results in a spectrophotometric shift from The volume flow rate during the kinetic measurements was À1 330 nm for the substrate to 360 nm for the product, hence sCD73 50 mLmin . Each lectin solution was injected over the sensor sur- activity was measured by following the increase in absorbance at face for 10 min to record association kinetics, followed by a 5 min 360 nm. running buffer wash-off phase. Finally, the sensor was cleaned of the remaining bound analyte with the regeneration buffer injec- HPLC analysis: Enzyme was incubated with AMP (0.1 mm), NMN (0.5 mm) or NAD (0.5 mm)at378C for 24 h and samples were taken at regular intervals. The samples were passed through a 10 kD cut-off filter to remove enzyme and frozen until further ana- Table 2. Data collection and refinement statistics. Values in parentheses lysed. The control samples without enzyme were treated equally. refer to highest-resolution shell. HPLC analysis was performed on an Agilent 1100 series instrument equipped with an Eclipse XDB-C18 reversed-phase column (Agi- CD73 (PDB ID: 4H1S) lent) (5 mm, 1504.6 mm). The column was run using conditions wavelength [] 0.9173 that have been described previously.[25] In short, solvent A consist- resolution range [] 29.86–2.197 (2.276–2.197) m ed of KH2PO4 (0.1 ) containing tetra-n-butylammonium sulfate space group P22121 (5 mm, pH 5.0); solvent B was solvent A containing methanol (30%, unit cell: a, b, c [] 54.91, 95.01, 230.29 v/v). The solvent program was a linear gradient starting at 100% a, b, c [8] 90, 90, 90 solvent A and increasing to 100% solvent B over 30 min. The flow total reflections 416 523 unique reflections 62 077 (5878) rate was held at 0.4 mLminÀ1, detection monitored at 260 nm and multiplicity 3.5 (3.3) injections were 5 mL. The elution times (in min) of the standards completeness [%] (99.46) (96.27) were: NR 3.4, NMN 4.2, ADP 11.7, NAD 13.6, adenosine 14.7 and mean I/s(I) 13.62 (3.59) AMP 16.2. Wilson B-factor 25.56

Rmeas 11.8 (54.7) Divalent metal reconstitution: Apo-sCD73 was produced by over- R 11.0 (50.8) m m merge night dialysis against MES (50 m , pH 7.5) and NaCl (100 m ) sup- R 4.5 (21.7) m pim plemented with EDTA (2 m )at48C followed by another round of Rfactor 0.1733 m m dialysis against MES (50 m , pH 7.5) and NaCl (100 m ). The apo Rfree 0.2154 enzyme was subsequently stored as aliquots and incubated over- water 570 night at 48C with either of the following divalent metal ions protein residues 1046 (0.1 mm): MgII,MnII,FeII,CoII,NiII,CuII and ZnII. Nucleotidase activity r.m.s.d. bonds [] 0.008 was measured as described above using AMP (0.1 mm) as sub- r.m.s.d. angles [] 1.17 Ramachandran favoured [%] 96.3 strate. Ramachandran outliers [%] 0.4 Lectin binding assay: Glycosylation analysis of sCD73 was per- Clashscore 6.7 formed by biophotonic microarray imaging. The imaging was per- average B-factor 25.50 macromolecules 25.40 formed using microarrays with 96 spots created by inkjet printing solvent 27.30 of gold nanoparticles.[12] The gold nanoparticles were functional-

2390 www.chembiochem.org 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2012, 13, 2384 – 2391 Crystal Structure of a Soluble Human CD73 tion. No significant degradation of the sensor surface activity was [8] a) T. Knçfel, N. Strater, J. Mol. Biol. 2001, 309, 255– 266; b) S. Garavaglia, observed after ten regeneration cycles. S. Bruzzone, C. Cassani, L. Canella, G. Allegrone, L. Sturla, E. Mannino, E. Millo, A. De Flora, M. Rizzi, Biochem. J. 2011, 441, 131 –141. sCD73 Crystallization, structure elucidation and molecular mod- [9] H. Zimmermann, Biochem. J. 1992, 285, 345– 365. elling: sCD73 crystals were obtained by sitting drop vapour diffu- [10] M. R. Klemens, W. R. Sherman, N. J. Holmberg, J. M. Ruedi, M. G. Low, sion at 48C. Drops (200 nL) were prepared by using protein L. F. Thompson, Biochem. Biophys. Res. Commun. 1990, 172, 1371 –1377. (1 mgmLÀ1) and mother liquor (200 nL) containing a succinic acid/ [11] C. Fini, C. A. Palmerini, P. Damiani, U. Stochaj, H. G. Mannherz, A. Floridi, phosphate/glycine (SPG) buffer system (0.1m, pH 4) and PEG 1500 Biochim. Biophys. Acta Protein Struct. Mol. Enzymol. 1990, 1038, 18– 22. (15%, w/v). Crystals were flash frozen in liquid nitrogen using [12] R. V. Olkhov, A. M. Shaw, Anal. Biochem. 2010, 396, 30– 35. [13] R. Burgemeister, I. Danescu, W. Gutensohn, Biol. Chem. Hoppe-Seyler PEG 200 (10%, w/v) as cryoprotectant. Data were collected from 1990, 371, 355 –361. a single cryofrozen crystal at Diamond (Harwell, UK). The data were [14] S. Serna, S. Yan, M. Martin-Lomas, I. B. H. Wilson, N. C. Reichardt, J. Am. [27] scaled and integrated using XDS and subsequently handled Chem. Soc. 2011, 133, 16495 –16502. [28] using the Phenix suite. Structure determination was carried out [15] a) R. Karamanska, J. Clarke, O. Blixt, J. I. MacRae, J. Q. Q. Zhang, P. R. by molecular replacement utilising 5’-nucleotidase precursor struc- Crocker, N. Laurent, A. Wright, S. L. Flitsch, D. A. Russell, R. A. Field, Gly- ture (PDB ID: 2Z1A) from T. thermophilus HB8 as the search model. coconjugate J. 2008, 25, 69–74; b) L. G. Baum, J. C. Paulson, Acta Histo- Refinement and model building were carried out using Phenix[28] chem. Suppl. 1990, 40, 35 –38. and COOT.[29] Structure validation with MOLPROBITY[30] was inte- [16] a) C. Fini, M. Coli, A. Floridi, Biochim. Biophys. Acta Gen. Subj. 1991, 1075, grated as part of the iterative rebuild and refinement procedure. 20–27; b) M. Picher, L. H. Burch, A. J. Hirsh, J. Spychala, R. C. Boucher, J. Biol. Chem. 2003, 278, 13468–13479. Data and refinement statistics are shown in Table 2. The surface [17] C. Fini, P. L. Ipata, C. A. Palmerini, A. Floridi, Biochim. Biophys. Acta Pro- and interface areas were calculated with the PDBePISA tool (http:// tein Struct. Mol. Enzymol. 1983, 748, 405 –412. www.ebi.ac.uk/msd-srv/prot_int/pistart.html). [18] C. St. Hilaire, S. G. Ziegler, T. C. Markello, A. Brusco, C. Groden, F. Gill, H. Carlson-Donohoe, R. J. Lederman, M. Y. Chen, D. Yang, M. P. Siegenthal- er, C. Arduino, C. Mancini, B. Freudenthal, H. C. Stanescu, A. A. Zdebik, Acknowledgements R. K. Chaganti, R. L. Nussbaum, R. Kleta, W. A. Gahl, M. Boehm, N. Engl. J. Med. 2011, 364, 432 –442. Supported by the Ramsay Memorial Foundation Trust (DPHMH) [19] P. B. Crowley, A. Golovin, Proteins Struct. Funct. Bioinf. 2005, 59, 231– 239. and the European Commission (MJW). We would like to thank [20] H. F. Dvorak, L. A. Heppel, J. Biol. Chem. 1968, 243, 2647 –2653. Dr. Tom Jowitt and Marj Howard of the Biomolecular Interactions [21] T. Knçfel, N. Strater, J. Mol. Biol. 2001, 309, 239– 254. Facility at the University of Manchester for their help with MALLS [22] D. P. Heuts, J. O. Gummadova, J. Pang, S. E. Rigby, N. S. Scrutton, J. Biol. analyses and David Mansell for his assistance with HPLC analy- Chem. 2011, 286, 29584 –29593. [23] a) A. Swaroop, B. L. Hogan, U. Francke, Genomics 1988, 2, 37–47; b) E. ses. NSS is an EPSRC Established Career Fellow and a Royal Soci- Pçschl, J. W. Fox, D. Block, U. Mayer, R. Timpl, EMBO J. 1994, 13, 3741 – ety Wolfson Merit Award holder. 3747. [24] E. Gasteiger, C. Hoogland, A. Gattiker, S. Duvaud, M. R. Wilkins, R. D. Appel, A. Bairoch, The Proteomics Protocols Handbook (Ed.: J. M. Walker), Keywords: 5’-nucleotidases · adenosine · CD73 · crystal Humana Press, Totowa, 2005, pp. 571 –607. structure · structural biology [25] L. Guida, L. Franco, E. Zocchi, A. De Flora, FEBS Lett. 1995, 368, 481 – 484. [1] a) J. M. Navarro, N. Olmo, J. Turnay, M. T. Lopez-Conejo, M. A. Lizarbe, [26] M. J. Weissenborn, R. Castangia, J. W. Wehner, R. Sardzik, T. K. Lindhorst, Mol. Cell. Biochem. 1998, 187, 121–131; b) R. Fabiani, G. Ronquist, Clin. S. L. Flitsch, Chem. Commun. 2012, 48, 4444 –4446. Chim. Acta 1993, 216, 175 –182. [27] W. Kabsch, Acta Crystallogr. Sect. D Biol. Crystallogr. 2010, 66, 125–132. [2] a) J. W. Goding, R. Terkeltaub, M. Maurice, P. Deterre, A. Sali, S. I. Belli, [28] P. D. Adams, P. V. Afonine, G. Bunkoczi, V. B. Chen, I. W. Davis, N. Echols, Immunol. Rev. 1998, 161, 11–26; b) H. Muller-Steffner, I. Schenherr- J. J. Headd, L. W. Hung, G. J. Kapral, R. W. Grosse-Kunstleve, A. J. McCoy, Gusse, C. Tarnus, F. Schuber, Arch. Biochem. Biophys. 1993, 304, 154 – N. W. Moriarty, R. Oeffner, R. J. Read, D. C. Richardson, J. S. Richardson, 162; c) I. J. Okazaki, J. Moss, J. Biol. Chem. 1998, 273, 23617 –23620. T. C. Terwilliger, P. H. Zwart, Acta Crystallogr. Sect. D Biol. Crystallogr. [3] G. Hask, J. Linden, B. Cronstein, P. Pacher, Nat. Rev. Drug Discovery 2010, 66, 213– 221. 2008, 7, 759 –770. [29] B. L. P. Emsley, W. G. Scott, K. Cowtan, Acta Crystallogr. Sect. D Biol. Crys- [4] a) B. Zhang, Cancer Res. 2010, 70, 6407 –6411; b) T. Eckle, T. Krahn, A. tallogr. 2010, 66, 486 –501. Grenz, D. Kohler, M. Mittelbronn, C. Ledent, M. A. Jacobson, H. Osswald, [30] V. B. Chen, W. B. Arendall, J. J. Headd, D. A. Keedy, R. M. Immormino, L. F. Thompson, K. Unertl, H. K. Eltzschig, Circulation 2007, 115, 1581 – G. J. Kapral, L. W. Murray, J. S. Richardson, D. C. Richardson, Acta Crystal- 1590. logr. Sect. D Biol. Crystallogr. 2010, 66, 12–21. [5] L. Wang, J. Fan, L. F. Thompson, Y. Zhang, T. Shin, T. J. Curiel, B. Zhang, J. Clin. Invest. 2011, 121, 2371– 2382. Received: June 22, 2012 [6] G. Forte, R. Sorrentino, A. Montinaro, A. Luciano, I. M. Adcock, P. Maioli- Please note: J.G. has been added back to the author list of this manuscript no, C. Arra, C. Cicala, A. Pinto, S. Morello, J. Immunol. 2012, 189, 2226 – since its publication in ChemBioChem EarlyView. The Editor. 2233. [7] T. Knçfel, N. Strater, Nat. Struct. Biol. 1999, 6, 448 –453. Published online on September 20, 2012

ChemBioChem 2012, 13, 2384 – 2391 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2391 12.1. SUPPORTING INFORMATION

12.1 Supporting Information

The complete supporting information can be found under http://onlinelibrary.wiley.com/doi/10.1002/cbic.201200426/suppinfo.

33 CHAPTER THIRTEEN

CHEMOENZYMATIC SYNTHESIS OF O-MANNOSYLPEPTIDES IN SOLUTION AND ON SOLID PHASE

Reproduced with permission from "R. Sardzik, A. P. Green, N. Laurent, P. Both, C. Fontana, J. Voglmeir, M. J. Weissenborn, R. Haddoub, P. Grassi, S. M. Haslam, G. Wid- malm, S. L. Flitsch, Chemoenzymatic Synthesis of O-Mannosylpeptides in Solution and on Solid Phase, J. Am. Chem. Soc. 2012, 134, 4521–4524. DOI: 10.1021/ja211861m" Copyright 2012 American Chemical Society.

The project was initiated by N. Laurent, R. Haddoub and J. Voglmeir, who also chemoenzymatically synthesised the trisaccharide 4. R. Sardzik and M. J. Weissenborn performed the first synthesis of the tetrasaccharide 5 on submilligram scale. As not sufficient analytical data were obtained from this small scale, R. Sardzik synthesised it in a bigger scale and purified it by HPLC. P. Grassi and S. M. Haslam determined the correct linkage of the sialic acid and galactose by using glycosidases, GC and mass spectrometry. R. Sardzik resynthesised and purified the tetrasaccharide 5 on a bigger scale, varied the peptide sequence and performed the synthesis of the tetrasaccharide on the chip. R. Sardzik and A.P. Green performed the one-pot cascade synthesis of 5. P. Both and J. Voglmeir expressed all employed enzymes. C. Fontana and G. Widmalm confirmed the correct linkages of the glycan 5. A. P. Green and S.L. Flitsch wrote the manuscript and R. Sardzik, P. Both and M. J. Weissenborn provided the graphics. A. P. Green, R. Sardzik, C. Fontana, G. Widmalm, P. Grassi and S. M. Haslam wrote the supporting information.

34 Communication

pubs.acs.org/JACS

Chemoenzymatic Synthesis of O-Mannosylpeptides in Solution and on Solid Phase † § † § † † ‡ Robert Sardž ík, , Anthony P. Green, , Nicolas Laurent, Peter Both, Carolina Fontana, † † † # # Josef Voglmeir, Martin J. Weissenborn, Rose Haddoub, Paola Grassi, Stuart M. Haslam, ‡ † Göran Widmalm, and Sabine L. Flitsch*, † School of Chemistry, Manchester Interdisciplinary Biocentre, The University of Manchester, Manchester M1 7DN, U.K. ‡ Department of Organic Chemistry, Arrhenius Laboratory, Stockholm University, S-106 91 Stockholm, Sweden # Division of Molecular Biosciences, Faculty of Natural Sciences, Imperial College London, London, SW7 2AZ, U.K.

*S Supporting Information

α-DG is a cell surface glycoprotein which acts as a receptor ABSTRACT: O-Mannosyl glycans are known to play an for both extracellular matrix proteins5,6 and a number of important role in regulating the function of α-dystroglycan arenaviruses, including Lymphocytic choriomeningitis virus (α-DG), as defective glycosylation is associated with (LCMV) and Lassa fever virus (LFV).7 In order to conserve its various phenotypes of congenital muscular dystrophy. biological functions, α-DG requires extensive post-translational Despite the well-established biological significance of these glycosylation. In addition to typical N-glycosylation and mucin- glycans, questions regarding their precise molecular type O-GalNAc glycans, a number of O-mannosyl glycans have − function remain unanswered. Further biological inves- been detected and characterized.8 10 This mode of post- tigation will require synthetic methods for the generation translational modification has rarely been described in of pure samples of homogeneous glycopeptides with mammals, but has an important role in the function of α-DG, diverse sequences. Here we describe the first total since defects in O-mannosyl glycosylation lead to loss of syntheses of glycopeptides containing the tetrasaccharide extracellular ligand binding activity, resulting in various α β β α NeuNAc 2-3Gal 1-4GlcNAc 1-2Man , which is reported phenotypes of congenital muscular dystrophy (CMD).11 To α to be the most abundant O-mannosyl glycan on -DG. date, mutations in six genes which encode for actual or putative Our approach is based on biomimetic stepwise assembly glycosyltransferases have been identified in patients with − from the reducing end and also gives access to the various forms of CMD.12 17 The molecular functions of two naturally occurring mono-, di-, and trisaccharide sub- of these putative glycosyltransferases, fukutin and fukutin- structures. In addition to the total synthesis, we have related protein (FKRP), remain unclear. It has recently been “ ” developed a one-pot enzymatic cascade leading to the reported that like-acetylglucosaminyltransferase (LARGE) is rapid synthesis of the target tetrasaccharide. Finally, solid- responsible for a post-phosphoryl modification of a phosphory- phase synthesis of the desired glycopeptides directly on a lated O-mannosyl glycan.18 An impressive total synthesis of this gold microarray platform is described. structure has recently been achieved using a traditional chemical approach.19 Protein O-mannosyl transferase 1 (POMT1), POMT2, and protein O-mannose β-1,2-N-acetyl- t is predicted that over 50% of proteins in the human body glucosaminyltransferase 1 (POMGnT1) are known to catalyze I are glycosylated.1 These complex glycans are known to play the first two steps in the biosynthesis of the O-linked a critical role in the regulation of a diverse range of biological tetrasaccharide NeuNAcα2-3Galβ1-4GlcNAcβ1-2Manα (Fig- processes.2 However, despite recent advances in the field of ure 1).20,14 This tetrasaccharide has been reported to be the glycomics, the precise function of these carbohydrates is poorly most abundant O-mannosyl glycan on α-DG from a range of understood, largely due to difficulties in obtaining pure samples different species and tissues, suggesting its relevance to the of homogeneous glycopeptides. Isolation of significant basic functional role of this glycoprotein.9,10,21 quantities of these structures from natural sources is extremely The binding of glycopeptides to their epitopes is often highly challenging. As a result, the development of new methodologies dependent upon peptide sequence as well as glycan structure. for the efficient synthesis of well-defined glycopeptides is of As a result, biological studies to illucidate the role of the O- great interest. Several elegant strategies for the synthesis of mannosyl glycans found on α-DG will require the synthesis of peptides containing both N- and O-linked glycans have been well-defined glycopeptides with diverse sequences. Although reported.3 O-Mannosyl peptides are an important class of two syntheses of the fully assembled tetrasaccharide linked to a 22,23 structures distinct from the usual O-glycans, but have thus far serine/threonine residue have been described previously, only been identified on one human protein, α-dystroglycan (α- the synthesis of the biologically relevant glycopeptides has not DG), a heavily glycosylated protein found in muscle and brain been reported. We envisioned an approach to the synthesis of tissue. Here we report the first total synthesis of glycopeptides containing an O-mannnosyl tetrasaccharide, which represents Received: December 20, 2011 the textbook example of this class of biomolecule.4 Published: February 28, 2012

desired β1,2 linkage using human POMGnT1, the enzyme naturally responsible for the attachment of a GlcNAc residue onto O-mannosylated protein in the α-DG biosynthesis. POMGnT1 was heterologously expressed in Pichia pastoris using a protocol recently developed in our laboratory (SI). The reaction was regularly monitored by HPLC and was shown to be complete in 5 days on a 50 mg scale, providing the disaccharide 3 in 85% yield after HPLC purification. It is important to note that the rate of this reaction can be greatly enhanced on a smaller scale by increasing the concentration of enzyme. The manno-threonine fragment 1 can be readily incorporated into SPPS to produce a range of natural and non- natural manno-peptide sequences. Interestingly, we have Figure 1. α-DG has a mucin-like domain containing the recently observed that the activity of POMGnT1 toward a tetrasaccharide NeuNAcα2-3Galβ1-4GlcNAcβ1-2Manα.Thefirst particular substrate is highly dependent upon peptide 25 two glycosidic linkages are formed by a POMT1/POMT2 complex sequence. Further studies may reveal the minimum sequence and POMGnT1, respectively. The enzymes responsible for the final requirements to maintain enzyme activity. two steps have yet to be defined. CFG nomenclature is used to The enzymes responsible for catalyzing the final two steps of represent the glycan structures. the biosynthesis of the tetrasaccharide have yet to be defined. Instead, enzymes were recruited from alternative biosynthetic these structures which closely mimics their biosynthesis. Our pathways to complete the synthesis. Bovine β1,4-galactosyl- approach commences with the chemical synthesis of mannosyl transferase (β1,4-GalT, EC 2.4.1.38) is well characterized and peptides (Figure 1). The installation of the α-linked mannose has been extensively used for chemoenzymatic synthesis of prior to peptide synthesis is synthetically straightforward and glycoconjugates owing to its broad substrate specificity.26 Using avoids the need for POMT1/POMT2 glycosyltransferase this enzyme, introduction of the galactose with the required catalysts and the expensive sugar donor mannosylphosphor- β1,4 linkage was achieved after 24 h of incubation at 37 °C, yldolichol. Subsequent use of three consecutive, enzymatic yielding trisaccharide 4 in 87% yield. Finally, a trans-sialidase glycosylation reactions to synthesize the tetrasaccharide would from Trypanozoma cruzi (TcTS) was used for attachment of lead to a very short, flexible, and efficient synthetic route the terminal sialic acid with the required α2,3 configuration.27 mimicking the putative biosynthetic pathway. Contrary to sialyltransferases, the TcTS does not require The synthesis commenced with the known manno-threonine 24 expensive CMP-NeuNAc donor and is able to catalyze trans- building block 1 (Scheme 1). This building block was glycosylation from the sialoprotein fetuin. However, careful incorporated into standard Fmoc-based SPPS (Supporting monitoring of the reaction is required, as hydrolysis of the Information (SI)). After cleavage from the resin, deprotection newly formed sialosidic linkage can occur over prolonged of the per-acetylated mannose moiety was achieved using reaction times. As a result, these reactions are difficult to drive NaOMe/MeOH (pH 10) to yield manno-peptide 2, containing to completion, and a mixture of starting material and product is α − a natural peptide sequence of -DG (amino acid residues 317 generally produced. Nevertheless, after 6 h incubation at 37 °C, 326). The GlcNAc moiety was then introduced with the the target glycopeptide 5 was successfully formed in 47% yield, along with 39% recovered trisaccharide. Scheme 1. Total Synthesis of Glycopeptide 5, Containing O- The use of three sequential enzymatic elongation steps Mannosyl Glycan NeuNAcα2-3Galβ1-4GlcNAcβ1-2Manα allows rapid assembly of the tetrasaccharide unit in a manner not possible using traditional chemical glycan synthesis, which is plagued with protecting group manipulations and selectivity issues. This synthesis has enabled production of the final tetrasaccharide in milligram quantities, allowing us to unambiguously characterize the position and stereochemistry of the glycan linkages, as well as the peptide sequence, using detailed NMR studies. The sequential addition of sugar residues is evident from the anomeric region of the 13C NMR spectra (Figure 2). The complete 1H and 13C NMR assignments of 3−5 (SI) were facilitated by first predicting the chemical shifts of the oligosaccharide-threonine structure by the computer program CASPER28 and subsequently analyzing the 2D NMR spectra. For sequential information on the peptide structure, a BS-CT-HMBC experiment29 proved highly informative as residues could be linked together via heteronuclear alternating intra-residue two-bond and inter- residue three-bond correlations. Further evidence of the tetrasaccharide structure was provided by extensive mass spectrometric analysis (SI). An additional advantage of our synthetic strategy is that it provides simple access to the corresponding mono-, di-, and trisaccharide-containing intermediates, which are frequently detected on α-DG due to

4522 dx.doi.org/10.1021/ja211861m | J. Am. Chem. Soc. 2012, 134, 4521−4524 Journal of the American Chemical Society Communication

with a wide range of binding partners, including proteins, viruses, and whole cells, leading to significant advances in the field of glycomics.33 Despite the numerous advantages of these glycan microarrays, their production is severely hampered by difficulties in obtaining pure samples of well-defined carbohydrates through either synthesis or isolation. Additionally, isolated carbohydrates must be further modified with a linker to allow attachment to the array surface. One approach to overcome this problem involves solid-phase synthesis of the required glycan directly on the microarray surface.34 We now report solid-phase synthesis of the O-glycan NeuNAcα2- 3Galβ1-4GlcNAcβ1-2Manα, attached to a natural peptide sequence of α-DG (amino acid residues 373−384), immobilized on a gold platform. The first step was the formation of an Figure 2. Anomeric region of 13C NMR spectra of disaccharide- N-hydroxysuccinimide (NHS)-functionalized self-assembled peptide 3 (top), trisaccharide-peptide 4 (middle), and tetrasaccharide- monolayer (SAM) (Figure 4). SAMs of alkanethiols on gold peptide 5 (bottom). Resonances are denoted by capital letters: (A) αMan, (B) βGlcNAc, (C) βGal, and (D) NeuNAc. heterogeneous glycosylation. Our synthesis has provided samples of the final tetrasaccharide 5 as well as intermediates 2, 3, and 4 as standards for NMR and glycomic databases. This may simplify the identification of O-mannosyl glycans on mammalian proteins other than α-DG, for example, in the gastrointestinal tract of mice.30 The use of “one-pot”, multiple glycosylations represents an attractive method for the synthesis of complex glycans.31 Taking advantage of the highly selective nature of the enzymes involved in our synthesis, we have developed a “one-pot” approach to the assembly of glycopeptide 5. Manno-peptide 2 (1 mg) was added to a premixed solution containing the three enzymes and the three sugar donors (UDP-GlcNAc, UDP-Gal, and fetuin), and the reaction was monitored by HPLC (Figure ° Figure 4. Solid-phase synthesis of target glycopeptide on a gold 3). Following incubation at 37 C for 24 h, peaks platform. Conditions: (a) RGAIIQT(Man)PTLGPOH, RT; (b) ° POMGnT1, UDP-GlcNAc, MnCl2, MES buffer, pH 7, 37 C; (c) β ° GalT1, UDP-Gal, MnCl2, MES buffer, pH 7, 37 C; (d) TcTs, fetuin, phosphate buffer, pH 7, 37 °C.

surfaces provide well-established platforms for carbohydrate microarrays.35 Attachment of the chemically synthesized mannopeptide was achieved through the amine at the N- terminus of the peptide by formation of an amide bond. The three enzymatic elongation steps described previously (see Scheme 1) were then successfully carried out on the immobilized glycopeptide to produce the desired tetrasaccharide attached to the gold surface. Each glycosylation step, as well “ ” as the attachment of the mannopeptide to the surface, was Figure 3. HPLC profiles monitoring the one-pot conversion of 36 manno-peptide 2 to tetrasaccharide 5. conveniently monitored using MALDI-TOF MS. The first two enzymatic steps proceeded to completion, providing access corresponding to glycopeptides 2−5 were observed. After 60 to uniform monolayers of the di- and trisaccharide structures. h, only traces of the mono- and disaccharide were detected, and Incomplete conversion in the final step led to a mixture of the anticipated tri- and tetrasaccharides were formed with good glycoforms on the gold surface, mimicking the biological conversion as a 50:50 mixture.32 This “one-pot” enzymatic heterogeneity encountered on the surface of cells. The use of cascade significantly reduces the time required to produce the POMGnT1 directly on a microarray platform is of particular desired tetrasaccharide and alleviates the need for the interest since this technology will allow the substrate specificity intermediate purification steps. This methodology should now of this important human enzyme to be probed. This solid-phase enable the rapid assembly of this tetrasaccharide on a range of synthesis has a number of advantages over solution-phase natural peptide sequences of α-DG. synthesis: Only minute quantities of material are required, Having developed an efficient solution synthesis, we turned minimizing the use of valuable enzymes and mannopeptides. our attention to preparation of these O-mannosyl glycopeptides HPLC purification steps associated with solution-phase syn- on a solid platform. In recent years, carbohydrate microarrays thesis are avoided since non-covalently bound reagents and have emerged as powerful tools to study glycoenzyme enzymes can be simply washed from the surface. This specificity and investigate the interactions of carbohydrates technology, coupled with our recent studies describing “spot

4523 dx.doi.org/10.1021/ja211861m | J. Am. Chem. Soc. 2012, 134, 4521−4524 Journal of the American Chemical Society Communication synthesis” of peptides directly on the array,37 has the potential (8) Chiba, A.; Matsumura, K.; Yamada, H.; Inazu, T.; Shimizu, T.; to provide rapid access to the tetrasaccharide and its truncated Kusunoki, S.; Kanazawa, I.; Kobata, A.; Endo, T. J. Biol. Chem. 1997, intermediates attached to a diverse range of α-DG peptide 272, 2156. sequences. These structures can be produced in parallel on the (9) Nilsson, J.; Nilsson, J.; Larson, G. R.; Grahn, A. Glycobiology array in a high-throughput manner to facilitate biological 2010, 20, 1160. (10) Stalnaker, S. H.; Hashmi, S.; Lim, J.-M.; Aoki, K.; Porterfield, studies. M.; Gutierrez-Sanchez, G.; Wheeler, J.; Ervasti, J. M.; Bergmann, C.; In summary, we have developed the first syntheses of Tiemeyer, M.; Wells, L. J. Biol. Chem. 2010, 285, 24882. glycopeptides containing the O-mannosyl glycan NeuNAcα2- (11) Michele, D. E.; K. P. Campbell, K. P. J. Biol. Chem. 2003, 278, 3Galβ1-4GlcNAcβ1-2Manα. The synthesis of the tetrasacchar- 15457. ide fragment was achieved in a highly efficient manner using (12) Beltran-Valeró de Bernabe,́ D.; et al. Am. J. Hum. Genet. 2002, three consecutive enzymatic glycosylations. We have demon- 71, 1033. strated these reactions both in a “one-pot” fashion in solution (13) van Reeuwijk, J.; et al. J. Med. Genet. 2005, 42, 907. and on solid phase, providing rapid access to the desired (14) Yoshida, A.; et al. Dev. Cell 2001, 1, 717. structures and their intermediates. This technology is currently (15) Kobayashi, K.; et al. Nature 1998, 394, 388. being used to produce a library of glycopeptides to investigate (16) Brockington, M.; Blake, D. J.; Prandini, P.; Brown, S. C.; Torelli, α S.; Benson, M. A.; Ponting, C. P.; Estournet, B.; Romero, N. B.; the role of this unusual glycan in the binding of -DG to its Mercuri, E.; Voit, T.; Sewry, C. A.; Guicheney, P.; Muntoni, F. Am. J. various receptors and to provide standards for NMR and Hum. Genet. 2001, 69, 1198. glycomic databases. (17) Longman, C.; Brockington, M.; Torelli, S.; Jimenez-Mallebrera, C.; Kennedy, C.; Khalil, N.; Feng, L.; Saran, R. K.; Voit, T.; Merlini, L.; ■ ASSOCIATED CONTENT Sewry, C. A.; Brown, S. C.; Muntoni, F. Hum. Mol. Genet. 2003, 12, *S Supporting Information 2853. Experimental procedures, characterization of new compounds, (18) Yoshida-Moriguchi, T.; Yu, L.; Stalnaker, S. H.; Davis, S.; Kunz, − S.; Madson, M.; Oldstone, M. B. A.; Schachter, H.; Wells, L.; and complete refs 12 15. This material is available free of Campbell, K. P. Science 2010, 327, 88. charge via the Internet at http://pubs.acs.org. (19) Mo, K.-F.; Fang, T.; Stalnaker, S. H.; Kirby, P. S.; Liu, M.; Wells, L.; Pierce, M.; Live, D. H.; Boons, G.-J. J. Am. Chem. Soc. 2011, 133, ■ AUTHOR INFORMATION 14418. Corresponding Author (20) Manya, H.; Chiba, A.; Yoshida, A.; Wang, X.; Chiba, Y.; Jigami, [email protected] Y.; Margolis, R. U.; Endo, T. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 500. Author Contributions § (21) Sasaki, T.; Yamada, H.; Matsumura, K.; Shimizu, T.; Kobata, A.; These authors contributed equally. Endo, T. Biochim. Biophys. Acta, Gen. Subj. 1998, 1425, 599. Notes (22) Matsuo, I.; Isomura, M.; Ajisaka, K. Tetrahedron Lett. 1999, 40, The authors declare no competing financial interest. 5047. (23) (a) Seifert, J.; Ogawa, T.; Ito, Y. Tetrahedron Lett. 1999, 40, ■ ACKNOWLEDGMENTS 6803. (b) Seifert, J.; Ogawa, T.; Kurono, S.; Ito, Y. Glycoconjugate J. 2000, 17, 407. This research was supported by grants from the EPSRC, the (24) Varon, D.; Lioy, E.; Patarroyo, M. E.; Schratt, X.; Unverzagt, C. BBSRC, the Royal Society (Wolfson Award to S.L.F.), the Knut Aust. J. Chem. 2002, 55, 161. and Alice Wallenberg foundation, the Swedish Research (25) Voglmeir, J.; Kaloo, S.; Laurent, N.; Meloni, M. M.; Bohlmann, Council, and the European commission. L.; Wilson, I. B. H.; Flitsch, S. L. Biochem. J. 2011, 436, 447. (26) Palcic, M. M. Curr. Opin. Biotechnol. 1999, 10, 616. ■ REFERENCES (27) Ferrero-García, M. A.; Trombetta, S. E.; Sanchez,́ D. O.; Reglero, A.; Frasch, A. C. C.; Parodi, A. J. Eur. J. Biochem. 1993, 213, (1) Apweiler, R.; Hermjakob, H.; Sharon, N. Biochim. Biophys. Acta, 765. Gen. Subj. 1999, 1473,4. (28) Lundborg, M.; Widmalm, G. Anal. Chem. 2011, 83, 1514. (2) (a) Spiro, R. G. Glycobiology 2002, 12, 43R. (b) Paulson, J. C. (29) Claridge, T. D. W.; Perez-Victoria,́ I. Org. Biomol. Chem. 2003, 1, Trends Biochem. Sci. 1989, 14, 272. (c) Hart, G. W.; Copeland, R. J. 3632. Cell 2010, 143, 672. (30) Ismail, M. N.; Stone, E. L.; Panico, M.; Lee, S. H.; Luu, Y.; (3) (a) Brocke, C.; Kunz, H. Bioorg. Med. Chem. 2002, 10, 3085. Ramirez, K.; Ho, S. B.; Fukuda, M.; Marth, J. D.; Haslam, S. M.; Dell, (b) Pratt, M. R.; Bertozzi, C. R. Chem. Soc. Rev. 2005, 34, 58. A. Glycobiology 2011, 21, 82. (c) Davis, B. G. Chem. Rev. 2002, 102, 579. (d) Arsequell, G.; Valencia, (31) (a) Bezay,́ N.; Dudziak, G.; Liese, A.; Kunz, H. Angew. Chem., G. Tetrahedron: Asymmetry 1999, 10, 3045. (e) George, S. K.; Int. Ed. 2001, 40, 2292. (b) Dudziak, G.; Bezay,́ N.; Schwientek, T.; Schwientek, T.; Holm, B. R.; Reis, C. A.; Clausen, H.; Kihlberg, J. J. Clausen, H.; Kunz, H.; Liese, A. Tetrahedron 2000, 56, 5865. Am. Chem. Soc. 2001, 123, 11117. (f) Mezzato, S.; Unverzagt, C. (32) After 48 h the tri- and tetrasaccharide were formed as a 65:35 Carbohydr. Res. 2010, 345, 1306. mixture. At this stage additional fetuin was added to increase the yield (4) Varki, A.; Sharon, N. In Essentials of Glycobiolgy, 2nd ed.; Varki, of the desired tetrasaccharide (SI). A., Cummings, R. D., Esko, J. D., Freeze, H. H., Stanley, P., Bertozzi, (33) (a) Horlacher, T.; Seeberger, P. H. Chem. Soc. Rev. 2008, 37, C. R., Hart, G. W., Etzler, M. E., Eds.; Cold Spring Harbor Laboratory 1414. (b) Feizi, T.; Fazio, F.; Chai, W.; Wong, C.-H. Curr. Opin. Struct. Press: Cold Spring Harbor, NY, 2008; p 14. Biol. 2003, 13, 637. (5) Ibraghimov-Beskrovnaya, O.; Ervasti, J. M.; Leveille, C. J.; (34) (a) Laurent, N.; Haddoub, R.; Flitsch, S. L. Trends Biotechnol. Slaughter, C. A.; Sernett, S. W.; Campbell, K. P. Nature 1992, 355, 2008, 26, 328. (b) Fazio, F.; Bryan, M. C.; Blixt, O.; Paulson, J. C.; 696. Wong, C.-H. J. Am. Chem. Soc. 2002, 124, 14397. (6) Gee, S. H.; Montanaro, F.; Lindenbaum, M. H.; Carbonetto, S. (35) Houseman, B. T.; Mrksich, M. Chem. Biol. 2002, 9, 443. Cell 1994, 77, 675. (36) Su, J.; Mrksich, M. Angew. Chem., Int. Ed. 2002, 41, 4715. (7) Cao, W.; Henry, M. D.; Borrow, P.; Yamada, H.; Elder, J. H.; (37) Laurent, N.; Haddoub, R.; Voglmeir, J.; Wong, S. C. C.; Gaskell, Ravkov, E. V.; Nichol, S. T.; Compans, R. W.; Campbell, K. P.; S. J.; Flitsch, S. L. ChemBioChem 2008, 9, 2592. Oldstone, M. B. A. Science 1998, 282, 2079.

4524 dx.doi.org/10.1021/ja211861m | J. Am. Chem. Soc. 2012, 134, 4521−4524 13.1. SUPPORTING INFORMATION

13.1 Supporting Information

The complete supporting information can be found under http://pubs.acs.org/doi/suppl/10.1021/ja211861m/suppl_ﬁle/ja211861m_si_001.pdf.

35 CHAPTER FOURTEEN

CONCLUSION AND OUTLOOK

14.1 Conclusion

The aim of the work presented in this thesis was to develop new coupling methods for preparation of arrays, develop novel techniques for analysis of peptide- and glycoarrays, and apply these arrays and techniques to studies of carbohydrate-protein interactions.

To pursue these goals we have developed a simple synthetic route for the synthesis of carbohydrate ligands carrying suitable linkers (chapter7). The efforts towards improving methods for array preparation resulted in development of an oxo-ester mediated NCL based method for immobilisation of molecules with N-terminal cysteines onto oxo-ester functionalised surfaces (chapter8). The coupling is chemoselective and reliably produces the desired arrays under physiological conditions. This coupling system was also successfully applied to peptides and mono- and trivalent carbohydrates and the peptide-ligand concentration necessary for efficient coupling was significantly lowered from 50 mM to 2 mM. Since it relies on the coupling with pentafluorophenol esters, this system should be compatible with any array surface presenting carboxylic acids including nitrogels on glass.

To develop a new analytical tool for the analysis of glass and polystyrene slides MALDI-ToF MS was employed and described in chapter9. The key feature of this method was the application of conducting tape to the back of glass and polystyrene plates. This enabled the analysis of non-covalent arrays of tritylated glycans on poorly conducting surfaces. The trityl group served as an ’internal matrix’ allowing for direct ionisation of the ligands without additional matrices. As the trityl groups were attached to thiols, these moieties were easily available by cleaving the trityl thioester. The free thiols were then used to form glycoarrays on gold surfaces employing the strong S-Au interaction or under thioester formation with maleimide functionalised surfaces (chapter 10). Any group working on non-covalent arrays that thus far solely relied on secondary protein binding could use this technique to analyse even small changes of attached ligands. Trityl groups

36 14.2. OUTLOOK offer reliable MALDI analysis as they do not depend on matrix co-crystallisation. More- over S-tritylated molecules are suitable for use on polystyrene, glass and gold surfaces.

The knowledge acquired via array formation and analysis was employed to determine carbohydrate-protein interactions. In collaboration with the Shaw group, the biophotonic scattering technique was shown to be suitable for the high throughput analysis of lectin binding to monosaccharides (chapter 11). The same principle was applied successfully to determine glycosylation patterns of known and unknown glycoproteins by specific lectin binding (chapters 11 and 12). This new analytical technique enables the quick determination of glycosylation patterns and could be used in the pharmaceutical industry for quality checks of their glycosylated therapeutic antibodies. In contrast to currently employed techniques, glycan cleavage from the protein or derivatisation is not required. Moreover the biophotonic scattering technique has the potential for high throughput analysis and is still able to obtain kinetic data. To study enzyme specificities towards these glycoproteins and peptides, the first solution-phase and on-chip chemoenzymatic synthesis of a four carbohydrates containing mannopeptide from α-dystroglycan was shown (chapter 13). The one-pot synthesis was successfully performed starting from the mannopeptide, and made this glycopeptide available for biological studies in tenth-of-a-milligram quantitites.

14.2 Outlook

Native chemical ligation is routinely used for the coupling of peptide fragments into long peptides and even proteins.38 As of yet, this has not been carried out on microarrays due to the difficulties in thioester formation in solid supported molecules. Since the oxo-ester NCL circumvents this problem, the established coupling protocol could be used to activate the C-terminus of a coupled peptide and couple an additional peptide. Such an approach has great potential for automation and it could lead to the first direct protein synthesis on microarrays. Another advantage of the oxo-ester mediated native chemical ligation method is the fact that it relies on the use of the natural amino acid cysteine. Chem- ical digests of peptides have been shown to afford peptides with N- terminal-cysteine groups.39 If a proteolytic digest could be performed on proteins that afford peptide fragments with N-terminal-cysteines, these fragments could be coupled specifically to the surface even out of complex mixtures. This methodology could be applied to determine O-glycosylation sites on proteins. Cleaving O-glycans from the proteins by β-elimination affords a double bond on the protein.40 This double bond could be converted into a thiol by adding hydrogen sulphide.41 The derivatised glycoprotein would then be subjected to the above discussed proteolytic digest and would hence cleave the protein at the natural and incorporated cysteine residues. These results would be significantly different to the non derivatised glycoprotein fragments and hence enable the determination of O-

37 14.2. OUTLOOK glycosylation sites of proteins.

The use of trityl groups as a ’self-matrix’ could be further explored by e.g. incorporation into SAMs. This should improve the reproducibility and sensitivity of the MALDI- ToF MS analysis. Tritylated compounds could furthermore be utilised as a labelling tool in biological or archaeological samples. In archaeological samples, e.g., collagen and non-collagenous proteins (NCPs) are found.42 The NCPs, however, are difﬁcult to analyse since the collagen is in great excess and overlays the NCP signal in the mass spectrometric analysis. In contrast to collagen, NCPs contain sulphur groups which could be chemoselectively functionalised with trityl groups. Once tritylated, these samples could be directly analysed by MALDI-ToF MS. Moreover, the tritylated label could be used as a tag for puriﬁcation. A complex mixture containing the tritylated ligands could be applied to a polystyrene slide. Rinsing the polystyrene plate with water should leave the tritylated samples in great excess and then could be analysed directly by MALDI-ToF MS. The glycoarrays on polystyrene, even though suitable for bacteria binding studies, were not feasible for enzymatic transformations. It has been assumed that only low conversion yields which are not detectable by MALDI-ToF MS were obtained. By analysing enzymatic reactions on different surface based system through lectin binding and MALDI- ToF MS it would be possible both to detect even low enzyme activity (lectin binding) and determine conversion for higher enzyme activity (MALDI-ToF MS).

The instrumentation for biophotonic scattering analysis is likely to be commercialised in the near future making it accessible to the general scientiﬁc community. The technique of glycosylation pattern analysis could be further explored to determine the entire glycan structure. This could be achieved by employing glycosidases followed by a lectin readout in multiple steps. This technique deserves additional attention from the glycomics and proteomics communities as it could ultimately enable automated high throughput analysis of glycosylation of large numbers of glycoproteins.

The work presented in this thesis succeeded in improving the formation, analysis and application of microarrays for the study of carbohydrate-protein interactions. The ﬁndings described here should increase the impact of arrays in the glycomics ﬁeld and the new analytical tools help with the discovery of disease-associated glycosylations.

38 CHAPTER FIFTEEN

Bibliography

[1] A. Varki and N. Sharon, Essentials of Glycobiolgy, Cold Spring Harbor, NY, 2008.

[2] M. Hartmann and T. K. Lindhorst, Eur. J. Org. Chem., 2011, 3583–3609.

[3] R. Sardzik, R. Sharma, S. Kaloo, J. Voglmeir, P. R. Crocker and S. L. Flitsch, Chem. Commun., 2011, 47, 5425–5427.

[4] P. R. Crocker, J. C. Paulson and A. Varki, Nat. Rev. Immunol., 2007, 7, 255–266.

[5] Z. Jehan, S. Uddin and K. S. Al-Kuraya, Curr. Med. Chem., 2012, 19, 3730–3738.

[6] N. Laurent, J. Voglmeir, A. Wright, J. Blackburn, N. T. Pham, S. C. C. Wong, S. J. Gaskell and S. L. Flitsch, Chembiochem, 2008, 9, 883–887.

[7] X. Zeng, C. A. S. Andrade, M. D. L. Oliveira and X.-L. Sun, Anal. Biochem., 2012, 402, 3161–3176.

[8] D. M. Ratner and P. H. Seeberger, Curr. Pharm. Des., 2007, 13, 173–183.

[9] C. D. Rillahan and J. C. Paulson, Annu. Rev. Biochem., 2011, 80, 797–823.

[10] K. Drickamer and M. E. Taylor, Gen. Biol., 2002, 3, 1–4.

[11] O. Bohorov, H. Andersson-Sand, J. Hoffmann and O. Blixt, Glycobiology, 2006, 16, 21C–27C.

[12] O. Blixt, S. Head, T. Mondala, C. Scanlan, M. E. Huﬂejt, R. Alvarez, M. C. Bryan, F. Fazio, D. Calarese, J. Stevens, N. Razi, D. J. Stevens, J. J. Skehel, I. van Die, D. R. Burton, I. A. Wilson, R. Cummings, N. Bovin, C. H. Wong and J. C. Paulson, Proc. Natl. Acad. Sci. U. S. A., 2004, 101, 17033–17038.

[13] S. Serna, J. Etxebarria, N. Ruiz, M. Martin-Lomas and N. C. Reichardt, Chem. Eur. J., 2010, 16, 13163–13175.

39 BIBLIOGRAPHY

[14] A. J. Ibanez, A. Muck and A. Svatos, J. Mass Spectrom., 2007, 42, 634–640.

[15] J. Su and M. Mrksich, Langmuir, 2003, 19, 4867–4870.

[16] A. Scherl, C. G. Zimmermann-Ivol, J. Di Dio, A. R. Vaezzadehl, P. A. Binz, M. Amez-Droz, R. Cochard, J. C. Sanchez, M. Gluckmann and D. F. Hochstrasser, Rapid Commun. Mass Spectrom., 2005, 19, 605–610.

[17] E. P. Bennett, U. Mandel, H. Clausen, T. A. Gerken, T. A. Fritz and L. A. Tabak, Glycobiology, 2012, 22, 736–756.

[18] S. A. Brooks, Expert Rev. Proteomics, 2006, 3, 345–359.

[19] T. Yoshida-Moriguchi, L. P. Yu, S. H. Stalnaker, S. Davis, S. Kunz, M. Madson, M. B. A. Oldstone, H. Schachter, L. Wells and K. P. Campbell, Science, 2010, 327, 88–92.

[20] L. Ban and M. Mrksich, Angew. Chem., Int. Ed., 2008, 47, 3396–3399.

[21] N. Laurent, R. Haddoub, J. Voglmeir and S. L. Flitsch, Methods Mol. Biol., 2012, 808, 269–84.

[22] Z. L. Zhi, N. Laurent, A. K. Powel, R. Karamanska, M. Fais, J. Voglmeir, A. Wright, J. M. Blackburn, P. R. Crocker, D. A. Russell, S. Flitsch, R. A. Field and J. E. Turnbull, Chembiochem, 2008, 9, 1568–1575.

[23] A. R. de Boer, C. H. Hokke, A. M. Deelder and M. Wuhrer, Glycoconjugate J., 2008, 25, 75–84.

[24] J. C. Love, L. A. Estroff, J. K. Kriebel, R. G. Nuzzo and G. M. Whitesides, Chem. Rev., 2005, 105, 1103–1169.

[25] E. Ostuni, R. Chapman, R. Holmlin, S. Takayama and G. Whitesides, Langmuir, 2001, 17, 5605–5620.

[26] C. E. Inman, S. M. Reed and J. E. Hutchison, Langmuir, 2004, 20, 9144–9150.

[27] R. Zenobi and R. Knochenmuss, Mass Spectrom. Rev., 1998, 17, 337–366.

[28] H. Wollnik, Mass Spectrom. Rev., 1993, 12, 89–114.

[29] J. Homola, S. Yee and G. Gauglitz, Sensors and Actuators B, 1999, 54, 3–15.

[30] J. Homola, Chem. Rev., 2008, 108, 462–493.

[31] M. Fais, R. Karamanska, S. Allman, S. A. Fairhurst, P. Innocenti, A. J. Fairbanks, T. J. Donohoe, B. G. Davis, D. A. Russell and R. A. Field, Chem. Sci., 2011, 2, 1952–1959.

40 BIBLIOGRAPHY

[32] T. Turbadart, Proc. Phys. Soc., 1959, 73, 40–44.

[33] R. Ritchie, Surface Science, 1965, 3, 497–&.

[34] D. A. Conant, J. Acoust. Soc. Am., 2002, 111, 2341–2341.

[35] F. Fazio, M. Bryan, O. Blixt, J. Paulson and C. Wong, J. Am. Chem. Soc., 2002, 124, 14397–14402.

[36] M. C. Bryan, L. V. Lee and C. H. Wong, Bioorganic & Medicinal Chemistry Letters, 2004, 14, 3185–3188.

[37] L. Zou, H.-L. Pang, P.-H. Chan, Z.-S. Huang, L.-Q. Gu and K.-Y. Wong, Carbohydr. Res., 2008, 343, 2932–2938.

[38] P. Dawson, T. Muir, I. Clarklewis and S. Kent, Science, 1994, 266, 776–779.

[39] J. Kang, J. P. Richardson and D. Macmillan, Chem. Commun., 2009, 407–409.

[40] B. Arroyo-Flores, C. Calvo-Mendez, A. Flores-Carreon and E. Lopez-Romero, Mi- crobiology, 1995, 141, 2289–2294.

[41] W. Langhout and H. Waterman, J. Appl. Chem., 1954, 4, 285–288.

[42] H. Roach, Cell Biol. Int., 1994, 18, 617–628.