Microfluidics and microarray based approaches to biological analysis 3

NHS-functionalized slides and epoxide functionalized slides were respectively used by Zhu et al and Schreiber et al to immobilize kinase substrates via their amine groups, resulting in

CHAPTER 3 MICROARRAY-BASED SCREENING OF KINASE ACTIVITY

3.1 Introduction

3.1.1 Protein Phosphorylation

Phosphorylation of proteins by kinases is one of the most important mechanisms for regulation of cell function.1,2 It has been estimated that more than one third of all proteins can be modified by phosphorylation in mammalian cells, and that more than 1

% of the genes in the human genome encode protein kinases.3 A major challenge in the signal transduction field has been to define sequence, structural and mechanistic features responsible for the substrate selectivity, regulation and cellular function of individual protein kinases Additionally, there is considerable interest in identifying potent and selective inhibitors for each of these enzymes since these can be used as potential therapeutic agents in the treatment of cancer, heart disease and immune-related conditions.4 However, few kinases have yet been identified and fully characterized

One of the most important features of protein kinases is their substrate specificity, which to a large extent is determined by the primary sequence around the phosphorylation site of their targeting proteins As a result, a number of methods have been developed to identify potential kinase substrates, including combinatorial synthesis of peptide libraries on membrane using the SPOT technology,5 one-bead-one-compound peptide libraries,6 positional-scanning combinatorial libraries,7 and peptide libraries using affinity-column selection.8

More recently, peptide-based microarrays have also been developedfor screening of kinase activity.9,10,11,12 Compared to the SPOT technology, the much higher density

of spots allowed in a peptide array makes it possible for simultaneous screenings of

tens of thousands of kinase substrates on a 3” x 1” glass surface Zhu et al analyzed the

activity of 119 of the 122 yeast kinases using microwells made of PDMS.10 They covalently attached the substrates to the microwells and incubated them with the kinase and radioactively-labeled ATP After phosphorylation had taken place, the kinases and ATP were washed away and the microwells scanned Most of the other

array-based kinase assays rely on standard microscope slides as support Schreiber et

al, in a proof-of-concept experiment, arrayed three kinase substrates onto glass slides

and incubated them with different kinase solutions.9 Incorporation of radioactively labeled ATP was also used for detection A very similar approach was also used by

Falsey et al.11 Surface plasmon resonance was also recently reported for detection of kinase activity.12

3.1.3 Issues to Be Addressed

Peptide-based microarray is one of the most promising technologies for the high throughput screening of protein kinase activity Because of the high density of enzyme substrates arrayed on a microscope slide, one can potentially screen for a wide range of substrates and kinase activity as well as antigen-antibody or ligand-receptor interaction and so on.13,14 However, the development of these technologies is still in its infancy since some issues remain to be addressed For efficient kinase assay to be performed, the kinase substrates need to be arrayed in a site-specific fashion so that efficient and optimum phosphorylation reaction can take place Most peptide and protein arrays

reported so far use either non-specific, covalent immobilization of molecules to the slide or not very stable site – specific immobilization strategies NHS-functionalized

slides and epoxide functionalized slides were respectively used by Zhu et al and Schreiber et al to immobilize kinase substrates via their amine groups, resulting in

random immobilization of proteins onto the glass surface.9,10 New strategies were

developed by Falsey et al as well as Housman et al for site-specific immobilization of

kinase substrates.11,12 However, in the first case, this resulted in an oxime bond which

is relatively unstable, and the five-membered-ring thiazolidine may present immobilized peptides in an unfavorably restricted orientation to interact with their targeting proteins In the second case, the developed strategy required the conjugation

of peptides with an unnatural cyclopentadiene moiety, making it synthetically challenging and not easily accessible Another issue to be addressed is the detection of substrate phosphorylation To date, most of the peptide arrays developed for kinase assay require the use of radioactive 32P for detection of substrate phosphorylation, presenting a serious threat to human health.9,10 Furthermore, the long exposure time (usually hours to days) needed for sensitive detection of 32P upon substrate phosphorylation does not lend itself to high-throughput applications

3.2 Results and Discussion

3.2.1 Fluorescent Antibody-Based Detection of Kinase Substrate Phosphorylation

All peptide arrays developed to date for kinase assay require the use of radioactive 32P for detection of substrate phosphorylation, presenting a potential risk to human health Furthermore, the long exposure time needed for sensitive detection of 32P upon

Fluorescently labeled antibodies have been reported, in a microarray format, for detection of protein/protein, protein/peptide and protein/small molecule interactions9 as well as cell assay.11 The use of poly- and monoclonal antibodies directed against phosphoamino acids have been widely used to detect phosphorylated proteins in gel electrophoresis.15 This method is extremely sensitive since antibodies can detect as little as a few fmol of phosphorylated epitopes.16 In addition, due to the highly specific nature of antibody-antigen recognition, little or no cross-reactivity of one phosphoamino acid antibody (e.g anti-phosphotyrosine) to other phosphoamino acids

or non-phosphorylated amino acids was observed.17

3.2.1.1 Qualitative Fluorescent Antibody-based Detection of Phosphorylated Amino Acids and Peptides

In order for peptide arrays to gain wider popularity for kinase screenings, it is imperative to develop an alternative detection method that poses less health risks than

32P, yet provides similar sensitivity for efficient detection of kinase activity in a microarray format In order to determine whether FITC-labeled anti-phosphoamino acids could be used to detect phosphorylation of kinase substrates in a peptide array,

we first tested their use for detection of phosphorylated amino acids and peptides Both phosphorylated and non-phosphorylated tyrosine amino acids were spotted on an amine functionalized glass slide, and detected using FITC-labeled anti-phosphotyrosine and anti-phosphoserine As shown in Figure 3.1, only the FITC-labeled anti-phosphotyrosine was able to detect the phosphotyrosine immobilized on the slide Neither binding of anti-phosphoserine to phosphotyrosine, nor that of anti-phosphotyrosine to non-phosphorylated tyrosine, was observed We also attempted to

detect the protected phosphotyrosine with both FITC-labeled antibodies and no binding was observed

a) TyrTyr P-TyrP-Tyr b) Tyr Tyr P-Tyr P-Tyr

Figure 3.1 Fluorescent antibody-based detection of phosphorylated amino acids

Phosphorylated (p-Tyr) and non-phosphorylated tyrosine (Tyr) were arrayed onto amine-functionalized slides and probed with (a) FITC-labeled anti-phosphotyrosine and (b) and anti-phosphoserine

The same experiment was then repeated with phosphorylated peptides instead of amino acids Both phosphorylated and non-phosphorylated peptide substrates of p60 tyrosine kinase (YIYGSFK) were synthesized with an additional CGG N-terminal linker for immobilization purpose Both peptides were arrayed onto glass slides functionalized with glyoxylic acid.11 After blocking with BSA, the slides were probed for phosphorylation by incubation with FITC-labeled anti-phosphotyrosine and anti-phosphoserine for an hour Only anti-phosphotyrosine was able to detect the phosphorylated p60 substrate as shown in Figure 3.2, confirming the high specificity

of this antibody-based detection of tyrosine phosphorylation

Figure 3.2 Fluorescent antibody-based detection of phosphorylated peptides

Phosphorylated (P-p60) and non-phosphorylated (p60) substrates (CGG-YIYGSFK) of the p60 tyrosine kinase were arrayed onto slides functionalized with glyoxylic acid and probed with (a) FITC-labeled anti-phosphotyrosine and (b) FITC-labeled anti-phosphoserine

3.2.1.2 Quantitative Fluorescent Antibody-Based Detection of Peptide Phosphorylation

Fluorescent antibodies allows for the rapid screening of phosphorylated amino acids and peptides However, substrate phosphorylation levels must be correlated with measured fluorescence intensity Increasing ratio of phosphorylated to non-phosphorylated p60 kinase substrates were mixed with a combined constant concentration, and were arrayed onto thioester-functionalized slides (see paragraph 3.2.2.2) After a 3-hour incubation, ensuring the total binding of the substrates, the slides were probed with the FITC-labeled anti-phosphotyrosine antibody for 1 hr The intensity of the spots increased as the ratio of phosphorylated substrate to non-phosphorylated p60 substrate increased (Figure 3.3) The fluorescence of any fluorescent substrate is directly proportional to its concentration, and a graph of fluorescence intensity versus the ratio of phosphorylated p60 substrate (Figure 3.4) showed a linear correlation for a p60 ratio ranging from 0 to 100 %, demonstrating the feasibility for on-chip quantitation of phosphorylated peptides using this method

1/100 25/75 50/50 75/25 100/0

1/100 25/75

50/50 75/25

100/0

Phosp60/p60 ratio

1/100 25/75 50/50 75/25 100/0

1/100 25/75

50/50 75/25

100/0

1/100 25/75 50/50 75/25 100/0

1/100 25/75

50/50 75/25

100/0 Phosp60/p60 ratio

Figure 3.3 Fluorescence of increasing ratios of phosphorylated to non-phosphorylated

p60 substrate Increasing ratios of phosphorylated/non-phosphorylated GGC p60 peptides, with a combined constant concentration, were arrayed onto thioester-functionalized slides, incubated for 3 hours, and probed with the FITC-labeled anti-phosphotyrosine antibody for 1 h

Fluorescence intensity vs amount of phosphorylated

substrate

0 5000 10000 15000 20000 25000

Figure 3.4 Fluorescence intensity vs amount of phosphorylated substrate Increasing

ratios of phosphorylated/non-phosphorylated GGC p60 peptides, with a combined constant concentration, were arrayed onto thioester-functionalized slides incubated for

3 hours, and probed with the FITC-labeled anti-phosphotyrosine antibody for 1 hr

3.2.1.3 Fluorescent Antibody-Based Detection of Kinase Activity

Current array-based kinase assays are time consuming due to the long time required for detection A detection relying on fluorescent antibodies is much faster, and a very fast screening of kinase activity can be performed by first incubating the slides with kinase, and subsequently with the fluorescent antibodies (Scheme 3.1)

Scheme 3.1 Very fast screening of kinase activity Glass slides were arrayed with

kinase substrates and after substrate phosphorylation with kinase, the phosphorylation level are rapidly detected with fluorescent antibodies

The two substrates of the serine protein kinase PKA and tyrosine kinase p60, ALRRASLG and YIYGSFK respectively, were synthesized with an additional GGC linker Both peptides were spotted onto the same slides functionalized with glyoxylic acid Following blocking with BSA, the slides were first incubated for an hour with the corresponding kinase (p60 tyrosine kinase for Figure 3.5 a) and PKA kinase for Figures 3.5 b)), and then detected with FITC-labeled anti-phosphotyrosine and anti-phosphoserine Only anti-phosphotyrosine was able to detect the tyrosine kinase activity of p60 Similarly, only anti-phosphoserine was able to detect the serine kinase activity of PKA The complete absence of cross-detection between the two FITC-labeled antibodies demonstrates the high specificity of the antibodies against their corresponding phosphorylated amino acids/peptides

PKA p60 PKA p60

b) a)

PKA p60 PKA p60

b)

Figure 3.5 Detection of kinase activity with FITC-labeled antibodies PKA (left

panel) and p60 (right panel) substrates were arrayed onto glyoxylic acid-functionalized slides Slide a) was incubated with p60 kinase and slide b) with PKA Slides were probed with both FITC-labeled anti-phosphotyrosine and FITC-labeled anti-phosphoserine, incubation of slide a) with only FITC-labeled anti-phosphotyrosine is shown and incubation of slide b) with only and FITC-labeled anti-phosphoserine is shown

The concentration- and time-dependent detection of peptide phosphorylation on chip were then studied Decreasing concentrations (3 mM, 1 mM, 0.3 mM and 0.1 mM) of GGC p60 substrate in PBS, pH 7.4, were arrayed onto a thioester-containing glass slide and incubated with p60 kinase for increasing periods of time (1, 5 and 12 hrs) The slides were incubated with the FITC-labeled anti-phosphotyrosine for 1 hour, washed, dried, scanned (Figure 3.6) and the fluorescence intensity of the spots was measured (Figure 3.7)

3 mM 1 mM 0.3 mM 0.1 mM

3 mM 1 mM 0.3 mM 0.1 mM

Figure 3.6 Antibody-based fluorescence measurement of kinase activity: Decreasing

concentrations (3 mM, 1 mM, 0.3 mM and 0.1 mM) of GGC p60 substrate in PBS, pH 7.4, were arrayed on thioester slides and incubated with the p60 kinase for increasing period (1, 5 and 12 hrs) The slides were incubated with the FITC-labeled anti-phosphotyrosine for 1 hour, washed, dried, scanned and the fluorescence intensity of the spots measured

The concentration-dependent kinase activity was confirmed by plotting the observed fluorescence intensity over the differing concentrations of the peptide spotted on the slide following the same incubation time (5 hrs) with the corresponding kinase (Figure 3.7) It was found that the fluorescence intensity was directly proportional to the concentration of the substrate, showing the feasibility for determination of concentration-dependent kinase activity Using this antibody-based fluorescence detection, kinase phosphorylation was readily detected even with 0.1 mM – which, with a spot size of 1 nL, corresponds to ~0.1 pmol of the peptide substrate By increasing the scanning time or changing the dye used to label the antibody, a 10- to 100-fold lower detection limit should be obtainable

concentration-dependant kinase activity

R 2 = 0.997 0

5000 10000 15000 20000 25000

Figure 3.7 Concentration-dependent kinase activity Peptide solutions of increasing

concentrations were spotted, incubated with the kinase solution for 5hrs and the fluorescence intensity measured

The time-dependent kinase activity was also determined by plotting the observed fluorescence intensity obtained after incubation of the peptide (3 mM spotting concentration) with the kinase solution for increasing period of time (Figure 3.8) The result indicated that the fluorescence intensity as a result of peptide phosphorylation correlates well with kinase incubation time This, together with our previous findings that the amount of phosphorylated substrates is linearly proportional to the fluorescence intensity detected using the FITC-labeled antibodies (Figure 3.4), indicates the feasibility of time-dependent on-chip quantitation of kinase activity Furthermore, in sharp contrast with existing radioactive 32P detection methods, this antibody-based fluorescence detection method is not only safe and highly sensitive, but also requires much shorter time (1 h or less) for detection of phosphorylated amino acids and peptides on chip

Time dependent kinase activity

0 5000 10000 15000 20000 25000

Time (hours)

Figure 3.8 Time-dependent kinase activity Peptide solutions (5 mM spotting

solution) were arrayed, incubated with the kinase solution for increasing periods of time and the fluorescence intensity measured

3.2.2 Developing New Site-Specific Immobilization Strategies for Efficient Based Screening of Kinase Activity

Array-Peptide-based microarray is one of the most promising techniques that allows for potential miniaturized, high-throughput screenings of enzymatic activities, antigen-antibody or ligand-receptor interaction and so on Proteins have well-defined three-dimensional structures, and for efficient screening of protein-protein, protein-peptide, protein-small molecules interactions, or enzyme activity, proteins need to be arrayed in

a site-specific orientation in order to ensure they retain their activity Peptides do not typically possess well-defined 3-dimensional structures, but immobilization of peptides on glass surfaces with a correct orientation is still imperative in order for the peptides to interact effectively with their targeting proteins Most peptide microarrays used for kinase screening use non-specific, covalent immobilization - kinase substrates were attached to NHS functionalized slides9 and epoxide functionalized slides10 via their amine functionalities - thereby failing to address this issue The site-specific ligation of glyoxylyl peptides onto glass slides functionalized with a semi-carbazide

sol-gel layer was recently reported.18 Falsey et al recently reported chemoselective

N-terminal attachment of peptides onto a slide.11 Using glass slides functionalized with glyoxylic acid, they were able to immobilize N-terminally cysteine-containing peptides via oxime-bond or thiazolidine-ring ligation reaction This strategy, while elegantly designed to site-specifically immobilize peptides, may be of limited applications as the oxime bond is relatively unstable, and the five-member-ring thiazolidine may present immobilized peptides in an unfavorably restricted orientation to interact with their

targeting proteins Houseman et al have recently reported the use of Diels-Alder

reaction for site-specific immobilization of peptides onto glass slides.12 However, this method requires the conjugation of peptides with an unnatural cyclopentadiene moiety, making it synthetically challenging and not easily accessible

In order for kinase assay to be performed efficiently and accurately in a microarray format, the substrates need to be attached at their terminal residues via a site-specific immobilization method Two new strategies based on avidin-biotin interaction and on native chemical ligation were hence developed for site-specific immobilization of kinase substrates onto functionalized glass slides

3.2.2.1 Site-Specific Immobilization via Avidin-Biotin Interaction

Taking advantages of the interaction between avidin and biotin, one of the strongest non-covalent known interactions (Kd = 10-15M), one can immobilize N-terminally biotinylated peptides onto a glass slide functionalized with avidin Avidin is also a highly stable protein that maintains its functions even under extremely harsh conditions,19 and therefore is an ideal candidate for slide functionalization Although the biotin-avidin interaction is well studied, it has however never been reported for the site-specific immobilization of peptides onto glass slides

Glass slides were functionalized with avidin by first derivatizing the slides with an epoxide containing silane and then with avidin Avidin is very stable and glass slides can be functionalized in advances and stored at room temperature for long period of times before being spotted

3.2.2.1.2 Application to Kinase Array

The extremely high binding affinity of biotin towards avidin allows for tight and efficient site-specific immobilization of the biotinylated peptide on the glass slide (Scheme 3.2) In addition, since the avidin-biotin binding is instantaneous, only a very short incubation time was needed for the immobilization of the peptide onto the slide,

in contrast with other existing methods which require long incubation times for immobilization to occur.10,11,12 The layer of avidin on the slide surface also provides a molecular layer between the glass surface and proteins interacting with the immobilized peptides, thereby eliminating the BSA blocking and minimizing nonspecific binding on the glass surface

S H

N

H

N O

Scheme 3.2 Site-specific immobilization of biotinylated substrates onto avidin

functionalized slides

To demonstrate the feasibility of this approach, an N-terminally biotinylated Janus Tyrosine Kinase (JAK) substrate with an additional GG spacer (Biotin-GG-KGTGYIKTG) was dissolved in PBS, pH 7.4, and arrayed on the avidin slide After a few minutes of incubation, the slide was washed with PBS, water and then dried Successful detection of the JAK peptide was accomplished by incubation for 1 hour with a Cy3-labeled antibody raised against JAK as shown in Figure 3.9

Figure 3.9 Site-specific immobilization of kinase substrate via avidin-biotin

interaction Biotinylated JAK peptide was arrayed onto a glass slide functionalized with avidin and after a very short incubation, and washing, was probed with a Cy3-labeled antibody raised against this peptide

3.2.2.2 Site-Specific Immobilization via Native Chemical Ligation

3.2.2.2.1 Description

Various chemical ligations were developed in the early 90s in order to help in the synthesis of long non-protected peptides Uniquely reactive functionalities are incorporated into each peptide by chemical synthesis to allow for the site-specific reactions of unprotected peptides This chemical ligation has proven to be easy to implement and a variety of ligation chemistries have been used, however resulting in unnatural groups such as oxime,20 thiazolidine ring,21 thioester However, these

peptide segments On the contrary, the native chemical ligation, developed by Kent, takes place between the N-terminal cysteine of one peptide and the thioester of a second peptide, eventually resulting in the formation of a stable peptide bond (Scheme 3.3)

Scheme 3.3 Principle of Native Chemical Ligation

A feature of the native chemical ligation is that ligation occurs at a unique N-terminal cysteine even if the two peptides contain other cysteine residues Uniquely, the thioester-linked intermediate involving the N-terminal cysteine residue is able to undergo nucleophilic rearrangement by a highly favorable intramolecular mechanism; this step is irreversible and gives a polypeptide product; which is linked by a stable native peptide bond