Microelectrode Arrays, Electrocatalysis, and the Need for Proper Characterization

Abstract

Indirect electrochemical methods are a powerful tool for synthetic chemistry because they allow for the optimization of chemical selectivity in a reaction while maintaining the advantages of electrochemistry in terms of sustainability. Recently, we have found that such methods provide a handle for not only the synthesis of complex molecules, but also the construction of complex, addressable molecular surfaces. In this effort, the indirect electrochemical methods enable the placement or synthesis of molecules by any electrode or set of electrodes in a microelectrode array. The success of these surface-based reactions are typically evaluated with the use of fluorescence labelling studies. However, these fluorescence-based evaluations can be misleading. While they are excellent for determining that a reaction has occurred in a site-selective fashion on an array, they do not provide information on whether that reaction is the one desired or how well it worked. We describe here how the use of a “safety-catch” linker strategy allows for a more accurate assessment of reaction quality on an array, and then use that capability to illustrate how the use of transition metal mediated cross-coupling reactions on an array prevent unwanted background reactions that can occur on a polymer-coated electrode surface. The method enables a unique level of quality control for array-based transformations.

Introduction

While modern synthetic chemistry has played a key role in the development of many new pharmaceuticals, lead compounds, and molecular probes for investigating biological pathways, it has played a much smaller role in the synthesis of the complex molecular surfaces that are central to the development of new medical diagnostics and bioanalytical tools. These devices are composed of three key elements; a solution phase molecular recognition event, a method for detecting, quantifying, and recording the molecular recognition event, and a polymer, monolayer, or alternative surface that provides the interface between the first two features.^1,2 The surface on the device must be stable with respect to time, and it must be compatible with both the chemistry needed for its construction and its deployment in subsequent analytical studies. In the end, it is the nature of this surface and the molecules that can be placed or synthesized on it that defines the scope of biological problems that can be studied with any given device.

Microelectrode arrays have great potential to serve as devices for molecular diagnostics.^3–9 They contain thousands of electrodes that can each be used to detect and record a change in current triggered by a molecular binding event.^10,11 In such experiments, one of the molecular participants involved in the binding event is fixed to the surface of the array by an addressable electrode or set of electrodes and the other participant is allowed to float over the entire surface of the array. Since the arrays contain thousands of electrodes, multiple potential binding partners for any given target or family of targets in solution can be placed on an array and monitored in a single analytical experiment. Of course, this can only be accomplished if the surface bound participants in the study can be located on the array proximal to individual addressable electrodes in the array and then kept at those locations.

To date, a variety of approaches have been used to address this challenge. Current state-of-the-art approaches range from the use of self-assembled monolayers^12–22 to the direct functionalization of carbon electrodes.²³ While these methods have advantages, they can have limited stability (SAMs) and take advantage of only a small subset of the synthetic chemistry toolbox used to build molecules of medicinal relevance. The result is that our existing tools for building surfaces on diagnostic devices lack the chemical versatility needed to tackle more-complex problems. Consider the construction of an addressable molecular library containing molecular probes that range from DNA-aptamers, to peptide derivatives, to small organic molecules for the purpose of detecting multiple metabolites in solution.^24–28 Such a surface would ideally have each of the molecules in the library placed on the array in triplicate so that statistical data on the interactions being monitored can be obtained and utilize multiple electrodes for each molecule in the library so that the signal obtained can be amplified and error reduced. The experiment requires the selective functionalization of hundreds to thousands of electrodes in a high density array with multiple different molecules. The construction of this type of library requires both that surface to be stable and compatible with the use of more than one or two selected synthetic methods.

Fortunately, microelectrode-array based approaches are not limited to a small subset of organic synthetic methods or a single approach to the surface supporting those reactions. The electrodes in the array are compatible with conducting a wide variety of mediated electrochemical processes that work by using the electrode to make or recycle chemical reagents, catalysts, or substrates.²⁹ Such reactions can be used to generate acids, bases, nucleophiles, electrophiles, transition metal catalysts, oxidant, reductants, hydrogen gas, etc. In turn, those reagents and catalysts can all be confined to selected electrodes in the array (Figure 1). The approach is straightforward.

Figure 1.

A strategy for conducting “site-selective” reactions on a microelectrode array

The electrode is used to oxidize or reduce a “precatalyst” in order to make a catalyst (or chemical reagent) at the surface of the electrode, and a confining agent that destroys that catalyst (or chemical reagent - usually by reducing it or oxidizing it back to the precatalyst) is added to the solution above the array. The rate of catalyst (or reagent) generation at the electrode is controlled (by adjusting the current at the electrode) relative to the rate of its destruction in the solution above the array to make sure that the reaction only occurs at the selected electrode. In this way, the electrochemical method and confinement strategy are used to locate the reaction at selected sites on the array. With this approach, much of modern synthetic chemistry can be utilized to construct a functionalized surface on a microelectrode array.^30,31

With that stated, it is important to note that building a highly functionalized surface on an array requires not only development of the necessary reactions, but also a method for assessing how well those reactions work. To date, this assessment has mainly been done with fluorescence labelling studies. In these studies, the substrate for a reaction is labelled with a fluorescent group and then placed on the surface of the selected electrodes in the array by the synthetic method being developed. The intensity of fluorescence by the selected electrodes and the absence of fluorescence at sites remote from the electrodes is used to determine how well the reaction was confined to the electrodes employed for the placement reaction. The method is very useful and the images it generates are straightforward to analyse. However, fluorescence studies do not actually indicate how well the reaction worked or even if it took place at all. They only indicate that the fluorescent label wound up exclusively by the selected electrodes. One then assumes the substrates were placed at those sites by the desired reaction. That assumption may not be accurate.

Consider the chemistry outlined in Scheme 1. The reactions shown were conducted as part of an effort to develop electrochemically cleavable protecting groups for use in array-based library synthesis. The arrays were coated with a diblock copolymer containing an arylbromide moiety,^32,33 and then a substrate containing a reductively cleavable protecting group was placed by every electrode in the array using a Cu(I) cross-coupling reaction.^11,34 The plan was then to cleave the protecting group by using nitrobenzene as an electrochemical mediator. Initially, the reaction looked like it worked well. For the deprotection reaction, a block of twelve electrodes was used to reduce nitrobenzene to the radical anion. The radical anion was confined to the electrodes used for its generation by the presence of water in the solution above the array. To this end, the water protonated the radical anion from nitrobenzene before it could migrate to a neighbouring electrode. Following the deprotection reaction, the entire array was used for a base-catalysed esterification reaction with a fluorescently labelled NHS-ester.³⁵ When the resulting array was examined using a fluorescence microscope, the polymer above the electrodes selected for the deprotection reaction were clearly labelled with the fluorescent group (Scheme 1b). However, this image was very misleading. A negative control experiment that took a picture of the array following the deprotection reaction, and prior to the coupling reaction with the fluorescent label, also showed fluorescence by the selected electrodes (Scheme 1c), apparently from a reduction product from nitrobenzene complexing to the polymer coating on the array. Subsequent solution phase reactions showed that the desired deprotection reaction did not proceed at all using the mediated electrolysis conditions, and it appeared clear that the first fluorescent image taken of the array (Scheme 1a) was almost certainly an entirely false positive result.

Scheme 1.

a) The overall reaction. b) An image of the array taken following the two step deprotection, coupling strategy. c) A control experiment taken following the deprotection reaction prior to the coupling reaction.

Background reactions like the one shown in Scheme 1 are not the only issue that can lead to a misleading fluorescence image. Consider the chemistry shown in Scheme 2.^34,36 In this study, an acrylate-based Michael acceptor was placed by a checkerboard pattern of electrodes in a microelectrode array having 12,544 electrodes/cm². The pattern was generated for a block of twenty electrodes by using the electrodes selected as cathodes to reduce Vitamin B₁₂ to a radical anion that then deprotonated methanol to form a base. The base catalysed an esterification reaction between an alcohol in an agarose coating on the surface of the array and an activated ester in solution. Excess activated ester in solution was used to confine the reaction to the selected electrodes by consuming the methoxide generated at the cathode before it could migrate to the neighbouring electrodes in the array. The resulting array was then incubated for 45 minutes with a fluorescently labelled peptide containing a cysteine C-terminal amide. This led to a thiol-based Michael reaction that placed the peptide and the fluorescent label onto the surface of the electrodes selected for functionalization. A check on the array using a fluorescence microscope clearly showed the desired checkerboard pattern. The array was then exposed to a second reaction to place the acrylate Michael acceptor at a remote location. This time, a dot in a box pattern was used for the placement reaction. The array was then placed in a PBS buffer solution at a pH of 7 with no additional peptide. Following a 45-minute incubation period, the array was again examined using a fluorescence microscope and the image shown in Scheme 2 was taken. The thio-Michael reaction proved to be reversible and some of the peptide from the checkerboard pattern migrated to the dot in a box pattern.

Scheme 2.

The reversibility of array-based thio-Michael reactions.

The ramifications of this migration precluded any use of the array for monitoring more than one peptide at a time. If the second pattern had been used to place a second peptide on the array, then the dot in a box pattern would be the result of the new peptide being placed on the array and some peptide that migrated from the original site of the checkerboard pattern. In a similar fashion, the electrodes associated with the original pattern would contain both the original peptide and some peptide from the new pattern. A fluorescence study done following each placement reaction would not detect the presence of these mixtures; mixtures that would invalidate any subsequent biological signalling study. An image of the array taken with a fluorescence microscope would simply show fluorescence nicely confined to both patterns.

The failure of fluorescence-labelling to provide an accurate view of the chemistry in these early studies led to questions about the Cu(I)- and Cu(II)-transition metal mediated reactions that we used to replace the thio-Michael reaction as the method of choice for the functionalization of a microelectrode array.^29,30 Were those transition metal reactions as clean as we believed them to be, or were they also plagued by background reactions? Fortunately, the answer to the second question is no. Transition metal mediated electrolyses do provide an excellent method for the site-selective functionalization of a high-density microelectrode array.

Results and Discussion:

Background Fluorescence and Site-Selective Cross-Coupling Reactions

The conclusion stated above was not initially obvious. Early fluorescence-based studies suggested that the transition metal mediated cross coupling reactions were also problematic. Consider the control-experiment shown in Scheme 3. In this reaction, a Cu(I)-based cross coupling reaction between an aryl bromide on the surface of an array and a fluorescently labelled alcohol nucleophile (pyrene butanol) was conducted.^11,34 In the first of two experiments (Scheme 3a), a small square of four electrodes was used to conduct a negative control for the reaction. This was done by replacing the pyrene butanol nucleophile needed for the cross coupling reaction with pyrene. The Cu(I) needed for the reaction was generated at the four electrodes by using the electors as cathodes and then confined to those electrodes with the use of oxygen in solution. The electrodes used for the experiment were cycled on and off to aid in confinement of the reaction. This works by slowing Cu(I)-generation at the electrodes so that the oxygen-based solution phase oxidation can keep up with the amount of catalyst generated. Following the reaction, the electrodes used for the electrolysis showed fluorescence even in the absence of any nucleophile. No fluorescence was observed on the remainder of the array showing that the confinement strategy worked beautifully. The cathodic electrolysis must have modified the polymer surface over the electrodes employed in a manner that allowed the polymer to absorb pyrene at those sites. On the same array, a larger 6X6 box of electrodes was then used to conduct the actual coupling reaction with the pyrene butanol nucleophile. An image of the array following this step showed the 6X6 pattern and a brighter pattern for the four electrodes that were supposed to be a negative control. It appeared that some of the nucleophile used for the reaction on the 6X6-pattern wound up by the first set of electrodes used. This suggestion was consistent with an experiment that reversed the order of the two reactions (Scheme 3b). In this case, the reaction using the 6X6 pattern of electrodes and the nucleophile was conducted first. This reaction was then followed by the negative control using the small 4-electrode square pattern with pyrene replacing the alcohol nucleophile. In this case, the image for the background reaction using the small 4-electrode square pattern was slightly less intense than the image observed by the electrodes in the 6X6 pattern. Again, no fluorescence was observed at any electrode not used for the electrolysis.

Scheme 3.

An initial control experiment lacking the solution phase nucleophile. All four reactions contained 25 mM of the Cu-catalyst.

Efforts to understand the nature of the background reaction began with an experiment to rule out involvement of the arylbromide. To this end, the chemistry was repeated using a polymer coating that lacked the arylbromide altogether (Scheme 4). Once again, the two square patterns were used. This time, the reaction employing the 4-electrode square pattern contained neither the fluorescent nucleophile nor the Cu-precursor needed for catalyst generation. The second reaction involving the 6X6 square pattern of electrodes utilized both – although no cross-coupling reaction was possible since there was no arylbromide on the surface of the electrodes. Once again, fluorescence was observed by each electrode used in both patterns. Clearly, the electrolysis reaction in the absence of both the Cu-catalyst and the arylbromide was modifying the polymer coating on the surface of the electrodes.

Scheme 4.

A control experiment lacking the surface bound coupling partner. The second reaction was run with 25 mM Cu-mediator using a 6×6 pattern of electrodes.

At this point, we speculated that in the absence of the Cu-mediator the hydrogen evolution reaction (HER) was occurring at the cathodes. The HER would lead to the formation of both hydrogen gas and hydroxide from the water solvent used in the reaction. The hydroxide produced could hydrolyze the polymer crosslinking groups that are attached to the polymer through ester linkages (Scheme 3). It was proposed that this loss of crosslinking led to a more porous polymer surface that better adsorbed the pyrene group. Since the cathodic reaction was required for the generation of hydroxide, the modification of the polymer only occurred next to electrodes selected for the reaction.

The appearance of a background reaction was not restricted to reduction reactions involving Cu(I)-based chemistry. The Cu(II)-mediated Chan-Lam coupling reaction between an arylborate ester surface on the array and solution phase nucleophiles used to set up a number of array-based biological studies also showed a background reaction in control experiments.^31,37,9 For these reactions, the selected electrodes in the array are used as anodes to generate the Cu(II)-mediator needed. The Cu(II)-mediator is confined to the electrodes used for its generation by adding excess nucleophile to solution. For a coupling reaction employing an alcohol as the nucleophile, the Cu(II)-mediated oxidation of the alcohol nucleophile in solution will consume Cu(II) in the solution above the array. In this way, the Cu(II) generated at any electrode in the array cannot migrate to a neighbouring electrode. In the experiment shown in Scheme 5, the 4-electrode square pattern was again used as a negative control that exposed the electrodes to the reaction conditions in the absence of any copper. Pyrene butanol was used as a solution phase nucleophile. Even without the Cu-mediator and no possibility for a Chan-Lam coupling reaction, the reaction gave rise to fluorescence at the electrodes. No fluorescence was observed at electrodes not used for the electrolysis showing that the initial polymer coating did not undergo any background reaction in the absence of the anodic oxidation reaction. The second image shown in Scheme 5 shows the same array after a second coupling reaction was run with a 6X6 pattern of electrodes, this time with the Cu-mediator present. In this experiment the fluorescence associated with the Chan-Lam coupling reaction having the Cu-mediator present was much more intense than the control experiment. The background reaction did appear to be minor relative to what was observed for the Cu(I)-based reactions.

Scheme 5.

Background fluorescence from a Cu(II)-Chan Lam Coupling Reaction. a) The background reaction without Cu. b) The Chan-Lam coupling reaction.

Yet while the intensity of product placed on an electrode using the Cu (I) and Cu(II) mediated reactions was more intense at the electrodes that had copper present, the increased intensity did not mean that at those electrodes none of the background reaction occurred. So, how much of the observed fluorescence at electrodes where the Cu-mediator was used was the result of the background reaction? Would these background reactions cause cross-contamination between the electrodes as in the Michael chemistry example that would interfere with a subsequent biological study? After all, in the Cu(I) experiments it appeared that the surface of the polymer was modified to a point where non-specific adsorption events were occurring. These are questions that the use of fluorescence studies cannot answer.

The Use of Safety-Catch Linkers and the Chan-Lam Coupling Reaction

One of the advantages of using microelectrode arrays as a platform for the analysis of molecular interactions is the synthetic handle the electrodes themselves provide. The electrodes can be used to not only place or build molecules at any site on an array and to analyse interactions between those molecules and targets, but also selectively recover for characterization the molecules associated with any electrode in the array. The last task is done with the use of a “safety-catch” linker.^38,39 “Safety-catch” linkers contain both an ester moiety to attach a molecule to a solid support and a masked nucleophile that is positioned either five- or six-atoms away from the carbonyl of that ester. Unmasking of the nucleophile leads to a cyclization reaction with the nucleophile attacking the carbonyl in an addition step that is followed by an elimination step the removes the molecule from the solid support while forming either a lactone or lactam product depending on the nucleophile used. The first array-based efforts along these lines utilized t-Boc protected alcohols as the masked nucleophile.³⁸ To cleave the “safety-catch” linker, selected electrodes in an array functionalized with the linker were used as anodes to generate acid. This led to deprotection of the alcohol, lactone formation, and cleavage of whatever molecule was located proximal to the selected electrodes from the array. The lactone product was then characterized by LCMS and compared to independently synthesized material. Since the acid for the deprotection reaction could be confined to the electrodes where it was generated, the method allowed for recovery and characterization of the molecules associated with any electrode or set of electrodes in the array.

A second generation approach took advantage of the oxidatively cleavable “safety-catch” linker strategy illustrated in Scheme 6.³⁹ The chemistry is illustrated in the context of the current studies with the linker placed on the array using a Cu(II)-mediated Chan-Lam coupling. The linker itself contains a monosubstituted olefin that is used to mask an alcohol nucleophile. Cleavage of the molecule from the array is accomplished with an electrochemically driven cis-hydroxylation reaction. The osmium oxidant needed for the cis-hydroxylation is generated using electrodes in the array and then confined to those electrodes by the addition of 4-phenyl-1-butene to the solution above the array. This solution phase olefin will undergo a rapid cis-hydroxylation reaction and in so doing consume any of the Os(VIII)-reagent that diffuses away from the electrode where it was generated. The cleavage strategy is highly selective and orthogonal to a wide variety of functional groups.

Scheme 6.

An oxidation based safety-catch linker strategy. a) A fluorescence image taken of the array after the step 1 with the background control represented by the 2X2 pattern in the middle. b) A fluorescence image of the same array after OsO₄ cleavage of the “safety-catch” linker.

The “safety-catch” linker used in Scheme 6 and the rest of this study was selected because it was straightforward to prepare (Scheme 7). The synthesis started with an N-acylation of allylglycine followed by ester formation on the C-terminus with a mono t-butyldimethylsilyl protected 1,4-dihydroxybutane. Removal of the silyl protecting group led to the pyrene labelled safety-catch linker.

Scheme 7.

Preparation of “safety-catch” linker substrates.

The substrate shown in Scheme 7 was used as a test nucleophile for both the Cu(II)-mediated Chan-Lam chemistry and the Cu(I)-catalyzed cross coupling reactions described above. As mentioned earlier, it was the Cu(II)-mediated Chan-Lam reaction that was used for the sequence illustrated in Scheme 6. In this example, the Cu(II)-mediated reaction was run for 1800 cycles (0.5 s on and 0.1 s off), a length of time that was optimal for placing the substrate on the surface of the electrodes. The experiment with the Cu-mediator present was conducted with a 6X6 pattern of electrodes. The background control without the Cu-mediator was then conducted using a 4-electrode square pattern of electrodes inside the large box. In the negative control, no Chan-Lam coupling reaction was possible. Once again, examination of the array using fluorescence microscopy showed that the fluorescence intensity associated with the electrodes where the Cu-mediator was present was much greater than that observed for the electrodes where the Cu-mediator was not used, but the background reaction did occur (Scheme 6a).

At this point, all of the electrodes used for the initial experiments were employed ro a cis-hydroxylation reaction (Scheme 6b). This reaction was run for 3,600 cycles in order to ensure that cis-hydroxylation of the double bond in the safety-catch linker proceeded to completion. HPLC analysis was used to compare the product lactone cleaved from the array with independently synthesized material, a comparison that confirmed that the lactone cleaved from the array had the desired structure.

Following cleavage of the “safety-catch” linker, no fluorescence remained on the array where the Cu-mediated Chan-Lam coupling had been conducted. In addition, little if any background fluorescence was observed at the electrodes used in the 4-electron control group. This result suggested that the background reaction that occurred under the oxidative Cu(II)-mediated conditions was not a non-specific binding event involving the pyrene. It involved placement of the linker on the surface in a manner that still allowed the cis-hydroxylation derived cleavage reaction to clip the pyrene from the surface.

This observation was consistent with the generation of acid from the oxidation of DMF or adventitious water in the absence of the Cu-mediator. Acid generated at the selected electrodes in the absence of copper could catalyse a reaction between the alcohol nucleophile in the safety-catch linker substrate and the borate ester on the surface of the array (Scheme 8). The resulting product would then undergo an oxidative cleavage of the “safety-catch” linker in the same manner that a substrate placed on the surface with a Cu(II)-mediated Chan-Lam coupling reaction would. If this were the case, then the background reaction would effectively accomplish the same goal as the desired Chan-Lam coupling reaction – selective placement of the intact substrate on the array. However, the acidic reaction was worrisome in the same way the earlier thiol-based Michael chemistry was because of its reversibility.

Scheme 8.

An acid catalysed background reaction during a Chan-Lam coupling reaction.

Two control experiments were conducted to both support the mechanistic suggestion forwarded in Scheme 8 and determine how significant this background reaction is when the Cu-mediator for the Chan-Lam coupling reaction is present. In the first (Scheme 9), the same pattern was established for an array by using the Cu(II)-mediated Chan-Lam to add a fluorescent alcohol to a borate ester coated array. The central pattern for the background reaction was then generated at four electrodes by conducting the reaction in the absence of the Cu-mediator. As usual only faint fluorescence was observed at the electrodes used for the background reaction. The electrodes in both patterns were then used as anodes to oxidize diphenyl hydrazine and generate acid. This was done in the presence of pinacol in order to reverse the acid catalysed background reaction shown in Scheme 8. The reaction led to a significant drop in fluorescence at the four central electrodes used for the background reaction with only a trace reduction in fluorescence at electrodes in the 6X6 pattern where the Cu(II)-mediated reaction had been conducted. This result was consistent with both the background reaction being an acid-catalysed transesterification reaction and that background reaction being minimal at electrodes that employed the Cu(II)-mediator. The result is consistent with the oxidation of Cu(I) to Cu(II) occurring more readily than the oxidative generation of acid at an anode.

Scheme 9.

Chan-Lam control experiment one.

In the second control experiment (Scheme 10), the experiment shown in Scheme 6 was repeated. The resulting array was then incubated in a solution of DMF with pyrene overnight to see if the surface of the array had been damaged in a manner that would lead to non-specific adsorption of pyrene at electrodes used for the electrolysis. The reaction led to no observance of fluorescence at the electrodes used for either the Chan-Lam coupling or the negative control experiment. So in direct contrast to the Cu(I)-reduction chemistry, no modification/decomposition of the polymer occurred during the oxidative process that led to the non-specific binding of pyrene.

Scheme 10.

Chan-Lam control experiment two.

In each case (Cu or no Cu), no fluorescence was ever seen at an electrode not used for the electrolysis - whether that electrode had been used in a previous experiment or not. Once a molecule is placed by an electrode using a Chan-Lam coupling reaction and that electrode is turned off, no additional molecules are added to that site. We did not incubate a functionalized array with acid to see if substrate deliberately placed on the array using the background reaction would migrate to other sited on the array because we know the background reaction is minimal in the presence of the Cu-mediator. Hence, use of the Chan-Lam coupling reaction avoids this issue altogether.

In the end, it was clear that the Chan-Lam coupling reaction can be used to construct a surface on an array with individual molecules placed by individual electrodes without any risk of their mixing post synthesis or significant background reactions. These observations render the Chan-Lam coupling reaction a leading candidate for building array surfaces that contain multiple molecular recognition elements.

A Re-evaluation of Cu(I)-Mediated Electrochemical Cross-Coupling Reactions

With those observations in place, attention was turned back to the Cu(I)-catalyzed cross coupling reactions. The key question here concerned the amount of the background reaction at electrodes where the actual cross-coupling reaction had been conducted. Was the desired reaction fast enough to exclude the background reaction? The answer to this question would be yes if the rate of Cu(II) to Cu(I) reduction at the electrode was sufficient to suppress the generation of base.

To test this idea, the chemistry shown in Scheme 11 was pursued. The overall format of the experiment was the same as that used to examine the Chan-Lam coupling reaction, except in this case the electrodes were used as cathodes in order to generate the Cu(I)-catalyst needed. As earlier, a small 2X2 pattern was used to probe the background reaction with a larger 6X6 pattern used to probe the desired Cu(I)-mediated process. The experiments were run on 12K-arrays since the synthetic chemistry using Cu(I) works better on the 12K-array (see below for a potential explanation). The result of the experiment shown in Scheme 11 was surprising in that the reaction led to no background fluorescence at the small 2X2 square pattern of electrodes used for the control without copper even when the brightness on the array was amplified to a great extent (the second image provided). The reaction utilizing both the copper catalyst and the nucleophile proceeded exactly as expected.

Scheme 11.

A 12K-array and the lack of a background reaction when the electrodes are used as cathodes. Both images are of the same array using different imaging intensities.

It is important to recognize that there is a significant difference between the reaction setup used to conduct reactions on a 1K-array where the background reaction was initially observed (Scheme 3) and that used for the 12K-array shown in Scheme 11.³⁵ 1K-Arrays are inserted into a reaction vial along with a remote Pt-wire as the counter electrode. Since reactions happen on the surface of the array and the Pt-wire is far away from that surface, the reactions are for all practical purposes divided cell electrolyses. Acid generated at the anode will not impact the surface of the array or any base initiated side reaction on that surface. The 12K-arrays are thin film flow cells where the anode and cathode are separated by only 1.5 mm. They are for all practical purposes undivided cells where the acid generated at the anode can influence the chemistry occurring at the cathodes in the array. A lack of background reaction with the 12K-arrays in Scheme 11 is consistent with the earlier suggestion that in the absence of a Cu-catalyst, the cathodic-reaction leads to the formation of base that then initiates side-reactions that alter the polymer coating on the array. With a 12K-array setup, acid from the anode would neutralize the base generated and prevent the background reaction. The presence of this acid may also be the reason that Cu(I)-cross coupling reactions proceeded better on the 12K-arrays. On a 1K array, the generation of base in the absence of acid from the anode might lead to hydrolysis of the ester linkers in the polymer, an overall reduction of crosslinking in the polymer, and removal of the polymer from the surface of the array. The result would be a loss of fluorescence on the electrodes that would trick one into thinking the cross-coupling reaction had not proceeded well. The presence of acid from the anode in a 12K-array reaction would prevent this complication.

Evidence that the Cu(I)-placement reaction had worked as planned for the 6X6 box of electrodes was obtained by taking advantage of a safety catch linker (Scheme12). The fluorescence image of the array taken after this cleavage reaction showed complete removal of fluorescence by the functionalized electrodes (the image in the center of the Scheme). In addition, HPLC analysis of the solution above the array following the cleavage reaction showed the expected lactone in a manner identical to that observed for the Chan-Lam coupling reaction.

Scheme 12.

Searching for a background reaction on a 12K-array used as a cathode.

The only way to see the background reaction for the Cu(I) process on the 12K-array was to utilize a very high potential for the reaction and then to turn the brightness of the microscope up to enhance the image. This is how the images shown in Scheme 12 were generated. Note how in this case a faint image for the background reaction could be seen (the fluorescence image in the left of the scheme). The image for the 6X6 pattern was much brighter since the Cu(I)-catalyst was present for the experiment run on those electrodes. After leavage of the “safety-catch” linker, it was difficult to see the background reaction at the 2X2 square in the middle of the pattern. At this intensity for the image, fluorescence due to non-specific binding of pyrene to the polymer itself was seen at every electrode not used in the experiment, and it is possible that at this intensity for the image this fluorescence obscured the background reaction. The very high intensity for the image was used to illustrate that there was essentially zero background reaction or adsorption of pyrene at the electrodes used for the actual Cu(I)-mediated cross coupling reaction.

The lack of even non-specific adsorption of pyrene onto the polymer coating the electrodes used for the cross-coupling reaction was not surprising since the cross- coupling reaction placed a hydrophilic safety-catch linker onto the surface of the electrode. The more hydrophilic surface decreases interactions between the polymer at those sites and pyrene. To support this claim, the rest of the polymer coating the array was converted to a hydrophilic surface by exchanging the bromide on the original polymer with a borate ester group using a Pd(0)-catalyzed cross coupling reaction.³³ Washing the array then led to the image included on the right hand side of Scheme 12 that showed no pyrene at any of the electrodes. Clearly, the hydrophilic polymer did not bind pyrene.

The chemistry shown in Schemes 11 and 12 demonstrate that the Cu(I)-catalyzed cross-coupling reaction proceeded nicely on a 12K-array with no evidence for a background reaction. So like the Cu(II)-mediated Chan-Lam coupling reaction, the Cu(I)-catalyzed reaction was ideally suited for placing molecules onto microelectrode arrays for subsequent signalling studies. At the present time, both reactions are employed.

Conclusions:

While initial fluorescence imaging approaches to understanding microelectrode array-based transition metal cross coupling reactions were misleading, the use of a safety-catch linker strategy enabled a much more thorough analysis that alleviated the initial concerns. Oxidative reactions were ideally suited for borate ester polymer coated arrays because the transformations left the core polymer structure undisturbed. In the absence of a Cu(II)-mediator, a reversible acid catalysed transesterification reaction did occur with alcohol nucleophiles. The substrate itself was not altered and underwent subsequent reactions as expected. In the presence of the Cu(II)-mediator, the background reaction was minimal. The complimentary Cu(I)-mediated coupling reactions utilizing an arylbromide based polymer coating on the arrays were more complicated. In these reductive reactions, a background reaction in the absence of the Cu(I)-mediator did destroy the polymer backbone leading to an irreversible adsorption of fluorophores into the polymer at any site used for the electrolysis. Fortunately, this background reaction was completely supressed in the presence of the Cu(I)-mediator. Hence, the reductive Cu(I)-cross coupling reaction can also be employed to synthesize functionalized arrays for biological studies. Finally, the use of the safety-catch linker strategy for evaluating the quality of array based reactions sets the stage for the exploration of new transformations and how we evaluate the efficiency and yield of those reactions. This foundation will be essential as we move toward total synthesis efforts that enable array to be utilized for monitoring binding events involving larger, more complex molecular libraries.

Experimental Section:

Microelectrode arrays.^11,34

The arrays and the power supply were purchased from CustomArray Inc. The ElectraSense^® reader used to conduct array-based reactions was made by Combimatrix Co (now CustomArray).

Fluorescence microscopy.³⁹

Fluorescence images were carried out using a Nikon Eclipse E200 microscope and a Nikon D5000 camera with X-Cite Series 120 Q as the burner. Optical filters used with the microscopy were as follows: CFW-BP01-Clinical-000 (Semrock) filter cube excitation 380–395 nm, emission 420–470 nm, ET-GFP (FITC/Cy2) (Chroma) filter cube excitation 450–490 nm, emission 500–550 nm, and TeRed-A-Basic-000 (Semrock) filter cube excitation 540–580 nm, emission 590–670 nm.

Diblock copolymer.

Both the arylbromide version and the p-tolyl version of the diblock copolymers were synthesized by Dr. Qiwei Jing. For a detailed discussion, see the supporting information of Hu, L. B.; Bartels, J. L.; Bartels, J. W.; Maurer, K.; Moeller, K. D. J. Am. Chem. Soc. 2009, 131(46), 16638–16639

Sample procedure for spin-coating arrays with the diblock copolymer.

The microelectrode arrays were spin-coated with a spin-coater MODEL WS-400B-6NPP/LITE (for 1-K arrays) or MODEL WS-650MZ-23NPPB (for 12-K array). The chip was inserted into a socket in the spinner and adjusted to be horizontal, then three drops of 0.03 g/mL (3 wt%) block copolymer solution in 1:1 xylene/THF were added onto the chip covering every electrode. The chip was then spun 1000 rpm for 40 s. The coating was allowed to dry for 15–20 min and subjected to irradiation using a 100W Hg lamp for 20 min before use.

General procedure for Cu(I) mediated cross-coupling reaction on 1-K array.¹¹

The alcohol substrate (8.0–15.0 mg) and 8.0 mg Bu₄NBr were dissolved in 100 μL of DMF in a 1.5 mL Eppendorf tube. Six μL of a 25 mM CuSO₄ solution in DMF and 6 μL of a 50 mM triphenylphosphine solution in DMF were added. The resulting solution was further dissolved into 1.5 mL of a 7:2:1 mix of MeCN/DMF/H₂O. A 1-K array chip was incubated in this solution. The array was set as cathode by setting it negative relative to the remote Pt-wire. A pattern for the active electrodes was then selected and turned on. The electrodes in the array were pulsed at a specific voltage relative to the remote Pt wire, cycling 0.5s on and 0.1s off for specific number of cycles. The array was then washed with EtOH and examined using a fluorescence microscope once it was dry.

General procedure for Cu(II) mediated Chan-Lam reaction on 1-K array.³⁷

The alcohol substrate (15 mg) and 75 mg Bu₄NPF₆ were dissolved in 1.5 mL DMF in a 1.5 mL Eppendorf tube. A Cu(OAc)₂ solution (75μL of 25mM in water) was then added. The solution was stirred for at least 4h before use. A 1-K array chip was incubated in the solution. The array was used as an anode by setting it at a positive potential relative to the remote Pt-wire. A pattern of electrodes was selected for the electrolysis and then those electrodes. The electrodes in the array were pulsed at a specific voltage relative to the remote Pt-wire, cycling 0.5s on and 0.1s off for specific number of cycles. The chip was then washed with EtOH and examined with a fluorescence microscope once it was dry.

Diol exchange reaction on 1-K array.³³

A mixture of 10 mg pinacol, 80 mg Bu₄NPF₆, 100 mg diphenylhydrazine and 100 μL pyridine were dissolved in 1.5 mL methanol. The electrodes in the array were used as anodes. In this case, the selected pattern of electrodes was set at a potential of +3.0V relative to a remote Pt wire. The electrodes were then cycled for 0.5s on and 0.1s off for 900 cycles. The chip was then washed with EtOH and examined with a fluorescence microscope once it was dry.

Cis-dihydroxylation reaction on 1-K array.³⁹

The procedure was modified from literature.³⁹ To a stirred solution of AD-mix (0.14 g, 0.7 g/mmol substrate) in a 1:1 mixture of t-BuOH/H₂O (2 mL, 0.1 M) in an Eppendorf tube was added 30μL 4-phenyl-1-butene, 400 mg K₂CO₃, and 20 mg MeSO₂NH₂. The mixture was stirred overnight at room temperature until both phases were clear. The tube was put in an ice bath and the solution stirred during the course of the experiment. A 1-K array was incubated in this stirred solution. The electrodes in the array were then used as anodes by setting them to a voltage of +2.0 V relative to the Pt-counter electrode. This was done for 3600 cycles (0.5 sec on and 0.1 sec off). After reaction, the array was repeatedly washed with a sat. Na₂SO₃ solution, DMF and EtOAc before examining the surface with a fluorescence microscope once it was dry.

General procedure for Cu(I) mediated cross-coupling reaction on 12-K array.¹¹

The solution was prepared the same way as for 1-K array, except for using safety-catch linker as the substrate. For this reaction, 18μL CuSO₄ DMF solution and 18μL PPh₃ DMF solution were added to the solution instead of 6 μL used previously. The array was placed into the ElectraSense^® reader after fitting it with a cap to make a thin film flow cell over the array. The cap was coated with Pt and served as the counter electrode. At this point, 110–120μL of the reaction solution was added to gap between the array and the cap with the use of a syringe. The electrodes in the array were used as cathodes by setting their potential to a negative value relative to the Pt-electrode cap. The electrodes in the array were turned on at a specific voltage for a period of 90 s and then turned off again for 180s. The reaction was repeated 4 times and then the array was washed with EtOH and examined with a fluorescence microscope once it was dry.

General procedure for Cu(II) mediated Chan-Lam reaction on 12-K array.³⁷

The reaction solution for the Chan-Lam coupling was prepared the same way as described above for the reaction run on a 1-K array. The array was covered with a Pt-coated cap, placed into the ElectraSense^® reader, and then 110–120μL of the reaction solution added to the space between the array and the cap. The electrodes in the array were then used as anodes by setting them to a potential of +2.4 V relative to the auxiliary electrode. The electrodes were turned on for 30 s and then off again for 10 s for a total of 20 cycles. The array was washed with EtOH and examined with the use of a fluorescence microscope once it was dry.

Cis-dihydroxylation reaction on 12-K array.³⁹

The reaction solution was prepared as described above for the experiment run on a 1-K array. The array was covered with a Pt-coated cap, placed into the ElectraSense^® reader, and then 110–120μL of the reaction solution added to the space between the array and the cap. Electrodes in the array were then used as anodes by setting them to a potential of +2.0V relative to the Pt counter electrode for a period of 90 s. The electrodes were then turned off for 180 s. The reaction was repeated 8 times, and then the array was repeatedly washed with a sat. Na₂SO₃ solution, DMF and EtOAc before examination using a fluorescence microscope once it was dry.