Using TIRF Microscopy to Understand the Mechanism of the Biginelli Reaction

(1)

Bachelor Thesis Chemistry

Using TIRF Microscopy to Understand the

Mechanism of the Biginelli Reaction

by

Teun IJntema

4

th

_{of July 2017}

Student Number 10717242

Research Institute Supervisor

Van ’t Hoff Institute for Molecular Sciences Prof. Dr. A.M. Brouwer

Research Group Daily Supervisor

(2)

(3)

Abstract

Mechanistic studies have been performed on the Biginelli reaction, the condensation of ethyl acetoacetate, urea and benzaldehyde forming dihydropyrimidones. Previous studies involved trac-ing the reaction with1_{H NMR spectroscopy and mass spectrometry, but they have not been}

suc-cessful in proving the mechanism unambiguously. By using total internal reflection fluorescence microscopy, the reaction can be followed on a single molecular level, which can give new insight into the mechanism. With the use of fluorophoric analogues for benzaldehyde and ethyl acetoac-etate, the association and dissociation to immobilised urea can be displayed. In this project, we found the reaction proceeds via a condensation of urea and benzaldehyde, followed by the reaction with ethyl acetoacetate, according to the proposal made by Kappe.

(4)

Samenvatting

De Biginelli reactie is uitgevonden in 1893 en is sindsdien steeds belangrijker geworden. De reactie producten, dihydropyrimidonen, vertonen een hoge biologische activiteit en worden daar-door veel gebruikt in de farmaceutische industrie. Onder de toepassingen vallen middelen die de groei van tumorcellen hinderen, maar ook helpen tegen bijvoorbeeld epilepsie en malaria. Het nadeel van deze reactie is dat de opbrengst laag is en dat de reactietijd erg lang is.

Om dit probleem op te lossen worden katalysatoren gebruikt, maar in deze reactie is dat lastig. Doordat er drie reactanten (ureum, benzaldehyde en ethyl acetoacetaat) met elkaar reageren is het erg lastig om er achter te komen welke reagentia de eerste reactie vormen. Hier achter komen is van uiterst belang, omdat je daarna je katalysatoren op een rationele manier kan ontwerpen. Dit kan veel tijd besparen in het onderzoek, waardoor de reactie eerder effici¨ent gemaakt worden. Er zijn drie mogelijke eerste reacties: ureum kan eerst reageren met benzaldehyde of eerst met ethyl acetoacetaat of de twee laatstgenoemde moleculen kunnen eerst met elkaar reageren. Volgens de literatuur is de kans het grootst dat ureum als eerste zal reageren met de andere moleculen, dus zal het onderzoek zich richten op deze reacties.

Met behulp van een nieuwe techniek binnen de scheikunde, totale interne reflectie fluorescentie microscopie, kan er op het niveau van individuele moleculen naar deze reactie gekeken worden. Als een fluorescerend molecuul in de buurt van een glazen plaatje komt, wordt er een signaal ontvangen in de detector. In dit onderzoek is ureum vastgezet aan een glazen plaatje. Vervolgens kunnen benzaldehyde en ethyl acetoacetaat daar dan wel of niet aan binden. Deze moleculen vertonen uit zichzelf geen emissie, daarom zijn er groepen aan gezet die dat wel doen. Op deze manier kan er dus gekeken worden naar welk van de twee moleculen liever met ureum kan reageren. Dit zal dan ook de meest voorkomende volgorde zijn in de Biginelli reactie.

In dit onderzoek zijn we er achter gekomen dat ethyl acetoacetaat uit zichzelf niet zal binden aan ureum, maar dat dit pas gebeurd als ook benzaldehyde in het reactiemengsel aanwezig is. Dit betekent dus dat benzaldehyde en ureum eerst met elkaar reageren, voordat de reactie met ethyl acetoacetaat ook mogelijk is.

Schematische weergave van een TIRF microscoop. Fluorofore moleculen in de buurt van het oppervlak vertonen fluorescentie, verder weg niet.

(5)

1 Introduction

Since the discovery of the Biginelli reaction in 1893 by Pietro Biginelli, a lot of research has been carried out; both on the mechanism and the products of the reaction. The one-pot condensation of urea, benzaldehyde and ethyl acetoacetate yields dihydropyrimidones, as shown in Figure 1. These reaction products are of huge interest in the pharmaceutical world. Modification of the substrates results in molecules with altered activity and applications, including anti-tumour activity, potassium channel antagonists, anti-HIV agents, anti-malarial drugs, antibiotics and several more.1

Figure 1. The Biginelli reaction.2

The efficiency of this reaction, however, is low. Brønsted acids catalyse the reaction substantially, but the reaction still takes several days.3Besides the enhancement of the reaction speed, catalysts are needed to control the enantioselectivity as well. The pharmaceutical applications of the dihydropy-rimidines require specific enantiomers instead of racemic mixtures. Organocatalysts have proven to be successful in providing this.3–5_{In order to enhance the reaction speed or the enantioselectivity,}

knowl-edge of the mechanism has to be gained. Several mechanisms have been proposed already, which in terms of reaction sequence comes down to three different mechanisms, which are displayed in Figures 2, 4 and 6. All mechanisms will be discussed in the following paragraphs.

The mechanism displayed in Figure 2 shows the condensation of ethyl acetoacetate and benzalde-hyde as the primary step, followed by the reaction with urea.6 _{The main argument presented was}

that the enone intermediate, as displayed in Figure 2, was prepared by the condensation of the two precursors with piperidine. This was followed by the addition of urea and catalytic hydrochloric acid in a separate step. As the Biginelli reaction traditionally is a one pot reaction, separating both steps will not represent the reactivity for the standard reaction. This might alter the reactivity of the molecules thus also the mechanism.

Figure 2. Reaction mechanism of the Biginelli reaction proposed by Sweet and Fissekis.6

Using thiourea instead of normal urea has shown that this sequence of the reaction is even more unlikely. Under standard Biginelli conditions, the substitution of sulphur for an oxygen results in a product shown in Figure 3a.7When the enolic intermediate (Figure 2) reacts with thiourea, a different product is formed. In this case, the sulphur takes the place of a nitrogen in the ring and forms the product displayed in Figure 3b. This is compelling evidence against the enolic intermediate in the Biginelli reaction, as this results in a different product than the original reaction.

(a) (b)

Figure 3. Reaction products of the Biginelli reaction with thiourea (a) and of the enolic intermediate with thiourea (b).

The second mechanism is displayed in Figure 4, where the aldehyde and urea initially condensate to an iminium intermediate.7 This is followed by the addition of the ethyl acetoacetate, resulting in the dihydropyrimidone. This mechanism has been proven experimentally using1H NMR spectroscopy and mass spectrometry.7,8 Even though no intermediates have been isolated or observed by1H NMR spectroscopy, a bisureide has been detected. This is a byproduct when an excess of urea was present.

(7)

The excess urea reacts with the iminium intermediate, as shown in Figure 5. This is compelling evidence for the iminium pathway, as this molecule can only occur when the benzaldehyde and urea react.

Figure 4. Reaction mechanism for the Biginelli reaction proposed by Kappe.7

Figure 5. Bisureide that is formed when an excess of urea is present in the Biginelli reaction.7

Intermediates have been detected using mass spectrometry in different studies.8 To prevent any fragmentation in the reaction mixture, the soft ionisation method electrospray ionisation (ESI) was used. Continuous snapshots of the reaction mixture have been made using on-line MS. The tan-dem version ESI/MS-MS can characterise the intermediates more profoundly. Little evidence for the pathway described in Figure 2 was found.

Besides the experimental evidence, multiple computational efforts confirm that the iminium route is most likely in terms of energy.8–10 _{The energies of the transition states of the iminium route are}

substantially lower than those of the enone route from Figure 2.

In the computational studies described above, a third mechanism was considered as well. This starts with the condensation of ethyl acetoacetate and urea, followed by the reaction with the aldehyde group, as displayed in Figure 6. The energies of the transition states are supposed to be in between the other two.8 _{Besides the calculations, there is also experimental evidence that this mechanism could}

exist. The 1H NMR- and ESI-MS spectra previously discussed didn’t reveal any intermediates, but later experiments performed by Cepanec et al. did.7,11

Figure 6. Reaction mechanism for the reaction proposed by Cepanec.11

Instead of Brønsted acids, the inorganic Lewis acid SbCl3 was used to perform this reaction. The

enamine intermediate displayed in Figure 6 was isolated in this study, while no proof of any other intermediates was found. Even though this thesis is focused on organocatalysis with Brønsted acids, this sequence of the reaction has to be considered.

The problem of finding the right sequence of the reaction could be solved by using total internal reflection fluorescence microscopy (TIRF microscopy). This novel method has shown its worth in bio-physical applications by e.g. clarifying the walking pattern of myosin V and extending the knowledge of RNA polymerase transcription.12,13It is relatively new in organic synthesis and catalytic reactions, but there is a lot of promise in this field.14

Total internal reflection is the most important physical basis for this experimental method.15This means that all the light shun on a surface is reflected and none of the photons penetrates the reflective surface. There are two physical requirements for this, the first being that the refractive index of the medium where the laser comes from, n3in Figure 7, needs to be higher than that of the other medium,

n1. This is the case for our experiments, as the 3rd medium is glass, which has a refractive index

higher than the solvents used.16 _{The second requirement is for the angle of the incident light needs}

to exceed the critical angle θ(c) of the two surfaces. The critical angle of two media can be calculated by Equation 1, where the n’s are the refractive indices as displayed in Figure 7.

θ(c) = sin-1(n1) (n3)

(8)

If both requirements are met, all the incident light is refracted.15 _{Even though no light penetrates}

the glass cover slip, a small electromagnetic field is created at the interface. This electromagnetic field, called the evanescent field, propagates perpendicular to the interface and covers several hundred nanometers into the solution. This evanescent field is able to excite molecules that are close to the surface. If those molecules are fluorophoric, they can emit light upon excitation, meaning that molecules in the vicinity of the surface light up, while fluorophores further away in the solution remain dark (Figure 7). The background in TIRF experiments is relatively low due to the limited number of excited molecules.

Figure 7. An illustration of the principle of TIRF microscopy. The critical angle (θc _{is exceeded,}

causing the total internal reflection, illustrated with the white arrow. The evanescent field is encircled, exciting the fluorophores near the surface (white), but not far away (black).15

The TIRF microscopy is thus appropriate for observing molecules that are close to a surface.15

In this project, the goal is to exploit that characteristic by attaching a reactive group to the surface. If a fluorophoric molecule attaches to this group, this group emits light from a specific position for a longer time. You can spot this on the microscope by a fluorescent spot on one position. When the fluorophore is released from the surface, the luminescent spot disappears. The result of this is constant blinking of all molecules that attach to and are released from the surface. It is important to note that after disengaging of the fluorophoric groups from the surface, they are still emitting light. However, due to the fast motion of the molecules, the results is an average background emission. This is distributed over many locations near the surface and thus less strong than the emission from one fixed position.

The focus of this project is to understand and clarify the reaction mechanism of the Bignelli reaction. Much earlier research using for example 1_{H NMR- and mass-spectrometry has not been}

conclusive, so other methods have to be used. The newer, less well known experimental method of total internal reflection fluorescence microscopy will be used to understand this mechanism. Therefore, the research question is as follows: Can the reaction mechanism of the Biginelli reaction be clarified using TIRF microscopy?

As explained earlier, the two most likely mechanisms for the Biginelli reaction are the ones proposed by Kappe and by Cepanec (Figures 4 and 6 respectively). This project focuses on distinguishing between those two and looking for a preferred mechanism using TIRF microscopy. Both mechanisms start by a condensation on urea, either with benzaldehyde or ethyl acetoacetate. To exploit the strengths of TIRF, we are immobilising urea on the surface, as this is present in the first step of both mechanisms.

When fluorophoric groups are present on ethyl acetoacetate and benzaldehyde, insight in this reaction can be provided. When looking at the on-time of the blinking, it can be determined which of the two molecules stays associated to the urea for a longer time. This longer on-time provides a larger possibility for a reaction to occur and. When similar concentrations of fluorophores are used, the amount of blinking can be compared, as this is related to the association constants. The absolute association constants cannot be determined, as there is no way of figuring out the total amount of

(9)

immobilised urea molecules on the surface. Both values, association constant as kbindand dissociation

constant as kunbind are displayed in Figure 8

Figure 8. Association and dissociation of a fluorophoric aldehyde on an immobilised urea. The R represents a fluorophoric group, for example one in Figure 9. The kbindand kunbindare the association

and dissociation constants for this reaction.

As an indication that the reaction is successful, we can look at whether any product is formed. The product will be attached to the surface, which results in permanently illuminated spots.

There are, however, several requirements for the use of the fluorophores. To make a comparison between both experiments, the same settings have to be used on the TIRF microscope. This means that the absorption spectra of the two fluorophores have to overlap, as well as the emission spectra. If the absorption spectra do not overlap, a different wavelength has to be used to excite the molecules, which makes a comparison harder. The same holds for the emission spectra, but then different filters have to be used.

Another requirements is a high quantum yield for the fluorophores. This is the fraction of photons that are emitted in respect to the number that are absorbed.17 _{If the quantum yield is too low, the}

signal of the blinking will be too low, causing an insufficient signal to give an accurate representation.

2 Experimental

2.1 Methods

First of all, several basic properties of the fluorophores were measured. These include the absorption and emission spectra, the molar absorption coefficient, the fluorescence quantum yield and the lifetime of the excited state.

For the TIRF experiment, fluorophoric analogues of molecules 1 and 3 have been used. The chemical properties of these molecules differ from the ones in the standard Biginelli reaction, which is why the reactivity in this reaction has to be tested. The progress of the reaction is monitored by measuring the emission spectra, which are expected to change upon the formation of the product.

Several reactions were tested, the reactants and solvents for each individual reaction can be found in Table 1. All were performed under the same conditions and concentrations, as described in the Supporting Information.3_{The reactants that have been used can be found in Figures 1, 9, 10 and 11.}

Table 1. The reactions that have been carried out to check the reactivity of the fluorophores. The reactants can be found in Figures 1, 9, 10 and 11. The concentrations and reaction conditions are described in the Supporting Information.

Reaction Ethyl Acetoacetate Urea Benzaldehyde Catalyst Acid Solvent

1 1 2 4 7 Present DMSO

2 1 2 4 7 Present THF/EtOH

3 1 2 4 No catalyst Not present THF/EtOH

4 1 11 4 7 Present THF

5 1 11 4 No catalyst Not present THF

9 1 2 4 No catalyst Not present THF/EtOH

13 Not present 2 5 7 Present DMSO

(10)

(a) 3-perylenecarboxaldehyde (b) BODIPY aldehyde (c) PMI acetoacetate

Figure 9. The Fluorophores used during this project. Figure (a) shows molecule 4, 3-perylenecarboxaldehyde. Figure (b) shows molecule 5, a BODIPY with a benzaldehyde side group. Figure (c) shows molecule 6, a perylene monoimide with an ethyl acetoacetate side group.

Figure 10. Organocatalysts used for the enantioselective Biginelli reaction.

The TIRF experiment has been performed according to a general procedure, which can be found in the Supporting Information. The reactants that have been used during the TIRF experiment are displayed in Table 2. The concentrations for the reactants are described in the Supporting Information as well.

Table 2. Reactants for the TIRF experiments. The concentrations and the conditions are given in the Supporting Information.

Experiment Ethyl Acetoacetate Urea Benzaldehyde Catalyst Acid Solvent 1 Not Present Immobilised 5 7 Present Toluene 2 1 Immobilised 5 7 Present Toluene 3 6 Immobilised Not Present 7 Present Toluene 4 6 Immobilised 3 7 Present Toluene 5 1 Not Present 5 7 Present Toluene

2.2 Equipment

Figure 11. The substituted urea that will be immobilised on a cover slip.

For the TIRF experiment, a stabilite 2017 laser was used at 476 nm with an average power of 51 mW. An emission filter blocking all light up to 496 nm was used, together with an extra notch blocking the light from 470 nm to 490 nm. For the recording of the microscope image, a Hamamatsu C11440 digital camera was used, recording movies of 1000 frames. For the absorption measurements, a Shimadzu UV 2700 UV-vis spectrophotometer has been used. For the emission measurements, several different spectrofluorometers have been used. For the general properties of the molecules, the Spex fluorolog 3-22 by Horiba Jobin Yvon has been used. Due to technical issues with this machine, a

Fluorolog 3 by Horiba Jobin Yvon has been used to measure the emission spectra for all the reactions in Table 1. This machine was equipped with a synapse CCD dectector. A Shimadzu LC 10-AT liquid chromatograph was used for the HPLC separation.

(11)

3 Results & Discussion

In this section, the results of the project will be displayed together with a discussion of these results. First of all, the basic properties of the fluorophores 4, 5 and 6 will be looked at. Those include the absorption and emission spectra, fluorescence quantum yields and lifetime of the excited state. After that, functionalisation of the cover slips will be examined, together with the emission spectra of the reactions from Table 1. The last part includes the analysis of the TIRF experiment.

3.1 General properties of the Fluorophores

Figure 12 shows the normalised absorption and emission spectra of molecules 4, 5 and 6 in toluene. The lines show the absorption spectra, where the dotted lines depict the emission spectra. The spectra were measured in DCM, DMSO and THF as well, which are included in the Supporting Information. The main difference is that the peaks are less well defined in polar solvents than for the apolar solvents.

Figure 12. Absorption- and emission spectra of molecules 4, 5 and 6 in toluene. The full lines represent the absorption spectra, where the dotted lines depict the emission spectra.

In Table 3 the absorption coefficients for molecules 5 and 6 are given. Those have been measured in duplo, due to the sensitivity of the analysis. As can be seen in the table, a difference between both measurements is observed. The cause of this is the low amount of fluorophores used in the analysis (below 2 mg) and the weighting scale is not very accurate at those values. Taking that into consideration, all duplo measurements for molecule 5 show reasonable similarities, but the ones for molecule 6 are further off.

Table 3. Molar absorption coefficients (ε) · 104 _{at λ}

max for molecules 5 and 6. The λmax are

displayed in brackets following the absorption coefficients. The measurements have been performed in duplo.

Molecule Toluene DCM THF DMSO 5 (1st _measurement) _{12.8 (506)} _{11.5 (504)} _{12.3 (503)} _{11.6 (504)}

5 (2nd _measurement) _{12.0 (506)} _{11.0 (504)} _{11.4 (503)} _{10.4 (504)}

6 (1st _measurement) _{4.59 (505)} _{4.50 (512)} _{3.81 (504)} _{4.51 (514)}

6 (2nd _measurement) _{5.94 (505)} _{6.48 (512)} _{6.07 (504)} _{5.66 (514)}

The fluorescence quantum yields of molecule 6 are displayed in Table 4, together with the lifetime of the excited state. The quantum yield has been calculated according to the method described in literature, with Rhodamine 6 G as a reference.16,18 _{As stated in literature, the quantum yield lowers}

with increasing polarity of the solvent and the same happens for the primary lifetime.17 _{There could}

be two explanations for there being two lifetimes. The first one is that there is an impurity in the sample, but HPLC has shown that this was not the case. The second explanation could be that this is a property of the molecule. The similarities in amplitude across all solvents for the τ1 and τ2 support

(12)

Table 4. Quantum yields and lifetimes (τ ) of molecule 6, measured in several solvents. The excitation wavelength used was 494 nm for the quantum yield measurements. For the lifetime experiments, the excitation wavelength was 488 nm and the emission wavelength was 610 nm.

Solvent Quantum Yield (Φ) τ1 in ns/amplitude τ2 in ns/amplitude

Toluene 75.5 % 4.69 (0.81) 0.28 (0.19) DCM 69.4 % 4.62 (0.79) 0.19 (0.21) THF 70.8 % 4.67 (0.79) 0.23 (0.21) DMSO 62.4 % 4.33 (0.79) 0.68 (0.21)

3.2 Characterisation of the cover slip

Table 5 shows the properties of the functionalised and the clean cover slips. Both cover slips have been cleaned according to the method described in the Supporting Information, the functionalised one has been silinised as well. The goal of performing the three described experiments was to determine if the immobilisation had been successful. The significant difference in contact angle shows that the surface of the cover slips has definitely changed. This is caused by the difference in hydrophilicity between the two surfaces. The TIRF experiment itself shows the most compelling evidence that the surface has been functionalised. Bright spots are visible for the functionalised cover slips, which means the fluorophores are attached to the surface. This is not the case for the clean cover slips, therefore the functionalisation has been successful.

Table 5. Characterisation of the functionalised and clean cover slips. The contact angle images can be found in the Supporting Information. The TIRF image for a functionalised cover slip can be found in Figure 18, the clean cover slip can be found in the Supporting Information.

Clean cover slip Functionalised cover slip Contact angle (o₎ _{11 ± 1} _{32 ± 2}

TIRF measurement No signal Blinking

3.3 Tracking of the Biginelli Reaction

For this part, all the result of the reactions described in Table 1 will be discussed in chronological order. Reaction 1 was performed in DMSO, since this is described as the best solvent in terms of yield and enantioselectivity.3 _{Molecule 4 turns out to be insoluble in DMSO, hence this reaction was}

unsuccessful. THF was described as the second best solvent, so this was chosen for reactions 2 to 9. Urea, however, is poorly soluble in this solvent, so 10 vol% ethanol was added to the THF to solve this problem. Reaction 3 was carried out as a control experiment, with the same reactants but without acid and catalyst.

Figure 13. Normalised emission spectrum for reactions 2 (full lines) and 3 (dotted lines). The excitation wavelength was 490 nm.

The spectra for these two reactions can be found in Figure 13. These spectra give the impression that this reaction is not successful. No shift in the emission spectra is seen and the spectra with and

(13)

without acid and catalyst overlap. The reaction does not proceed without the acid, so the similar spectra show that this reaction is unsuccessful without it as well.3 The only shift that can be seen is from reaction 2 after 4 days, but the change is insignificantly small, so this will not be discussed.

When the reaction is successful, a change in emission for molecule 4 is expected. The conjugated substituent changes drastically upon forming the reaction product, which influences the wavelength of the emission. Not seeing this change is a good indication that this reaction is not possible, but no evidence. This evidence can be obtained by performing HPLC,1_{H NMR or mass spectrometry with}

soft ionisation methods. This was not carried out due to time limitations, which also applies for the other reactions.

Reactions 4 and 5 were performed to check whether the reactivity of the urea molecules with silicon tails (molecule 11) is different from normal urea. The reactants in those reacions were ethyl acetoacetate 1, urea 11 and benzaldehyde 4. Reaction 4 did contain catalyst 7 and trifluoroacetic acid (TFA), where reaction 5 did not. After several minutes a precipitate formed is both samples, most likely due to an aggregation product of the silicon molecules due to a condensation reaction. For this reason, the reaction was not tracked.

Other catalysts (8, 9 and 10) have been tested for the reaction of benzaldehyde 4 with molecules ethyl acetoacetate 1 and urea 2 in reactions 6, 7 and 8. Reaction 9 is a control experiment for those reactions. The emission spectra show no significant change for all cases, which suggests that the Biginelli reaction with molecule 4 is not possible. As this is no evidence that those reactions do not work, it was tried to measure an1H NMR spectrum of the reaction mixtures. The concentrations in which these reactions have been performed turned out to be too low to obtain a good spectrum.

(a) Reaction 6 (b) Reaction 7

(c) Reaction 8 (d) Reaction 9

Figure 14. Normalised emission spectra for reactions 6 to 9 for the excitation wavelength of 450 nm. All spectra are zoomed in on the peaks, the full spectra can be found in the Supporting Information. The normalised emission spectra for reactions 10 and 13 can be found in Figure 15. In both reactions urea 2, benzaldehyde 5, catalyst 7 and trifluoroacetic acid were present. The difference was the presence of ethyl acetoacetate 1 in reaction 10, where it was absent in reaction 13. As shown in Figure 15a, there is a shift of the emission spectrum, as described in literature.3 _{This indicates that}

the reaction is possible with molecule 5 as a replacement for molecule 3. The control reaction 13, which is the same as reaction 13 but without ethyl acetoacetate, shows that this does not happen

(14)

when molecule 1 is absent. This indicates that the Biginelli reaction is successful with benzaldehyde 5 as a fluorophore. The small shift that is visible in Figure 15b might be caused by the condensation of molecules 2 and 5.

Figure 15. Normalised emission spectra of the Biginelli reaction with fluorophores for the excitation wavelength of 490 nm. Both pictures are zoomed in on the peaks, the full spectra can be found in the Supporting Information.

In reaction 11 and 12, ethyl acetoacetate 6, urea 2, catalyst 7 and TFA were used. In reaction 11 benzaldehyde 3 was present, whereas benzaldehyde 5 was present in reaction 12. As shown in Figure 16a, no shift of the emission spectrum was observed for reaction 11. As explained earlier, this doesn’t mean that the reaction is not possible.

(a) Reaction 11

(b) Reaction 12

Figure 16. Normalised emission spectra of reactions 11 and 12 for excitation wavelength 490 nm. The reaction with molecule 6 needed to work for the TIRF experiment to be successful, so extra effort was put in to proving the reaction works. This is why reaction 12 has been carried out, where

(15)

fluorescent analogues for both 1 and 3 were used. Since the absorption of molecule 6 overlaps with the emission of molecule 5, F¨orster resonance energy transfer (FRET) should be observed when the product is formed. This would result in a significant increase in the emission peak around 575 nm, while the peak at 515 nm should decrease in intensity. As shown in Figure 16b, this phenomenon is not observed. The intensities of the peaks relative to each other stays the same, but there is still a small shift for the peak around 515 nm. So FRET is not observed, but there is another indication that the reaction happens.

To make a definitive conclusion, HPLC was performed on the reaction sample. First of all, the HPLC graphs of molecules 5 and 6 were measured separately, they can be found in the Supporting Information. As shown in Figure 17, these two molecules were still present after the reaction, but a new product showed up at 14 minutes. This could very well be the reaction product, but if it was, the BODIPY peak is expected to be more prominent due to its higher absorption coefficient. It still has to be investigated what the cause is of this peak.

Figure 17. HPLC graph of reaction mixture 14. A normal phase column was used with a 2% ethyl acetate in DCM as eluent and a UV-Vis detector. The absorption spectrum of the peak at 14 min is displayed on the left.

3.4 TIRF microscopy

(a) (b)

Figure 18. TIRF figures of the experiments with BODIPY aldehyde 5. (a) Is the image of the experiment 1, with urea 11 on the surface, BODIPY aldehyde 5, catalyst 7 and TFA. (b) is the image of experiment 2, where ethyl acetoacetate was added to the reactants of experiment 1. The R is BODIPY. A time trace of those experiments is included in the Supporting Information

(16)

For the final part of the discussion, the results of the TIRF microscopy experiment will be discussed. The first experiment was performed to look at the association of BODIPY aldehyde to the immobilised urea, in presence of catalyst 7 and TFA. As shown in Figure 18a, a signal was found for these two reactants. Some of the spots were on for a time period of up to multiple seconds, as shown in the Supporting Information. This might indicate the formation of the condensation product of urea and molecule 5, as shown in Figure 18a. However, most of the dots were merely blinking in the time frame of hundreds of milliseconds.

For experiment 2, a new cuvette with ethyl acetoacetate 1, urea 11 on the cover slip, benzaldehyde 5, catalyst 7 and TFA was observed under the TIRF microscope. When comparing the videos of this experiment to the ones of experiment one, it was clear that the amount of blinking molecules did not increase. What did happen was that the luminescence of the spots was more intense. The product of this experiment, which is shown in Figure 18b, has a higher fluorescence quantum yield than the reaction product of experiment 1.3 _{Permanently illuminated spots appeared as well. The lifetime of}

this luminescence was up to 50 seconds, after which bleaching killed the fluorescence.

More telling were the TIRF experiments with PMI acetoacetate 6. This fluorophore was present in reactions 3 and 4, together with urea 11 immobilised on the surface, catalyst 7 and TFA. The difference can be found in the presence of benzaldehyde 3 in reaction 4, where it was absent in reaction 3. As shown in Figure 19a, no signal is observed in this reaction. This means that the is no association between those molecules and that condensation does not work and there is little to no attraction between them.

As shown in Figure 19b, permanent fluorescent sport were observed after the addition of ben-zaldehyde 3 in reaction 4. This observation suggests that for the product to be formed, urea and benzaldehyde need to react first. Ethyl acetoacetate reacts with the condensate and thus does not participate in the first step.

(a) (b)

Figure 19. TIRF figures of the experiments with PMI acetoacetate 6. (a) is experiment 3 with urea 11 on the surface, PMI acetoacetate 6, catalyst 7 and TFA. (b) is experiment 4, where benzaldehyde was added to the reactants of experiment 3. The R is perylene monoimide. A time trace of this experiment is included in the Supporting Information

In the introduction, more objective ways of identifying a preferred reaction path were discussed, such as the association and dissociation constant. Those values were not obtained during this exper-iments, as not enough molecules were blinking. This means not enough data could be gathered on the values. Besides that, a vast amount of the blinking molecules were not bright enough to analyse, especially for experiments 3 and 4.

The reason for both limitations could be found in the that the silinisation was not optimal. The method used has been optimised for another molecule instead of molecule 11. The silinisation has to be optimised for this molecule itself. Another reason is that the fluorescence quantum yield for molecule 5 is lower than optimal. This means that not all excited molecules actually fluoresce, so the amount of of illumination is limited.

(17)

4 Conclusion

From the TIRF microscopy experiment the conclusion can be drawn that the association between molecule benzaldehyde and urea is the first step of the Biginelli reaction. The fluorophoric analogue of benzaldehyde shows blinking on surface on which urea has been immobilised. This happens im-mediately after the addition, which implies that there is an attractive force between these molecules. After one hour, a condensation product in the form of a permanently illuminated spot was found as well. Both these phenomenon were not the case for the fluorophoric analogue for ethyl acetoacetate with the immobilised urea.

The main argument for this sequence of the reaction is that permanently illuminated spots are found after the addition of benzaldehyde to the fluorophoric analogue of ethyl acetoacetate and the immobilised urea. This means that benzaldehyde reacts with the immobilised urea first, followed by the attachment of the fluorophore. Therefore, the correct mechanism for the Biginelli reaction is the one proposed by Kappe.7

5 Outlook

The main issue we encountered during this project was the low amount of blinking molecules during the TIRF experiment. The solution for this problem can be found in three parts. First of all, the silinisation method has to be optimised for molecule 11. The method that was used has been optimised for an amine instead of the urea that was used here. Optimisation should result in a higher amount of immobilised urea on the surface, which will improve the amount of molecules that can attach to the surface during the experiment. Another solution could be the use of different fluorophoric groups with a higher fluorescence quantum yield. This results in more emission when a molecule is attached thus a better signal. Simply using a higher concentration of fluorophore could also solve the problem. The problem with the last two methods is that the background signal will also increase, so the first method is recommended.

Once the problem of the low signal has been solved, other, more objective measurements can be carried out. As described in the introduction, the goal was to find dissociation and reaction constants. Those values can give a better insight in the reaction mechanism than the comparative study we have done so far.

Another interesting characteristic would be getting to know how much silanised groups are present on the surface. If this is known, the association constants can be measured instead of comparing them using similar concentrations of fluorophores. This, together with the measurement of the dissociation constant, could actually prove which of the sequences of the reactions is more likely to be the right one.

Finally, what could still be done is checking if the reactions from Table 1 have actually failed or succeeded. Measuring the emission spectra could indicate that a change or reaction has happened, but is no unambiguous proof. This could be done by performing 1H NMR spectroscopy or mass spectrometry on the reaction samples.

(18)

Acknowledgements

First of all, I would like to thank prof. dr. Fred Brouwer for giving me the opportunity to do this fascinating project to finish of my bachelor. The project has been very inspirational and I enjoyed my time thoroughly. Secondly, I would like to thank my daily supervisor Dongdong Zheng for his help every day, the interesting discussions and finally for great company during the last three months. I would also like to thank Dina Petrova, Michiel Hilbers and Chia-Ching Huang for the help with the TIRF-, lifetime- and emission measurements and Hans Sanders and Mina Raeisolsadati Oskouei for the preparation of the fluorophores. I would also like to thank Nicole Oudhof for the excellent company, the free coffee and reviewing my thesis. For the last reason, I would like to thank Christiaan van Campenhout and Rosa Brakkee as well. My final gratitude is for the whole Molecular Photonics group for the great atmosphere during and after the working hours. I really enjoyed my stay at this amazing research group.

(19)

References

[1] Sandhu, J. S.; Suresh, Arkivoc 2012, Part (i), 66–133. [2] Biginelli, P. Gazz. Chim. Ital 1893, 23, 360–416.

[3] Raeisolsadati Oskouei, M. Fluorogenic Organocatalytic Reactions. Ph.D. thesis, University of Amsterdam, 2017. [4] Chen, X.-H.; Xu, X.-Y.; Liu, H.; Cun, L.-F.; Gong, L.-Z. J. Am. Chem. Soc. 2006, 128, 14802–14803.

[5] Wang, Y.; Yang, H.; Yu, J.; Miao, Z.; Chen, R. Adv. Synth. Catal. 2009, 351, 3057–3062. [6] Sweet, F.; Fissekis, J. D. J. Am. Chem. Soc. 1973, 95, 8741–8749.

[7] Kappe, C. O. J. Org. Chem. 1997, 62, 7201–7204.

[8] De Souza, R. O.; da Penha, E. T.; Milagre, H.; Garden, S. J.; Esteves, P. M.; Eberlin, M. N.; Antunes, O. A. Chem. Eur. J. 2009, 15, 9799–9804.

[9] Ma, J. G.; Zhang, J. M.; Jiang, H. H.; Ma, W. Y.; Zhou, J. H. Chin. Chem. Lett. 2008, 19, 375–378. [10] Lu, N.; Chen, D.; Zhang, G.; Liu, Q. Int. J. Quantum Chem. 2011, 111, 2031–2038.

[11] Cepanec, I.; Litvić, M.; Filipan-Litvić, M.; Grüngold, I. Tetrahedron 2007, 63, 11822–11827.

[12] Yildiz, A.; Forkey, J. N.; McKinney, S. A.; Ha, T.; Goldman, Y. E.; Selvin, P. R. Science 2003, 300, 2061–2065. [13] Kapanidis, A. N.; Margeat, E.; Ho, S. O.; Kortkhonjia, E.; Weiss, S.; Ebright, R. H. Science 2006, 314, 1144–1147. [14] Zhang, Y.; Song, P.; Fu, Q.; Ruan, M.; Xu, W. Nat. Commun. 2014, 5, 4238.

[15] Axelrod, D. Methods Cell Biol. 2008, 89, 169–221.

[16] Refractive Index. http://macro.lsu.edu/HowTo/solvents/Refractive%20Index.htm, accessed on 29th_{of June 2017.}

[17] Lakowicz, J. R., Ed. Principles of Fluorescence Spectroscopy; Springer US: Boston, MA, 2006; pp 205–235. [18] W¨urth, C.; Grabolle, M.; Pauli, J.; Spieles, M.; Resch-Genger, U. Nat. protoc. 2013, 8, 1535–1550.

[19] Suhina, T.; Weber, B.; Carpentier, C. E.; Lorincz, K.; Schall, P.; Bonn, D.; Brouwer, A. M. Angew. Chem. 2015, 127, 3759–3762.

(20)

Supporting Information

General Procedure 1: Tracking the Biginelli Reaction

Ethyl acetoacetate (1.3 · 10-5 _{mole, 1 eq), benzaldehyde, (1.3 · 10}-5 _{mole, 1 eq) and urea (1.3 · 10}-5

mole, 1 eq) were dissolved in 0.2 mL solvent. For the cases where the fluorophoric analogues of ethyl acetoacetate and benzaldehyde were used, the same amount was added as for the standard molecules. Afterwards, a catalyst from Figure 10 (2.6 · 10-6_{mole, 0.2 eq) and trifluoroacetic acid (2.6 · 10}-6_mole,

0.2 eq) were added. For the control experiments, this step was not carried out. The reaction was run at 35o_{C with the use of a water bath. The emission spectrum was measured immediately after adding}

the TFA or the last component in the case of the control experiments and several days of reacting onwards.

General Procedure 2: Cleaning and Silanisation of Cover Slips

19

The cover slips were washed in 0,3 % Extran AP 12 solution by sonication for 30 minutes at 40o_C,

followed by sonication in deionised water for 10 minutes and in ethanol for 30 minutes. The cover slips were dried in an oven for 1 hour and further cleaned in an ozone photoreactor for 2 hours. The cuvettes and stops used in the TIRF experiment were cleaned according to the same procedure. Afterwards, the cover slips were silanised with 2% (volume) 1-[3-(Trimethoxysilyl)propyl]urea in 96% ethanol in which 2% water was added. The pH of this solution was adjusted to approximately 5 by addition of acetic acid. A teflon rack with cover slips was kept for 1 hour in this solution with stirring. The cover slips were sonicated afterwards for three times in ethanol (20 minutes in total) and washed with acetone and DCM subsequently. Finally, the cover slips were dried in air and put in an oven for 3 hours at 110oC to keep them clean.

General Procedure 3: TIRF microscopy experiment

Cuvettes were glued to the functionalised cover slips using Devcon two component epoxy glue. The first step for all experiments was checking the microscope’s image for just the solvent. 0.2 mL of the solvent was deposited in the cuvette and observed under the microscope. The second step was the addition of the reaction components, as displayed in Table 2. The concentrations in the cuvettes were as follows: the concentration of the fluorophores (molecule 5 or 6) was 10-10M, of the other reactant (molecule 1 or 3) it was 10-4M and for the catalyst and the trifluoroacetic acid 2·10-5M. The reaction mixture was observed using the TIRF microscope immediately and after one to two hours.

(21)

Graphs and Images

Figure 20. Absorption- and emission spectra of molecules 4, 5 and 6 in DMSO. The full lines represent the absorption spectra, where the dotted lines depict the emission spectra.

Figure 21. Absorption- and emission spectra of molecules 4, 5 and 6 in DCM. The full lines represent the absorption spectra, where the dotted lines depict the emission spectra.

Figure 22. Absorption- and emission spectra of molecules 4, 5 and 6 in THF. The full lines represent the absorption spectra, where the dotted lines depict the emission spectra.

(22)

(a) (b)

Figure 23. Contact angle of a functionalised and a clean cover slip (Figures (a) and (b) respectively).

Figure 24. TIRF image of a cover slip cleaned according to the method described in the Supporting Information. No silinisation was performed. Benzaldehyde 5 was used as a fluorophore.

(23)

Figure 26. Histogram of the lifetime measurement of molecule 6 in DCM, excited at 610 nm.

Figure 27. Histogram of the lifetime measurement of molecule 6 in THF, excited at 610 nm.

(24)

(c) Reaction 8 (d) Reaction 9

Figure 29. Emission spectra for reactions 6 to 9. The excitation wavelength was 450 nm.

(25)

Figure 31. Emission spectrum for reaction 13, excited at 490 nm.

Figure 32. HPLC graph of benzaldehyde 5. A normal phase column was used with a 2% ethyl acetate in DCM as eluent and a UV-Vis detector. The absorption spectrum of the peak at 3 minutes is displayed in small.

Figure 33. HPLC graph of benzaldehyde 6. A normal phase column was used with a 2% ethyl acetate in DCM as eluent and a UV-Vis detector. The absorption spectrum of the peak at 6.5 minutes is displayed in small.

(26)

Figure 34. Time trace of a blinking spot during TIRF experiment 1.

Figure 35. Time trace of a permanently fluorescing spot until bleaching occured during TIRF experiment 2.

Figure 36. Time trace of a permanently fluorescing spot until bleaching occured during TIRF experiment 4.

Using TIRF Microscopy to Understand the Mechanism of the Biginelli Reaction

Bachelor Thesis Chemistry