Production and characterization of fructose-6-phosphate aldolase mutants

(1)

Uppsala University

Bachelor Thesis

Production and characterization of fructose-6-phosphate aldolase mutants

Author:

Yannick Hajee

Supervisors:

Mikael Widersten Gina Chukwu

June 9, 2020

(2)

Abstract

Aldolases are enzymes that offer a green and uniquely stereo-selective method to catalyze aldol reactions, reactions which are used to form carbon-carbon bonds in synthetic organic chemistry. This makes their use very attractive in the synthesis of complex, chiral and biologically active compounds such as pharmaceuticals. Fructose-6-phosphate aldolase (FSA) is a promising aldolase since it does not need phosphorylated donor substrates and is surpris- ingly thermostable. Understanding the reaction mechanism of the enzyme helps in making improved versions of it, but aside from a catalytic lysine residue it is not yet clear which amino acid residues contribute to catalysis in FSA and how. Tyrosine 131, glutamine 59 and threonine 109 have been suggested but not definitively proven to contribute. This project set out to help elucidate the role of these residues by producing FSA mutants containing Q59A, T109A and Y131F substitutions respectively and analyzing their catalytic properties under the steady state and where applicable pre-steady-state phases of the reactions. FSA mutants containing Q59A and T109A substitutions respectively were successfully produced and expressed in Escherichia coli. FSA Q59A was purified to a total of 180 mg of highly pure enzyme. The enzyme kinetics were investigated spectrophotometrically using coupled reactions. FSA Q59A was somewhat slower in the aldol addition reaction and more than seven times slower in the retroaldol reaction than wild-type FSA, confirming that Q59 plays a role in the reaction. The half-saturation point for dihydroxyacetone is twice as high and the half-saturation point for fructose-6-phosphate a little over one fourth as high in FSA Q59A as in wild-type FSA. The enzyme’s pre-steady-state kinetics were also investigated with stopped-flow methods but the signal was not clear enough to be useful.

(3)

1 Abbreviations

2TY A rich bacterial growth media, see Buffers and Media

6-Pgdl 6-Phosphonoglucono-delta-lactone, alternatively D-6-phosphoglucono-1,5-lactone A₂₈₀ Absorption over 1 cm at 280 nm, a measure of protein concentration

DERA 2-deoxyribose-5-phosphate aldolase

DHA Dihydroxyacetone

DNA Deoxyribonucleic acid E. Coli Escherichia coli e-flasks Erlenmeyer flasks F6P Fructose-6-phosphate

FSA Fructose-6-phosphate Aldolase G3P Glyceraldehyde 3-phosphate

G6PDH Glucose-6-phosphate dehydrogenase

GPDH Glycerol-3-phosphate dehydrogenase, alternatively α-glycerophosphate dehydrogenase GPI Glucose-6-phosphate isomerase

Gro3P Glycerol-3-phosphate

HA Hydroxyacetone

IMAC Immobilized Metal Ion Affinity Chromatography

LB Lysogeny broth (or other alternative names). See Buffers and Media.

LB-Amp LB with the antibiotic ampicillin.

MQ Milli-Q water, which is very pure water

OD₆₀₀ Optical density at 600 nm, a measure of cell density

O/N Overnight

PCR Polymerase Chain Reaction

pUC A type of bacterial plasmid backbone or the associated sequencing primers.

rpm Rotations per minute

SDS-PAGE Sodium dodecyl sulphate – polyacrylamide gel electrophoresis

TEA Triethanolamine

TIM Triose-phosphate isomerase

wt Wild-type, the version of an enzyme that is present in nature.

(5)

2 Introduction

2.1 Aldolases and FSA

Aldol reactions (see Figure 1 for a generalized aldol reaction) are a powerful way to form carbon- carbon bonds in synthetic organic chemistry. Aldolases are enzymes that offer a green and uniquely stereo-selective method to catalyze these aldol reactions. This makes their use very attractive in the synthesis of complex, chiral and biologically active compounds such as pharmaceuticals.^{1, 2} Yield of such compounds could theoretically be four times as high in chiral reactions using stereo-selective aldolases as when using non-stereo-selective conventional organic synthesis methods. Furthermore the need to separate the different enantiomers of a racemic product could be reduced or eliminated which cuts cost and saves time.

Figure 1: Reaction formula for a generalized aldol addition reaction. Up to two new chiral centers (*) can be formed during this reaction.

Fructose-6-phosphate aldolase (FSA) is an enzyme that catalyzes aldol addition between a non- phosphorylated donor and an aldose acceptor.³ Two isoenzymes exist, FSA A and FSA B, and here FSA A is studied. FSA is found naturally in E. coli but its physiological role is not known.³ It can accept a (somewhat narrow) variety of substrates but the reaction that it gets its name from is the reversible aldol addition of dihydroxyacetone (DHA) and glyceraldehyde 3-phosphate (G3P) which yields fructose-6-phosphate (F6P), see Figure 2.¹ It is the first reported enzyme that can catalyze this reaction, and one of two enzymes, besides 2-deoxyribose-5-phosphate aldolase (DERA) that can catalyze the aldol addition of two aldehydes.¹

Figure 2: Reaction formula for the reaction catalyzed by FSA that is investigated in this project. The forward reaction is called the aldol addition reaction and the backwards reaction is called the retroaldol reaction.

FSA has a catalytic lysine residue which generates a Schiff base intermediate after reacting with a ketone substrate. The Schiff base intermediate then deprotonates the imine alcohol, which transforms it into a nucleophilic hydroxyenamine that reacts with electrophilic aldehyde substrates.¹

(6)

A few characteristics make FSA particularly interesting as a bio-catalyst. The quarternary struc- ture of FSA is a decamer which consists of two pentameric rings, where each sub-unit forms a typical TIM barrel.¹ It is stabilized by non-covalent forces and this makes FSA unusually thermostable for a mesophilic (from organisms that prefer moderate temperatures) protein.¹ Unlike many other aldolases, such as fructose-1,6-bisphosphate aldolase in glycolysis, FSA does not need phosphorylated donor substrates, which gives it a lot of freedom in which kinds of compounds it could make. Yet, it has a relatively narrow substrate scope but it has already been engineered to accept new substrates or a generally larger scope of substrates.^{1, 4}

In order for it to be an attractive option to use an aldolase for the synthesis of a specific compound, a few requirements need to be met. Firstly, the aldolase needs to be able to catalyze the reaction in question, the aldol addition of the specific desired substrates. Aldolase enzymes, like most enzymes, are selective in which compounds they can use as their substrates. Secondly, characteristics such as reaction rate, stability and stereo-selectivity must be sufficiently high. High stability increases the useful life of the enzyme, and thus increases the total amount of product each enzyme molecule can make during its lifespan. High reaction rate similarly increases the total amount of product an enzyme can produce during its lifespan and saves reaction time. The stereo-selectivity of aldolases is one of their selling points but the fractions of different enantiomers produced can differ between aldolases.⁵ Usually a more stereo-selective aldolase is better because it can reach a higher yield of a specific enantiomer and it saves time and cuts cost by reducing the need for separation of the enantiomers.

The wild-type enzymes can be used as a starting point for making different aldolases with optimized characteristics for specific reactions. Enzymes with slightly different amino acid sequences could offer better substrate specificity, stereo-selectivity, stability, reaction rate, etc. Different strategies can be employed to get these improved enzymes. Rational design of an improved version of a protein is an option, which requires you to understand or at least be able to simulate how changes in specific residues will impact the properties of the protein. Another option is to use artificial evolution, which consists of repeated rounds of introducing random mutations to the protein and selecting for the variants with desirable traits. Different combinations of rational design and artificial evolution can be employed as well, such as using artificial evolution on an enzyme which has already been improved with rational design,⁵ or restricting random mutations to specific residues which are known to be important.¹

In order to do this rational or semi-rational design, there needs to be an understanding of how the enzyme works. In the case of FSA, there are good models for what the reaction mechanism looks like but still more research needs to be done on the details.^{1, 3, 4, 6} See Figure 3 for an illustration of the active site of FSA. The catalytic lysine residue at position 85 is the main actor in the catalysis and covalently attaches to the substrate. Y131 has been proposed to act as a general acid/base with a catalytic water molecule also hydrogen bonded to Q59 and T109 supposedly participating in proton transfer during the reaction.^{1, 3, 4, 6} Earlier studies have shown decreased activity towards F6P for FSA with substituted Q59, T109 and no activity for FSA with substituted Y131.^{1, 3, 4} However, FSA Y131F is reported to have retained activity towards different substrates and hydrogen-bond breaking mutations of Q59 do not affect that reaction negatively either, which casts doubt onto how catalytically important these residues are. The authors hypothesize that another residue might partially take over the function of Y131 in this case.⁴ Further investigation of the kinetic effects of substitutions at Q59, T109, Y131 is warranted.

(7)

Figure 3: An illustration of the active site of FSA during the first step in the catalysis. Adapted from a larger published figure.³

2.2 Aim

Aside from the catalytic lysine residue it is not yet clear which amino acid residues contribute to catalysis in FSA and how. Tyrosine 131, glutamine 59 and threonine 109 have been suggested but not definitively proven to contribute. The aim of this project is to help elucidate the role of these residues by producing FSA mutants containing Q59A, T109A and Y131F substitutions respectively and analyzing their catalytic properties under the steady state and where applicable pre-steady-state phases of the reactions. As can be seen in Figure 4, these substitutions constitute a loss of the functional groups that can hydrogen-bond or act as a general acid-base, which are the supposed roles of these residues in the reaction.

Figure 4: An illustration of the different amino acids to show the change in side-chain functional groups the substitutions bring.

(8)

2.3 Site-directed mutagenesis

The Quikchange site-directed mutagenesis kit has a good figure in its manual⁷ explaining the process behind it, see Figure 5. It uses a polymerase chain reaction (PCR) to copy the template plasmid and incorporate oligonucleotide primers containing the desired mutations in the new plasmids. The old template DNA that does not have the desired mutation is different from the newly synthesized DNA in two respects. It does not have a ’nick’ and it is methylated. If the methylated DNA is transformed into E. coli along with the new mutated DNA the cell will recognize the methylated DNA as the older ’original’ DNA and reverse the mutation in the new DNA. In order to reliably get transformants with the mutated plasmid, the Dpn I restriction endonuclease enzyme is used to specifically digest only the methylated DNA. Then a transformation is done and the E. coli cells repair the nicks in the mutated plasmids. The successful transformants are selected by exposure to an antibiotic that the plasmid gives resistance to and if all went well there is a transformant and the process is finished.

Figure 5: An illustration of how site-directed mutagenesis works, taken from the Quikchange site- directed mutagenesis kit manual.⁷

2.4 Protein expression

E. coli is the go-to model organism because it is so easy to work with, and different strains are optimized for specific uses. E. coli XL1-Blue is optimized for successful transformation with

(9)

mutated DNA and is a part of the QuikChange site-directed mutagenesis kit. The mutant enzymes are not expressed in E. coli XL1-Blue however, but in E. coli BL21. The extra transformation step is needed because that strain is optimized for expression and works well with the pGEM-3z-(f+) plasmid that was used in this case. This plasmid contains a very powerful sp6 promoter that requires the bacterial host to have the sequence for viral T7 RNA polymerase in its DNA. E. coli BL21 has this and L-(+)-Arabinose triggers the expression of the viral T7 RNA polymerase and consequently the expression of the mutant.

2.5 Coupled reactions

The kinetics of FSA were measured using coupled reactions. This means that the products of the reactions that FSA catalyzes were measured indirectly. Two additional enzymatic reactions translated an increase in [F6P] into an increase in [NADPH] (Figure 7) or an increase in [G3P]

into a decrease in [NADH] (Figure 6). NADH and NADPH absorb light strongly at 340 nm and, using a spectrophotometer, changes in their concentration can easily be measured in real time in solution. This is ideal for measuring the initial reaction velocity of the FSA retroaldol or aldol addition reaction.

Figure 6: Reaction formula for the retroaldol reaction including the coupled reaction. The compound in red, NADH, is measured by spectrophotometry.

Figure 7: Reaction formula for the aldol addition reaction including the coupled reaction. The compound in red, NADPH, is measured by spectrophotometry.

It would be possible to directly measure F6P, G3P or DHA, but it would require a lot more work.

They could be detected using liquid chromatography or possibly molecular absorption spectroscopy but those techniques cannot be performed in real-time while the reaction is occurring. It would require quenching many reactions after different times to get enough data points to determine the reaction rate at one substrate concentration, which can be determined with one easy measurement using the coupled reactions.

There are some points to keep in mind when using coupled reactions, however. Firstly, the reaction rate that is being measured should not depend on how fast the coupled reactions are. Therefore the coupled reactions need to proceed at a much faster rate as compared to the aldolase catalyzed

(10)

reactions in order not to be rate-limiting.⁸ An associated consideration is that the stated coupling enzyme concentration in any literature cannot be trusted to be high enough. The exact amounts of enzymes used in the measurements are given in this thesis but the actual activity of your enzymes can vary a lot depending on age. Therefore whether or not the coupled reactions are rate-limiting at Vmax should always be investigated and the amount of coupling enzymes increased if they are. Secondly, it takes some time for the coupled reactions to reach equilibrium, which is visible as a ’lag’ phase in the beginning of the measurement where the change in absorbance does not yet reflect the change in product of the reaction of interest.^{8, 9} See Figure 8 for one of the FSA Q59A aldol addition absorption graphs, an example of a measurement with a lag phase.

The reaction rate should be measured after the lag phase, in the period where the reaction rate is constant.

Figure 8: FSA Q59A aldol addition absorption graph, an example of a measurement with a lag phase.

The measurement is in blue, and a straight line in red shows when the reaction has reached equilibrium.

A third consideration applies to kinetics measurements in general, not just to those that use coupled reactions. The kinetics measurements done here measure the initial reaction rate, which is measured as the slope of the absorbance curve. The reaction rate must not change significantly during the measurement period in order for the slope to be a reliable measurement of the initial reaction rate, because then small changes in how quickly the reaction mixture is mixed and inserted into the spectrophotometer would have a significant effect on the measured initial reaction rate.

That means that the concentrations of the substrates must not change significantly within the measurement period, which means the reaction rate must not be too high. See Table 1 for the reaction rates can be at maximum in order for the concentrations of different compounds not to change more than 10% over the duration of the measurement.

(11)

Table 1: The calculated maximum reaction rates in order for concentrations not to change more than 10% within 70 s

Compound Lowest initial concentration (mM) 10% (mM) Max reaction rate (Abs/min)

NADH 0.4 0.04 0.10

F6P 2.5 0.25 0.63

NADP⁺ 1 0.1 0.26

G3P 2.5 0.25 0.64

DHA 5 0.5 1.29

Another reason to decrease the reaction rate is that less enzyme and less coupled reaction enzymes are needed per measurement.⁸ The motivation to have a high reaction rate is that it gives a high signal to noise ratio, so a balance is sought between signal to noise ratio and consumption of enzyme, within the constraints of the maximum reaction rates.

A last consideration is that different cuvettes and different initial concentrations should be used for NADH/NADPH or NAD⁺/NADP⁺ in order to get optimal precision. In the retroaldol reaction (Figure 6) the initial NADH concentration of 400 µM is as high as possible without giving an absorbance much higher than 1.5 in a 0.5 cm cuvette and subsequently reducing accuracy. In the aldol addition reaction (Figure 7) the NADPH concentration instead increases over time during the reaction. This means that the precursor NADP⁺ concentration can be higher (settled on 1 mM) and that a 1 cm cuvette is used instead of a 0.5 cm cuvette to increase the signal from the low NADPH concentration.

2.6 Stopped-flow kinetics

Stopped-flow kinetics is a technique that is used to study fast chemical reactions. In a stopped- flow instrument solutions are rapidly mixed by being forced from syringes into a mixing chamber.

Milliseconds later the mixed solution is transferred to the observation cell and a sensing switch, connected to the piston that moves the solution, triggers the measuring device and stops the flow suddenly. The measuring device can vary depending on the reaction that is being measured, but it needs to detect changes in the reaction rapidly. In this case, measuring the binding of hydroxyacetone (HA) to FSA, the measuring device detects fluorescence. In the ’free’ state of FSA, fluorescence of Y131 is suppressed by energy-transfer to a nearby tryptophan residue. Binding of HA to FSA alters the conformation of the enzyme, weakens the link to the tryptophan residue and allows Y131 to fluoresce more strongly, which is what is being measured.

3 Methods

3.1 Buffers and media

2TY medium consisted of 16 g/L tryptone, 10 g/L yeast extract and 5 g/L NaCl in milli-Q water (MQ). LB agar was made by autoclaving a solution of 10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl and 15 g/L Agar in MQ. It was reliquefied by heating it in the microwave. Once it had cooled down to a temperature that did not burn the hands (<50 °C) ampicillin was added to a final concentration of 100 µg/mL and the mixture was poured into plates and allowed to solidify giving LB-Amp plates. Triethanolamine (TEA) buffer consisted of 50 mM TEA in MQ,

(12)

pH-adjusted to pH 8.0. Dialysis buffer consisted of 20 mM Na₂HPO₄, 0.5 M NaCl, 10% glycerol (v/v) in MQ, pH-adjusted to pH 7.5. Binding buffer consisted of 20 mM Na₂HPO₄, 0.5 M NaCl, 20 mM imidazole in MQ, pH-adjusted to pH 7.5. Washing buffer and Elution buffer were the same as binding buffer but with imidazole concentrations of 60 mM and 300 mM respectively.

Lysis buffer was made based on binding buffer. Lysozyme was added to a final concentration of 0.1 mg/mL and DNAse to a final concentration of 10 µg/mL and 1 EDTA-free protease inhibitor tablet was added per 50 mL. The lysis buffer was made fresh just before the resuspension of cell pellets and subsequent protein purification. A 20% (w/v) L-(+)-Arabinose solution in water was made and sterile-filtered through a 0.2 µm filter.

3.2 Mutation and transformation

A pGEM-3z-(f+) plasmid with the protein-coding sequence of the wild-type FSA was altered by mutagenic PCR with the QuikChange Site-Directed Mutagenesis Kit, following QuikChange protocol.⁷ The protein-coding sequence of the wild-type FSA already contained a His-tag to facilitate easy purification by immobilized metal ion affinity chromatography (IMAC) but for all other intents and purposes it is the wild-type FSA. See Figure 9 for a simple overview of the plasmid. The mutagenic primers used can be found in Table 2.

Figure 9: A simple overview of the pGEM-3z-(f+) plasmid with the protein-coding sequence of the wild-type FSA, to be mutated by site-directed mutagenesis.

The altered plasmid was transformed into E. Coli XL1-Blue. When transferring the bacteria, clipped-off pipette tips were used so as not to rupture any cells. For each successfully transformed

(13)

mutant 4 colonies were picked and cultured in 5 mL 2TY O/N for ∼ 16 h. The DNA was harvested from the cultures with a Thermo Scientific GeneJET plasmid miniprep kit, following the included protocol.¹⁰ Subsequently part of this DNA was prepared for sequencing according to the instructions on the eurofins genomics Mix2Seq kit and sent in for sequencing. The sequencing primers used were pUC-fwd and pUC-rev. Each plasmid was sequenced in the forward and reverse directions. The consensus sequence from the forward and reverse sequencing was compared to the wild-type FSA sequence to identify which colonies had the correct mutations.

After an unsuccessful transformation a few different steps were taken to try and get the Y131F mutation to work. Firstly, the transformation was repeated with the same PCR products and the same protocol. Secondly, the transformation was repeated with four times as much PCR product (4 µL instead of 1 µL of PCR product). Thirdly, the mutagenic PCR was repeated with the same primers and 5, 20 and 50 ng of template DNA (initial procedure only did 20 ng of template DNA).

Then a transformation was done using 4 µL of PCR product again. Fourthly, the mutagenic PCR was repeated with annealing temperatures of 60 °C and 65 °C instead of 55 °C to avoid primers forming secondary structures. A transformation was done using 5 µL of PCR product. Finally, three new sets of primers were ordered and the PCR and transformation were repeated with these primers and the unaltered QuikChange protocol.

Table 2: The DNA sequences of the mutagenic primers used to produce specific mutations in the FSA-wt plasmid, from 5’ to 3’ and separated into codons. The lowercase letters in red indicate the bases that mismatch with the template sequence. Multiple attempts were made with different Y131F primers.

Mutation Direction Sequence

Q59A Fwd GG CGT CTG TTT GCC gcG GTA ATG GCT ACC AC

Rev GT GGT AGC CAT TAC Cgc GGC AAA CAG ACG CC

T109A Fwd CCG ACG CTG GGAgCC GCG GTA TAT G

Rev C ATA TAC CGC GGc TCC CAG CGT CGG

Y131F Fwd GCG GAA TAT GTT GCG CCT TtC GTT AAT CGT ATT GAT GCT C Rev G AGC ATC AAT ACG ATT AAC GaA AGG CGC AAC ATA TTC CGC Fwd G GAA TAT GTT GCG CCT TtC GTT AAT CGT ATT GAT GC

Rev GC ATC AAT ACG ATT AAC GaA AGG CGC AAC ATA TTC C Fwd CG GAA TAT GTT GCG CCT TtC GTT AAT CGT ATT GAT GC Rev GC ATC AAT ACG ATT AAC GaA AGG CGC AAC ATA TTC CG Fwd GCG GAA TAT GTT GCG CCT TtC GTT AAT CGT ATT GAT GC Rev GC ATC AAT ACG ATT AAC GaA AGG CGC AAC ATA TTC CGC

3.3 Expression and Purification

The plasmid DNA that had the desired mutations was transformed into the E. coli BL21 A1 strain which is optimized for expression. 2 µL of miniprepped DNA was added to 50 µL aliquots of E.

coli BL21 A1 chemically competent cells and incubated on ice for 30 min. They were then heat shocked at 42 °C for 10 s and incubated on ice for 5 min. The cells were transferred to 950 µL of room temperature 2TY and incubated at 37 °C with shaking for 1 h. Finally, 200 µL of each of the transformants was plated on LB-Amp plates and incubated at 37 °C O/N for ∼ 16 h.

Colonies from these plates were used to inoculate 1 mL pre-cultures of 2TY with 100 µg/mL ampicillin in culture tubes. They were incubated at 30 °C and 225 rpm O/N for ∼ 16 h. Medium

(14)

cultures were made by inoculating 35 mL of 2TY with 100 µg/mL ampicillin with 200 µL of the pre-culture in 250 mL e-flasks. These were also incubated at 30 °C and 225 rpm O/N for ∼ 16 h.

Large cultures were made by inoculating 750 mL 2TY with 50 µg/mL ampicillin with 7.5 mL of the medium culture in 2 L e-flasks. Three such cultures were inoculated per mutant. They were incubated at 30 °C and 225 rpm for a few hours until they reached an OD600 of ∼0.8. In this specific case the OD₆₀₀ was measured after 4 hours and was ∼1.75-2.00, a bit too high. Protein expression was induced by adding L-(+)-Arabinose to a final concentration of 0.1 % (w/v) and the expression cultures were subsequently incubated at 30°C and 225 rpm O/N for ∼ 16 h.

The cells were harvested by centrifugation for 20 min. at 6700 g and 4 °C. The centrifugation was done in two rounds, pouring off the supernatant from the first round and adding more culture to get a bigger pellet. The bottles with pellets were put in -20°C for 1 h so the pellets would become harder and easier to transfer completely. The pellets were then moved to falcon tubes and stored at -80 °C.

Pellets were resuspended in fresh lysis buffer in a weight ratio pellet:lysis buffer of 1:9. A ho- mogenizer was used to ensure no cell clumps remained. The sample was kept on ice as much as possible. The sample was put through a french press cell disruptor two times and subsequently centrifuged for 1 h at 27000 g, 4 °C to get rid of non-soluble proteins and cell debris.

The pellet was discarded and the lysate was filtered through a 5 µm and 1.2 µm filter with a syringe to get rid of any remaining particles. ∼ 5 mL of chelating sepharose beads charged with Ni²⁺-ions were washed in a column with 10 volumes of elution buffer, 10 volumes of MQ and 10 volumes of binding buffer in that order. Then the beads were added to the ∼70-80 mL lysate and incubated on a slow shaker at room temperature O/N for ∼ 16 h. The beads were divided into two columns and the lysate flow-through collected for later analysis on SDS-PAGE. The beads were washed with 10 volumes of washing buffer. The absorbance over 1 cm at 280 nm of the flow-through was measured (A₂₈₀) and this washing and measuring was repeated three times until A₂₈₀ ≤ 0.06. The protein was eluted with 10 mL of elution buffer in each of the two columns.

It was concentrated to 5 mL by centrifugation in a vivaspin 100 kDa column at 3000 g for 70 min.

The concentrated eluate was put in a dialysis bag with 3.5 kDa pores. The ends of the bag were folded twice and closed with plastic clamps. It was left floating in 1.5 L slowly mixing dialysis buffer and equilibrated O/N for ∼ 16 h at 4 °C. The next day the dialysis buffer was exchanged for 1.5 L fresh and left to equilibrate for another 6 h. The dialysed protein solution was transferred to a falcon tube and stored O/N at 4 °C. The solution was centrifuged at 2400 g, 4 °C for 12 min to pellet the precipitate and the supernatant was transferred to a new falcon tube which was stored at 4 °C. The A280 was measured and the protein monomer concentration calculated with the help of the protein’s physical characteristics as calculated by ProtParam¹¹ with the amino acid sequence.

The purified protein and the lysate flow-through were analyzed by SDS-PAGE. A 60x diluted purified protein sample and an 8x diluted lysate flow-through sample were made in SDS sample buffer. Both were boiled for 10 min at 83 °C. 10 µL of these samples (the purified protein corresponded to 7.5 µg) and of Thermoscientific Pageruler low range unstained protein ladder were loaded onto a pre-cast SDS-PAGE gel which was run at 200 V for ∼30-45 min. The gel was stained with Coomassie Brilliant Blue R-250 by boiling in the microwave and shaking on a rocker- board for 15 min. Then the solution was switched to destaining solution and left on rocker-board

(15)

O/N for ∼ 16 h, after which an image was taken.

Before doing kinetics measurements the protein solution was buffer-exchanged to TEA buffer with a commercial G25-sepharose PD-10 desalting column. The resulting FSA in TEA buffer was also stored at 4 °C and the protein concentration was calculated from the A280.

3.4 Characterization

3.4.1 Kinetics by spectrophotometry

The reaction rates of both the aldol addition reaction and retroaldol reaction were measured by spectrophotometry at differing substrate concentrations. The reaction mixtures were based on TEA buffer and after pipetting together all components with FSA Q59A added last, the cuvette was inverted twice to mix before quickly starting the measurement. Initial measurements were done to find a substrate concentration that gave V_max, to tweak the FSA Q59A concentration to give a Vmax of at most 0.1 Abs/min but high enough for a good signal to noise ratio and to make sure that the coupling enzyme concentrations were high enough to not be rate-limiting at V_max. In subsequent measurements only the substrate concentration was varied. The change in NADH (retroaldol reaction) or NADPH (aldol addition reaction) concentration was measured at 340 nm, 30°C for 70 s. The reaction rate was obtained from the slope of the absorbance line once it became a straight line, around 20 s into the measurement.

The retroaldol activity of FSA Q59A was measured by spectrophotometry in a 0.5 cm quartz cuvette. A new NADH stock solution was made every day. The final reaction mixture contained 50 mM TEA (pH 8.0), 400 µM NADH, between 2.5-20 mM F6P substrate, 8-20 U/mL of TIM and 0.8-2 U/mL of GPDH and 2 µM of FSA Q59A. A consistent amount of coupling enzymes (TIM, GPDH) was used but there was a large error margin in the stock concentration. Each substrate concentration was measured at least 5-fold.

The aldol addition activity of FSA Q59A was measured by spectrophotometry in a 1.0 cm quartz cuvette. A new NADP⁺stock solution was made every day. This reaction contains two substrates so during measurements one of the two substrates was kept constant while varying the concentration of the other. In this case the G3P concentration was kept constant at 2.5 mM while the DHA concentration was varied between 5-300 mM. The final reaction mixture contained 50 mM TEA (pH 8.0), 1 mM NADP⁺, 0.1 µM of FSA Q59A. Most DHA concentrations were measured in duplicate except for a final measurement at [DHA] = 300 mM.

A Michaelis-Menten curve was fitted to the data with SimFiT software¹² (mmfit program). From this the kinetic parameters K_M and V_max were obtained with a confidence interval. Subsequently k_cat was calculated from V_max and the total enzyme concentration E_tot as k_cat = ^V_E^max

tot. A rational function equivalent to the Michaelis-Menten equation was also fitted to the data with Simfit (rffit program) in order to obtain k_cat/K_M with a confidence interval.

3.4.2 Stopped-flow

Solutions of 0.5, 1, 2, 10, 20, 50, 100, 250 and 500 mM hydroxyacetone in TAE buffer were made. A 20 µM solution of FSA Q59A in TAE buffer was made. Since the final reaction mixture consisted of a 1:1 mix of enzyme and substrate solutions, the final substrate and enzyme concentrations were half of those in the stock solutions. Stopped-flow fluorescence measurements were done

(16)

at 30.1 °C at 276 nm, with a 295 nm filter imposing a cut-off for signals. The fluorescence was measured over a 250 ms interval, informed by measurements at the highest substrate concentration aimed at seeing how quickly the fluorescence signal stabilized. The sample pipes were cleaned by doing 10 measurements on MQ before beginning actual measurements. For each measurements 2 syringes were rinsed twice with MQ and then filled with the substrate solution respectively the enzyme solution. Air bubbles were removed, the syringes emptied into the loading chamber and the solutions degassed by repeatedly applying sub-atmospheric pressure with the syringes. A stable baseline signal was established by driving five measurements’ worth of solution through the system. Then 15 measurements were then done in sequence, automatically. For each measurement 100 µL of both substrate and enzyme solutions was forced together and mixed and the mixture’s fluorescence was quickly measured over the specified time interval. The syringes were refilled, emptied and the solutions degassed before doing an additional 15 measurements, giving a total of 30 measurements per substrate concentration. Of these 30 measurements any clear outliers were deleted and the rest of the graphs averaged. The reaction constants were obtained by fitting the average curve to a single or double exponential curve.

4 Results & Discussion

4.1 Mutation and transformation

Plasmids with the sequences for FSA Q59A and FSA T109A were successfully produced and transformed into E. coli but no FSA Y131F transformant was produced. Two out of the three sequenced FSA Q59A colonies had the correct mutation and in the third it was unknown because the site of the mutation fell outside of the consensus sequence of the forward and reverse sequencing results. Three out of the four sequenced FSA T109A colonies had the correct mutation and the fourth had a big deletion in the protein-coding sequence.

It is unclear why the mutation and transformation of FSA Y131F failed. When troubleshooting many possible causes were considered. The simplest explanation was that a mistake had been made in the procedure, for example pipetting the wrong amount or forgetting to pipette something.

Simple repeats of parts of the procedure failed too, however, and all steps and components of the procedure were identical between the successful Q59A and T109A and the unsuccessful Y131F mutagenesis and transformation except for the mutagenic primers. Perhaps the mutagenic primers for Y131F formed secondary structures that prevented them from binding properly to the plasmid, causing the whole PCR reaction to produce less DNA. That could be fixed by increasing the annealing temperature to make it less likely for the primers to form secondary structures. This did not suffice however. This does not exclude the possibility that the problems were caused by the formation of secondary structures but it was at least not fixable by increasing the annealing temperature, or it was not the only problem. The final attempt was made by repeating the procedure with different sets of mutagenic primers. They were shorter than before and less likely to form secondary structures, but did not work either. The possibility that the Y131F mutation was somehow lethal was considered. This should not be the case since the FSA Y131F protein should not be expressed until the plasmid is transformed into the E. coli BL21 strain, which has the necessary viral polymerases, and L-(+)-Arabinose is added to induce expression. A next step could be to repeat the mutagenic PCR and transformation with the new primers, but using increased annealing temperatures and multiple different template concentrations for the PCR and multiple different amounts of PCR product for the transformations.

(17)

4.2 Expression and Purification

Figure 10: SDS-PAGE gel of a low-range protein ladder, the lysate after incubation with Ni²⁺-sepharose beads and the purified protein.

For each of the E. coli BL21 FSA Q59A or T109A cultures 3×750 mL of expression cultures were made which yielded approximately 15 g of cell pellet for each mutant. Due to a lack of time only FSA Q59A was purified. It was purified from 8.27 g of cell pellet and yielded 4 mL of 1.88 mM, 44.8 mg/mL FSA Q59A monomer solution, for a total of ∼180 mg of protein. The SDS-PAGE gel showed that it was very pure and of the correct molecular weight (Figure 10). Many different proteins were present in the lysate but the purified protein consists overwhelmingly of a protein with a M_w of between 20-25 kDa. This agrees with the expected M_w for FSA Q59A of 23.8 kDa.

4.3 Kinetics

4.3.1 Michaelis-Menten

A Michaelis-Menten curve and its equivalent rational function fit fairly well to the data from the retroaldol kinetics measurements and very well to the data from the aldol addition measurements, see Figures 11 and 12. The steady-state kinetic parameters K_M and V_max were obtained from the fitted curves and used to calculate k_cat. k_cat/K_M was obtained from the fitted rational curve. These values can be found in Table 3 where they are compared to wild- type FSA. Ideally the kinetics of wild-type FSA and FSA Q59A would have been measured in as close to identical conditions as possible, to make sure any differences are inherent and not caused by differences in measurement conditions. The kinetics of wild-type FSA were being measured in our lab using near-identical conditions to the measurements done on FSA Q59A but these measurements were not complete yet when writing this thesis. Therefore the kinetic parameters of FSA Q59A were compared to published data about wild-type FSA.¹

(18)

Figure 11: Michaelis-Menten curve fitted to FSA Q59A retroaldolase activity at varying [F6P]. [E]tot

= 2 µM, T = 30 °C. R² = 0.8056

Figure 12: Michaelis-Menten curve fitted to FSA Q59A retroaldolase activity at varying [F6P]. [E]tot

= 0.1 µM, T = 30 °C, [G3P] = 2.5 mM. R²= 0.9916

(19)

Table 3: Steady-state kinetic parameters (95% confidence) for the retroaldol cleavage of F6P and aldol addition of G3P and DHA by FSA Q59A. It is compared to FSA wild-type data from¹

Enzyme

Formation of F6P from G3P and DHA Cleavage of F6P into DHA and G3P k_cat K_M^DHA k_cat/K_M k_cat K_M^{F 6P} k_cat/K_M (s⁻¹) (mM ) (s⁻¹M⁻¹) (s⁻¹) (mM ) (s⁻¹M⁻¹) FSA Q59A 5.4 ± 0.5 60 ± 14 90 ± 14 0.29 ± 0.03 5.3 ± 2 54 ± 13

FSA wt 8.0 ± 0.5 31 ± 10 275 ± 95 2.3 19 119

FSA Q59A is less active than wild-type FSA in both the aldol addition and retroaldol reactions.

This confirms the suggestion by earlier research^{1, 3} that residue Q59 plays a role in the catalytic mechanism, at least for the substrates used here. The effect is much stronger in the retroaldol reaction, where k_cat is more than 7 times smaller, than in the aldol addition reaction, where k_cat is not even two times smaller. Since the comparison is with published data that was obtained in not identical conditions a factor 2 difference in k_cat is not quite enough to draw conclusions from, but a factor 7 difference in k_cat is indicative of a real difference between FSA Q59A and wild-type FSA. Judging by K_M, FSA Q59A seems to bind F6P between 3-4 times more strongly than the wild-type and dihydroxyacetone about half as strongly as the wild-type. In a multi-step reaction like this, however, K_M is not simply the binding affinity so one should be careful interpreting it that way.

There is a striking difference in the precision of the aldol addition and retroaldol measurements.

Just by looking at Figures 11 and 12 it is clear that the aldol addition measurements were much more precise. They were precise enough that the measurements only needed to be done in duplicate, whereas the retroaldol measurements were done in 5-fold. The volumes of substrate, FSA or coupling enzyme solution that were pipetted were not very different so it seems unlikely that pipetting error was the culprit. The same equipment was used so that cannot be the cause either.

Perhaps the difference in precision was caused by the fact that the retroaldol reaction measured a decrease and the aldol addition reaction measured an increase in absorbance over time. The high concentration of NADH in the retroaldol reaction meant a high absorbance and a 0.5 cm cuvette, both of which lead to less precise absorbance measurements when compared to the low absorbance and 1.0 cm cuvette of the aldol addition reaction.

4.3.2 Stopped-flow

The measurements that were done did not give a clear trend because they had a very low signal-to- noise ratio. It would be disingenuous to calculate kinetic parameters of the initial imine-forming reaction between the enzyme and hydroxyacetone from this data. In order to do that these measurements should be repeated with a higher enzyme concentration, which would give more activity and therefore a clearer signal. A 10 times higher final enzyme concentration of 100 µM FSA Q59A would be a good starting point. To do enough measurements with such a high enzyme concentration would require a very large amount of enzyme, however, so it might not be feasible.

(20)

5 Conclusion

The FSA Q59A and FSA T109A mutants have been successfully produced. IMAC allowed for the purification of a large amount (∼180 mg) of very pure and concentrated FSA Q59A. This mutant was active, but less so than the wild-type FSA in both the aldol addition and retroaldol reactions. This confirms the suggestion by earlier research^{1, 3} that residue Q59 plays a role in the catalytic mechanism, at least for F6P, G3P and DHA substrates. The effect is much stronger in the retroaldol reaction, where kcatis more than 7 times smaller, than in the aldol addition reaction, where k_cat is not even two times smaller. Judging by K_M, FSA Q59A seems to bind F6P between 3-4 times more strongly than the wild-type and dihydroxyacetone about half as strongly as the wild-type. In a multi-step reaction like this, however, K_M is not simply the binding affinity so one should be careful interpreting it that way. Stopped-flow measurements were done on a combination of FSA Q59A and hydroxyacetone, but the enzyme concentration used in the measurements was too low to get good resolution and no conclusions could be drawn from the data.

6 Acknowledgements

First and foremost I thank my supervisors Mikael Widersten and Gina Chukwu for all their help.

In addition to that I want to thank all the other people that also helped me out in the lab, chief among them Annika S¨oderholm and Alberto Zavarise.

7 References

1Huan Ma, Sarah Engel, Thilak Reddy Enugala, Derar Al-Smadi, Candice Gautier, and Mikael Widersten. New Stereoselective Biocatalysts for Carboligation and Retro-Aldol Cleavage Re- actions Derived from d-Fructose 6-Phosphate Aldolase. Biochemistry, 57(40):5877–5885, 2018.

Publisher: ACS Publications.

2Pere Clap´es, Wolf-Dieter Fessner, Georg A. Sprenger, and Anne K. Samland. Recent progress in stereoselective synthesis with aldolases. Current Opinion in Chemical Biology, 14(2):154 – 167, 2010.

3Lena Stellmacher, Tatyana Sandalova, Sebastian Leptihn, Gunter Schneider, Georg A. Sprenger, and Anne K. Samland. Acid–Base Catalyst Discriminates between a Fructose 6-Phosphate Al- dolase and a Transaldolase. ChemCatChem, 7(19):3140–3151, 2015. eprint: https://chemistry- europe.onlinelibrary.wiley.com/doi/pdf/10.1002/cctc.201500478.

4Xiaohong Yang, Lidan Ye, Aipeng Li, Chengcheng Yang, Huilei Yu, Jiali Gu, Fei Guo, Ling Jiang, Fan Wang, and Hongwei Yu. Engineering of D-fructose-6-phosphate aldolase A for improved activity towards cinnamaldehyde. Catalysis Science & Technology, 7(2):382–386, 2017.

Publisher: Royal Society of Chemistry.

5Richard Obexer, Alexei Godina, Xavier Garrabou, Peer RE Mittl, David Baker, Andrew D Griffiths, and Donald Hilvert. Emergence of a catalytic tetrad during evolution of a highly active artificial aldolase. Nature chemistry, 9(1):50, 2017. Publisher: Nature Publishing Group.

(21)

6Kai Tittmann. Sweet siblings with different faces: the mechanisms of FBP and F6P aldolase, transaldolase, transketolase and phosphoketolase revisited in light of recent structural data.

Bioorganic chemistry, 57:263–280, 2014. Publisher: Elsevier.

7QuikChange Site-Directed Mutagenesis Kit Instruction Manual Catalog # 200518 (30 reactions) and 200519 (10 reactions) Revision C.

8Andrew C Storer and Athel Cornish-Bowden. The kinetics of coupled enzyme reactions. applica- tions to the assay of glucokinase, with glucose 6-phosphate dehydrogenase as coupling enzyme.

Biochemical Journal, 141(1):205–209, 1974.

9John S Easterby. A generalized theory of the transition time for sequential enzyme reactions.

Biochemical Journal, 199(1):155–161, 1981.

10GeneJET Plasmid Miniprep Kit. Library Catalog: www.thermofisher.com.

11Elisabeth Gasteiger, Christine Hoogland, Alexandre Gattiker, Marc R Wilkins, Ron D Appel, Amos Bairoch, and others. Protein identification and analysis tools on the ExPASy server. In The proteomics protocols handbook, pages 571–607. Springer, 2005.

12W.G. Bardsley. SimFiT Windows V7.3.7 Academic 64-bit, https://simfit.org.uk.

Production and characterization of fructose-6-phosphate aldolase mutants

Uppsala University

Bachelor Thesis