Directed Enzyme Evolution: Biocatalysis Explained

[00:00:12]Welcome to the MOOC “Tailored materials and  enzymes for industrial processes”. My name is Robert Kourist and this is the fifth unit of  the basic module and deals with directed enzyme evolution. So the previous unit I explained that  the catalytic properties of enzymes are stored in the primary sequences and we want to improve  and we use rational protein design or directed evolution in order to improve these properties  for applications. And in rational protein design we can bring in mutations at selected  sites that for instance we identified based on structure elucidation or molecular modeling  and then we predict that the effect of a certain amino acid exchange we incorporate this mutation  alter the structure and then alter the function.

[00:01:06]The problem is that it is exceedingly difficult  to make an accurate prediction of a change of an amino acid sequence, but in some cases and I  explained one the last unit this works reasonably well. The case was the one of the aryl malonate  decarboxylase which produces optical pure alpha substituted carboxylic acids. In this case the  enzyme decarboxylates malonates, it cleaves carbon dioxide and in the wild type the enzyme produces  the R- enantiomer by protonation from one side and exchange of this Cysteine into a serine, which  does not protonate anymore and the introduction of the cysteine at the opposite side of the active  site completely inverted the enantioselectivity.

[00:01:54]This is a classical case of rational design,  because a rational prediction was introduced was, investigated experiment and could be confirmed.  Unfortunately, the activity of the resulting mutant was here G74C/C188S was by 18,000 fold  reduced. So somehow this cysteine or maybe the serine or maybe both they interfered with activity  and at this stage the researchers had no clue, how they could recover the activity of this mutant  and therefore they went for directed evolution. In rational protein design we only need to introduce  very few mutations and this can be done in a very controlled way, it's very simple. In direct  Evolution the generation of the mutants means it influences what kind of amino acids we can  access and how they're distributed. So this is a crucial element of the experiment and there  are basically two approaches. One is purchase chemically synthesized libraries, the other is  to do it yourself in the lab. Chemical synthesis is very expensive, cost several thousands of  euros but of course it's not work intensive and mutagenesis in the lab can be done with  very simple ingredients but it needs a skilled coworker. So there's personal cost and then  the decision is just a question of priorities.

[00:03:26]However, also and chemically synthesized like  libraries and PCR generated libraries are a bit different. So it's not exactly the same, so  it's a mixture of priorities and also what kind of libraries we want to have. In mutagenesis we  have PCR based methods and we have other methods and for the sake of simplicity I will focus here  on polymerase Chain Reaction based mutagenesis, because this is now the most frequently used  method. There are others there are chemical reagents that induce mutations and we don't like  to work with them, because they're cancerogenic, or there’s UV mutagenesis could be also used  and there are mutator strains that are deficient of repair systems, so that each cell doubling  they would cause a certain number of mutations, also a bit difficult to control. So I would think  I think polymerase chain a reaction is the most frequently used method right now to generate  libraries for directed evolution. Error prone PCR was one of the very first methods used for  mutagenesis for directed evolution. It is quite a simple method and can be done in every lab  that has the ingredients and machines for a regular PCR. So in the regular PCR a double strand  the DNA is our template, then there are specific primers that bind and the polymerase amplifies  the two strands and then the product of the PCR serves as a template for the next cycle. So the  product of the first cycle is also the template of the next so in every step in every cycle the  concentration of the DNA doubles and because of this, this is called a chain reaction. So  the polymerase has a very high accuracy, so it has an error rate of one in 100,000 or one  in a million and then polymerases have usually a second unit at attached to them which slide along  the DNA and check the double strand for errors and this also has an error rate of one and a thousand.  So this gives us totally an error rate in the billions. With this high accuracy it would be  difficult to insert mutations. However, there are ways to increase the error rate of a polymerase  and this is done in so-called error prone PCR and these are quite easy quite simple ways. So  we can use a polymerase that does not have the proof reading capacity so this already increases  the error rate a lot and we can add manganese instead of magnesium so both are bivalent ions and  magnesium is in the active site of the polymerase to coordinate the nucleotides and if there is  manganese, somehow the enzyme has a higher error rate. We can also spike a nucleotide concentration  this means if we have more for instance of dCTP, then the likelihood that dCTP comes into the  active site of the enzyme is higher leading also to a higher error rate and if we combine all  this this can increase the error rate to 1 to 2%.

[00:06:41]So in this case we would still have our template,  primers bind and then the amplification causes errors which I have here denoted with these red  crosses and these errors would of course be then again in the next cycle as a template and the  polymerase then if that amplifies conserves these mistakes but also introduces new errors in  every cycle. So with the number of cycle and the conditions of this we can also adjust the mutation  frequency and this is a widely used method. It has some limitation and usually it is done for the  generation of very large libraries between 50,000 and typically 100,000 clones. Sometimes we have  already concrete idea for an amino acid position and we want to insert all 20 amino acids into one  position and in the last unit I presented site directed mutagenesis where a certain nucleotide  exchange is introduced by a mutagenic primer, which is a primer which has one different  nucleotide or several different nucleotides, but still binds because the large part of the  primer is still complementary. For saturation mutagenesis we can use primers with a so-called  N and N stands for four bases G A T and C and so a primer which is possess it has an N is actually  a mixture of four primers 25% of each base and if this is then Incorporated in the amplification and  the template is degraded we have now the N in our Gene which means we have a mixture of genes with  each 25% of different base and by using one or several bases with an N we can put in every of the  64 codons, which means we can put in every amino acid at a certain position provided of course  we have a certain idea which of our 200 or 300 positions to set to mutate. The last component  for a directed evolution experiment is the high throughput screen. So here we need a simple way  to measure a reaction in a microtiter plate and the activity of the the AMDase can be determined  by the quantification of substrate and product for instance in a high HPLC. But this would mean  for each mutant we need to make at least one HPLC measurement which would be a large number of  HPLC measurements already, but in this case we can if we decarboxylate a malonate and produce a mono  acid we have a pH shift. And because here we have two carboxylic groups only one so the medium gets  more basic and then we can use a buffer with a low buffer capacity and you see this here. In these  wells, a reaction took place and the shift to the basic a pH was visualized with addition of a pH  indicator and here in these wells no reaction took place. Then there was the wild type enzyme was  added at certain points to show to see for of all these four measurements all the six measurements  have the same rate and then we get a So-called mutant landscape. So this would be here inactive  variants or lower active variants, this will be here our wild type. You see we have quite of a  spread so this essay is not so accurate and then we might have some where we have a higher activity  so with such as green. We can now efficiently screen thousand of variant 100 per plate and there  also microplates that have 364 wells which even give us many more clones to screen. So coming  back to the experiment so the mutant G74C C188S has the cysteine on the other side is inverse  but has lower activity and the determinants of the activity were unknown but it is known that the  enzyme always cleaves the same carboxylic group, which is this one which one pointing out towards  us and this is accommodated in a hydrophobic pocket, which results in unfavorable interactions  and this leads to the cleavage of carbon dioxide.

[00:10:55]So it was thought that a variation of this  hydrophobic pocket might change the activity and these are here residues in this active site.  So we have a methionine, we have a valine 156, we have valine 43 we have a tyrosine 48, a  leucine 40 and a leucine 77. Here is the catalytic cysteine. This should not be altered of course and  here we have the serine which is the cysteine in the wild type. They constitute this hydrophobic  pocket and now by saturation mutagenesis each of these residues was saturated so first of all this  serine which is now not anymore needed was altered and it was shown that the glycine in this position  gives a five times higher activity and this was then used as a template and next here different  residues were saturated and here for instance if leucine 77 is is altered methionine was the  most active variant and interestingly here the substitution of methionine 159 by leucine led to  a 100 fold almost 100 fold activity increase and then when this was used as a template again the  hit Y for the F led to a 900 fold high activity so by one to three cycles of mutagenesis the activity  could be increased 900 fold. So this is a typical experiment of a randomization protein engineer.  So the the effect is striking we have a 900 fold activity increase but it is very difficult to  understand why substitution of M959 with a leucine which is not much bigger leads to such a striking  activity increase. One thing we can see here, we can see here that all the hits all the  improved variants have hydrophobic amino acids, which makes a lot of sense because all of these  amino acids are hydrophobic. So there was no arginine, lysine, serine or glutamate found and  actually in hindsight it is also possible to do this experiment now with the restriction of the  amino acids put and putting in only hydrophobic amino acids. So there are lot of degrees of  freedom but it is also not known if this here is the best variant of the enzyme, but directed  evolution brings produces improved variants without an accurate understanding why here for  instance these changes of these hydrophobic amino acids lead to this activity increase. This  makes it such a powerful method. So if you compare the rational design and the randomization we have  several important factors. One factor is how well do we understand our enzyme. If we have a very  good hypothesis making a very good prediction one mutant can be sufficient and indeed in the AMDase  example a double mutant inverted selectivity completely. If we have very good knowledge then  we need to screen only few. If we have not so much knowledge understand our enzyme rather poorly  then we need a tremendous screening effort and this is an equation of experimental capacities. So  nowadays we have a lot of ways to get structures without experimental structure information. In  many cases we can make we have similar enzymes proteins that have been crystallized so we can  make homology models and we can also make very plausible de novo predictions and also often  we know something about the mechanism even if it's a very new enzyme. On the other hand if  it's an interesting enzyme often it's quite new so we do not have so much experience. So we are  often here we have some knowledge but not enough knowledge to make accurate predictions. This  is a typical example of the hydrophobic pocket so the researchers doing this Miyamoto and  Ohta they identified this hydrophobic pocket as activity driver but they could not pinpoint  and predict mutations so they could use this for randomization. We have a similar situation with  a screening effort. It is quite easy to screen a few thousand variants. We can do this with an  HPLC, with a pH screen so this is quite simple.

[00:15:23]To screen millions or even billions this would  mean we need a selection assay or fluorescent assay and they are available for only few enzymes.  So therefore the quite usual situation is that we have some idea for instance we can predict the  active side residues and we have a medium through screen and this means we can screen 100 to 200  variants or maybe 5,000 variants and this is now a way how enzyme engineering is done often nowadays.  So with this I introduced you to a few basic concepts of molecular biotechnology, biocatalysis  so the way to produce enzymes and optimize them by rational protein design and directed enzyme  evolution and the following the following units will then discuss how enzymes can be improved  for the use of multi cascade reactions by immobilization to carriers and with this I  would like to thank you for your attention.