The Process

As the first drug to exceed $10 billion in sales with annual demands surpassing 440,000 lbs, a more economic and green method of manufacturing Lipitor and its API atorvastatin essential. Codexis, Inc. faced more two major challenges in the process of manufacturing atorvastatin (Fig. 2) a green-by-design method. Using recombination based directed evolution of three enzymes, Codexis, Inc. managed not only to create a simpler and green process but also employed the use of ProSAR (statistical analysis method) to make the enzymes sufficiently active to enable practical, commercial scale manufacture. This process is now being used to manufacture 100 metric tons (mT) of hydroxynitrile a year!

Fig. 2: Atorvastatin calcium. (Permission for reuse granted by Roger Sheldon of Delft University of Technology).
Let's see how they solved the two challenges at hand:

Problem #1: Design a green method of creating the key chiral compound in atorvastatin, ethyl (R)-4-cyano-3-hydroxybutyrate 1, or hydroxynitrile

The traditional method as shown in Figure 3 only produced 50% yield, resulted in extensive byproducts, and required the use of high-vacuum fractional distillation to purify the final product which further decreased yield (, 2014). In the final step of synthesis using the previous commercial method, manufacturers reacted halohydrin with a cyanide ion under basic conditions at high temperatures, resulting in lengthy by-products due to the base-sensitive nature of the substrate and product of the reaction. In order to recover usable product from this reaction, the next step was to extract the product using high-vacuum fractional distillation which ended in additional loss of yield and more waste products (Ma, 2009).

Fig. 3: The multi-step process initially used for commercial manufacturing of HN. (Permission for reuse granted by Roger Sheldon of Delft University of Technology).

Solution #1: Codexis, Inc. develops a two-step, three-enzyme process

Though researchers at Codexis, Inc. could not avoid the use of cyanide, they were able to facilitate bioinformatic technology to enhance enzymes that would perform the same reaction in mild pH conditions, in cooler temperatures, under less pressure, and use water as a reaction medium. They achieved this process in two steps:

Step 1) The reduction of ethyl-4-chloroacetoacetate was performed using a ketoreductase (KRED) enzyme in combination with glucose and NADP-dependant glucose dehydrogenase (GDH).The product of this reaction, (s)ethyl-4-chloro-3-hydroxybutyrate, was achieved at a 96% yield, nearly a 40% increase (Ma, 2009).

Step 2) Team members at Codexis, Inc. were then able to pinpoint and enhance enzymes called halohydrin dehalogenases (HHDH's) (Fig.4) that both catalyzed the desired reaction and accepted cyanide as a non-natural nucleophile resulting in a fixed formation of beta-hydroxynitriles, the desired chiral compound (Fig. 5) (Ma, 2009).

Fig. 4: HHDH converting (S)-4-chloro-3-hydroxybutyrate to hydroxynitrile via cyanation reaction under markedly more mild conditions than initial process (Permission for reuse granted by Dr. Gjalt Huisman).
There were many advantages of using enzymes as catalysts:

-Function under neutral conditions in water
-Biodegradable and nontoxic
-Decrease steps needed in a reaction
-Made from renewable resources
-Chemo- and enantio- selectivity

There were also many limitations:

-Inadequate reactivity toward non-natural substrates like cyanide
-Inadequate activity under high substrate loading
-Low operational stability under manufacturing conditions

Fig. 5: Two-step three enzyme process developed by Codexis, Inc. (Permission for reuse granted by Roger Sheldon of Delft University).

Problem #2: Researchers at Codexis, Inc. could not produce the necessary demand of HN largely due to the limitations of using enzyme catalysts

Fig. 6: The process by which enzyme mutations were analyzed via ProSAR technology. 50 mutations at a time were put into the hopper, sequenced, and analyzed. Once analyzed, they were sorted into four categories: deleterious, neutral, potentially beneficial, and beneficial. This process was repeated several times (Permission for reuse given by Dr. Gjalt W. Huisman).

Solution #2: Use statistical analysis platform to develop recombination directed evolution of biocataytic enzymes, increasing volumetric productivity of the cyanation process by ~4,000 times (Ma et al., 2007)

Researchers at Codexis, Inc. used ProSAR technology to isolate and recombine the enzymes harnessing the best properties of all mutated variants (Fig. 6). ProSAR, or protein sequence activity relationships, is a method of statistical analysis that enables the best benefits of all variants to be identified and enhances the overall properties of the enzymes involved. Once the best variants were identified, they performed a "traditional hit shuffle" which recombined them together. Researchers discovered that 86% of all variants were beneficial, and each round of shuffling resulted in a 1.5x increase in productivity than the round before. The end product was approximately 4,000 times more productive than the starting product, indicating the enhanced enzymes were finally suitable for commercial production (Fox, 2007).

Atorvastatin Home