(214a) Modeling Continuous Enzymatic Reactive Crystallization of ?-Lactam Antibiotics

McDonald, M. - Presenter, Georgia Tech
Rousseau, R. W., Georgia Institute of Technology
Bommarius, A., Georgia Institute of Technology
Grover, M., Georgia Tech
Modeling continuous enzymatic reactive crystallization of β-lactam antibiotics

Matthew A. McDonald*, Andreas S. Bommarius*, Martha A. Grover*, and Ronald W. Rousseau*

*School of Chemical & Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia


Keywords: reactive crystallization, β-lactam antibiotics, continuous manufacturing



β-lactam antibiotics, despite a long and storied history, find continued widespread use in the fight against infections. In 2010 the two most common classes of β-lactam antibiotics, the penicillins and cephalosporins, accounted for nearly 60% of world antibiotic consumption, with delivery of over 20 billion doses of penicillin-derived antibiotics.1 Traditionally, semi-synthetic β-lactam antibiotics have been made batch-wise which lends itself to sharing equipment, such as crystallizers, between different active pharmaceutical ingredients (APIs).2 β-lactam antibiotics can also cause a severe allergic reaction; a 2012 study found that 92 out of 11,761 patients at Mount Sinai Hospital in New York City suffered anaphylaxis from penicillin, a prevalence of 0.7%.3 This rate is significant enough for the U.S. Food and Drug Administration (FDA) recently to require that all penicillin-derived antibiotics be manufactured, processed, and packaged in facilities separate from those used for any other pharmaceuticals, removing the ability to share batch equipment between APIs.4 The need to establish new β-lactam-specific facilities presents an opportunity to implement novel, more efficient, processes such as enzymatic synthesis. However, a new design concept is required to overcome some drawbacks of the enzymatic route. The key to this process is enzymatic reaction enhanced by simultaneous crystallization to manufacture semi-synthetic β-lactam antibiotics while decreasing the risk of cross-contamination and without creating large amounts of waste.

We use ampicillin and cephalexin as a model penicillin and cephalosporin, respectively. The enzyme, penicillin G acylase (PGA), catalyzes the condensation of 6-aminopenicillanic acid (6-APA) and phenylglycine methyl ester (PGME) and 7-aminodesacetoxy-cephalosporanic acid (7-ADCA) and PGME into ampicillin and cephalexin, respectively. PGA also hydrolyzes both antibiotics and PGME into phenylglycine, a sparingly soluble contaminant.


Batch Processing

Catalytic hydrolysis of the β-lactam product by the same enzyme that catalyzes its synthesis renders the desired product an intermediate in a kinetically-controlled system. By selectively removing this intermediate from the solution via a reactive crystallization process we have shown that the process selectivity can be enhanced approximately 1.5-fold compared to an equivalent non-crystallizing batch process, resulting in an equivalent increase in yield. The increase in yield and selectivity stems from the concentration of product in solution, and hence the concentration available to hydrolysis by the enzyme, remaining at the solubility limit while the actual amount of product formed (dissolved and crystallized) far exceeds this level.5

Continuous Processing

More recently we have explored continuous reactive crystallization as a way of increasing the productivity and efficiency of semi-synthetic β-lactam synthesis. Continuous manufacturing may enable greater sustainability, product consistency, and access to different chemistries. The accumulation of phenylglycine by-product—a potential contaminant—poses the greatest challenge to increasing conversion. Continuous operation could increase conversion over batch by tailoring the conditions to optimize reactive crystallization; for example, a plug flow reactor/crystallizer can operate with excess 6-APA or 7-ADCA relative to PGME to hinder hydrolysis of the PGME while a recycle stream could act to seed the plug flow apparatus to slow antibiotic hydrolysis via crystallization.

The reaction kinetics are predicted via a previously described model5 and the crystallization kinetics have been determined via online chord-length distribution and concentration measurements, by focused beam reflectance measurement (FBRM) and attenuated total-reflectance Fourier-transform infrared (ATR-FTIR) spectroscopy, respectively.6 Mapping the operation space requires reconciliation of the kinetically controlled enzymatic synthesis and crystallization of the antibiotic. Many of the parameters that are typically used to control crystallization are restricted by the simultaneous reaction; the enzyme prevents use of anti-solvents or temperature changes and the pH changes both the enzyme activity and ampicillin solubility. At times, the reaction and crystallization processes can be at odds with each other; high concentrations of 6-APA have been shown to inhibit nucleation and growth of ampicillin trihydrate crystals while increasing enzyme selectivity.7 Our model examines these interactions and allows for prediction of a wide range of operating conditions for continuous reactive crystallization synthesis of β-lactam antibiotics.


[1] Gelbrand, H., Miller-Petrie, M., Pant, S., Gandra, S., Levinson, J., Barter, D., White, A., Laxminarayan, R., Ganguly, N. K., and Kariuki, S. (2015) The State of the World’s Antibiotics 2015, Wound Healing Southern Africa 8, 30-34.

[2] Elander, R. P. (2003) Industrial production of β-lactam antibiotics, Applied microbiology and biotechnology 61, 385-392.

[3] Albin, S., and Agarwal, S. Prevalence and characteristics of reported penicillin allergy in an urban outpatient adult population, 6 ed., pp 489-494, OceanSide Publications, Inc.

[4] (2015) 21 C.F.R § 211.42 (The Code of Federal Regulations of the United States), (Administration, F. a. D., Ed.).

[5] McDonald, M. A., Bommarius, A. S., and Rousseau, R. W. (2017) Enzymatic reactive crystallization for improving ampicillin synthesis, Chemical Engineering Science 165, 81-88.

[6] Encarnación-Gómez, L. G., Bommarius, A. S., and Rousseau, R. W. (2016) Crystallization Kinetics of Ampicillin Using Online Monitoring Tools and Robust Parameter Estimation, Industrial & Engineering Chemistry Research 55, 2153-2162.

[7] Ottens, M., Lebreton, B., Zomerdijk, M., Rijkers, M., Bruinsma, O. S. L., and van der Wielen, L. M. (2004) Impurity effects on the crystallization kinetics of ampicillin, Industrial & engineering chemistry research 43, 7932-7938.