Development of the First Version of Liberase™ HI Purified Enzyme Blend 

Most development projects rarely follow a straight line from concept to product launch. The story of the Liberase™ HI Purified Enzyme Blend is no exception; it’s a tale of starts, stops, and unexpected opportunities that breathe life back into a shelved project. The project was re-invigorated when Camillo Ricordi visited Boehringer Mannheim in 1992 to advocate that Boehringer fund a project to improve the consistency and reliability of enzymes used to recover human islets that were subsequently used for experimental islet transplant. 

The story began in March 1987 when Boehringer Mannheim GmbH made a significant investment to expand their Indianapolis site to include R&D and manufacturing capabilities. I was hired as a Research Investigator to focus on protein purification for the Boehringer Mannheim Biochemicals (BMB) business unit. BMB was a highly respected supplier of nucleic acid restriction enzymes at that time. The new U.S. R&D group focused on developing and manufacturing cell biology-associated research products. An early development of this initiative was licensing a serum-free media formulation that led to the manufacture of the Nutridoma™ product line. These serum-free media formulations were designed to support the in vitro growth of hybridoma cells, enabling scientists to harvest antibodies from cell cultures and simplify the downstream protein purification process. 

I spent the initial months planning rather than performing projects since laboratory construction would not be completed until June 1987. During this time, I met with John Hopkinson, the newly hired manager responsible for marketing cell biology products. John previously worked at Worthington Biochemicals when Amicon owned the company, so he was familiar with the “collagenase business.” Linda Jacobsen, hired for the Nutridoma project, joined these discussions. She was a collagenase user when she managed a core laboratory at Purdue University that supplied rat hepatocytes for toxicology and cancer research studies. 

As you guessed, these discussions immediately led to projects designed to improve the reproducibility of collagenase used to isolate cells. I reviewed the literature and began to define what was needed to perform the work and a budget to complete the project. The published literature on the enzymes responsible for tissue dissociation was confusing and often conflicting. My logic was first to review the key characteristics of crude collagenase. Traditional crude collagenase products are derived from Clostridium histolyticum fermentation supernatants. The proteins in these supernatants are precipitated with saturated ammonium sulfate, followed by extensive dialysis to remove the salts and low molecular weight culture media components. 

Earlier studies identified two classes of C. histolyticum collagenase, class I & II. Chromatographic methods could separate these isoforms, with each having distinctive enzyme activities. Class I (C1) had strong gelatinase activities and low peptidase activity when using a collagen peptide as substrate. (1, 2) By contrast, class II (C2) had low to moderate gelatinase activity and strong peptidase activity. Crude collagenase also contained two endoprotease enzymes (clostripain and C. histolyticum neutral protease). Multiple publications indicated that there could be anywhere from one to three distinct forms of C2 enzymes and from one to four forms of C1 activity.(2) 

The first question I asked was whether separate genes express the multiple forms of collagenase. Or could one gene be expressed for each class and the multiple collagenase forms generated by proteolytic degradation of the intact enzymes? The biology of C. histolyticum must be placed in the proper context to address this question.  

C. histolyticum is an anaerobic bacterium surviving at the bottom of the food chain. The secreted collagenase and protease enzymes digest protein found in dead organic matter to create foodstuffs for growth and nourishment. It is a minimalist organism, so it is doubtful that the organism will have up to seven genes for degrading collagen when only two would suffice. Both collagenase isoforms are very large, with the highest molecular weights of C1 and C2 collagenase being 118 kDa and 110 kDa, respectively.(2) My training and academic experience correlating primary amino acid sequence to secondary and tertiary protein structure led me to view these large collagenase enzymes as multi-domain proteins. Typically, protein domains are linked together by amino acid sequences with little to no secondary structure. I assumed that the presence of clostripain and C. histolyticum neutral protease led to degradation of the intact enzyme into different proteolyzed collagenase forms. Clostripain proteolyzes proteins at arginine and lysine residues, and the neutral protease cleaves proteins at leucine and phenylalanine residues.(3) This was the most straightforward conclusion to explain the conflicting published data.  

In June 1987, the R&D laboratories were finished, but other priorities led to further delays in completing the project plan to develop purified collagenase products. In October 1988, I shared the costs and timeline for this project with John Hopkinson. The proposal focused on adopting the Wunsch assay for measuring collagenase peptidase activity, the trypsin BAEE substrate to detect clostripain activity, and a generic casein substrate to measure neutral protease activity. After establishing these assays, I would perform small-scale purification runs, using different chromatographic resins to determine if I could separate collagenase from protease activities.  

Camillo Ricordi’s visit to Indianapolis in 1992 gave new life to the project. Camillo advocated that BMB fund an R&D project to identify the key enzymes in Collagenase P responsible for successful human islet isolation (see earlier blog post).  

Immediately after Camillo’s visit, a project plan to develop a consistent and reliable collagenase product for human islet isolation was developed. I shared my earlier plans that defined what was needed to purify and characterize the collagenase and protease enzymes. What was unexpected was the broad scope of the new project proposal. The project would be divided into two teams where five individuals would work on enzyme purification and characterization and five others were responsible for establishing a porcine islet isolation method and applying it to screen purified enzyme formulation to recover islets. Ultimately, the islet isolation team performed 26 human islet isolations to confirm the enzyme formulation used in Liberase HI.(4) 

One critical component that led to the successful development of the Liberase HI product was the unique characteristics of Collagenase P. Collagenase P was introduced by BMB in 1990. Shortly after its introduction, Camillo Ricordi evaluated the product for porcine and human islet isolation and found that one-third of the lots could be used for islet isolation. The percentage of “good” lots was superior to those manufactured by Sigma. 

The Collagenase P lots used for human islet isolation were categorized as “good” and “bad” based on the feedback BMB received from customers. Fortunately, the U.S. sold more of this enzyme than any other country, so I could obtain samples of each lot of Collagenase P for further analysis. I analyzed these lots for collagenase peptidase activity using the Wunsch assay, trypsin-like activity using untreated (i.e., trypsin-like activity) or treated by pre-incubation with a reducing agent (e.g., dithiothreitol or 2-mercaptoethanol) to detect total clostripain activity, or general protease activity using fluorescein-labeled casein.(3) These analyses showed no statistical differences in the biochemical properties of good and bad lots.   

I then looked for alternative methods to analyze collagenase and found reports from Hefley’s lab where they characterized different lots of crude collagenase by using a small-scale, high-performance, strong anion exchange chromatographic procedure that separated C1 from C2 collagenase. Analysis of multiple crude collagenase lots using this method showed that lots that showed poor recovery of cells from the membranous bone of mouse calvaria had very low concentrations of C2 collagenase. They then purified C1 and C2 collagenase from this “bad” lot and prepared a C1-C2 enzyme mixture with a C1:C2 ratio similar to those found in a good lot. This artificially prepared enzyme mixture made a “bad” lot of collagenase “good.” The bone cell yields equivalent to those obtained with a “good” lot of collagenase.(5) 

I adopted Hefley’s anion exchange chromatographic method and analyzed lots of good and bad Collagenase P used for human islet isolation. These results showed that good and bad lots of Collagenase P generated similar chromatographic profiles.  

Stuck, I decided to go back and see if any common features were found in the good Collagenase P lots. To my surprise, there was a pattern as summarized in the table below.  

Enzyme Activity	Assay (Reference)	Range of Enzyme Units/Pancreas
Collagenase	Wunsch	1,500 – 2,500
Clostripain	BAEE	2,000 – 35,000
Neutral protease	Caseinase	50,000 – 140,000

The ratio of C2 to total collagenase (C1 + C2) was between 0.25 to 0.5, while the mean caseinase activity was about 70,000 units (U). 

Later, during the development process, analysis of endotoxin contamination showed that the bad Collagenase P lots contained higher levels of endotoxin than those found in the good lots. This was a surprising finding since it was reported that C. histolyticum was gram-positive and should not contain endotoxin. However, C. histolyticum is gram-variable, so depending on culture conditions, it can contain endotoxin. 

The project began in January 1993 and by August 1993, the islet isolation group confirmed that the purified collagenase-protease enzyme mixture containing a 60:40 C1:C2 collagenase mixture with ≈ 1600 Wunsch U, ≈ 7000 BAEE U of clostripain, and ≈ 70,000 neutral protease U of thermolysin was as effective in recovering porcine islets as a “good” lot of Collagenase P.  

The bench scale purification process was then immediately transferred to Operations to manufacture the enzyme mixture. For the next year, I worked closely with the Operations staff to write and review manufacturing documents. In January-February 1994, members of the islet isolation team visited Ray Rajotte’s lab at the University of Alberta in Edmonton. They found that the first version of the Liberase HI product was very effective in recovering islets from human pancreata. 

In November 1994, the product was launched and the Liberase project team invited Camillo Ricordi and Ray Rajotte to join the celebration. 

In retrospect, the pressure to launch a product ASAP, led to insufficient time to develop methods to remove clostripain from the purified enzyme mixture. Subsequent versions of Liberase HI  first minimized clostripain contamination (version 2) then decreased the neutral protease activity (version 3).(6) 

A standardized enzyme formulation for human islet isolation remains undefined for several reasons. The primary reason is that these formulations represent incremental changes to the original Liberase HI formulation. The hidden bias is that this formulation replicated the enzyme compositions from good lots of Collagenase P. Replication of enzyme formulations that C. histolyticum uses to breakdown dead organic matter puts into perspective how little we know about optimization of enzyme compositions for cell isolation. 

VitaCyte was fortunate to receive a Phase II Small Business Innovation Research award to prospectively evaluate the effect of different doses of recombinant C1 and C2 C. histolyticum collagenase on human islet isolation using a design of experiment (DOE) approach. A fixed dose of BP Protease (i.e., a Dispase equivalent enzyme) was used in all the enzyme formulations. These studies showed that the C1 to C2 activity ratio within the DOE did not affect human islet yield. Moreover, the collagenase dose could be reduced by half using a 30:70 C1:C2 ratio compared to islets isolated using a 60:40 C1:C2 natural purified collagenase mixture containing the same dose of BP Protease.(7, 8) The latter formulation was incremental improvement to version 3 of Liberase HI (substitution of Thermolysin with BP Protease). These studies indicated that these islets had improved in vitro functional activity compared to those isolated using natural purified collagenase and BP Protease.(8)  

Subsequent reports at the University of Alberta in Edmonton confirmed the improved in vitro functional response of human islets using an enzyme mixture containing a 50:50 C1:C2 ratio of recombinant collagenase and the same dose of BP Protease as used above.(9) Here, islets isolated with a lower dose of recombinant collagenase gave a significantly better glucose-stimulated insulin response than those isolated with the standard dose of 60:40 C1:C2 ratio of natural purified collagenase and thermolysin. In addition, a significantly higher number of islets were recovered after short-term cell culture, leading to more islets being transplanted into patients. 

The goal of any future efforts to standardize an enzyme formulation for human islet isolation must focus on minimizing the unknowns in the enzyme composition. This is only achieved after rigorous characterization of the enyzmes used for islet isolation followed by subsequent correlation of these characteristics with islet yield and function. 

References