Identifying Effective Enzyme Activity Targets for Recombinant Class I and Class II Collagenase for Successful Human Islet Isolation

Background

Isolation following a good manufacturing practice-compliant, human islet product requires development of a robust islet isolation procedure where effective limits of key reagents are known. The enzymes used for islet isolation are critical but little is known about the doses of class I and class II collagenase required for successful islet isolation.

Methods

We used a factorial approach to evaluate the effect of high and low target activities of recombinant class I (rC1) and class II (rC2) collagenase on human islet yield. Consequently, 4 different enzyme formulations with divergent C1:C2 collagenase mass ratios were assessed, each supplemented with the same dose of neutral protease. Both split pancreas and whole pancreas models were used to test enzyme targets (n = 20). Islet yield/g pancreas was compared with historical enzymes (n = 42).

Results

Varying the Wunsch (rC2) and collagen degradation activity (CDA, rC1) target dose, and consequently the C1:C2 mass ratio, had no significant effect on tissue digestion. Digestions using higher doses of Wunsch and CDA resulted in comparable islet yields to those obtained with 60% and 50% of those activities, respectively. Factorial analysis revealed no significant main effect of Wunsch activity or CDA for any parameter measured. Aggregate results from 4 different collagenase formulations gave 44% higher islet yield (>5000 islet equivalents/g) in the body/tail of the pancreas (n = 12) when compared with those from the same segment using a standard natural collagenase/protease mixture (n = 6). Additionally, islet yields greater than 5000 islet equivalents/g pancreas were also obtained in whole human pancreas.

Conclusions

A broader C1:C2 ratio can be used for human islet isolation than has been used in the past. Recombinant collagenase is an effective replacement for the natural enzyme and we have determined that high islet yield can be obtained even with low doses of rC1:rC2, which is beneficial for the survival of islets.

Successful allo- or autoislet transplantation requires recovery of a sufficient number of functional islets from cadaveric or severely fibrotic pancreata.^1,2 The dose and composition of the enzymes used in the islet isolation process is a critical factor that impacts the number and quality of islets released from tissue. Several studies indicate the biochemical characteristics of purified collagenase or the choice of neutral protease are associated with lower islet yields, poorer islet recovery after a short culture period, or lower islet viability.^3,4 For many centers, about 50% of the islet isolations do not generate a sufficient number of islets for single donor transplants.^5-7 Poor islet recovery is a critical issue that must be addressed because it continues to plague the economic viability and widespread adoption of islet transplantation.

Further improvements in the human islet isolation process will be required because this process evolves to become a cost-effective, islet therapeutic product manufactured for use in clinical transplantation. Anticipated regulatory oversight will require that a safe, pure, and potent islet product is consistently manufactured using a standardized and validated process.⁸ To achieve this level of control, acceptance criteria that include tolerance limits should be established for the enzymes used in the tissue dissociation process. These essential reagents should be consistently manufactured and thoroughly characterized to determine the effect of their biochemical characteristics on islet quality and yield.

This report summarizes results from a statistically designed experiment using defined amounts of Clostridium histolyticum recombinant class I (rC1) and class II (rC2) collagenase activity to investigate the impact of these activities on human islet yield and function. In these experiments, a fixed dose of a Dispase equivalent neutral protease was used in all the enzyme mixtures. Both split pancreas and whole pancreas models were used to test enzyme targets (n = 20). Donor characteristics matched historical islet isolation (with different enzymes) results (n = 42) were compared with recombinant enzyme islet isolations. The design of experiment (DOE) approach provides a richer insight into enzyme variables that affect recovery of functional human islets because the independent effects of varying levels of C1 and C2 on the release of islets can be assessed while simultaneously determining the interaction between C1 and C2 on islet recovery and function. These results emphasize the importance of validating critical reagents to ensure a robust manufacturing process is used to create a cellular therapy product to treat diabetic patients.

MATERIALS AND METHODS

Donor Pancreas

Human cadaveric donor pancreases (n = 20) were obtained through organ procurement organizations from brain-dead donors after informed consent had been obtained as part of multiorgan procurement. The procured pancreases were shipped in cold University of Wisconsin solution or histidine tryptophane ketoglutarate from the donor center to the islet isolation laboratory.⁹

Islet Isolation Enzymes

Recombinant collagenases were prepared by first synthesizing the gene sequences for intact class I (C1) and intact class II (C2) collagenases. Each gene was incorporated into a vector containing the T7 promoter and an antibiotic resistance gene, then transformed into a low protease Escherichia coli host strain engineered for recombinant protein expression. After antibiotic selection, specific clones containing either the C1 or C2 gene were selected, expanded, and the cells pelleted by centrifugation and stored frozen in glycerol at −70°C. Recombinant collagenase expression was induced by adding synthetic isopropyl β-D-1-thiogalactopyranoside to the cloned cells growing in log phase in culture media free of animal products. On completion of the fermentation, the cells were pelleted by centrifugation, washed, and frozen at −20° C. The frozen cells were thawed, diluted with buffer, and a bacterial cell extract prepared using a hydraulic Microfluidizer. The cell extracts were clarified by high-speed centrifugation and each enzyme purified using a combination of anion exchange, and hydrophobic interaction chromatography resins. No animal products were used in any step of the purification process. The characteristics of the purified recombinant class I (rC1) and recombinant class II (rC2) collagenases are summarized in Table 1.

TABLE 1.

Biochemical characteristics of the various tissue dissociation enzymes used for human islet isolation

The following commercially available enzymes were also used in this study: CIzyme Collagenase HA (natural collagenase blend [NCB]) and CIzyme Bacillus polymyxa protease (BP protease) (VitaCyte, Indianapolis, IN) and NB neutral protease (NP; SERVA, Heidelberg, Germany). The NCB and NP were purified from Clostridium histolyticum culture supernatants whereas the BP protease (a Dispase equivalent enzyme) was purified from Paenbacillus polymyxa culture supernatants. No animal products were used in the P. polymyxa cultures. The key characteristics of these enzymes are also summarized in Table 1.

Human Islet Isolation

Islet isolations were performed as previously reported.⁴ On arrival, the pancreas was trimmed, cannulated, and distended with digestive enzymes. Different combinations of digestive enzymes were used as described below. Enzymes were prepared in 175 mL or 350 mL in Hank Solution for split pancreas or full pancreas isolation. After ductal perfusion of the enzymes, the pancreas was digested in a 250-mL (split pancreas) or 500-mL (full pancreas) Ricordi chamber using a modification of Ricordi's semiautomated method. The digested tissue was then purified by continuous iodixanol (OptiPrep; Axis-Shield, Oslo, Norway) density gradient with a COBE-2991 cell processor (Terumo BCT, Lakewood, CO). The purified islets were cultured in CMRL-1066 supplemented medium (Corning/Mediatech Inc., Manassas, VA) for the indicated number of days prior to functional testing.

Evaluation of BP Protease for Human Islet Isolation

Four human pancreases were used to assess the performance of BP protease relative to NP in human islet isolation. Each pancreas was split into 2 lobes, the head and the body/tail, and perfused with a fixed dose of 20 Wunsch Unit NCB/g combined with either NP (1.75 dimethyl casein U/g) or an equivalent amount of BP protease (24 000 fluorescein isothiocyanate-bovine serum albumin [FITC-BSA] U/g). Each enzyme mixture was alternated between lobes to control for different islet densities within the organ. The order in which the lobes were digested was alternated from 1 pancreas to the next to prevent confounding with cold ischemia time.

Recombinant Collagenase DOE

Twelve human pancreases were used to evaluate high and low target activities of rC1 and rC2. For this DOE, only the body/tail portion of the pancreas was used for the isolations with recombinant enzymes. Doses for rC1 and rC2 were targeted as collagen degradation activity (CDA) units and Wunsch Units per gram, respectively. High (+) and low (−) targets for rC1 (200 000 and 100 000 CDA Units/g, respectively) and rC2 (20 and 12 Wunsch Unit/g, respectively) were randomly assigned to a 2 × 2 factorial design. All rC1rC2 formulations were tested in combination with a fixed dose of 24 000 FITC-BSA Unit/g BP protease. Treatment groups are designated as follows: rC1rC2− −, 100 000 CDA Unit/g and 12 Wunsch Unit/g; rC1rC2+ −, 200 000 CDA Unit/g and 12 Wunsch Unit/g; rC1rC2− +, 100 000 CDA Unit/g and 20 Wunsch Unit/g; and, rC1rC2 + +, 200 000 CDA Unit/g and 20 Wunsch Unit/g. Three human isolations were performed for each treatment group. For each pancreas, the digestion of the body/tail portion was always performed first. Subsequently, the head portion of these same 12 pancreases was digested with our standard composition of natural collagenase in combination with neutral protease (NCB + NP at fixed target doses of 20 Wunsch units and 1.75 DMC units per gram of pancreas, respectively⁴). This isolation served as a control to assess the quality of the pancreas for islet recovery.

After performing DOE, subsequently, we have selected and tested the low activity target formulation (rC1rC2 − −; 100 000 CDA Unit/g, 12 Wunsch Unit/g) in whole human pancreases (n = 4).

Islet Functional Assessments

Isolated islets from both control and experimental groups were assessed for viability using fluorescein diacetate/propidium iodide.¹⁰ Islet potency was monitored by oxygen consumption rate (OCR/DNA)¹¹ measurement. A glucose-stimulated insulin release (GSIR) test was used to assess the islet functional quality and subsequently the stimulation index was calculated.¹²

Comparison With Historical Enzymes Used for Islet Isolation

Islet isolation results of recombinant test enzymes were compared with historical enzymes (n = 42). Donor characteristics of age, body mass index, cold ischemia time matched historical islet isolation results were selectively used and statistically compared (Table 5).

TABLE 5.

Islet isolation aggregate results of recombinant enzymes (D) (n = 12) were compared with donor-characteristics matched historical enzymes (A, B, and C) (n = 42)

Statistical Design and Analysis

For BP protease evaluation, results were compared between the isolations performed with NP and BP protease using 1-way analysis of variance. For the recombinant collagenase DOE, response variables were tested for statistical significance by least-squares analysis of variance for a 2 × 2 factorial design using JMP 8 software¹³(SAS Institute Inc., Cary, NC). The linear model included the main effects of target dose for class I (CDA activity) and class II collagenase (Wunsch activity) and the interaction between CDA and Wunsch activity. Each treatment combination was tested in triplicate in the body/tail portion of the pancreas. Treatments were assigned randomly to blocks using the DOE feature of JMP such that all 4 treatment combinations were tested before any were repeated. All enzyme formulations were blinded such that those performing the islet isolation and subsequent evaluation were not aware of which enzyme activity targets were being used.

RESULTS

Islet Yield With BP Protease

In a split pancreas experiment, BP protease and NP were alternated between the 2 lobes. Digestion profiles, islet morphology, digestion times, islet yield, and functional data were similar between isolations performed with BP protease and NP (Table 2) in combination with natural collagenase indicating BP protease could be used as a protease source for human islet isolation. Glucose-stimulated insulin release and OCR of islets were similar in both groups (Table 2). Subsequently, BP protease was used as the protease for evaluation of rC1 and rC2 at a fixed target dose of 23 400 FITC-BSA U/g pancreas.

TABLE 2.

Evaluation of BP protease for isolating human islets using a split pancreas model

Islet Yield With rC1rC2 in DOE

Results of the DOE indicated that varying the Wunsch and/or CDA target dose, and consequently the C1:C2 mass ratio, had no significant effect on the digestion characteristics measured (Table 3). Similarly, digestions performed with high target doses of Wunsch and CDA activity resulted in comparable islet yields, pre- and postpurification, to those obtained with 60% and 50% the Wunsch and CDA target activities, respectively. Islet function, measured as OCR/DNA and GSIR, was similar among islets isolated with each recombinant enzyme formulation (Table 3). Factorial analysis revealed no significant main effect of Wunsch target activity (rC2 dose) or CDA target activity (rC1 dose) for any parameter measured (Figures 1 and 2). Similarly, the interaction between rC1 and rC2 target dose was not significant for any response variable measured (data not shown).

TABLE 3.

Digestion characteristics and yield and function of islets isolated from the body/tail portion of the human pancreas with high (+) and low (−) target doses of rC1 and rC2 collagenase and a fixed dose of BP protease

FIGURE 1.

Box and whisker plots for the main effect of CDA target activity (rC1 dose). Twelve human pancreases were subjected to a 2 × 2 factorial arrangement of treatments in a DOE. Treatments included high and low target levels for Wunsch and CDA activity. Islets were isolated using a modified Ricordi procedure. Shown in the individual panels are the main effects of CDA target activity for several response variables. Symbols depict values obtained from each individual pancreas. Boxes extend from the 25th to the 75th percentiles. Whiskers represent the minimum and maximum range. Horizontal line within each box represents the median. Response variables were tested for statistical significance by least-squares analysis of variance for a full factorial using JMP 8 software (SAS Institute Inc., Cary, NC). P values presented in each panel are the result of the effects test for CDA target activity. N = 6 pancreases per group.

FIGURE 2.

Box and whisker plots for the main effect of Wunsch target activity (rC2 dose). Twelve human pancreases were subjected to a 2 × 2 factorial arrangement of treatments in a DOE. Treatments included high and low target levels for Wunsch and CDA activity. Islets were isolated using a modified Ricordi procedure. Shown in the individual panels are the main effects of Wunsch target activity for several response variables. Symbols depict values obtained from each individual pancreas. Boxes extend from the 25th to the 75th percentiles. Whiskers represent the minimum and maximum range. Horizontal line within each box represents the median. Response variables were tested for statistical significance by least-squares analysis of variance for a full factorial using JMP 8 software (SAS Institute Inc., Cary, NC). P values presented in each panel are the result of the effects test for Wunsch target activity. N = 6 pancreases per group.

Islet isolation data from the body/tail portion of the pancreas digested with recombinant collagenase mixtures and BP protease (rC1rC2 + BP) were compared with similar results obtained in a split pancreas model using a standard composition of natural collagenase with neutral protease (NCB + NP, Figure 3). Although the rC1rC2 isolations used roughly 40% to 75% the total collagenase dose, islet yields trended higher in the rC1rC2 isolations (5209 ± 522 IEQ/g) relative to NCB + NP isolations (3 616 ± 739 IEQ/g) performed in the same pancreas segment (Figure 3).

FIGURE 3.

Digestion characteristics and islet yields obtained with recombinant collagenase mixtures (rC1rC2) and CIzyme BP protease (BP) compared with data generated in a split pancreas model with a standard composition of natural collagenase [CIzyme Collagenase HA (NCB) and SERVA NB NP at fixed target doses of 20 Wunsch units and 1.75 DMC units per gram of pancreas, respectively]. Mean ± SEM. NCB + NP-head, N = 14 isolations (12 from current study and 2 archival isolations from a previous study⁴); NCB + NP-body/tail, N = 6 isolations (2 from the current study and 4 archival isolations from a previous study⁴); rC1rC2 + BP-body/tail, N = 12 isolations (all from the current study). Bars without a common letter (superscript) are statistically different, P < 0.05.

Isolated islets from all enzyme groups were assessed with GSIR assay and OCR analysis. The results demonstrated that there was no detrimental effect on islet function with recombinant enzymes. The islet OCR/DNA and GSIR stimulation indexes were similar to control islets isolated with natural enzymes.

When we tested the low activity target formulation (rC1rC2; 100 000 CDA Unit/g, 12 Wunsch Unit/g) in whole human pancreases (n = 4), the digestion profiles, including the digestion time (19.1 ± 0.6), were normal although we used enzyme target activities that were only 50% to 60% those of a traditional formulation. Islet yield (IEQ/g pancreas) at prepurification and postpurification was 5918 ± 817 and 5098 ± 258, respectively, for the low-dose rC1rC2 enzyme blend. Islet viability was greater than 90% in all isolations.

Islet Yield From Head Portion of the Pancreas

Shown in Table 4 are the results from the digestions performed in the head portion of the same 12 pancreases used to evaluate the recombinant collagenase formulations. These digestions served as a reference to judge the suitability of the organ for islet isolation and were originally intended for normalizing the islet yield data via analysis of covariance. Unfortunately, a linear relationship between the islet yields in the head and tail could not be statistically established, thus invalidating a covariate analysis. Therefore, a more qualitative assessment of organ quality was performed based on the extensive experience of the islet isolation team. Results of the reference digestions revealed no significant bias in organ quality for any of the four groups (Table 4). Isolated islets from head or body/tail portion of the pancreas showed no difference in functional analysis (Tables 3 and 4).

TABLE 4.

Digestion Characteristics and islet assessment in the reference digestions performed with NCB and C. histolyticum NP in the head portion of the human pancreas to confirm suitability of the organ

Islet Yield Comparison With Historical Enzymes

Historical enzymes A (SERVA Collagenase NB-1 + SERVA Neutral Protease NB), B (VitaCyte CIzyme Collagenase HA + VitaCyte CIzyme Thermolysin), C (VitaCyte CIzyme Collagenase HA + SERVA Neutral Protease NB) were compared (n = 42) with D (cumulative Recombinant enzymes). Both prepurification (6328 ± 2214) and postpurification (5209 ± 2267) islet yields were significantly higher in the recombinant enzyme group when compared to historical enzymes (A and B) as shown in Table 5.

DISCUSSION

The experiments summarized in this report were designed to investigate the use of recombinantly sourced class 1 (rC1) and class 2 (rC2) collagenase on recovery of functional human islets. Experimental design software was used to plan and execute these experiments with high and low target doses of rC1 and rC2 in a factorial arrangement. Consequently, 4 different enzyme formulations with divergent C1:C2 collagenase mass ratios (about 27:73, 38:62, 43:57, and 55:45) were assessed for their effectiveness to recover functional human islets while keeping the neutral protease activity constant. A statistical approach utilizing a DOE to assess the individual activities of C1 and C2 on human islet isolation showed that these 4 enzyme mixtures were equivalent in recovering functional islets from the body/tail of the human pancreas. This empirical strategy is superior to earlier approaches that simply used fixed proportions of C1 and C2 as a single independent variable. This approach is flawed in that it ignores the fact that the 2 classes of collagenase may act interdependently in degrading native collagen. By using a factorial approach, we were able to assess the independent effects of varying levels of C1 and C2 on the release of islets while simultaneously determining interaction between C1 and C2.

Progress in correlation of the biochemical characteristics of collagenase and neutral protease to human islet isolation relies on collaboration of islet isolation experts with commercial manufacturers of purified enzymes. In most cases, laboratories use purified enzyme products because they are supplied by the manufacturer, which accounts for the low number of reports in the literature on this topic. To place the results in our report in context, the purified enzyme products commonly used for human islet isolation can be separated into 2 categories. The first category uses a 60:40 ratio of purified C1:C2 collagenase based on mass.¹⁴ To achieve this ratio, each class of enzyme is separately purified, recombined in a 60:40 C1:C2 mass ratio, and the amount bottled based on the total Wunsch activity. The commercial enzymes in this category are Liberase HI purified Enzyme Blend,¹⁵ Liberase MTF^16,17 (Roche Diagnostics), and CIzyme Collagenase HA¹⁸ (VitaCyte). This ratio, initially used in the Liberase HI product, was determined after biochemical analysis of multiple lots of crude collagenase showed that most lots with approximately this ratio were successfully used to isolate human islets. The second category is purified collagenase where the C1:C2 ratio is not rigorously defined. The SERVA NB-1 collagenase uses a process to separate C1 from C2 but these ratios are not defined by the manufacturer.¹⁹ For all these products, animal containing proteins are used in the C. histolyticum fermentation media to stimulate high levels of collagenase production. These manufacturers also supply purified or enriched bacterial neutral protease which is required for accelerating pancreatic tissue dissociation. These enzymes are purified thermolysin (Roche, VitaCyte), purified Dispase equivalent enzyme (CIzyme BP protease, VitaCyte), or an enriched CHNP (NB Protease, SERVA). Thermolysin and CHNP are purified from culture media containing animal protein, whereas BP protease is purified from culture supernatants of Paenibacillus polymyxa grown in an animal free media. Liberase HI contained about 12 mg of purified Thermolysin and 2200 Wunsch Units of a 60:40 purified C1:C2, but later products sold the purified collagenase and protease separately, leading many laboratories to use different collagenase-protease enzyme compositions for human islet isolation.^20,21

Kin et al²² retrospectively analyzed multiple lots of Liberase HI or the collagenase and thermolysin components used to manufacture this product and found that those lots with lower C2:C1 activity ratios resulted in higher success rates in recovering human islets. The C2:C1 ratios were determined after chromatographic separation of each class of enzyme from a portion of the enzyme solution used in the islet isolation procedure with subsequent assay of enzyme activity using a fluorescein isothiocyanate gelatin substrate. In a subsequent study, this group obtained purified C1, C2, and thermolysin from Roche and found that a 50:50 mass ratio of C1:C2 was not as effective as a 60:40 mass ratio in recovering human islets when approximately the same total Wunsch activity was used in the tissue digestion process.²³ Surprisingly, results from our study showed the mass ratio of C1 and C2 had no bearing on the isolation outcome as the mass of C2 was actually higher than that of C1 in 3 of 4 formulations tested. However, in both of these earlier studies, a confounding factor was the lack of information on the molecular form of C1 or C2 and, thus, the true collagen degradation activity was unavailable. Truncated molecular forms of C1 only possess approximately 10% the collagen degradation activity of the intact protein.^18,24 Additionally, thermolysin was used in these earlier studies rather than BP protease. Together these data suggest that the balance between the 2 classes of collagenase is a relevant variable to consider, but imply the contribution to total collagen degradation activity rather than actual mass should be examined. Interestingly, C2:C1 collagen degradation activity ratios for all formulations assessed in our study were below 0.11, consistent with the observations of Kin et al in which lots of collagenase with C2:C1 activity ratios below 0.204 had increased odds of successful outcomes.

Friberg et al²⁵ reported that only 30% of the total FALGPA activity (measure of C2 activity) is absorbed by the pancreas during the perfusion process, suggesting a vast amount of enzyme is not used with islet isolation protocols considered standard in practice. Data from our study supports this observation as comparable isolation results were observed with C2 and C1 target activities that were approximately 60% and 33%, respectively, of those provided in a traditional 60:40 blend of natural collagenase (eg, new enzyme mixture [NEM]). Furthermore, neither the main effect of Wunsch (C2 activity) nor CDA (C1 activity) approached statistical significance for any parameter measured in the current study, even though low target activities were reduced by approximately 60% and 50%, respectively, relative to the high target activities. This would indicate a minimum threshold of enzyme activity has yet to be defined and strongly suggest further optimization of enzyme formulations for human islet isolation is needed. A potential caveat to this conclusion is that the inherent variability in donor pancreas diminishes the statistical power of our DOE and only allows relatively large differences in any given response variable to be detected at a significance level of P = 0.05.

Subsequent to DOE, we tested the low activity target formulation (rC1rC2 − −; 100 000 CDA Unit/g, 12 Wunsch Unit/g) in 4 whole human pancreases and observed tissue digestion kinetics and average islet yields (5098 ± 258 IEQ/g trimmed pancreas) comparable to those obtained with the more traditional enzyme formulations (≥250 000 CDA Units/g, ≥ 20 Wunsch Units/g; unpublished observations).

Brandhorst et al²⁶ collaborated with Roche to assess the ability of recombinant C1 and C2 collagenase in combination with thermolysin to recover human islets. Their results showed no significant differences in islet yield, islet size distribution, or islet function when data from isolations performed with a recombinant collagenase formulation similar to Liberase HI were compared to those obtained with Liberase HI. The results presented in our report also confirmed the efficacy of the recombinantly sourced collagenases but demonstrated comparable results could be obtained with much lower Wunsch target activities than those used by Brandhorst et al. The rC1 and rC2 used in our report have slightly different amino acid sequences than those used by Brandhorst et al. with four and thirteen amino acid sequence differences for rC1 and rC2, respectively.^27-29 Whether the differences between the primary sequences can account for this presumed increase in potency is unknown, but the recombinant C1 used in the current study also possesses a higher CDA specific activity in vitro relative to natural C1 manufactured by VitaCyte.

A secondary objective of this study was to assess the effectiveness of the purified collagenase-BP protease mixture relative to similar reports which evaluated the ability of various enzyme mixtures to recover functional human islets. The majority of these reports compare 1 commonly used enzyme mixture against another enzyme mixture. In contrast, a previous report compared human islet isolation results obtained from 9 different purified enzyme mixtures (n = 99).⁴ Their results showed that collagenase products that contained higher proportions of intact C1 gave higher islet yields than those comprised of more truncated C1 and that NP was superior to thermolysin in recovery of human islets. From this analysis, an NEM comprised of VitaCyte's CIzyme Collagenase HA and SERVA’s NB protease was created. This NEM formulation was reported to be superior to all other enzyme mixtures for recovering human islets. Aggregate results from the four different recombinant collagenase-BP protease formulations used in our study gave a 44% higher yield in the body/tail of the pancreas (n = 12 isolations) when compared with those from the same pancreas segment using the NEM formulation (n = 6). Higher islet yield per gram pancreas was also obtained with rC1 and rC2 enzymes compared with historical enzymes (Table 5).

Our results demonstrating the effectiveness of BP protease in combination with either natural or recombinantly sourced collagenase for use in human islet isolation confirms an earlier report which demonstrated the superiority of BP protease relative to thermolysin for isolation of human islets.³⁰ Here, islets isolated with a purified natural collagenase-BP protease mixture had significantly higher viabilities after 2 days of culture at 37°C relative to those isolated with collagenase-thermolysin. An added advantage for using the recombinant collagenase-BP protease enzyme mixture is the absence of animal components used to generate the raw material or during the manufacturing process.

These results indicate that more work is required to develop a robust enzyme formulation for human islet isolation. Commercial manufacturers of collagenase containing tissue dissociation enzymes play an essential role for providing appropriately characterized reagents for the scientific community to use for islet isolation. The results in this report show a broader range of C1:C2 can be used successfully for human islet isolation but the limits of this range were not identified. The neutral protease used for human islet isolation is critical for successful recovery of islets.^3,4 A rigorous study determining the selection and optimal dose of neutral protease has not been performed, and it is likely that combinations of neutral proteases may be more effective than a single neutral protease to improve islet recovery. These questions should be addressed because manufacturing human islet therapeutic products are used in approved cellular transplantation procedures to manage adult type 1 diabetic patients.

In conclusion, recombinant collagenase is an effective replacement for natural enzyme when used for human islet isolation. A lower-dose recombinant collagenase in combination with BP protease was successfully used to recover greater than 5000 IEQ/g pancreas. Results from this study indicate the effectiveness of this animal-free enzyme mixture to recover islets at doses of collagenase that are approximately 50% of those used commonly for human islet isolation.