With the proof of concept demonstrated, significant effort was required to scale up the extraction and purification of placental glucocerebrosidase to make it clinically and commercially viable. The magnitude of this challenge can be appreciated by the fact that it took ~20,000 placentas to extract enough p-glucocerebrosidase to treat 1 patient for 1 year. Genzyme Corporation adapted the purification and carbohydrate remodeling schemes developed by Brady, et al., to improve both the purification efficiency and purity of the final product. Since the source material for the enzyme was human placentas, a great deal of care was taken to ensure that the final production process had sufficient viral reduction steps to eliminate any risk of viral contamination. In addition to scaling up the p-glucocerebrosi-dase purification, it was also essential to scale up the production of the three carbohydrate-remodeling enzymes and implement additional chromatographic steps for their removal. The resulting highly purified carbohydrate-remodeled p-glucocerebrosidase was formulated with 1% human serum albumin and sold commercially as Ceredase (Alglucerase).
The clinical trial that led to the approval of Ceredase was conducted on 12 patients, 4 adults and 8 children who were classified with Type I Gaucher Disease. Patients were given 60 IU of Ceredase/kg of body weight by intravenous infusion once every 2 weeks for between 9 and 12 months. Part way through the trial, the dose of two severely affected children was increased to 60 IU/kg every week. All patients in the trial showed a clinical response with a reduction in spleen volume and an increase in hemoglobin. Other clinical improvements observed in several patients included a reduction in liver size, reduction in plasma glucocerebroside and serum levels of tartrate-resistant acid phosphatase (a lyso-somal enzyme that is elevated in some lysosomal storage diseases), and an increase in platelet count. Some evidence of an improvement in the bone disease was observed in three of the patients .
Since Ceredase was a natural product of human origin, a higher level of host protein impurities was allowed than would be typical of a recombinant product today. Detection of impurities was also challenging since the protein was formulated with a significant molar excess of human serum albumin. Following product approval and during the development of reversed-phase high-performance liquid chromatography (HPLC) methods to determine protein purity, two prominent (relative to other impurity peaks) protein impurity peaks were detected in the Ceredase preparations. Isolation and N-terminal sequence analysis identified the two peaks as the a and p subunits of human chorionic gonadotropin (HCG). Although product containing HCG had been given to several patients for an extended period of time, no serious adverse events were reported [58,59]. The low level (~2%) of HCG contamination and the fact that it was subjected to the carbohydrate remodeling process most likely contributed to the lack of adverse events associated with the presence of HCG. The carbohydrate remodeling step also resulted in the oligosaccharides of HCG terminating in mannose residues, producing a very short half-life relative to native HCG. Once the HCG impurity was identified, the purification scheme was modified to eliminate this impurity from later preparations.
Although the presence of a low level of placental protein impurity was acceptable, as the impurities were also human proteins and would not be expected to be immunogenic, complete removal of the three nonhuman-derived enzymes required for carbohydrate remodeling was essential, and development of specific and sensitive assays for detecting residual enzymes was required. The efficiency of the purification scheme in removing these proteins is demonstrated by the low level of immune response in patients who received Ceredase at doses often exceeding 100 mg per infusion .
Even before Ceredase received approval as a therapy for Gaucher Disease, its safety and efficacy were readily apparent. But it was also apparent that the clinical supply of Ceredase would eventually be limited by the availability of human placentas. Another important consideration at the time was the potential risk of viral contamination in a product derived from human tissue. This risk was extremely low in the case of Ceredase as placentas were screened for viral contamination, and the harsh detergent-based extraction procedure as well as the purification steps were very effective viral reduction steps. Given these considerations, development of a recombinant product, Cerezyme (Imiglucerase), was initiated even before the approval of Ceredase.
Since P-glucocerebrosidase is a large glycoprotein and carbohydrates are essential for correct folding and activity, a mammalian expression system was chosen. The cDNA for P-glucocerebrosidase was cloned into CHO cells using a dihydrofolate reductase (DHFR) expression system. Following amplification, highest producing clones were adapted to an anchorage-dependent serum-free microcarrier spinner culture system for production evaluation . The highest producing line was then progressively scaled up from spinner flasks to production bioreactors of increasing size up to 2000 L. The bioreactors were operated in a continuous perfusion mode to maximize production efficiency and reduce the potential for proteolytic degradation or other posttranslational modifications of the product due to cell lysis. The Cerezyme purification process was similar to that used for the production of Ceredase and was based on standard chromatographic procedures. Since the biore-actors were harvested in a serum-free mode during production, the initial impurity burden on the purification scheme was less than that for the Ceredase purification process. But because Cerezyme was produced using recombinant technology in a CHO cell line, a significantly higher level of purity was required, necessitating modifications of the purification scheme. These increased purity requirements also necessitated the development of additional analytical tools such as an enzyme linked immunosorbent assay (ELISA) to detect host cell proteins, reversed-phase HPLC, and peptide-mapping methods.
In developing the recombinant product, it was also desirable to eliminate the human serum albumin that was used as a stabilizer in Ceredase to reduce further the risk of viral contamination. In the absence of a protein stabilizer, P-glucocerebrosidase is prone to aggregate at the concentrations required for a commercial product. This required a change from the liquid formulation used in Ceredase to the lyophilized formulation used today for Cerezyme.
One of the challenges of developing recombinant protein products is the potential for differences between the recombinant product and the naturally occurring protein. These differences can result in increased immunogenicity, altered biodistribution, efficacy, and safety. In many cases, direct comparison of the recombinant protein with the natural protein is not feasible owing to a lack of sufficient natural protein or processing of the enzyme during extraction or isolation, as may occur in plasma proteins isolated from urine. The availability of highly purified human placental P-glucocerebrosidase (Ceredase) allowed a detailed comparison of the recombinant product with to the placental protein. This comparison was subsequently used as the basis of comprehensive analyses for demonstrating product comparability during manufacturing scale-up and other process changes.
The biochemical comparison of the two proteins was based on detailed structural protein analysis, enzymatic properties, and in vitro binding and uptake studies. One of the key methods incorporated in this analysis was peptide mapping. When the Cerezyme peptide-mapping procedure was developed, electrospray mass spectrometry was in its infancy and was just being developed for protein applications. Peptide mapping is a powerful tool for studying posttranslational modifications as well as differences in amino acid sequence and proteolytic processing. Without mass spectrometry, the power of the peptide map hinges on the resolution and reproducibility of enzymatic digestion, chromatographic separation, and identification of peptides by N-terminal sequencing.
For Cerezyme and Ceredase, trypsin was chosen for peptide mapping because it resulted in peptides of suitable length for both recovery from the HPLC column and N-terminal sequence analysis. Digestion of Cerezyme with trypsin produces 44 theoretical tryptic peptides (Table 6.3); however, the actual number of peptides observed differed from the theoretical value due to partial and nonspecific cleavages. When the peptide maps of Cerezyme and Ceredase were compared (Figure 6.2), the tryptic maps obtained were very similar even at the level of minor peaks, with the exception of four peaks (labeled A to D). Interestingly, the four differences observed between the Cerezyme and Ceredase tryptic maps all arise from a single amino acid substitution at position 495. The cDNA sequence published by Sorge  was used for the development of the Cerezyme-producing cell line. This sequence was subsequently found to contain a sequence error arising from a cloning artifact that resulted in the encoding of a histidine residue instead of an arginine residue . The protein sequence of Cerezyme therefore differs from that of Ceredase at residue 495. In Ceredase, the C-terminal sequence is WRRQ and in Cerezyme the corresponding sequence is WHRQ. This results in tryptic cleavage at residue 495 in Ceredase compared with cleavage at residue 496 in Cerezyme. In the case of both Ceredase and Cerezyme, incomplete cleavage of the C-terminal peptide is observed to give rise to peak C in the Cerezyme map and peak D in the Ceredase map. The complete cleavage product, peak B in the Cerezyme map, is not present in the Ceredase map, presumably due to the presence of double arginines that apparently affect the tryptic cleavage. Peak A in the Cerezyme map arose from a chymotryptic-like cleavage at Y487, while a corresponding cleavage was not observed in the Ceredase map.
This single amino acid difference between the natural (Ceredase) and the recombinant protein (Cerezyme) does not have any significant effect on the enzymatic activity of the proteins since the two are enzymatically indistinguishable (Table 6.4).
The oligosaccharide structures present on placental P-glucocerebrosidase had previously been shown to be a combination of complex and oligomannose oligosaccharides, but the site-specific distribution of these structures had not been determined . The carbohydrate remodeling process (Figure 6.3) used in the production of Ceredase remodels the complex structures down to the GlcNAc2 Man3 core structure but does not affect the oligo-mannose oligosaccharides, resulting in a combination of mannose-terminated core structures and oligomannose chains.
Amino Acid Number
Was this article helpful?