Summary. Eight preparations of recombinant human erythropoietin (EPO) with differing isoform compositions were produced by using different culture conditions and purification procedures. The N-glycan structures of these EPOs were analysed using a recently developed profiling procedure and identified using matrix-assisted laser desorption ionization mass spectrometry. The specific activities of each of the EPOs were estimated by in vivo and in vitro mouse bioassays. The eight EPOs were found to differ in their isoform compositions (as judged by isoelectric focusing), their N-glycan profiles, and in their in vivo and in vitro bioactivities. N-glycan analyses identified at least 23 different structures among these EPOs, including bi-, tri-and tetraantennary N-glycans, with or without fucosylation or N-acetyllactosamine extensions, and sialylated to varying degrees. Mass spectrometry also indicated the presence of N-glycans with incomplete outer chains, terminating in N-acetylglucosamine residues, and of molecular masses consistent with phosphorylated or sulphated oligomannoside structures. The tetrasialylated tetra-antennary N-glycan contents of the eight rEPOs were found to be significantly and positively correlated with their specific activities as estimated by mouse in vivo bioassay, and significantly and negatively correlated with their specific activities as estimated by mouse in vitro bioassay. It was concluded that the tetrasialylated tetra-antennary N-glycan content of EPO is a major determinant for its in vivo biological activity in the mouse.

The International Standard (IS) for Recombinant DNA-Derived (rDNA) Erythropoietin (EPO) (in ampoules coded 87/684) and three other rDNA EPO preparations in ampoules coded 87/690, 87/696 and 88/574 respectively, were compared with two preparations of highly purified human urinary (HU) EPO and the 2nd International Reference Preparation of Human Urinary Erythropoietin for Bioassay (2nd IRP) by 26 laboratories in 11 countries using a wide range of in-vivo and in-vitro bioassays and immunoassays. These EPO preparations were also compared by electrophoresis and isoelectric focusing. Estimates of EPO content in terms of the 2nd IRP by all in-vivo bioassay methods gave combined unweighted geometric means (with 95% fiducial limits) of: 86 (75-99) IU/ampoule for the IS, 81 (70-94) IU/ampoule for 87/690, 58 (48-71) IU/ampoule for 87/696 and 120 (100-143) IU/ampoule for 88/574. Mean estimates of EPO content in terms of the 2nd IRP by in-vitro bioassays (except receptor assays) were larger than, and those by immunoassays were similar to, the mean estimates by in-vivo bioassays. The use of purified rDNA or HU EPO as standards in place of the 2nd IRP reduced the inter-laboratory variability of estimates of purified EPO preparations by in-vivo and in-vitro bioassays and by immunoassays, and reduced the variability of overall mean estimates for each of these preparations between the three types of method. The inter-laboratory variability of immunoassay estimates of human serum EPO was similar whether the 2nd IRP or one of the purified EPOs was used as standard. Significant differences in in-vivo and in-vitro biological, immunological and physicochemical properties were found between these four rDNA EPO preparations and between them and the HU EPO in the two purified preparations and in the 2nd IRP. There were also differences between the immunoreactivities of the two serum EPO samples included in the study, and between them and the immunoreactivities of the purified EPOs. The differences between rDNA EPOs appeared to be related to differences between the cells used for their biosynthesis, but may also be the result of differences in purification methods and of inter-batch variations. Significant differences in assay specificity were observed within each of the three general types of method. The specificity of the in-vivo bioassays was influenced by the route of hormone administration. The specificities of the mouse spleen cell in-vitro bioassays differed from that of the mouse spleen receptor-binding assay.(ABSTRACT TRUNCATED AT 400 WORDS)

Assays have been developed for the isoforms of erythropoietin (EPO) based on their binding to eight different lectins. These assays were used to compare the isoform compositions of two preparations of human urinary EPO (uEPO) and four preparations of recombinant DNA-derived human EPO (rEPO), which had been shown to differ in their biological and immunological properties and in their isoform composition as judged by isoelectric focusing and electrophoresis. Agarose-bound Ricinus communis agglutinin I (RCA), Erythrina cristagalli agglutinin (ECA), Maackia amurensis leukoagglutinin (MAL), Sambucus nigra agglutinin (SNA), Lycopersicon esculentum agglutinin (LEA), concanavalin A (Con A), Phaseolus vulgaris agglutinin-L4 (L-PHA) and Agaricus bisporus agglutinin (ABA) were used to bind EPO isoforms possessing: N-glycans containing non-sialylated outer Gal beta 1-4GlcNAc (RCA and ECA), NeuAc alpha 2-3Gal beta 1-4GlcNAc (MAL), NeuAc alpha 2-6Gal (SNA), or repeating Gal beta 1-4GlcNAc sequences (LEA); biantennary N-glycans (Con A); tetraantennary and 2,6-branched triantennary N-glycans (L-PHA); and O-glycans containing NeuAc alpha 2-6GalNAc (SNA) and Gal beta 1-3GalNAc (ABA). Free EPO was measured by mouse spleen cell bioassay or immunoassay. Estimates from most lectin-binding assays were reproducible between assays and batches of lectin-agarose, although batches of MAL- and ABA-agarose, and to a lesser extent LEA-agarose, differed in their EPO-binding. Lectin-binding assays showed differences between the isoform compositions of all EPOs, including the two Chinese hamster ovary cell-derived rEPOs, with RCA- and ECA-binding assays being the most discriminating. Lectin-binding estimates provided evidence that uEPO differs from these rEPOs in its lower content of isoforms with biantennary N-glycans and higher content of those with multiantennary N-glycans, and in its lower content of isoforms with N-glycans possessing repeating Gal beta 1-4GlcNAc sequences and of those with O-glycans containing Gal beta 1-3GalNAc. Lectin-binding estimates also indicated that, contrary to some reports, uEPO possesses Gal beta 1-3GalNAc-containing O-glycans but not NeuAc alpha 2-6GalNAc-containing O-glycans or NeuAc alpha 2-6Gal-containing N-glycans. Most groups of lectin-bound EPO isoforms did not differ in their relative bioactivities and immunoreactivities. However, estimates for ABA-bound EPO isoforms suggested that O-glycans might influence the bioactivity of EPO differently to its immunoreactivity. Furthermore, the bioactivities of some ECA-bound EPO isoforms were higher, and those of some of the MAL-bound EPO isoforms lower, than their immunoreactivities, consistent with the reported enhancement of EPO in vitro bioactivity by desialylation.

The preparation and nature of the International Reference Preparation of Human Chorionic Gonadotrophin (HCG) for Immunoassay (IRP), as well as that of a second batch of ampoules (HCG 75/589) prepared identically from the same HCG preparation, are described.A collaborative study of these materials was carried out by 11 laboratories in eight countries, using different bioassay and immunoassay methods. Using the various in-vivo and in-vitro bioassays and receptor assays, the mean log potency estimates for each method within each laboratory of the HCG content of ampoules of the IRP, in terms of the Second International Standard of Human Chorionic Gonadotrophin for Bioassay (IS), were homogeneous and gave an overall weighted geometric mean (95% confidence limits) of 650 (632\p=n-\669) International Units (i.u.)/ampoule. There was considerable heterogeneity of potency estimates of the IRP in terms of the IS both within and between many of the immunoassay systems (reflecting the impurity of the IS), and hence attempts to calibrate the IRP with immunoassay systems of different specificities were invalid.Immunoassay estimates of the HCG content of preparations of serum and urine, in terms of the IRP, showed considerable heterogeneity between assay systems (although the degree of this heterogeneity was no greater than that observed using the IS as standard), but the ranking order between preparations was consistent.Confirmation was obtained that contamination of the IRP with HCG-\g=a\ and HCG-\g=b\ subunits was insignificant. Accelerated degradation studies of the IRP stored at increased temperatures suggested that its stability under normal storage conditions would be satisfactory.It was agreed that the IRP was suitable to serve as an international reference preparation for immunoassay, and it was assigned a unitage of 650 i.u./ampoule on the basis of bioassay calibration.Since the ampoules of HCG (75/589) did not differ significantly from the IRP in any of the assay systems studied, it appeared to be equally suitable as a reference preparation.The International Reference Preparations of the \ g=a\ and \ g=b\ Subunits of Human Chorionic Gonadotrophin for Immunoassay are also described.

No abstract