The differences between the N-glycans of human erythropoietin (hEPO), extracted from serum, and recombinant human erythropoietin (rhEPO) reported by Skibeli et al1 have received attention in the editorials of 2 other journals. One suggested that these differences might be involved in the etiology of the autoimmune pure red cell aplasia found in some patients treated with rhEPO.2 The other suggested that these differences might form the basis for a test to detect the doping of athletes with rhEPO.3 A review of the data of Skibeli et al, however, indicates that the differences they have reported between theN-glycans of hEPO and rhEPO are not consistent with the differences between the physicochemical properties of these EPOs found in other laboratories. This discrepancy may be a consequence of the effects on the N-glycans of hEPO of the conditions used by Skibeli et al to extract hEPO from serum in order to compare it with unextracted rhEPO.
Skibeli et al reported that hEPO lacked the tetra-acidic (tetra-sialylated) N-glycans found in the rhEPOs, and was also lower in its content of tri-acidicN-glycans.1 These findings are summarized in their Table1,1 which also permits comparison of the N-glycan charges of different EPOs, as calculated, for example, by the method of Hermentin et al.4 This suggests that the total negative charge of the N-glycans of hEPO is < 70% that of the rhEPOs, and implies that hEPO is less acidic than the rhEPOs. Thus the sialic acid residues of the N-glycans of EPO represent up to 12 of a possible total of 14 sialic acid residues in EPO, and make a major contribution to the net negative charge of EPO, as indicated by the fact that the pI of intact EPO is in the range of about 2.5 to 4.0,5 and the pI of desialylated EPO is about 8.5.6 However, other studies have consistently shown hEPO to be more acidic than rhEPO,7-9 and, indeed, this is the basis for an established test for doping.9
The discrepancy between the findings of Skibeli et al and those of others is probably due to the conditions used to extract the hEPO from serum to compare it with unextracted rhEPOs. Although the data in Figure 8 of Skibeli et al is said to demonstrate “the similarity of sugar profiles from rhEPO with or without a bead-extraction step,” comparisons of the areas under the 2 elution profiles indicates that the extraction procedure has reduced the recovery of all oligosaccharides with elution times of 60 minutes or more, representing tri- and tetrasialylated N-glycans, and has increased the relative recovery of most of the neutral, mono- and di-sialylatedN-glycans.1 This effect of the extraction procedure probably represents some desialylation of theN-glycans of EPO, since the 20min-treatment with 10mmol/l HCl, used to dissociate the antibody-bound EPO, is also commonly used, albeit at 80°C rather than ambient temperature, for the quantitative desialylation of N-glycans.10
Inherent charge properties of the isolated sugar parts of human serum erythropoietin
Drs Storring and Yuen claim our data1-1 to be inconsistent with other reports describing the molecular differences between endogenous human erythropoietin (hEPO) and recombinant human EPO (rhEPO) due to desialylation of human serum EPO caused by the extraction conditions used in our study. But they do not take into consideration that we analyzed EPO from human serum, while the reports they refer to1,2 all investigated EPO from human urine, a completely different matrix.
This is important as charge profiles of glycoproteins undergo changes during renal secretion.1-7,1-8 In fact, 2 of the papers Storring and Yuen refer to1-2,1-5 described human serum EPO as more basic than human urinary EPO when analyzing the intact glycoprotein by isoelectric focusing. Furthermore, one of the other papers referred to as contradictory to our study1-3 showed a more basic charge pattern of human urinary EPO than the one obtained for rhEPO. In addition, Storring has used this urinary EPO preparation in his own work when comparing the isoelectric pattern of different batches of rhEPO with human urinary EPOs,1-4 reproducing the findings of Imai et al.1-3 Storring and Yuen also mentioned the shift in isoelectric point to 8.5 caused by the desialylation of hEPO,1-3 as an illustration of the contribution of sialic acids to the net charge of hEPO without commenting that Imai et al used recombinant hEPO and not endogenous hEPO.
Furthermore, during discussion of our results, Storring and Yuen did not take into consideration the fact that our findings of a reduced sialylation of the glycans of human serum EPO referred to the isolated sugar part, which cannot be directly compared to studies of the charge pattern of the intact glycoprotein. This point has previously been elaborated by Tsuda et al,1-9 who reported that the glycans from rhEPO contained more sialic acids than glycans from human urinary EPO, indicating that sugar from human urinary EPO is more basic than sugar from rhEPO.
In addition, we have presented results obtained from the analyses of serum EPO from anemic patients that must be taken into consideration when interpreting our results. In our paper all relevant reports, including the papers mentioned by Drs Storring and Yuen, were thoroughly referred to and discussed.