Researchers have designed a fusion protein that joins human erythropoietin (EPO) to an engineered variant of human albumin, and the design has produced striking results in mouse models. When the fusion, called EPO-QMP, was given intravenously to mice expressing human FcRn, its plasma half-life was 5.2-fold longer than that of an identical fusion built on wild-type albumin, reaching 5.7 days compared with 1.1 days. The work, published in Nature Communications, also showed that the engineered fusion could cross mucosal barriers: after intranasal administration, mice carried five times more EPO-QMP in their circulation than the wild-type fusion, and the treated animals showed a prolonged rise in erythrocytes compared with mice given the clinical drug darbepoetin alfa.
EPO products are widely used to stimulate red blood cell production in patients with anemia, including anemia caused by chronic kidney disease, cancer chemotherapy, or zidovudine treatment after HIV infection. The problem with existing recombinant EPO is its short persistence in the blood: the first marketed products have a plasma half-life of only 4 to 11 hours after intravenous administration in humans, rising to about a day after subcutaneous injection. Patients therefore receive injections up to several times per week, which burdens patients and affects compliance. Earlier engineering efforts have addressed this, notably darbepoetin alfa, which carries five amino acid substitutions creating two extra N-glycan sites and extends the human half-life to 26.3 hours after intravenous dosing, and methoxy PEG-epoetin beta, which reaches 134 hours through attachment of a large PEG moiety.
The new strategy relies on the biology of the neonatal Fc receptor, FcRn, a recycling receptor that protects albumin and IgG from lysosomal degradation and gives both proteins a plasma half-life of roughly three weeks in humans. Albumin and IgG enter endothelial cells by fluid-phase pinocytosis, bind FcRn in acidified endosomes at pH 5 to 6, and are released back into the blood at neutral pH. FcRn also mediates bidirectional transcytosis across polarized epithelial layers at mucosal barriers, and previous work by the same group showed that albumin is transported more efficiently than IgG from the luminal to the basolateral side after intranasal delivery in human FcRn transgenic mice. The team fused EPO to the N-terminus of a QMP albumin variant carrying three substitutions, E505Q, T527M and K573P, which had previously been shown to improve pH-dependent binding to human FcRn 180-fold. Two control fusions were also made, one with wild-type albumin and one with an AQ variant engineered for reduced receptor binding.
All three fusions were produced in human cell lines and purified to stable monomeric fractions of the expected 84.9 kilodalton size, with melting temperatures between 59.3 and 61.8 degrees Celsius, about 16 degrees higher than that of darbepoetin alfa. Functional testing confirmed that the fusion did not damage EPO’s activity. Surface plasmon resonance against immobilized human EPO receptor yielded sub-nanomolar equilibrium dissociation constants of 0.09 to 0.12 nanomolar for all three fusions, in the same range as the 0.07 nanomolar measured for darbepoetin alfa, and similar sub-nanomolar binding was observed against the mouse form of the receptor. In a proliferation assay using the human TF-1 erythroblast cell line, all three fusions stimulated cell growth to a similar extent as darbepoetin alfa, while unfused albumin alone had no effect.
Binding studies confirmed that the QMP engineering survived the fusion. In an ELISA and in SPR measurements at pH 5.5, EPO-QMP bound human FcRn with a dissociation constant of 0.7 nanomolar, essentially the same as unfused QMP at 1.0 nanomolar and far stronger than EPO-WT at 228.5 nanomolar, while the AQ fusion showed negligible binding. Analytical FcRn affinity chromatography, which mimics receptor release during cellular recycling, showed that EPO-QMP was released at pH 8.0, the same as unfused QMP and at a higher pH than the WT counterparts, indicating that fusing EPO did not alter the pH-dependent binding behavior. A cellular recycling assay in FcRn-overexpressing endothelial cells showed that the amount of EPO-QMP recycled out of the cells was ten to eleven times higher than that of the WT fusion.
These properties translated into markedly improved pharmacokinetics in vivo. In human FcRn transgenic Tg32 mice given 2.0 milligrams per kilogram intravenously, EPO-QMP was cleared at 60.3 milliliters per day per kilogram, 1.6-fold slower than EPO-WT, and the AQ variant was cleared rapidly at 193.9. EPO-QMP remained detectable in plasma for up to 27 days after administration, whereas EPO-WT fell below detection after 10 days. Subcutaneous administration of EPO-QMP at the same dose produced a maximum plasma concentration 6.7-fold lower than the intravenous route, peaking at 24 hours rather than 5 minutes, but the resulting half-life of 4.9 days was not statistically different from the intravenous value. Testing in a Tg32 mouse strain that also lacks mouse albumin revealed the importance of competition: without endogenous mouse albumin competing for FcRn, the half-lives of EPO-QMP and EPO-WT became similar, at 8.1 and 7.4 days respectively, showing that the QMP advantage in standard Tg32 mice arises because mouse albumin competes more effectively against wild-type than engineered albumin for receptor binding.
The mucosal delivery results followed from the same receptor biology. In a transwell system using polarized MDCKII cells overexpressing human FcRn, 4.3-fold more EPO-QMP than EPO-WT was transported from the apical to the basolateral side over four hours, although the fusion transported about 2-fold less efficiently than unfused QMP albumin. In mice, intranasal delivery of 80 micrograms of EPO-QMP produced five times more circulating fusion protein than the same dose of EPO-WT, measured as area under the curve over 9 to 96 hours. Lower amounts of EPO reached the blood after intranasal than after intravenous delivery, but the albumin signal was comparable between routes, and the study demonstrates that the engineered fusion can be delivered non-invasively across mucosal surfaces in this model.
Comparisons with the licensed drug darbepoetin alfa favored the new design in this mouse system. After equimolar intravenous dosing in Tg32 mice, both EPO-QMP and darbepoetin alfa increased erythrocyte counts and hematocrit at day 7, but the darbepoetin alfa group then declined while EPO-QMP-treated mice maintained high levels through day 14, and by day 24 the EPO-QMP group still had significantly more erythrocytes than both the darbepoetin alfa and untreated groups. The pharmacokinetic difference was visible in the drug levels themselves: darbepoetin alfa concentrations fell 11-fold between days 2 and 4, versus a 5-fold decrease for EPO-QMP. After intranasal delivery, EPO-QMP again outperformed darbepoetin alfa on both erythrocyte and hematocrit measures, with five times more EPO detected in whole blood at day 3 for the fusion group.
The authors note several caveats. The in vivo work was conducted in mice expressing human FcRn, with mouse albumin present as a competing ligand and mouse EPO receptor acting as an antigen sink, all factors that shape the measured pharmacokinetics, and the elimination curve for EPO-QMP displayed several phases suggesting multiple clearance mechanisms. The team states that further studies are needed to compare EPO-QMP’s efficacy with other erythropoiesis-stimulating agents, and that the potential of the QMP design warrants testing in non-human primates and with broader dosing regimens. On safety, no anti-drug antibodies against the EPO-albumin fusions or unfused QMP albumin were detected in plasma samples collected from the mice on day 27, and the authors point to prior clinical experience with an approved wild-type albumin fusion protein in which neutralizing antibodies were not detected in a phase 3 trial.
The overall conclusion from the study is that EPO fused to engineered QMP albumin is an attractive format to explore for anemic conditions, both through traditional injection routes and through delivery across mucosal barriers. For patients needing long-term or lifelong treatment, less frequent dosing and the possibility of self-medication at home could be meaningful benefits, although these remain goals for future research rather than demonstrated clinical outcomes. The work illustrates how a detailed understanding of FcRn-mediated recycling and transcytosis can be turned into a rational protein design, and it supports the broader use of engineered human albumin as a carrier for therapeutic proteins where extended half-life and improved patient compliance are priorities.
Subject of Research: Engineering of an erythropoietin-albumin fusion protein with prolonged plasma half-life and mucosal delivery capability
Article Title: A human erythropoietin-albumin fusion protein tailored for enhanced plasma half-life and delivery across mucosal surfaces
Article References: Anthi, A. K., Bejan, D., Mester, S., Bethur, N., Gonzalo-Gil, E., Benjakul, S., Aaen, K. H., Schlothauer, T., Sandlie, I., & Andersen, J. T. (2026). A human erythropoietin-albumin fusion protein tailored for enhanced plasma half-life and delivery across mucosal surfaces. Nature Communications, 17(1), Article 10780. https://doi.org/10.1038/s41467-026-78243-z
Image Credits: AI Generated
DOI: 10.1038/s41467-026-78243-z
Keywords: erythropoietin, albumin fusion, FcRn, anemia, plasma half-life, intranasal delivery, darbepoetin alfa, transcytosis, pharmacokinetics, biotechnology, protein engineering, mice study
News Source: Ophelia Keating. (October 28, 2026). Engineered albumin fusion extends erythropoietin half-life and enables nasal delivery in mice. Scienmag.
