Second, Ser25 did not appear to form a hydrogen bond across the helical interface, but rather hydrogen bonds with the main chain carbonyl of Ile21 on the same helix, consistent with the mutational data shown in Figure 7 that a hydrogen-bonding side-chain at position 25 is not required for activity. After selection with puromycin, growth factors were removed from the medium and viable cells were counted over time. Although pL was inactive, cells Hederagenin expressing pL-GIPSF, the construct containing the interface predicted by the Put3 model, conferred growth factor independence, demonstrating that the predicted interface residues are sufficient to confer biological activity. To determine if the interface residues were sufficient for dimerization, cell extracts were prepared from BaF3/HA-hEPOR cells expressing pL and the interface add-back construct. The samples were then immunoprecipitated with aE5, subjected to SDS-PAGE under reducing and non-reducing conditions, and immunoblotted with aE5. As shown in Figure 9C, in the presence of reducing agents, both constructs were expressed at similar levels, demonstrating that the inactivity of pL was not due to poor expression. In the absence of reducing agents,Betulonic acid pL migrated primarily as a monomer, while the add-back construct migrated primarily as a dimer. This result demonstrated that the predicted interface residues, Gly15, Ile18, Pro22, Ser25, Phe29, restoring the GxxxG motif, are sufficient in a poly-leucine context for homodimer formation and biological activity. Protein engineering and directed evolution are powerful approaches to design, optimize, and analyze biologically active proteins. In previous work, we isolated an artificial, dimeric, 44amino acid transmembrane protein, TC2-3, which activates the hEPOR and supports erythroid differentiation of primary hHPCs in the absence of EPO, even though it bears no sequence similarity to EPO. However, TC2-3 is much less active than EPO in inducing erythroid differentiation. To examine the basis for hEPOR activation by transmembrane proteins as well as to gain a better understanding of the structure of hEPOR traptamers, we isolated and characterized a more active version of TC2-3. By subjecting a library of TC2-3 mutants to more stringent selection conditions, we isolated a mutant, EBC5-16, which differs from TC2-3 by only a single amino acid but supports erythroid differentiation with activity comparable to EPO, as assessed by cell-surface GpA expression.