Fig 1: Design strategy of novel granulopoietic proteins.(A) Structure of the human G-CSF:G-CSFR complex (PDB: 2D9Q), showing site II and site III as the contact points for proposed receptor dimerization [40]. Surface residues at site II or site III that were reported to have a more than 2-fold impact on G-CSF’s activity are shown in blue [27]. G-CSF is shown in orange, the G-CSF receptor CRH domains in red, and the Ig-like domain in pink. (B) The critical residues at site II were disembodied and used as a geometric search query against the entire PDB to retrieve structurally compatible scaffolds. The top 6 compatible scaffolds structures are shown in cartoon representation. (C) After retrofitting the binding epitope, the most structurally stable scaffolds, named Moevan and Sohair, were identified by molecular dynamics simulations. The scaffold of Moevan and Disohair were adopted from an uncharacterized protein from Bacillus halodurans (PDB: 2QUP) [30] and a de novo designed, homodimeric, 4-helical, coiled-coil bundle (PDB: 5J73) [31], respectively. The blue-to-yellow 2D diagrams represent the topological chain paths of the G-CSF fold compared to the designs. (D) The retrofitted scaffolds were further optimized for their core packing and solvent-exposed residues. diSohair1 and diSohair2, dimeric variants of Sohair, were constructed to harbor 2 copies of the binding epitope in a C2-symmetric fashion, while the single-chain variant Sohair was made through de novo design of an additional loop across the dimeric interface of the 2 chains. G-CSF, granulocyte colony-stimulating factor; Ig, immunoglobulin; PDB, Protein Data Bank.
Fig 2: The designs bind to the human G-CSF receptor.SPR titration sensorgrams of the binding kinetics of A rhG-CSF (Filgrastim), B Boskar3, C Boskar4, and D Boskar4_t2. The binding kinetics analysis (association curve fits shown in spectrum colours) indicates higher apparent binding affinity of inherently bivalent ligands, whereby rhG-CSF and Boskar4_t2 represent two distinct modes of bivalent binding to the G-CSFR.
Fig 3: The potency of designs can be greatly enhanced by the optimal tandem spacing.The creation of two-copy Boskar4 tandems with different flexible spacers, through A a 24-residue-linker (Boskar4_t2), and B a short 6-residue-linker (Boskar4_st2). C Dose–response curves in NFS-60 cells show Boskar4_t2 (EC50 = 180 pM), and Boskar4_st2 (EC50 = 7.6 pM) to be 11- and 263-fold more active than the single domain Boskar4, respectively. Datapoints and error bars represent mean ± standard deviations of 3 biologically independent replicates. D, E This greatly enhanced potency is also evident in the dose- and time-dependent NFS-60 proliferation kinetics, highlighting the requirement for an optimal G-CSFR dimerisation spacing to achieve maximal G-CSFR activation. The experiments in C, E were performed as described in figure legend to Fig. 2.
Fig 4: The designs display potent and specific activity in cell-based assays.A Dose–response curves of NFS-60 proliferation upon 48 h treatment with Filgrastim (rhG-GCSF), Boskar3, or Boskar4. Datapoints and error bars represent mean ± standard deviations of three biologically independent replicates. B Time-dependent proliferation trajectory of surface-immobilised NFS-60 cells over a 10-day treatment. Data points and shades represent mean and standard deviation values of three biologically independent replicates. Experiments were performed three times in triplicates. Data of one representative experiment is shown. C, D Dose- and time-dependent proliferation trajectories over a 5-day treatment of free-floating NFS-60 cells, under the influence of Boskar3 and Boskar4 treatments, respectively. Data points and shades represent mean and standard deviation values of three biologically independent replicates. E G-CSFR-deficient primary stem cells (G-CSFR KO), show abolished proliferative responses to either rhG-CSF or the designs. Experiment was performed twice in triplicates. Data points and shades represent mean and standard deviation values of three biologically independent replicates. F Intracellular levels of phospho-AKT (Thr308), phospho-ERK1/2 (p44/42 MAPK), phospho-STAT3 (Tyr705), and phospho-STAT5 (Tyr694) in CD34+ HSPCs treated with rhG-CSF or the designs (see the “Methods” section). Geometric mean of the expression intensity of each phospho-protein (GeoMean intensity) is shown on the y-axis. The experiment was performed twice.
Fig 5: Proliferative activity of Moevan, Moevan_t2, and diSohair2 in murine NFS-60 cells.(A) Time-dependent proliferation trajectory of surface-immobilized NFS-60 cells over a 10-day treatment. Data points and shades represent mean and standard error (concentrations used: rhG-CSF: 532 pM; Moevan: 258 nM; diSohair2: 367 nM). (B) Surface-immobilized G-CSFR-deficient NFS-60 cells (G-CSFR KO), show abolished proliferative responses to either rhG-CSF or the designs. Data points and shades represent mean and standard error. (C, D) Dose- and time-dependent proliferation trajectories over a 5-day treatment of free-floating NFS-60 cells, under Moevan (C) or diSohair2 (D) treatments, respectively (concentration ranges: Moevan: 258 nM to 0.66 pM; diSohair2: 367 nM to 0.94 pM). Data points and shades represent the median and median absolute deviation from 3 separate measurements, respectively. (E) Dose- and time-dependent proliferation trajectories over 90 h for the tandemly repeated Moevan construct, Moevan_t2, compared against rhG-CSF. (F) The unmutated design templates (Moevan_control and diSohair_control) show no proliferative activity on NFS-60 cells. Data points and shades represent mean and standard error (S1 Data). G-CSFR KO, granulocyte colony-stimulating factor receptor knockout; G-CSFR WT, granulocyte colony-stimulating factor receptor wild type; rhG-CSF, recombinant human G-CSF.
Supplier Page from R&D Systems, a Bio-Techne Brand for Recombinant Human G-CSFR/CD114 Protein, CF