Engineering high-potency R-spondin adult stem cell growth factors

Abstract

Secreted R-spondin proteins (RSPOs1-4) function as adult stem cell growth factors by potentiating Wnt signaling. Simultaneous binding of distinct regions of the RSPO Fu1-Fu2 domain module to the extracellular domains (ECDs) of the LGR4 G protein-coupled receptor and the ZNRF3 transmembrane E3 ubiquitin ligase regulates Wnt receptor availability. Here, we examine the molecular basis for the differing signaling strengths of RSPOs1-4 using purified RSPO Fu1-Fu2, LGR4 ECD, and ZNRF3 ECD proteins in Wnt signaling and receptor binding assays, and we engineer novel high-potency RSPOs. RSPO2/3/4 had similar signaling potencies that were stronger than that of RSPO1, whereas RSPO1/2/3 had similar efficacies that were greater than that of RSPO4. The RSPOs bound LGR4 with affinity rank order RSPO4 > RSPO2/3 > RSPO1 and ZNRF3 with affinity rank order RSPO2/3 > > RSPO1 > RSPO4. An RSPO2-4 chimera combining RSPO2 ZNRF3 binding with RSPO4 LGR4 binding was a "Superspondin" that exhibited enhanced ternary complex formation and 10-fold stronger signaling potency than RSPO2 and efficacy equivalent to RSPO2. An RSPO4-1 chimera combining RSPO4 ZNRF3 binding with RSPO1 LGR4 binding was a "Poorspondin" that exhibited signaling potency similar to RSPO1 and efficacy equivalent to RSPO4. Conferring increased ZNRF3 binding upon RSPO4 with amino acid substitutions L56F, I58L, and I63M enhanced its signaling potency and efficacy. Our results reveal the molecular basis for RSPOs1-4 activity differences and suggest that signaling potency is determined by ternary complex formation ability, whereas efficacy depends on ZNRF3 recruitment. High-potency RSPOs may be of value for regenerative medicine and/or therapeutic applications.

Fig. 1.

Recombinant protein production. (A) Superdex200 HR gel-filtration chromatograms for the four wild-type MBP-Th-RSPO Fu1–Fu2-H₆ proteins produced in E. coli. (B) Nonreducing SDS-PAGE showing selected purified proteins used in this study. 2.5 μg of each protein was loaded per lane and the gel was stained with Coomassie brilliant blue. Molecular mass markers are shown in kilodaltons. Lanes are as follows: 1, marker; 2, MBP-Th-RSPO1 Fu1–Fu2; 3, MBP-Th-RSPO2 Fu1–Fu2; 4, MBP-Th-RSPO3 Fu1–Fu2; 5, MBP-Th-RSPO4 Fu1–Fu2; 6, MBP-Th-RSPO2–4 Fu1–Fu2; 7, MBP-Th-RSPO4–1 Fu1–Fu2; 8, glycosylated MBP-Th-RSPO4 Fu1–Fu2 produced in HEK293T cells; 9, biotin-ZNRF3 ECD; 10, MBP-Th-LGR4 LRR1–14. All recombinant proteins other than N-glycosylated MBP-Th-RSPO4 Fu1–Fu2 were produced in E. coli.

Fig. 2.

Signaling and receptor binding activities of bacterially produced recombinant MBP-Th-RSPOs1–4 Fu1-Fu2 proteins. (A) Potentiation of Wnt3a activation of the canonical Wnt/β-catenin pathway. HEK293T cells transfected with TOPFLASH firefly luciferase reporter and Renilla luciferase control plasmids were treated with the indicated concentrations of MBP-Th-RSPO Fu1–Fu2 proteins in 1:6 diluted Wnt3a conditioned media. (B) Binding of MBP-Th-RSPOs1–4 Fu1–Fu2 proteins to the LGR4 ECD in vitro by TR-FRET competition assay. Tb-chelate–labeled MBP-Th-LGR4 LRR1–14 (20 nM) and AF488-labeled MBP-Th-RSPO2 Fu1–Fu2 (125 nM) were incubated with the indicated concentrations of unlabeled MBP-Th-RSPO Fu1–Fu2 proteins. (C) AlphaLISA luminescent proximity assay for ZNRF3 ECD binding. The indicated concentrations of MBP-Th-RSPO Fu1–Fu2 proteins were incubated with 3 nM biotin-ZNRF3 ECD. Donor and acceptor beads were at 20 μg/ml each. Data shown for (A–C) are representative of at least three independent experiments each performed in duplicate. The error bars represent the S.E.M. of the experiment. (D, E, and F) Vertical scatter plots showing the pEC₅₀, E_max, and pK_I values obtained from replicate independent signaling and LGR4 binding experiments as in (A and B). The mean values and error bars representing the S.E.M. of the replicates are denoted. Statistical significance (**P < 0.01 and ***P < 0.001) from one-way ANOVA with Tukey’s test is shown for comparison with RSPO2 in (D and E) and for comparison with RSPO4 in (F). ns, not significant.

Fig. 3.

Signaling and LGR4 binding activities of recombinant RSPO4 Fu1–Fu2 proteins produced in bacteria or HEK293T cells and commercial full-length RSPO4 produced in CHO cells. (A) Signaling assay for MBP-Th-RSPO4 Fu1–Fu2 proteins produced in bacteria or HEK293T cells (glycosylated). Data shown are representative of three independent experiments each performed in duplicate and the error bars represent the S.E.M. of the experiment. (B) Signaling assay for bacterially produced MBP-Th-RSPO4 Fu1–Fu2 and MBP-free RSPO4 Fu1–Fu2 compared with commercial full-length RSPO4 produced in CHO cells (R&D Systems). Data shown are representative of two independent experiments each performed in duplicate and the error bars represent the S.E.M. of the experiment. (C) TR-FRET LGR4 ECD competition–binding assay as in Fig. 2B comparing bacterially produced MBP-Th-RSPO4 Fu1–Fu2 and commercial full-length RSPO4 (R&D Systems). Data shown are representative of three independent experiments each performed in duplicate, and the error bars represent the S.E.M. of the experiment.

Fig. 4.

Architecture of the ternary complex and chimera design. (A) Crystal structure of the ternary RSPO1 Fu1–Fu2:LGR5 ECD:RNF43 ECD complex (PDB ID 4KNG). (B) Topology of the RSPO Fu1–Fu2 module highlighting ZNRF3/RNF43 and LGR4/5/6 interacting regions and chimera fusion point. Disulfide bonds are shown as brown connecting lines, cyan circles show conserved RSPO residues critical for interaction with LGR4/5/6, magenta circles highlight conserved RSPO residues critical for interaction with ZNRF3/RNF43, and the red square indicates the chimera junction point. Residue numbers correspond to RSPO1.

Fig. 5.

Signaling and receptor binding activities of the chimeric RSPO2–4 “Superspondin.” (A) TOPFLASH Wnt signaling assay with the indicated MBP-Th-RSPO Fu1–Fu2 proteins. Data shown are representative of at least five experiments each performed in duplicate. The error bars represent the S.E.M. of the experiment. (B) ZNRF3 binding AlphaLISA assay as in Fig. 2C except that donor and acceptor beads were at 15 μg/ml each. Data shown are representative of three independent experiments each performed in duplicate. The error bars represent the S.E.M. of the experiment. (C) LGR4 binding TR-FRET competition assay as in Fig. 2B. Data shown are representative of at least three independent experiments each performed in duplicate and the error bars represent S.E.M. of the experiment. (D, E, and F) Vertical scatter plots showing the pEC₅₀, E_max, and pK_I values obtained from replicate independent signaling and LGR4 binding experiments as in (A and C). The mean values and error bars representing the S.E.M. of the replicates are denoted. Statistical significance (**P < 0.01 and ***P < 0.001) from one-way ANOVA with Tukey’s test is shown for comparison with RSPO2 in (D and E) and for comparison with RSPO4 in (F). ns, not significant.

Fig. 6.

AlphaLISA assay for ternary complex formation. The indicated concentrations of bacterially produced MBP-free RSPO2 and 2–4 chimera Fu1–Fu2 proteins were incubated with 5 nM biotin-ZNRF3 ECD, 5 nM MBP-Th-LGR4 LRR1–14 and 15 μg/ml each donor and acceptor beads. Data shown are representative of two independent experiments each performed in duplicate. The error bars represent the S.E.M. of the experiment.

Fig. 7.

Signaling and LGR4 binding activities of the chimeric RSPO4-1 “Poorspondin.” (A) TOPFLASH Wnt signaling assay with the indicated MBP-Th-RSPO Fu1–Fu2 proteins. Data shown are representative of three independent experiments each performed in duplicate. The error bars represent the S.E.M. of the experiment. (B) LGR4 binding TR-FRET competition assay as in Fig. 2B. Data shown are representative of at least three independent experiments each performed in duplicate. The error bars represent the S.E.M. of the experiment. (C, D, and E) Vertical scatter plots showing the pEC₅₀, E_max, and pK_I values obtained from replicate independent signaling and LGR4 binding experiments as in (A and B). The mean values and error bars representing the S.E.M. of the replicates are denoted. Statistical significance (***P < 0.001) from one-way ANOVA with Tukey’s test is shown for comparison with RSPO1 in (C and D) and for comparison with RSPO4 in (E).

Fig. 8.

Increasing RSPO4 signaling potency and efficacy with amino acid substitutions that increase affinity for ZNRF3. (A) Superimposed crystal structures of RSPO1 and -2 Fu1–Fu2 bound to ZNRF3 ECD highlighting conserved and nonconserved interactions involving the second RSPO β-hairpin. RSPO1 and -2 are dark and light green, respectively, and the ZNRF3 ECDs are shown in shades of magenta. Modeled RSPO4 residues L56 and I58 are shown in orange (PDB ID 4C9R and 4CDK). (B) Amino acid sequence alignment of the four human RSPOs for the ZNRF3/RNF43-interacting region. Conserved RSPO residues that are critical for ZNRF3/RNF43 interaction are highlighted with green stars and the variable positions in β-strand 3 that contact ZNRF3/RNF43 are highlighted with orange hexagons. (C) ZNRF3 binding AlphaLISA assay as in Fig. 2C except that donor and acceptor beads were at 15 μg/ml each. Data shown are representative of at least two independent experiments each performed in duplicate. The error bars represent the S.E.M. of the experiment. (D) TOPFLASH Wnt signaling assay with the indicated MBP-Th-RSPO Fu1–Fu2 proteins. Data shown are representative of at least four independent experiments each performed in duplicate. The error bars represent the S.E.M. of the experiment.