S 1 Supporting Information

1. Center for Theoretical Biological Physics (CTBP) and Departments of Physics and Astronomy, Chemistry and Biosciences, Rice University, Houston, Texas, United States 2. Department of Chemistry and Biochemistry, The University of California, San Diego (UCSD), La Jolla, California, United States *Corresponding author: Patricia A. Jennings, Email: pajennings@ucsd.edu Phone: (858) 534-6417 and José N. Onuchic, Email: jonuchic@rice.edu, Phone: (713) 348-4197.

Evolutionary Search. Pfam version 28.0 1 was used for an evolutionary search of the leptin sequences. The result is plotted as a Weblogo plot in Figure S2 (http://weblogo.berkeley.edu/logo.cgi 2,3 ).
Evaluation of the native structure. A native 1 H-15 N HSQC of the wild type and wt +5 protein was collected to compare the structure of the two proteins ( Figure S6). The data show that adding five residues to the Nterminal region has no significant effects on the overall structure of the protein, as the two spectra coincide. The wild-type backbone assignments were used to evaluate the structure of the loop variants, seen as an inset in Fig. 2. Overlapping peaks in the center of the spectra or peaks where the assignments were unclear was not used for the structural evaluation plotted in the inset structure. Peaks that overlap are represented in blue and peaks that have moved are represented in red in the structure in Fig. 2.
Comparing the 1 H-15 N TROSY-HSQC between the loop variants reviles a shift of the side chain of the tryptophan residue (W138, the peak at 130 and 10 ppm 1 H and 15 N respectively) used as our fluorescence probe in the thermodynamic experiments in vitro. This is expected as it is positioned in the modified region, in the end of helix D. This could potentially have an effect on the measured fluorescence in vitro, as it is solvent exposed and placed in a region where with the largest shifts of the structures are seen in the TROSY-HSQC's. To assure that the measured fluorescence is not affected from the loop switch we overlaid the full spectra from 300 to 450 nm for all proteins and the data coincides well ( Figure S7). Thus, we conclude that the structure is conserved after modification of the covalent loop with minor shift at the bottom of the four-helix bundle. However, these shifts have no significant effect on tryptophan fluorescence used in our thermodynamic experiments.
Modeling the New Versions of Leptin. As crystallization of the loop variants was unsuccessful, we collected structural information from 1 H-15 N TROSY-HSQC spectra of the protein and designed the new structures from Swiss PDB viewer. The five extra residues (MGSGVP) were added manually to the Nterminal of the 1AX8 wild-type crystal structure. To form the new disulphide bridge, we used the program Macro Molecular Builder to relax the structure and to for the new bond between the two cysteines (MMB2.15https://simtk.org/home/rnatoolbox). The final structures are shown in Fig. 1.
Imaging and Image Analysis. Cells were seeded at 30,000 cells/well onto 96-well plates (Corning) coated with collagen 3 mg/ml (Life Technologies), BSA 1 mg/ml (New England Biolabs), fibronectin 10 mg/ml (Sigma-Aldrich) solution. Cells were serum starved for 24 h. Prior to imaging, media were changed to imaging media (FluoroBrite DMEM, Life Technologies) and incubated with 50 ng/ml Hoechst (Sigma) to allow nuclear segmentation. Cells were imaged for the course of 2 h acquiring a frame every 3 min using a Nikon Plan Apo λ 10X/0.45 objective with a 0.7 x demagnifier and Nikon Eclipse Ti microscope with a sCMOS Zyla camera controlled by custom automated software written using MATLAB and Micro-Manager 4 . Temperature (37° C), CO 2 (5 %), and humidity were held constant