Determinants of BH3 binding specificity for Mcl-1 versus Bcl-xL

Abstract

Interactions among Bcl-2 family proteins are important for regulating apoptosis. Prosurvival members of the family interact with proapoptotic BH3 (Bcl-2-homology-3)-only members, inhibiting execution of cell death through the mitochondrial pathway. Structurally, this interaction is mediated by binding of the alpha-helical BH3 region of the proapoptotic proteins to a conserved hydrophobic groove on the prosurvival proteins. Native BH3-only proteins exhibit selectivity in binding prosurvival members, as do small molecules that block these interactions. Understanding the sequence and structural basis of interaction specificity in this family is important, as it may allow the prediction of new Bcl-2 family associations and/or the design of new classes of selective inhibitors to serve as reagents or therapeutics. In this work, we used two complementary techniques--yeast surface display screening from combinatorial peptide libraries and SPOT peptide array analysis--to elucidate specificity determinants for binding to Bcl-x(L)versus Mcl-1, two prominent prosurvival proteins. We screened a randomized library and identified BH3 peptides that bound to either Mcl-1 or Bcl-x(L) selectively or to both with high affinity. The peptides competed with native ligands for binding into the conserved hydrophobic groove, as illustrated in detail by a crystal structure of a specific peptide bound to Mcl-1. Mcl-1-selective peptides from the screen were highly specific for binding Mcl-1 in preference to Bcl-x(L), Bcl-2, Bcl-w, and Bfl-1, whereas Bcl-x(L)-selective peptides showed some cross-interaction with related proteins Bcl-2 and Bcl-w. Mutational analyses using SPOT arrays revealed the effects of 170 point mutations made in the background of a peptide derived from the BH3 region of Bim, and a simple predictive model constructed using these data explained much of the specificity observed in our Mcl-1 versus Bcl-x(L) binders.

Figure 1

Screening a combinatorial BH3 peptide library using yeast surface display. (A) Schematic of the yeast-display system used to study interactions of BH3 peptides with pro-survival proteins Bcl-x_L and Mcl-1. Expression of the BH3 peptide as a fusion to the yeast cell surface protein Aga2p was monitored by immunofluorescence detection of a FLAG tag located at the carboxyl terminus of the BH3 peptide (FITC fluorescence); binding of a pro-survival protein (Bcl-x_L or Mcl-1) was monitored by detection of a c-myc tag located at the amino terminus (PE fluorescence). (B) Positions of Bim-BH3 that were varied in the yeast-display or SPOT studies are shown as sticks in a structure of Bim bound to Mcl-1 (PDB code:2PQK). Residues varied in the yeast-display library are in cyan; additional residues substituted in the SPOT arrays are in green. Mcl-1 is shown using a surface representation. (C) Alignment of representative BH3 sequences from human BH3-only proteins. Residue positions randomized in the yeast-display library or mutated in the SPOT substitution analysis are shaded; numbering using a heptad convention is shown at the top. (D) Schematic of the screening scheme for isolating BH3 peptides with different binding specificities.

Figure 2

Characterization of clones from the yeast-display screen. (A) Sequenced Mcl-1 specific peptides are binned according to their specificity indices (S.I.) measured using yeast surface display, where S.I. = (mean fluorescence for binding to 10 nM Mcl-1)/(mean fluorescence for binding to 1 μM Bcl-x_L). (B) Bivariate flow cytometric plots of two Mcl-1 specific clones and wild-type Bim-BH3 are shown with their respective S.I. values. At left, binding in the presence of 10 nM Mcl-1; at right, binding in the presence of 1 μM Bcl-x_L. (C-F) Sequence logos for different populations. In (C), the library prior to sorting with composition weighted by codon degeneracy; wild-type Bim residues are boxed at the top; in (D) Mcl-1 specific peptides; in (E) Bcl-x_L specific peptides; in (F) peptides that bound to both Bcl-x_L and Mcl-1.

Figure 3

Characterization of specific peptides using an in vitro fluorescence polarization assay. Competition of (A) Mcl-1 specific peptides and (B) Bcl-x_L specific peptides with fluorescently labeled Bim-BH3 for binding to Mcl-1 and Bcl-x_L, respectively. Binding curves from representative experiments are shown. The concentration of Bcl-x_L or Mcl-1 was 50 nM. Wild-type Bim-BH3 is shown for comparison. The higher concentration points for Mcl-1 specific peptide MG1 and Bcl-x_L specific peptide XF8 were excluded due to low solubility, which led to light scattering (see materials and methods for additional notes about curve fitting). (C) Inhibition constants for Bim-BH3 and selected Mcl-1 and Bcl-x_L specific peptides measured using the competition binding assay. Average values from a minimum of two experiments are shown with errors as standard deviations over replicates.

Figure 4

X-ray structure of Mcl-1 specific peptide MB7 bound to Mcl-1. (A) Close up of the hydrophobic groove of human Mcl-1 (surface), with the MB7 peptide helix in slate blue. Residues that differ from wild-type Bim-BH3 are labeled. Small structural differences with respect to the structure of a wild-type Bim-BH3—Mcl-1 complex are observed in the vicinity of position 4a (B) and 2d (C). The MB7 complex is in blue and the native Bim-BH3 complex is in green (PDB entry: 2PQK).

Figure 5

SPOT array substitution analysis of Bim-BH3 peptides binding to Mcl-1 and Bcl-x_L. Data are for Mcl-1 binding in (A) and (C) and for Bcl-x_L binding in (B) and (D). The top and bottom panels used 100 nM and 1 μM protein, respectively. All spots in the leftmost column of each membrane show binding to the wild-type Bim-BH3 peptide. All other spots are point substitutions or a single repeat of the wild-type sequence in each row, with rows defining residue positions and columns indicating residue identities.

Figure 6

A model built using the SPOT array data captures the specificities of sequences identified using yeast display. (A and B) A section of the library arrays showing position 4a substitutions. (A) Each boxed set of three spots shows substitution at position 4a with Phe, Val or Asn. Mutations were made with different residues at position 3d, as indicated, with all other residues identical to wild-type Bim-BH3. SPOTS in the top or bottom rows were probed with 100 nM Mcl-1 or 100 nM Bcl-x_L, respectively. (B) Same as (A) but for mutations made in the context of Asp at 3b. (C) Effect of a Phe-to-Val substitution at position 4a in Bim-BH3 on binding to Mcl-1 (C) or Bcl-x_L (D) in fluorescence competition binding assays as described in Figure 3. (E) Engineered BH3 peptide sequences from the yeast screen were scored using a PSSM based on the Bim-BH3 substitution array data. The points plotted correspond to: Mcl-1 specific peptides (red circles), Mcl-1 specific peptides with Val at position 4a (red filled circles); Bcl-x_L specific peptides (blue squares), peptides that bound to both proteins (green triangles). (F) The same plot constructed with a PSSM that included the SPOT library array data; this model gave better separation of Mcl-1 binders vs. non-binders along the Mcl-1 score axis.