The number of COSMIC mutations after every filtration step, starting from the initial number of mutations with repetitions for different samples.

A Mutation analysis results from Maestro. Each line shows a different mutation, with the number of samples in COSMIC containing the mutation, predicted changes in Ab-Ag affinity and Ag stability between the original and mutated states. ΔG Affinity of the S288 mutations is close to or bigger than 3 Kcal/mol, suggesting that they might disrupt HER2 binding by Pertuzumab. S288 in the PDB structure numbering corresponds to S310 in the sequence numbering. B ELISA results for Pertuzumab binding to canonical HER2 and the two HER2 mutants, with Kd values. Both S310 mutations abrogated HER2 binding to Pertuzumab.

The interface between HER2 and Pertuzumab (PDB 1S78). HER2 is in blue, Pertuzumab light and heavy chains are in light and dark grey, respectively. Serine 310 (288 in the PDB numbering) is shown in cyan. All 14 residues chosen for the library are colored by the criteria that led to their selection for the library.

A Fourteen putative residues from Pertuzumab heavy chain that were chosen to be mutated. Each residue is classified for the reason it was chosen: in direct contact with SER310, adjacent to a SER310 contact, or a general contact with HER2 that was chosen to strengthen binding. Each residue was assigned to one of four groups based on its sequence position. B Pertuzumab heavy chain from the 1S78 PDB structure. Highlighted are the fourteen residues to be mutated, colored by the same color scheme as in Figure 2. The residues are divided into four groups based on their sequence position. C The number of residues selected for mutation from each group of residues for a specific library sequence. Groups 1 and 2 which contain four and six residues, respectively, had two mutations each in every library sequence, with all possible combinations of two mutations from each group. Groups 3 and 4, having two residues each, had one mutation in every library sequence. Each library sequence will have a total of six mutations.

Selections scheme. Two rounds of MACS, one against canonical HER2 and one against the S310Y mutant reduced the size of the initial library. A round of FACS (round 3) was used to gate putative binders that were taken into the fourth and final round of selection, yielding putative binders to be sequenced.

A Binders population after two round of MACS – one against canonical HER2 and the second against S310Y mutant underwent a round of sorting by FACS (R3). Gates were set as follows: I no antigen for control. II 200nM of canonical HER2 III 200nM of S310Y mutant IV 200nM of S310F mutant. The populations from the top right quadrant that represent S310Y and S310F putative binders, were collected for the next round of selection. B The populations from Figure 5A III and IV were taken through an additional selection round (R4). Top panels describe the results of R4 where R3 was selected against the S310Y mutants; bottom panels are R4 where R3 was against the S310F mutants. FACS gates were set as follows: I,V Null, negative control, where no antigen is added, II,VI 50nM of canonical HER2 III,VII 50nM of S310Y mutant IV,VIII 50nM of S310F mutant. Gated populations are marked in rectangles.

A,B Positions and variations that contribute to multi-specific binding according to selection results. The first row of each table shows the fourteen positions from the original Pertuzumab sequence selected for variation, as shown in Figure 3A, numbered as in the PDB 1S78. Positions are colored according to the color scheme in Figure 2: Direct contacts in yellow and peach, contact-adjacent in orange, and general antigen contacts in brown. A Three sequences from round 4 of selection, that were found to bind two different variants. On the right side of the table, for each sequence, the antigen it binds is specified. The colony number from which the sequence originated is in parenthesis. B Two colonies from round 4 of selection. Each of these sequences is composed of residues that appear in binders of all three antigens: canonical HER2, S310F, and S310Y. For example, according to Supplementary Table 1, in position 58 there can only be Isoleucine or Leucine. Both sequences have Isoleucine in this position. C,D,E ELISA results for Pertuzumab and four IgGs from the tested monoclones, against three antigens: C canonical HER2 D S310Y mutant E S310F mutant. F Kd values in nM for each IgG described in Figures 3C-E, against all antigens tested. No binding was detected for Pertuzumab and both mutants.

A Schematic representation of HER2 and HER3. HER2 S310 is represented by a yellow star in domain II. HRG ligand (purple) promotes HER2-HER3 dimerization and trans-phosphorylation, Pertuzumab (green) disrupts the dimerization complex and abrogates tyrosine phosphorylation. B-D HER3 phosphorylation (pHER3) due to HER2-HER3 dimerization, in antibody concentrations of 1 and 100nM for Pertuzumab and the two engineered IgG85 and IgG41. Different HER2 variants were used for dimerization: B canonical HER2 C S310Y mutant D S310F mutant. Quantification is based on cross-referenced western blot intensity densitometry from the mean ± SEM of three repeats. While phosphorylation was restored for the two HER2 S310 mutants in the presence of Pertuzumab, IgG 85 showed lower phosphorylation for S310F, and IgG 41 inhibited both HER2 mutants as well as canonical HER2.