Priming HIV-1 broadly neutralizing antibody precursors in human Ig loci transgenic mice by Devin Sok, Bryan Briney, Joseph G. Jardine, Daniel W. Kulp, Sergey Menis, Matthias Pauthner, Andrew Wood, E-Chiang Lee, Khoa M. Le, Meaghan Jones, Alejandra Ramos, Oleksandr Kalyuzhniy, Yumiko Adachi, Michael Kubitz, Skye MacPherson, Allan Bradley, Glenn A. Friedrich, William R. Schief, and Dennis R. Burton Science Volume 353(6307):1557-1560 September 30, 2016 Published by AAAS

Fig. 1 Frequency analysis of VRC01-class light chains and precursors in Kymab HK mice. Frequency analysis of VRC01-class light chains and precursors in Kymab HK mice. (A) Light-chain L-CDR3 length distributions in humans and HK mice. (B) Frequencies of five–amino acid L-CDR3s in humans and HK mice. (C) Frequencies of known VRC01-class bnAb light-chain Vκ genes in humans and HK mice. (D) Frequencies of known VRC01-class bnAb Vκ genes with five–amino acid L-CDR3s in humans and HK mice. (E) Modeled distributions of the number of VRC01-class precursors per HK mouse for average precursor frequencies of 0.2 or 1.3 per HK mouse, corresponding to frequencies of 1 in 360 million or 1 in 60 million HK B cells, respectively. One in 60 million is likely to be an upper bound on the frequency, and so the distribution is likely to be shaped more like that shown for 1 in 360 million (see text). In (B) to (D), points represent frequencies for individual humans or mice (each sequenced once), and bars represent mean ± SD for n = 4 humans or n = 5 HK mice. Devin Sok et al. Science 2016;353:1557-1560 Published by AAAS

Fig. 2 B cell sorting analysis reveals reproducible epitope-specific responses to eOD-GT8 60mer prime. B cell sorting analysis reveals reproducible epitope-specific responses to eOD-GT8 60mer prime. (A) Overview of two immunization experiments in Kymab mice. (B) Frequencies of epitope-specific memory B cells at days 14 or 42 after priming in experiment 1. (C) Same type of data as in (B) but for experiment 2, in which analysis was carried out at day 42 only. In (B) and (C), points represent frequencies measured for a single mouse, each measured once, and bars represent mean ± SD for all data in each column of the graph. Devin Sok et al. Science 2016;353:1557-1560 Published by AAAS

Fig. 3 Analysis of antibody sequences from epitope-specific memory B cells. Analysis of antibody sequences from epitope-specific memory B cells. (A) Summary of all sequence information obtained from experiments 1 and 2. Of the 33 VRC01-class pairs, 28 were identified as unique in nucleotide sequence. (B) Number of eOD-GT8 60mer-immunized or control mice from which at least one VRC01-class pair was isolated by B cell sorting (+) or no VRC01-class pairs were isolated (-). Data are aggregated from all animals and conditions in experiments 1 and 2, with a total of 90 mice. The P value was calculated by using Fisher’s exact test. (C) Number of eOD-GT8 60mer-immunized HK, HL, or HKL mice for which at least one VRC01-class pair was isolated (red, with percentages listed in white) or no VRC01-class pairs were isolated (gray). In (D) and (E), the same analysis was performed as in (B) and (C), respectively, but with five–amino acid L-CDR3 light chains instead of VRC01-class pairs. (F) L-CDR3 sequence logos for VRC01-class bnAbs (top row), eOD-GT8 60mer-induced VRC01-class paired antibodies (second row), eOD-GT8 60mer-induced five–amino acid L-CDR3s (third row), and naïve Kymab mice (bottom row), shown separately for kappa light chains (left column) and lambda light chains (right column). (G) L-CDR1 length distribution for eOD-GT8 60mer-induced VRC01-class antibodies (red) and all LCs in (A) (black). (H) Light chain Vκ and Vλ gene usage for eOD-GT8 60mer-induced VRC01-class antibodies. Red bars denote genes used by, or highly similar to those used by, known VRC01-class antibodies. (I) H-CDR3 length distribution for eOD-GT8 60mer-induced VRC01-class antibodies. Devin Sok et al. Science 2016;353:1557-1560 Published by AAAS