Supplemental Information EXTENDED EXPERIMENTAL PROCEDURES Statistical Coupling Analysis

Statistical coupling analysis (SCA) was carried out on a multiple sequence alignment (MSA) of 418 members of the DHFR protein family using a process described in (Halabi et al., 2009) and implemented in version 4.0 of the SCA MATLAB toolbox (available from the Ranganathan laboratory website). Briefly, the basic analytic procedure involves three steps: (1) compute a conservation-weighted correlation matrix that estimates the degree of coevolution between each pair of sequence positions in the MSA, (2) carry out spectral (or eigenvalue) decomposition of this matrix and identify its statistically significant top eigenmodes, and (3) examine the pattern of positional contributions to these top eigenmodes to deduce the number and composition of groups of coevolving amino acid positions (sectors). In the case of DHFR, the spectral analysis indicates five significant top eigenmodes that represent a single protein sector. To systematically determine the composition of the sector, we fit histograms of the positional weights in these top five eigenmodes to the Students t-distribution and used probably density cutoffs in the tail of the distributions ranging from 0.005 to 0.015 to identify the sector residues. The t-distribution was selected empirically by quality of fit to the top eigenmodes and is simply meant to provide a well-defined basis for sector identification; no mechanistic basis for the use of this distribution is implied. Importantly, sectors identified by this method are consistent with previous definitions (Halabi et al., 2009; Lee et al., 2008). Consistent with the basic principle of sectors defined in (Halabi et al., 2009), the DHFR sector comprises a relatively small subset of total amino acid posi-tions (13 to 31 % based on the cutoffs indicated) and comprises a physically contiguous network of residues in the tertiary structure (Figure 1C). A MATLAB script comprising SCA for the DHFR family and the MSA will be made available for download on the Ranga