On the distribution of protein refractive index increments

Abstract

The protein refractive index increment, dn/dc, is an important parameter underlying the concentration determination and the biophysical characterization of proteins and protein complexes in many techniques. In this study, we examine the widely used assumption that most proteins have dn/dc values in a very narrow range, and reappraise the prediction of dn/dc of unmodified proteins based on their amino acid composition. Applying this approach in large scale to the entire set of known and predicted human proteins, we obtain, for the first time, to our knowledge, an estimate of the full distribution of protein dn/dc values. The distribution is close to Gaussian with a mean of 0.190 ml/g (for unmodified proteins at 589 nm) and a standard deviation of 0.003 ml/g. However, small proteins <10 kDa exhibit a larger spread, and almost 3000 proteins have values deviating by more than two standard deviations from the mean. Due to the widespread availability of protein sequences and the potential for outliers, the compositional prediction should be convenient and provide greater accuracy than an average consensus value for all proteins. We discuss how this approach should be particularly valuable for certain protein classes where a high dn/dc is coincidental to structural features, or may be functionally relevant such as in proteins of the eye.

Figure 1

Differential refractometry of peptides A (triangles) and B (circles). Concentrations were measured by dry weight determination. After wavelength correction to 589 nm, the slopes correspond to dn/dc values of 0.179 ± 0.003 ml/g (solid line) and 0.194 ± 0.003 ml/g (dashed line).

Figure 2

Histogram of calculated dn/dc values for the 62,378 predicted proteins from the UCSC human genome browser for the human Feb. 2009 (GRCh37/hg19) assembly. The best-fit Gaussian is indicated as a black solid line, with mean of 0.1899 ml/g and standard deviation of 0.0030 ml/g.

Figure 3

Two-dimensional histogram of proteins from Fig. 2 and their dn/dc value as a function of the fraction of amino acid residues being either Arg, Asp, Cys, His, Met, Phe, Trp, or Tyr (this fraction being termed high dn/dc aa index in the axis label).

Figure 4

Histogram of dn/dc values from known human proteins with molecular mass below 10 kDa (top) and above 100 kDa (bottom), respectively. The solid line is the best-fit Gaussian, with a mean of 0.1902 ml/g and standard deviation of 0.0053 ml/g for small proteins, and a mean of 0.1888 ml/g and a standard deviation of 0.0025 ml/g for the large proteins.

Figure 5

Two-dimensional histogram of the human proteins from Fig. 2 sorted according to protein dn/dc and the partial-specific volume.

Figure 6

(Top panel) Histogram of the computed dn/dc values of all 6306 proteins obtained searching the NCBI protein sequence database for fatty acid hydroxylase. The solid line is the best-fit Gaussian, with a mean of 0.1971 ml/g and standard deviation of 0.0029 ml/g. As a visual reference, the Gaussian distribution from the analysis of all known and predicted human proteins is shown as a dotted line. (Bottom panel) Histogram of the computed dn/dc values of the 1514 proteins in NCBI classified as crystallins.