A Simple and General Method for Determining the Protein and Nucleic Acid Content of Viruses by UV Absorbance

Abstract

UV spectra of viruses are complicated by overlapping protein and RNA absorbance and light scattering. We describe and validate methodology for estimating RNA and protein concentration from such spectra. Importantly, we found that encapsidation did not substantially affect RNA absorbance. Combining absorbance data with a known T number, we confirmed that brome mosaic virus packages 3100 nucleotides/capsid, consistent with its genome. E. coli-expressed hepatitis B virus (HBV) packages host RNA based on capsid charge and volume. We examined HBV capsid protein (Cp183, +15 charge) and a less basic mutant (Cp183-EEE, +12 charge) that mimics a phosphorylated state. Cp183-EEE packaged ~3450 nucleotides per T=4 capsid and Cp183 packaged ~4800 nucleotides, correlating to the size of HBV’s RNA pre-genome and mature DNA genome, respectively. The RNA:protein charge ratio (about 1.4 phosphates per positive charge) was consistent with that of other ssRNA viruses. This method is generalizable to any virus-like particle.

Introduction

There are numerous examples of virus-like particles assembled in vitro and in heterologous systems where the nucleic acid content is of interest but has not been rigorously determined (Annamalai and Rao, 2005; Grieger and Samulski, 2005; Johnson et al., 2002; Krishna, Marshall, and Schneemann, 2003; Ludwig and Wagner, 2007; Ma and Vogt, 2002; Mukherjee et al., 2007; Pattenden et al., 2005; Satheshkumar et al., 2005; Stockley et al., 2007). This sampling of systems ranges from bacteriophages to retroviruses; they package RNA and DNA. In some cases, packaging is sequence specific, in other cases it is promiscuous. The nucleic acid may function as structural scaffolding (Tihova et al., 2004) or as an allosteric effector (Stockley et al., 2007). The nucleic acid content may be of particular interest as genetic information in virus-based transfection systems, especially in the context of gene therapy (Grieger and Samulski, 2005). One of the most accessible methods for defining the nucleic content of a sample is UV absorbance.

Accurate interpretation of the absorbance spectra of viruses, and other large nucleo-protein complexes, are complicated by several factors: the UV absorption of protein and nucleic acid overlap, light scattering systematically distort the spectra, and there is the possibility that structural change of the nucleic acid may alter its hyperchromicity. The overlap in the absorption of proteins and nucleic acids can be resolved (Gonen and Rytwo, 2009; Kalb and Bernlohr, 1977; Mukherjee et al., 2010). Methods have also been described to evaluate the effect of light scattering on absorbance spectra (Cox, DeWeerd, and Linden, 2002; Leach and Scheraga, 1960). An extinction coefficient for a large nucleoprotein complex can be experimentally determined that explicitly accounts for all of these features, as has been done for adenovirus (Sweeney and Hennessey, 2002). However, an ad hoc extinction coefficient will not always yield a meaningful measurement of concentration and has no predictive power for differing conditions and/or mutations. When the stoichiometry of chromophores is unknown and/or the light scattering contribution to absorbance is variable, a self-consistent method for determining the concentrations of protein and nucleic acid is required.

In this paper we describe and test a simple method for analyzing absorbance spectra of viruses, correcting for light scattering and determining accurate concentrations of both protein and nucleic acid. To demonstrate the utility of this method we examine two viruses. Brome Mosaic Virus (BMV) packages an average of about 3100 nucleotides/capsid in vivo and virus-like particles assembled in vitro (Fox et al., 1998). In vivo, Hepatitis B Virus (HBV) in a phosphorylated state packages the ~3200 nucleotide RNA pre-genome (Seeger, Zoulim, and Mason, 2007). We determine the RNA content of two versions of E. coli expressed HBV capsids: Cp183, a full-length 183 residue capsid (core) protein, and Cp183-EEE, which incorporates three Ser to Glu mutations on each RNA-binding domain to mimic the phosphorylated state and effectively decrease the net charge of each RNA-binding domain from +15 to +12, demonstrating the relationship between protein charge and packaged nucleic acid.

Results

Nucleic acid and protein spectra can be differentiated qualitatively and quantitatively (Fig 1A). Stacked purines and pyrimidines absorb light with an absorption maximum at 260nm and a 260nm/280nm ratio of ~2.0 (Glasel, 1995). Protein absorbance is dominated by tryptophans, tyrosines, and disulfide bonds. Protein absorbance has a peak near 280nm and a characteristic shoulder at 290nm. The 260nm/280nm ratio for protein is ~0.6 (Glasel, 1995; Goldfarb, Saidel, and Mosovich, 1951). Here, we consider the absorbance spectrum to be a sum of protein and nucleic acid components, determined algebraically from the UV absorption at 260nm and 280nm after removing the background light scattering. (A spreadsheet application of the following equations is included in Supporting Information.)

Figure 1.

Standard, experimental, and corrected absorbance spectra. (A) Typical UV absorbance spectra for pure protein, pure RNA, and a mixture of protein and RNA. These spectra can be differentiated qualitatively by shape and quantitatively by 260/280nm ratio. Quantitative evaluation of UV absorbance spectra requires that the contribution of light scattering to the apparent absorbance be accounted for at each wavelength. (B) Spectra from pure protein samples show varying severity of light scattering and baseline offsets. The degree of light scattering was estimated using the apparent absorbance at 340nm and 360nm, where protein and RNA absorbance is minimal, assuming a λ⁻⁴ relationship between light scattering and wavelength. (C) The spectra from panel B were corrected for light scattering and baseline offset. The corrected spectra have much more consistent 260nm/280nm ratios. The spectra in panels B and C were normalized to a 280nm absorbance of one for comparison to highlight the improvement of the 260nm/280nm ratio.

Light scattering corrections

Light scattering increases the 260nm/280nm absorbance ratio for large complexes, e.g. for viruses. Thus, for accurate absorbance-based nucleoprotein concentration calculations, the contribution of light scattering to the apparent absorption must be estimated and subtracted from the spectra. According to the Rayleigh approximation, light scattering of spherical solutes is proportional to λ⁻⁴; the amount of scattering at a given wavelength is proportional to the molecular mass of the solute. When the diameter of the scatterer is much less than the wavelength of the incident light, the amount of light scattering is well predicted by the Rayleigh approximation (Cox, DeWeerd, and Linden, 2002; Young, 1982). Figure 1B shows absorbance spectra from several pure HBV capsid protein samples with varying amounts of light scattering. The variation in light scattering may be attributed to the amount of capsid and aggregation.

We find that a nucleo-protein absorbance spectrum can be corrected for light scattering using a two point approximation. Measured absorbance can be partitioned into light that is actually absorbed (A_corrected,λ) and the apparent absorption that is due to light scattering at a given wavelength (A_LS,λ). A_LS,λ for a solution of particles that are small compared to the length of scattered light can be expressed as a function of λ.

$$A_{LS,\lambda }=c_1{(\lambda )}^{-4}+c_2$$ (1)

Where c₁ is a coefficient for wavelength-dependent light scattering and c₂ accounts for baseline offset. Though a more rigorous fit can be accomplished, for simplicity, we estimated these constants from the apparent absorbance at 340nm and 360nm. At these wavelengths, there is no significant absorbance by either protein or nucleic acid so that absorbance is essentially A_LS. The light scattering at these wavelengths can therefore be expressed as

$$A_{340}=c_1{(340nm)}^{-4}+c_2$$ (2)

$$A_{360}=c_1{(360nm)}^{-4}+c_2$$ (3)

c1 and c2 can be solved from this set of equations

$$c_1=\frac{A_{340}-A_{360}}{{(340nm)}^{-4}-{(360nm)}^{-4}}$$ (4)

$$c_2=A_{340}-c_1{(340nm)}^{-4}$$ (5)

Values of c₁ and c₂ from equations 4 and 5 can be combined with equation 1 to evaluate the light that is actually absorbed, A_corrected, from measured absorbance (A_λ) and A_LS.

$$A_{\mathit{corrected},\lambda }=A_{\lambda }-A_{LS,\lambda }$$ (6)

$$A_{\mathit{corrected},\lambda }=A_{\lambda }-c_1{\lambda }^{-4}-c_2$$ (7)

Equations 4 and 5 can be combined with equation 7 to yield an absorbance corrected spectrum.

$$A_{\mathit{corrected},\lambda }=A_{\lambda }-\left(\frac{A_{340}-A_{360}}{{(340nm)}^{-4}-{(360nm)}^{-4}}\right){\lambda }^{-4}-A_{340}+\left(\frac{A_{340}-A_{360}}{{(340nm)}^{-4}-{(360nm)}^{-4}}\right)340n{m}^{-4}$$ (8)

Purified empty HBV capsids, free of nucleic acid, demonstrate the effectiveness of light scattering correction (Fig 1). In uncorrected spectra (Fig 1B), the 260nm/280nm ratios vary from 0.59 to 0.71. The same spectra following light scattering correction (Fig 1C) show a consistent 260nm/280nm ratio of about 0.6, as expected for purified capsid protein. The slight mismatches in the spectra indicate that the Rayleigh approximation does not perfectly account for light scattering. However, the corrected 260nm/280nm ratios are consistent with one another and the expected value for protein. We have found that correcting for light scattering can have a substantial effect on the 260nm/280nm ratio and is necessary for reliable determination of protein and RNA concentrations by absorbance.

RNA and Protein concentration calculations

For concentration calculations from corrected spectrum, extinction coefficients for both RNA and protein are required. RNA extinction coefficients are given per nucleotide (Table 1). Extinction coefficients for RNA are nominally dependent on ionic strength, pH, and base pairing. For short oligonucleotides, extinction coefficients can be calculated based on sequence (Cavaluzzi and Borer, 2004; Kallansrud and Ward, 1996; Murphy and Trapane, 1996; Tataurov, You, and Owczarzy, 2008). For longer sequences, the well established and commonly used extinction coefficient at 260nm (ε_RNA260) is a per nucleotide average of 8,000 M⁻¹cm⁻¹ (Maniatis, Fritsch, and Sambrook, 1980). A 260nm/280nm ratio for pure RNA of 2.0 (Glasel, 1995) was used to calculate an ε_RNA280nm of 4,000 M⁻¹cm⁻¹.

Table 1.

Extinction coefficients for RNA and viral core protein

	RNA (nt)	BMV monomer	T=3 capsid	HBV monomer	T=3 capsid	T=4 capsid
ε260 (M−1cm−1)	8,000	14,400	2,592,000	18,300	4,392,000	3,294,000
ε280 (M−1cm−1)	4,000	24,000	4,320,000	30,500	7,320,000	5,490,000

Accurate extinction coefficients for protein at 280nm (ε_protein280) can be estimated based on sequence or calculated from the absorbance and chemically determined amino acid composition of a sample (Benson, Suruda, and Talalay, 1975; Edelhoch, 1967; Gill and von Hippel, 1989; Pace et al., 1995). We used the method of Pace et al. to estimate the ε_Protein280nm (Pace et al., 1995) which yielded values of 30,500 M⁻¹cm⁻¹ per HBV capsid protein monomer and 18,300 M⁻¹cm⁻¹ per BMV capsid protein monomer. A 260nm/280nm ratio of 0.6 for pure protein (Glasel, 1995) was used to calculate ε_Protein260nm for HBV (18,300 M⁻¹cm⁻¹) and BMV (14,400 M⁻¹cm⁻¹). Table 1 includes the extinction coefficients per protein monomer as well as for T=3 (180 monomer) and T=4 (240 monomer) capsids.

In a light scattering corrected spectrum, the absorbance at each wavelength is the sum of the absorbance by RNA and protein, assuming no conformationally related changes in their respective extinction coefficients.

$$A_{\mathit{corrected},260}={\epsilon }_{\mathit{RNA}260}[\mathit{RNA}]L+{\epsilon }_{\mathit{protein}260}[\mathit{protein}]L$$ (9)

$$A_{\mathit{corrected},280}={\epsilon }_{\mathit{RNA}280}[\mathit{RNA}]L+{\epsilon }_{\mathit{protein}280}[\mathit{protein}]L$$ (10)

Here, L is the absorbance path length. The concentrations of RNA and Protein can be algebraically calculated from this set of equations, yielding equations 11 and 12.

$$[\mathit{protein}]=\frac{A_{\mathit{corrected},260}-\left(\frac{{\epsilon }_{\mathit{RNA}260}}{{\epsilon }_{\mathit{RNA}280}}\right)A_{\mathit{corrected},280}}{\left({\epsilon }_{\mathit{protein}260}-\left(\frac{{\epsilon }_{\mathit{RNA}260}}{{\epsilon }_{\mathit{RNA}280}}\right){\epsilon }_{\mathit{protein}280}\right)L}$$ (11)

$$[\mathit{RNA}]=\frac{A_{\mathit{corrected},260}-\left(\frac{{\epsilon }_{\mathit{protein}260}}{{\epsilon }_{\mathit{protein}280}}\right)A_{\mathit{corrected},280}}{\left({\epsilon }_{\mathit{RNA}260}-\left(\frac{{\epsilon }_{\mathit{protein}260}}{{\epsilon }_{\mathit{protein}280}}\right){\epsilon }_{\mathit{RNA}280}\right)L}$$ (12)

Dissecting absorbance of Brome Mosaic Virus

To test this method, the RNA content of BMV capsids was determined. The BMV capsid protein has a molecular mass of 20,385Da, so that the 180 subunit T=3 capsid has a molecular mass of 3,669,300Da. BMV has a tripartite genome, composed of RNA1 (~3200nt), RNA2 (~2900nt), RNA3 (~2100nt) and a subgenomic RNA4 (~900nt). Capsids contain either one RNA1, one RNA2, or the combination of one RNA3 plus one RNA4 (for a total of 3000 nt). We examined BMV samples with all three kinds of particles at a 5:5:1 ratio. The expected average molecular mass of a BMV capsid containing genomic RNA would be 4.63 MDa, corresponding to an average RNA molecular mass of 979,000 Da per capsid, assuming an average of 3045 nucleotides per capsid at an average mass of 322 per nucleotide.

BMV mass was determined by Size Exclusion Chromatography – Multi Angle Laser Light Scattering (SEC-MALLS). Triplicate SEC-MALLS experiments yielded a weight average molecular mass (Mw) of 4.64 ± 0.06 MDa (Fig 2A, C).

Figure 2.

Light scattering corrected UV absorbance of BMV capsids is consistent with a virus with a uniform RNA content. (A) SEC-MALLS – Elution of BMV capsids was observed by differential refractive index (line). The weight averaged molecular mass (Mw), calculated by SEC-MALLS, for the peak is shown with error bars. (B) UV absorbance spectrophotometry – The symbols and error bars represent fractions were collected from the peak in panel A and evaluated by absorbance for concentrations of RNA and BMV capsid protein. The molar ratio of RNA to capsid protein was used to calculate the RNA content per T=3 capsid and the molecular mass of each particle. These estimations of molecular mass are overlaid on the differential refractive index chromatograph. (C) There is good agreement between the average theoretical molecular mass of RNA-filled BMV capsids, the average Mw from SEC-MALLS and the molecular mass calculated by UV absorption.

Fractions collected from the SEC-MALLS experiments were also analyzed for protein and nucleic acid content by UV absorption spectroscopy. Typical calculated light scattering at 260 nm was 15% of the total signal, which would have led to a significant overestimation of RNA absorbance. From triplicate experiments, each with measurements from 13 fractions across the capsid peak (Fig 1B), the average RNA to protein ratio was 18.8 nucleotides per capsid protein monomer. This is equivalent to an RNA mass of 1.09 MDa per capsid, for a 180 protein T=3 capsid. This yields a total particle mass for BMV of 4.78±0.16 MDa (Fig 2B, C). Mass estimates by SEC-MALLS and the UV absorption are thus in excellent agreement with the expected value of 4.63 MDa. This confirms the ability of light scattering corrected UV absorbance to accurately determine RNA and protein concentrations and by extension, RNA content and total capsid molecular mass. The agreement of expected mass, SEC-MALLS, and absorbance, suggests that any change in RNA absorbance associated with packaging in a virus capsid (a function of base stacking and secondary structure content) was within experimental error.

Dissecting the RNA content of Hepatitis B Virus

E. coli-expressed HBV capsids were a more complicated example. These capsids contained an unknown amount of random RNA from the expression system. They were a mixture of T=4 and T=3 morphologies (Crowther et al., 1994; Wingfield et al., 1995) and, unlike BMV, RNA-filled HBV capsids had a tendency to aggregate. This mixture was not well resolved by SEC with a Superose 6 column; aggregates migrated in the void volume while T=3 and T=4 capsids co-eluted shortly thereafter.

SEC-MALLS experiments were performed on E. coli-expressed RNA-filled capsids of HBV Cp183 and the phosphorylation mimic Cp183-EEE (Fig 3A and 3B, respectively). By EM (Fig 3D), the two samples were indistinguishable. However, comparison of light scattering and the differential refractive index (dRI) signal, which is proportional to concentration, indicated that the Cp183 RNA-filled particles were more prone to aggregation (Fig 3C). The Mw determined for the void volume by SEC-MALLS is in 10’s of MDa, much greater than the mass for a capsid with a genomic RNA (ca 5 MDa). Further complicating analysis, HBV capsids adsorbed weakly to size exclusion media and thus peaks were relatively broad and tailed. After the void peak, the Mw estimates gradually plateaued between elution volumes of 9mL and 10.5mL, which indicated unaggregated particles. The average Mw calculations were conducted using this volume range (9mL to 10.5mL). In this range, SEC-MALLS of Cp183 and Cp183-EEE capsids showed a clear difference in their respective average molecular masses which was attributable to the relative ratios of T=3 and T=4 particles and their RNA content.

Figure 3.

E. coli-expressed HBV capsids are a mixture of T=3 and T=4 particles with a constant RNA content. Molecular weights of (A) Cp183 HBV capsids and (B) Cp183-EEE HBV capsids were determined by SEC-MALLS (open symbols) and calculated from UV absorption assuming all capsids had a T=4 morphology (solid triangles) or a T=3 morphology (solid squares). The mass estimates are overlaid on differential refractive index chromatogram from SEC-MALLS. (C) A comparison of SEC-MALLS Mw for Cp183 (open circles) and Cp183-EEE (open squares) capsids, highlighting the increased molecular mass of the average Cp183 capsid, is consistent with a greater amount of encapsidated RNA in the wild type protein. Error bars in A–C, from triplicate experiments, are not displayed because they are obscured by data points but were less than 3% of the molecular mass for each measurement based on triplicate experiments. (D) A typical example of the micrographs taken from fractions between 9mL and 10.5mL that were used to calculate the proportion of T=3 and T=4 capsids in the fractions used for molecular mass determinations. For Cp183, 29.4% (133/452) of the capsids from these fractions had a T=3 morphology. For Cp183-EEE, 32.2% (280/870) of the capsids from these fractions had a T=3 morphology. (E) The average molecular masses for capsids from Cp183 and Cp183-EEE experiments, determined by SEC-MALLS and UV absorption spectroscopy, show good agreement. The UV absorption estimates were weighted by the T=3/T=4 proportions from panel D. The predicted length of RNA encapsidated by T=3 or T=4 capsids is given, assuming all capsid protein is associated with an equal amount of nucleic acid.

UV absorption spectra of SEC fractions were used to calculate RNA and protein concentrations and estimate capsid mass (Fig 3A and 3B). For Cp183, the average molar ratio was found to be 20.0 nucleotides/protein monomer and was consistent for the 15 fractions used for molecular mass calculations. For Cp183-EEE, the ratio was 14.4 nucleotides/monomer, also consistent across the peak. Accounting for the number of capsid proteins per capsid (240 for T=4 capsids or 180 for T=3 capsids) gave the ratio of nucleotides of RNA per capsid. However, UV absorption cannot a priori determine if the capsids have a T=4 or T=3 morphology. The estimated capsid masses from UV absorption, assuming that all capsids have a T=4 morphology (triangles) or a T=3 morphology (squares), bracket the observed masses determined by SEC-MALLS (circles) (Fig 3A and 3B). Interestingly, the RNA-derived molecular mass estimates were low for fractions near the void volume (~7.5mL), which may result from capsids with a smaller RNA content, association of free dimer from the in vitro assembly, or a breakdown in the Rayleigh approximation due to the large size of the aggregated particles that would cause an underestimation of light scattering and an overcorrection of the absorbance spectra.

The absorbance-based mass estimates were corrected for the relative amounts of T=3 and T=4 capsid for comparison to SEC-MALLS, which gives a weight average molecular mass (Mw in figure 3A). The proportion of T=3 and T=4 capsids were determined by negative stain electron micrographs of Cp183 and Cp183-EEE SEC fractions from 9 to 10.5ml. Of a total of 452 Cp183 capsids, 133 (29.4%) had a T=3 morphology. For Cp183-EEE capsids, 280 (32.2%) of a total of 870 capsids, were T=3.

Weight average molecular masses were calculated from the absorbance data and the ratio of T=3 to T=4 capsids. For Cp183, a 240 protein capsid with 20.0 nt/capsid protein will have a mass of 6.59 MDa. An RNA-filled T=3 Cp183 particle will have a mass of 4.94 MDa. For Cp183-EEE, which has only 14.4 nucleotides/protein, a T=4 particle will have a mass of 6.16 MDa and a T=3 particle will have a mass of 4.61 MDa. Accounting for the T=3 and T=4 percentages, the resulting weight average masses per capsid were 6.10 MDa and 5.66 MDa for Cp183 and Cp183-EEE, respectively. These masses were in good agreement with masses for Cp183 determined by SEC-MALLS as well as earlier studies using analytical ultracentrifugation and scanning transmission electron microscopy (STEM) density analysis (Wingfield et al., 1995).

Discussion

Here we have presented a general method for evaluating the protein and nucleic acid content of large macromolecular complexes and verified our results by comparison to SEC-MALLS. The method presented here rapidly and accurately determined the RNA and protein concentration of virus capsids using light scattering corrected UV absorption spectroscopy. We examined the RNA content of BMV capsids containing their native genome and HBV capsids containing an ill-defined mixture of RNA derived from the E. coli expression system. This approach was straightforward to implement and required small volumes at moderate concentration, which could be recovered from the cuvette.

Using this technique, the calculated molecular mass of RNA-filled BMV capsids was within 4% of the mass determined by SEC-MALLS and the expected mass (Fig 2C). In a heterogeneous experimental system, such as E. coli expressed HBV, where the RNA content is poorly characterized (Birnbaum and Nassal, 1990), this method allowed quantification of the different RNA content of the Cp183 and Cp183-EEE HBV capsids and estimation of the masses of T=3 and T=4 capsids that were in good agreement with the weight average molecular masses determined experimentally by SEC-MALLS. Differences in nucleic acid UV absorbance related to packaging were apparently minimal, suggesting that the net packaging-induced changes to RNA secondary structure were also minimal, at least in HBV and BMV.

These results also allowed us to address the role of internal charge on RNA packaging. This question is of particular interest with HBV, where the RNA-binding domain is partially phosphorylated when the virus initially assembles and the phosphorylation state changes as DNA is reverse transcribed and the RNA is hydrolyzed by the encapsidated reverse transcriptase (see (Perlman et al., 2005) and references therein). We were able to use E. coli expressed HBV capsid protein to examine the effect of charge on RNA packaging, conditions where mainly random host RNA is packaged (Birnbaum and Nassal, 1990; Nassal, 1992), while excluding the complications of fitness. The RNA-binding domain of the 183-residue HBV capsid protein (Cp183) has a +15 charge while the mutant Cp183-EEE has three Ser to Glu mutations corresponding to the significant phosphorylation sites S155, S162, and S170; Cp183-EEE successfully packages genomic RNA in cell culture (Lan et al., 1999).

The RNA content determined for Cp183 and Cp183-EEE agreed well with the predictions of the counterion condensation theory, Poisson-Boltzmann theory, and the trend observed for ssRNA packaged (Belyi and Muthukumar, 2006; Hagan, 2009; Manning, 2001). If viruses have evolved to package RNA at a rote to achieve an electrostatic energetic minimum, and the physical predictions of that minimum are correct, there should be 1.4 to 1.6 RNA phosphates per RNA-binding domain positive charge. Each T=4 Cp183 capsid has 3,600 positive charges, suggesting that they should package 5100 to 5800 nucleotides, slightly in excess of our observed RNA content of 4,800 nucleotides. This amount is consistent with the amount of partially double stranded DNA packaged in a mature capsid, which is expected to be largely unphosphorylated (Perlman et al., 2005). In Cp183-EEE the positive charge is +12 per protein leading to 2,880 positive charges per T=4 capsid. If theory is correct, we should expect 4000 to 4600 packaged nucleotides. We observed 3450 nucleotides per T=4 Cp183-EEE capsid, about the amount of RNA in a polyadenylated genome, and not far short of the predicted amount. Thus, the amount of packaged RNA is proportional to the charge of the RNA-binding, consistent with the electroneutrality hypothesis (Chua et al., 2009; Newman et al., 2009), though the actual charge ratio differs.

There are several limitations to this method. Foremost, it relies on good absorption spectra free from baseline errors and contaminating chromophores. This approach assumes that the nucleic acid and protein components have the same absorbance free and assembled; the possibility of changes in nucleic acid hypochromicity due to RNA folding should not be overlooked though there is little if any change in RNA aborbance for BMV, Cp183, and Cp183-EEE. Hyperchromicity is generally a small effect, but an awareness of the possibility may help identify RNA folding associated with packaging in a nucleoprotein complex.

The method also relies on a two point estimation of light scattering, which, in spite of its success (Fig 1), may over or underestimate the amount of light scattering due to noise or errors in the scattering model (eg the 4^th power wavelength dependence). For rod-like polymers whose length is long compared to λ, a third power dependence is appropriate (Berne, 1974). Alternative estimates of c1 and c2 can be obtained by fitting the apparent absorbance between 340nm and 400nm using log-log plots of apparent absorption and wavelength (Leach and Scheraga, 1960; Winder and Gent, 1971). For very large nucleoprotein complexes, the Rayleigh approximation does not predict the wavelength dependence of light scattering well. In these instances, better estimates may be obtained using Mie theory (Cox, DeWeerd, and Linden, 2002; Drake and Gordon, 1985; Matzler, 2002–08; Weiner, Rust, and Donnelly, 2001). Also, the calculated molecular mass from UV absorbance is very sensitive to variation in the ratio between ε_RNA260nm and ε_RNA280nm, which is a function of pH, sequence, secondary structure, and other factors (Laqua, Melhuish, and Zander, 1988; Wilfinger, Mackey, and Chomczynski, 1997).

The investigations presented here have not been exhaustive, however, they validate a consistent and simple assay for concentrations of protein and RNA from a mixture. This method is very sensitive to differences in nucleic acid content and in our hands has functioned as a useful measure of the RNA content of virus capsids (Mukherjee et al., 2007; Porterfield et al., 2010). This method should be applicable to other viruses and nucleoprotein complexes and may be particularly useful for characterizing artificial virus-like particles. Absorbance may also be a useful approach for examining the effect of DNA packaging at very high concentrations, as in dsDNA bacteriophages; where there is known DNA content, solving equation 12 for ε_260,DNA (assuming the usual ε_260,DNA=1.8ε_280,DNA) will offer into the three dimensional structure and base stacking of the DNA.

Materials and Methods

Capsid Expression and Purification

BMV capsids were purified from infected plants as previously described (Gopinath, Dragnea, and Kao, 2005; Yi et al., 2009) and stored at −80°C in SAMA Buffer (50mM sodium acetate, 8mM magnesium (acetate)₂, pH 4.5). For HBV, the adyw wildtype HBV capsid protein sequence (Cp183), codon optimized for expression in E. coli, and the S155E/S162E/S170E (Cp183-EEE) mutant were recently described (Porterfield et al., 2010). Purified HBV capsids were stored in 5% sucrose, 5mM EDTA, 2mM DTT, 50mM HEPES pH 7.5 at −80°C but were dialysed into 150mM NaCl, 50mM HEPES pH 7.5 for experiments. Due to a tendency to aggregate, HBV capsid samples were passed through a 0.2um syringe filter immediately before injection into SEC-MALLS.

SEC-MALLS

Size Exclusion Multi Angle Laser Light Scattering (SEC-MALLS) experiments were conducted using an analytical Superose 6 10/30 column inline with a Wyatt DAWN HELEOS II light scattering detector and a Wyatt Optilab Rex refractive index detector. The system was calibrated using 65 kDa Bovine Serum Albumin (Sigma) and 820kDa GroEL. Injections of 100–200ug capsid samples were separated by SEC at a flow rate of 0.5 mL/min. The SEC-MALLS system was pre-equilibrated in SAMA buffer for BMV experiments and in 150mM NaCl, 50mM HEPES pH 7.5 for HBV experiments. Weight-averaged molecular weight, Mw, calculations were automated using ASTRA, using a dn/dc value of 0.186. The dn/dc value for RNA is ~0.16, however, given the mass of RNA per capsid protein, the error in the dn/dc value is less than 3%. Fractions were collected at 20s intervals for analysis by UV absorption spectroscopy. The delay between the RI detector and the elution of the fractions for UV absorption was estimated by comparing the onset and maxima of peaks.

Measurement and Analysis of UV Absorption Spectra and Light Scattering

UV absorption spectra were collected using an Agilent 8453 UV-Visible spectrophotometer blanked with the appropriate buffer used for SEC-MALLS experiments. Quartz cuvettes were thoroughly washed with Milli-Q water and 70% ethanol and then dried completely under vacuum between samples.

For RNA polynucleotides, we used an extinction coefficient of 8,000 M⁻¹cm⁻¹ per nucleotide at 260nm (ε_RNA260nm) (Maniatis, Fritsch, and Sambrook, 1980). A 260nm/280nm ratio for pure RNA of 2.0 (Glasel, 1995) was used to calculate an ε_RNA280nm of 4,000 M⁻¹cm⁻¹. Though not used in this paper, for DNA viruses, extinction coefficients of ε_RNA260nm = 7000 M⁻¹ cm⁻¹ and ε_RNA280nm = 4400 M⁻¹ cm⁻¹ would be appropriate.

An excel spreadsheet calculator is available in Supporting Material to facilitate automating these calculations.

Electron Microscopy

Carbon coated copper grids (EMS), were glow discharged and 5uL of HBV capsid samples were applied and immediately blotted dry. The grids were negatively stained with 2% uranyl acetate for 1min and then blotted dry. Images were obtained at 50,000x magnification using a JEOL JEM-1010 transmission electron microscope and recorded using a Gatan 4k x4k CCD camera.