|Molecular Vision 2005;
Received 29 June 2005 | Accepted 15 September 2005 | Published 16 September 2005
Screening of crystallin-crystallin interactions using microequilibrium dialysis
Aldo Ponce, Larry Takemoto
Division of Biology, Kansas State University, Manhattan, KS
Correspondence to: Larry Takemoto, Division of Biology, Ackert Hall, Kansas State University, Manhattan, KS, 66506; Phone: (785) 532-6811; email: email@example.com
Purpose: It has been hypothesized that short-range, protein-protein interactions of crystallin are necessary for the maintenance of lens transparency. Because of their probable weak nature, it has been difficult to both detect and quantitate the nature of these interactions. To determine if interactions exist between α-crystallin and γ-crystallin under true equilibrium conditions, we have used microequilibrium dialysis.
Methods: Total α-crystallin and γ-crystallin were prepared from soluble proteins of fetal bovine lenses by HPLC and gel filtration chromatography. The proteins were added to one side of a microequilibrium dialysis cell, comprised of two chambers separated by a membrane with 100 kDa molecular weight cut-off. After reaching equilibrium, the amount of free γ-crystallin and the amount of γ-crystallin bound to α-crystallin was determined by HPLC and reverse phase analysis of both chambers. Selected γ-crystallin that bound to α-crystallin was further purified by ion exchange chromatography, and then incubated with α-crystallin, to verify the specificity of their binding.
Results: Analysis of both microequilibrium dialysis chambers incubated at different times at 37 °C indicated that equilibrium was reached at 4 days. When total α-crystallin and γ-crystallin were incubated for this time period, significant binding was observed between α-crystallin and the IIIA, II, and IVA species of γ-crystallin. These interactions were confirmed by microequilibrium dialysis determinations containing α-crystallin and purified γ-crystallin species.
Conclusions: These results show that microequilibrium dialysis can be used to demonstrate significant noncovalent interactions of α-crystallin and γ-crystallin under true equilibrium conditions.
A common characteristic of all ocular lenses is very high cytosolic concentrations of crystallin proteins, which comprise at least 35% of the lens wet weight . The very high protein concentrations result in a change in refractive index, assisting the lens in focusing incident light. In spite of these high concentrations of macromolecules, the lens of younger organisms is transparent. Studies by Delaye et al.  have strongly suggested that this transparency results from the short-range interactions of proteins, in a manner analogous to dense liquids and glasses. An understanding of the molecular basis of these possible interactions would therefore provide a basis for understanding the mechanisms involved in loss of transparency during aging and cataractogenesis.
Since α-, β-, and γ-crystallin comprise almost all of the water soluble proteins of the mammalian lens, it is these proteins that might be involved in putative interactions that result in lens transparency. Although the αA- and αB-crystallin chains associate to form hetero-oligomeric complexes of approximately 800 kDa in the bovine and human lens , in the intact lens there is controversy whether γ-crystallins associate either with themselves, or with other crystallins.
Previous studies have investigated a possible interaction between α-crystallin and γ-crystallin. Light scattering and gel permeation chromatography of bovine lens crystallin have demonstrated that small amounts of γ-crystallin inhibit α-crystallin aggregation , and use of a two-hybrid system have suggested possible interactions between specific α-crystallins and γ-crystallin [4,5]. Recently, a filtration assay has been used to quantitate possible α-/γ-crystallin interactions at equilibrium , although the assay cannot be used for all γ-crystallin species because of adsorption to the filter. Surface plasmon resonance has the potential to study interactions of γ-crystallin from old bovine lenses with α-crystallin from fetal bovine lenses , although it is necessary that one of the interacting species still be covalently immobilized to a solid substratum.
Microequilibrium dialysis is ideally suited to study these crystallin interactions, since it directly quantitates the interaction of two different molecules when at equilibrium, in a solution whose pH, ionic strength, and temperature can be specified. The two chambers of the dialysis apparatus are separated by a membrane that allows equilibrium passage of only one of the molecules studied. There are no possible artifacts due to derivatization of either molecule, since both molecules are in their native state, and a minimal amount of material is necessary, since the equilibrium chambers are as small as 25-50 μl in volume.
To study possible interactions of bovine α-crystallin and γ-crystallin, we have taken advantage of the large molecular weight difference of the α-crystallin oligomeric complex (800 kDa) and the γ-crystallin monomers (20 kDa), by using a 100 kDa cut-off membrane in the dialysis apparatus that allows selective passage of only γ-crystallin molecules. Using this methodology, we are able to quantitatively determine the binding of α-crystallin with specific γ-crystallin, under equilibrium conditions.
Preparation of crystallinFetal bovine lenses were obtained from Antech, Inc. (Tyler, TX), and stored at -80 °C until use. Three lenses were thawed at room temperature, decapsulated, then homogenized in 6.0 ml of ice cold TSK Buffer (0.06 M sodium phosphate, 0.1 M sodium sulfate, pH 7.0), in a Dounce homogenizer, followed by centrifugation for 15 min at 14,000 rpm in an Eppendorf 5402 refrigerated centrifuge. The supernatant was filtered using a 0.22 μm filter (Millipore, Bedford, MA), and 300 μl of the filtrate containing approximately 1 mg of protein was resolved at room temperature on a TSK G3000SW gel filtration column (7.8 mm x 300 mm; Tosoh Bioscience, Montgomeryville, PA) at a flow rate of 1.0 ml/min, using TSK buffer. The α- and γ-crystallin fractions were placed in 12,000-14,000 kDa cut-off dialysis tubing (Spectra/Por, Spectrum lab., Inc., Rancho Dominguez, CA), then concentrated using Aquacide I (Calbiochem, La Jolla, CA). To further purify the γ-crystallin fraction, it was dialyzed at 4 °C extensively against Buffer A (0.02 M Tris, 1 mM disodium ethylenediamine tetraacetate, 0.1 mM dithiothreitol, 3 mM sodium azide, pH adjusted to 6.0 with acetic acid) then resolved on a strong cation exchange column (S300, 4.6 mm x 250 mm; Eichrom Technologies, Darien, IL) equipped with a SCX cartridge (Phenomenex, Torrance, CA), using a gradient of Buffer A and Buffer B (0.02 M Tris, 1 mM disodium ethylenediamine tetraacetate, 0.1 mM dithiothreitol, 3 mM sodium azide, 0.5 M sodium acetate trihydrate, pH adjusted to 6.0 with sodium hydroxide).
Protein samples were dialyzed extensively against ED Buffer (10 mM Tris, 0.1 mM dithiothreitol, 3 mM sodium azide, 0.15 M sodium chloride, pH 7.4), then analyzed for protein using bovine serum albumin as standard , followed by loading into an microequilibrium dialysis apparatus containing two chambers of 50 μl volume (Nest Group, Southborough, MA). α- and γ-crystallin were loaded in various molar ratios, assuming an average monomeric molecular weight of 20,000 for these proteins. This average molecular weight was used in all subsequent calculations. The amounts of α- and γ-crystallin added to the microequilibrium dialysis chambers were based on calculations considering a Kd in the micromolar range , an approximate monomeric molar binding ratio of 1:1 from previous studies , and the presence of sufficient amounts of bound γ-crystallin to be accurately quantitated by HPLC analysis. A 100 kDa cut-off regenerated cellulose filter (Millipore, Bedford, MA), was cut to the inside dimensions of the microequilibrium dialysis apparatus, soaked 2-3 h in distilled water, then 1-2 h in ED buffer. The filter was inserted between the two chambers, and 50 μl of ED buffer with or without protein was carefully injected into each chamber, using a 50 μl, blunt end syringe. The chamber into which the protein solution was injected was defined as the "full" chamber, and the other chamber defined as the "empty" chamber. Care was taken to prevent introduction of air bubbles into the chambers. The apparatus was then incubated at 37 °C with mild shaking for various periods of time.
Protein analysis after microequilibrium dialysis
Samples from each chamber were removed using a 100 μl capacity, gel loading pipet tip (Fisher Scientific, Pittsburgh, PA), and 5 μl injected into a C18 reverse phase column (4.6 mm x 250 mm, 300 Å pore size; Phenomex, Torrance, CA). Protein was eluted from the column using a 36-40% linear gradient formed between water and acetonitrile containing 0.1% (v/v) trifluoroacetic acid for 35 min at a flow rate of 1 ml/min, and the elution was monitored at 215 nm. Peak areas were determined using PeakSimple software (SRI instruments, Torrance, CA). To identify the γ-crystallin species present in each peak, the peaks were collected, lyophilized, and analyzed using a Kratos Axima CFR MALDI-TOF mass spectrometer in linear mode at the Proteomic Mass Spectrometry facility at the University of Massachusetts Medical School, Worcester, MA.
Figure 1 shows the elution profile of total soluble proteins from fetal calf proteins, when resolved on a TSK G3000SW gel filtration column. As previously described , the first major peak (5.945 min) eluting at the void volume is comprised of α-crystallin, while the last major peak (11.138 min) is comprised of γ-crystallin. These two peaks were concentrated using Aquacide I, and then extensively dialyzed against ED buffer prior to microequilibrium dialysis.
Since equilibrium dialysis is dependent upon the passage and equilibration of γ-crystallin across the 100 kDa cut off membrane from the "full" to the "empty" chamber, studies were done to determine the minimum time to reach equilibrium. Figure 2 shows the reverse phase elution profile of γ-crystallin in the empty chamber after one (Figure 2A) and four days (Figure 2B) of incubation. The results show that more γ-crystallin species pass into the empty chamber after four days of incubation compared to one day of incubation. As expected, no detectable α-crystallin was present in the empty chamber when α-crystallin alone was injected into the filled chamber and incubated for 4 days (results not shown).
Figure 3 shows the quantitation of each peak in the full chambers compared to the empty chambers at one and four days. After one day (Figure 3A), most of the peaks were found in larger amounts in the full chamber, while at four days (Figure 3B), the amounts of each peak in each chamber are much closer in amount. Statistical analysis using Student's t-test indicated that with the exception of peak 3, which is a minor peak, there is no statistical difference in the amounts of the remaining peaks in the full and empty chambers at four days. Consistent with these results, all subsequent microequilibrium dialysis experiments were incubated for a minimum of four days.
Based upon Figure 3, it should be possible to use microequilibrium dialysis to quantitate the possible interactions of α-crystallin with γ-crystallin under equilibrium conditions, since the 100 kDa membrane will allow free γ-crystallin, but not α-crystallin, to equilibrate across the two chambers. As a result, any associations of γ-crystallin with α-crystallin will be detected as an increase in the area of the γ-crystallin peak in the full chamber, relative to the empty chamber. Table 1 shows the results of such analysis, using two different molar ratios of α-crystallin, relative to γ-crystallin, that were initially added to the full chamber. For γ-crystallin, the initial overall concentration was 8 mg/ml, and the values at an α-/γ-crystallin ratio of 0:1 represent the area of the γ-crystallin peaks when no α-crystallin is initially added to the full chamber. A decrease in these values in the equilibrated "empty" chamber when γ-crystallin is added in the presence of α-crystallin to the "full" chamber indicates associations of the two crystallins. At a molar ratio of 1.9:1 (α-/γ-crystallin), peak 1 and peak 4 show significant associations with α-crystallin, while at an increased ratio of α-/γ-crystallin (2.8:1), peak 1 and peak 4, plus peak 2 show significant associations.
The results of Table 1 show that multiple species of γ-crystallin associate with α-crystallin under equilibrium conditions. Since the initial amounts of each γ-crystallin species are different, it is difficult to directly compare the relative binding of each species with the α-crystallin. For this purpose, and to verify that specific γ-crystallin species were indeed associating with α-crystallin, selected γ-crystallin species in their native state were purified by ion exchange chromatography, then incubated with α-crystallin using microequilibrium dialysis. Figure 4 shows the resolution of γ-crystallin species, using a strong cation exchange column. The peaks designated as S300-5, S300-7, and S300-8 were shown to be pure using reverse phase chromatography, and were therefore used for further microequilibrium dialysis studies.
Table 2 summarizes the results of microequilibrium dialysis using these purified γ-crystallins. The increased amounts of γ-crystallin species in the full chamber compared to the empty chamber was used to calculate the percentage of α-crystallin that bound to γ-crystallin at equilibrium. All three purified gamma species associate with α-crystallin, with peak S300-8 having the lowest amount of association (10%) and peak S300-7 having the highest amount of association (18%). Analysis of peaks S300-5, S300-7, and S300-8 by MALDI-TOF indicated that they corresponded to γIIIA-, γII-, and γIVA-crystallin, respectively.
In spite of very high concentrations of α-, β-, and γ-crystallin, the young mammalian lens is still transparent. Although short-range, crystallin interactions are thought to be responsible for this transparency , with the exception of the well-known interactions of α-crystallin with themselves and β-crystallin with themselves, there is limited evidence of heterologous interactions between crystallin in these lenses. Light scattering, gel chromatography, and the two-hybrid system [3-5] have demonstrated possible interactions between α-crystallin and γ-crystallin, but a quantitative analysis of this interaction under true equilibrium conditions has yet to been determined.
Microequilibrium dialysis provides the ideal system for this determination, since it uses a minimal amount of protein, whose interactions are measured under true equilibrium conditions. There is no need to derivatize any of the binding proteins, and the binding conditions (pH, ionic strength, temperature, etc.) can be varied. Furthermore, the ratio of the two proteins involved in a possible interaction can be varied, to simulate ratios that occur in vivo. Microequilibrium dialysis only requires that the membrane separating the two chambers be permeable to the free, but not bound protein. In this respect, a 100 kDa cut-off membrane is ideal for studying the putative interaction of the γ-crystallin, with a monomeric molecular weight of approximately 20 kDa, with the α-crystallin, with an oligomeric molecular weight of approximately 800 kDa. Furthermore, by changing the molecular weight cut-off range of the membrane, it should be possible to quantitate the interactions of other crystallin species present in the intact mammalian lens.
Interactions can either be studied between heterogeneous mixtures of proteins such as total γ-crystallin shown in Table 1, or interactions with purified proteins as shown in Table 2. Analysis of heterogeneous mixtures of proteins will provide a rapid means of simultaneously screening many proteins for the presence of protein species that bind to another protein such as α-crystallin. Table 1 shows that α-crystallin binds to some, but not all species of γ-crystallin, when incubated at α-/γ-crystallin ratios of 1.9:1 and 2.8:1. Proteins that do bind (i.e., peaks S300-5, S300-7, and S300-8), can then be purified and their binding to α-crystallin confirmed in a microequilibrium dialysis assay where the concentrations of all γ-crystallin species are kept constant.
The results of such an assay, shown in Table 2, demonstrate significant binding (10-18%) of all purified γ-crystallin species with α-crystallin. These results are consistent with other studies, which have suggested a weak interaction of α- and γ-crystallin [3-7]. An advantage of the microequilibrium dialysis described in this report is its ability to readily obtain absolute numbers for the amounts of α- and γ-crystallin involved in a noncovalent complex. The magnitude of this binding suggests that it could not be the result of the well-known chaperone activity of α-crystallin , since the α-crystallin target protein interaction is irreversible and not found in perceptible quantities in the fetal bovine lens. The results strongly suggest a significant percentage of γ-crystallin is associated with α-crystallin under true equilibrium conditions. The association constant for this interaction must be weak, since the proteins resolve as individual α-crystallin or γ-crystallin species when resolved by the TSK G3000SW gel permeation column. To ascertain the weak interaction of these crystallins, the same amounts of α- and γ-crystallin species were incubated either individually or as a mixture at 37 °C for four days, and then quantitatively analyzed by TSK G3000SW chromatography. The results show that the same amounts of individual α- and γ-crystallin species were present, irregardless of whether they were incubated as individual α- and γ-crystallin species, or as a mixture of the two types of crystallin (results not shown). The results indicate that α-/γ-crystallin interactions shown in Table 1 and Table 2 must be of a weak nature that can be readily reversed by dilution in the TSK G3000SW column.
The results of Table 2 do indicate significant interactions of γ-crystallin with α-crystallin when incubated at a molar ratio of 5:4, respectively. By using microequilibrium dialysis, it should be possible to determine the percentage of γ-crystallin bound at different ratios of γ-/α-crystallin, which varies in different parts of the lens. Most importantly, since the α-crystallins and γ-crystallins are known to undergo extensive posttranslational modifications during aging and cataractogenesis [12-14], it should be possible using microequilibrium dialysis to determine the effects of these posttranslational modifications upon the associations of these two proteins, to ascertain whether they affect the crystallin/crystallin interactions thought to be necessary for the maintenance of lens transparency.
We wish to thank K. Albin and S. Jackson for assistance in preparation of proteins, and J. Leszyk for the MALDI-TOF analysis of γ-crystallin. This research was supported by grant EY02932 from the National Eye Institute.
1. Bloemendal H. The vertebrate eye lens. Science 1977; 197:127-38.
2. Delaye M, Tardieu A. Short-range order of crystallin proteins accounts for eye lens transparency. Nature 1983; 302:415-7.
3. Mach H, Trautman PA, Thomson JA, Lewis RV, Middaugh CR. Inhibition of alpha-crystallin aggregation by gamma-crystallin. J Biol Chem 1990; 265:4844-8.
4. Fu L, Liang JJ. Detection of protein-protein interactions among lens crystallins in a mammalian two-hybrid system assay. J Biol Chem 2002; 277:4255-60.
5. Fu L, Liang JJ. Alteration of protein-protein interactions of congenital cataract crystallin mutants. Invest Ophthalmol Vis Sci 2003; 44:1155-9.
6. Biswas A, Das KP. Role of ATP on the interaction of alpha-crystallin with its substrates and its implications for the molecular chaperone function. J Biol Chem 2004; 279:42648-57.
7. Peterson J, Radke G, Takemoto L. Interaction of lens alpha and gamma crystallins during aging of the bovine lens. Exp Eye Res. In press 2005.
8. Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 1976; 72:248-54.
9. Wang K, Spector A. The chaperone activity of bovine alpha crystallin. Interaction with other lens crystallins in native and denatured states. J Biol Chem 1994; 269:13601-8.
10. McFall-Ngai MJ, Ding LL, Takemoto LJ, Horwitz J. Spatial and temporal mapping of the age-related changes in human lens crystallins. Exp Eye Res 1985; 41:745-58.
11. Horwitz J. Alpha-crystallin can function as a molecular chaperone. Proc Natl Acad Sci U S A 1992; 89:10449-53.
12. Takemoto LJ. Oxidation of cysteine residues from alpha-A crystallin during cataractogenesis of the human lens. Biochem Biophys Res Commun 1996; 223:216-20.
13. Takemoto L, Boyle D. Deamidation of specific glutamine residues from alpha-A crystallin during aging of the human lens. Biochemistry 1998; 37:13681-5.
14. Yang Z, Chamorro M, Smith DL, Smith JB. Identification of the major components of the high molecular weight crystallins from old human lenses. Curr Eye Res 1994; 13:415-21.