Alteration of Protein–Protein Interactions of Congenital Cataract Crystallin Mutants

Abstract

purpose. A recent study demonstrated the presence of protein–protein interactions among lens crystallins in a mammalian cell two-hybrid system assay and speculated about the significance of these interactions for protein solubility and lens transparency. The current study extends those findings to the following crystallin genes involved in some congenital cataracts: CRYAA (R116C), CRYAB (R120G), and CRYGC (T5P).

methods. A mammalian two-hybrid system was used to assay the protein–protein interactions. Congenital cataract crystallin genes were cloned and fused into the two-hybrid system vectors (target and prey proteins). Together, with the third vector containing a reporter gene, chloramphenicol acetyltransferase (CAT), they were cotransfected into human HeLa cells. The presence of protein–protein interactions and the strength of these interactions were assayed by CAT ELISA.

results. The pattern of changes in protein–protein interactions of those congenital cataract gene products with the three major crystallins, αA- or αB-, βB2-, and γC-crystallins, differed. For the T5P γC-crystallin, most of the interactions were decreased; for the R116C αA-crystallin, the interactions with βB2- and γC-crystallin decreased and those with αB-crystallin and heat-shock protein (Hsp)27 increased; and for the R120G αB-crystallin, the interactions with αA- and αB-crystallin decreased, but those with βB2- and γC-crystallin increased slightly. An attempt was made to interpret the results on the basis of conformational change and disruption of dimeric interaction involving β-strands.

conclusions. The results clearly indicate that crystallin mutations involved in congenital cataracts altered protein–protein interactions, which may contribute to decreased protein solubility and formation of cataract.

A great deal of attention has recently been paid to autosomal dominant congenital cataracts in which point mutations occur in specific crystallin genes—for example, CRYAA (R116C) in zonular central nuclear cataract, ¹ CRYAB (R120G) in desmin-related myopathy, ² CRYBB2 (truncation of 51 amino acids from the C terminus) in cerulean cataract, ^{3 4} CRYGC (T5P) in Coppock-like cataract, ⁵ CRYGD (R58H) in aculeiform cataract, ⁵ and CRYGD (R14C) in juvenile-onset punctate cataract. ^{6 7} Many reports indicate that mutations in αA- or αB-crystallin cause structural and/or functional changes. ^{8 9 10 11 12} We also observed a conformational change in and destabilization and insolubilizeation of the T5P γC-crystallin mutant. ¹³ However, mutations in γD-crystallin (R14C or R58H) cause an increase in phase-separation temperatures or an enhancement in crystallization rather than a conformational change. ^{14 15} It is our belief that those mutations are likely to cause changes in protein–protein interactions that are observed in vivo. 

The mechanism of lens transparency was traditionally thought to arise from short-range order among crystallins. ^{16 17 18} Many studies have been conducted to examine specific protein–protein interactions among crystallins, mostly by spectroscopic or light-scattering measurements. ^{19 20 21 22 23 24 25 26 27 28 29} However, these studies require purified crystallins at high concentrations and are complicated by excluded-volume effects. ³⁰ To study protein–protein interactions in vivo, a two-hybrid system, either in yeast or in a mammalian cell, has been used. ^{31 32 33} Indeed, these studies suggest the presence of protein–protein interactions, either homogeneous or heterogeneous, among crystallins. The significance of these findings is that they may serve as baselines for studying the effects of gene mutations. In the present study, we extended the findings in our previous two-hybrid study of crystallins to congenital cataract crystallin mutants, with R116C αA-, R120G αB-, and T5P γC-crystallins used as models. Our results indicate that these mutations affect protein–protein interactions among crystallins. 

Materials and Methods

Subcloning of Cataract Mutant Genes into the Two-Hybrid System Vectors

Subcloning of R116C αA-crystallin and R120G αB-crystallin mutants was performed with a mutagenesis kit (Quick-Change; Stratagene, La Jolla, CA) and various constructs (BD-αA and BD-αB or AD-αA and AD-αB) as templates. The two forward primers (TTCCCGTGAGTTCCACTGCCGCTACCGCCTGCC and CTCCAGGGAGTTCCAC G GGAAATACCGGATCCC) and the two reverse primers (GGCAGGCGGTAGCGGCAGTGGAACTCACG GGAA and GGGATCCGGTATTTCCCGTGGAACTCCCTG GAG) for R116C αA- and R120G αB-crystallin mutations were custom synthesized (Life Technologies). The mutation sites are shown in bold. 

For subcloning of T5P γC-crystallin, PCR was performed on AD-γC and BD-γC constructs with two primers: the forward primer GGAATTCATGGGGAAGATCCCCTTC and the reverse primer CGGTAGTGTTAATCTAGATTAAT. The underlined sequences are the restriction sites for EcoRI and XbaI, respectively. 

A mammalian two-hybrid system assay kit was used (Clontech, Palo Alto, CA). ^{33 34} The first test protein (bait) was fused into the GAL4 DNA-BD in the pM vector, and the second test protein (prey) was fused into the VP16 AD in the pVP16 vector. During setup of the two-hybrid system, a positive and a negative control were used to ensure that the system works. The control experiments compared the interactions between the p53 protein and two other proteins: the SV40 large T antigen, which is known to interact with p53, and a polyoma virus coat protein (CP), which does not interact with p53. 

Cotransfection and CAT ELISA Assays

After construction of the two-hybrid system plasmids pM-X and pVIP16-Y, these two vectors, along with the pG5CAT reporter vector, were cotransfected into the human HeLa cells using a transfection reagent (Lipofectamine; Life Technologies, Rockville, MD). ³³ After being cultured for 72 hours at 37°C in 5% CO₂, the cells were harvested and lysed. Interactions between proteins x and y were assayed by measuring the expression of the CAT gene with the CAT enzyme-linked immunosorbent assay kit (CAT ELISA; Roche Molecular Biochemicals, Indianapolis, IN). The CAT readings were normalized in comparison with total protein concentration and expressed as multiples of increase of CAT activity relative to the control. The controls were the expression systems with constructs devoid of DNA inserts. Three independent transfections were performed, and each CAT experiment was performed in duplicate. Also included were experiments in which vectors were switched (X-BD/Y-AD and Y-BD/X-AD), to ensure that there was no vector specificity. 

After protein interactions were detected through the two-hybrid system screen, we used coimmunoprecipitation (co-IP) to confirm the interactions. ³³  

Results

Interactions Involving T5P γC-crystallin

In the two-hybrid system assay, the interactions between T5P mutants themselves remained unchanged, but those between the T5P mutant and wild-type (WT) γC-crystallin and other crystallins decreased (Fig. 1A) . CAT activities became negligible between the T5P mutant and WT γC-crystallin and other crystallins compared with those between the WT γC-crystallins themselves and between the WT γC-crystallin and other crystallins. In a separate experiment, we have found expression of fusion T5P mutant in the HeLa cells in both the soluble and insoluble fractions. The decreased CAT activities are not due to decreased solubility of the T5P mutant. In bacterial expression, most of the T5P was found in the inclusion body. ¹³  

Interactions Involving R116C αΑ-crystallin

Protein–protein interactions between the R116C mutant and other crystallins changed dramatically, but those between R116C themselves or between the R116C mutant and WT αA-crystallin showed little change (Fig. 1B) . The most noticeable changes were a large increase in the interactions between R116C and αB-crystallin and between R116C and heat-shock protein (Hsp)27 and a decrease in the interactions between R116C and βB2-crystallin and between R116C and γC-crystallin. 

Interactions Involving R120G αΒ-crystallin

Changes in protein–protein interactions involving the R120G αΒ-crystallin mutant differ from those involving the R116C αA-crystallin mutant (Fig. 1C) . There was an appreciable decrease only in interactions between the R120G mutant and αA- or αB-crystallin, a small increase in the interactions between the R120G mutant and βB2-crystallin, and essentially no change in the interactions between the R120G mutant and γC-crystallin or Hsp27. 

Co-IP and Western Blot Analysis

As in our previous report, ³³ Co-IP (Fig. 2A for R116C αA-crystallin and Fig. 2B for R120G αB-crystallin) confirmed that the presence of interactions detected from two-hybrid system assays, and Western blot indicated that levels of expression essentially did not differ between the WT and the mutants. The Co-IP for the T5P γC-crystallin is not included; it showed virtually no staining, because there was no interaction with any other crystallins. The differences shown in CAT data between WT and R116C αΑ-crystallin or between WT and R120G αB-crystallin were not due to the differences of expression levels (Fig. 3A) . Similarly, the increased CAT levels for Hsp27 in cotransfected cells were not due to increased expression levels (Fig. 3B) . 

Discussion

We used three congenital cataract crystallin genes as models for studying the effect of mutations on protein–protein interactions. Each mutation exerts a unique effect on the protein–protein interactions but its mechanism can be explained in terms of disruption of the β-strand structure or dimerization by mutations. We believe that observed changes of interactions are due to mutation rather than fusion that may affect the ability of the GAL 4 DNA-BD to bind to the reporter gene, because we observed the same results when switching the vectors for fusion. 

CRYGC (T5P) is one of the many γ-crystallin mutant genes for autosomal dominant congenital cataracts. This mutation is associated with Coppock-like cataract and has the phenotype of a dustlike opacity of the fetal lens nucleus. ⁵ On cloning and overexpression of the mutant gene, most of the T5P γC-crystallin is in the inclusion body and must be solubilized by GdnHCl. Spectroscopic measurements indicate that the T5P mutation changes conformation and decreases conformational stability. ¹³ The R116C αA-crystallin mutation is related to congenital cataract, ¹ and the R120G αB-crystallin mutation causes desmin-related myopathy and cataract. ² Both of these inherited diseases are autosomal dominant. Many reports indicate that these mutants have undergone conformational change, high-molecular-weight (HMW) aggregation, increased membrane binding, and reduced chaperone activity. ^{8 9 10 11 12}  

The aforementioned biophysical studies of congenital cataract crystallin mutants, however, did not examine the effect of individual mutation on other crystallins. We believe that our study of protein–protein interactions provides such information. Many two-hybrid system studies have shown interactions among crystallins in either homogeneous or heterogeneous systems. ^{31 32 33} These interactions are true interactions detected in vivo and arising from the two expressed proteins’ being in close physical contact and are unlike in vitro studies of protein–protein interactions that require high protein concentrations, in which detected interactions are mostly due to excluded volume effects. In the case of T5P γC-crystallin, the interactions between the T5P mutant and WT γC-crystallin and other crystallins were disrupted, but not those between T5P γC-crystallins themselves, indicating that homogeneous interaction sites or domains differ from those of heterogeneous interactions. Our recent spectroscopic study indicates that the mutation partially unfolds the protein, ¹³ and the imperfect structure must affect the interactions with crystallins observed in the current study. To look further at the structural change by T5P mutation, we refer to the three-dimensional structure of γ-crystallin, which is characterized by the presence of four Greek key motifs: motifs 1 and 2 in the N-terminal domain and motifs 3 and 4 in the C-terminal domain. ^{35 36} The four motifs form four β-sheets: two (the β1- and β3-sheets) lie on the outside of the molecule, and two (the β2- and β4-sheets) are in partial contact (domain association). The Thr-5 residue is in the β1-strand. T5P mutation destroys the β1-strand ³⁷ and thus also destroys the β1-sheet, which in turn disrupts the highly symmetrical structure of γ-crystallin. This change of tertiary structure must affect the interactions with other crystallins. The partial unfolding of the T5P mutant presents an interesting question about why the chaperone binding of αA- or αB-crystallin does not increase with T5P mutant compared with the WT γC-crystallin. The possible answer is that either the chaperone-binding sites differ from the two-hybrid–interaction sites or that α-crystallins do not function as chaperones in the nucleus. This question should be investigated further. 

Changes in interactions involving R116C αA-crystallin are quite different from those involving R120G αB-crystallin. Interactions between R116C and WT αA-crystallins and between R116C αA-crystallins did not differ from those between WT αA-crystallins, but showed increased interactions between R116C αA- and αB-crystallin or Hsp27 and decreased interactions between αA- and βB2- or γC-crystallin. For the R120G mutant of αΒ-crystallin, only the self interactions of R120G αΒ-crystallins and interactions between R120G αΒ-crystallin and αA- or αB-crystallin decreased significantly. This striking difference seems difficult to explain. To explore the cause for these differences, we turn to the known three-dimensional structures of two small heat shock proteins (sHsps): one from Methanococcus jannaschii, Mj Hsp16.5, and the other from wheat (w)Hsp16.9. ^{38 39} Mj Hsp16.5 has a hollow spherical structure formed by 24 monomers, ³⁸ and wHsp16.9 has a structure of a dodecamer consisting of two disks, each comprising six α-crystallin domains organized in a trimer of dimers. ³⁹ Both have a twofold structure, a dimer that is the building-block for oligomerization. Based on homology to Mj Hsp16.5, and wHsp16.9, the Arg-116 residue of αA-crystallin, and the Arg-120 residue of αB-crystallin are located at equivalent positions and are highly conserved among sHsps (Arg-108 in both wHsp16.9 and Mj Hsp16.5). These amino acid residues are in the β7-strand, which participates in the twofold structural formation of wHsp16.5 but not Mj Hsp16.5. The two β-strands involved in the twofold structure are different: a β1-strand from one subunit and a β6-strand from a second subunit in Mj Hsp16.5 and a β7-strand from each subunit in wHsp16.9. Therefore, if the structures of αA- and αB-crystallins resemble wHsp16.9, both R116C and R120G mutations would affect the twofold structure. However, if the structures of αA- and αB-crystallins resemble Mj Hsp16.5, then these two mutations should not have much effect on the structure. A structure similar to Mj Hsp16.5 was constructed for αB-crystallin on the basis of homology and site-directed mutations. ⁴⁰ However, a closer look at the homology between Mj Hsp16.5 and αA- or αB-crystallin indicated the absence of amino acid residues in the equivalent sequences of the β6-strand in αA- and αB-crystallins. ³⁸ In the structure of wHsp16.9, β7-strand is involved in the dimerization. ³⁹ The observation of prominent changes in protein interactions by these two mutations may indicate that the structures of αA- and αB-crystallins resemble wHsp16.9 more than Mj Hsp16.5. Unlike the T5P mutation of γC-crystallin, the R116C mutation of αA-crystallin and the R120G mutation of αB-crystallin may not involve disruption of the β-strand, but may rather involve changes of interactions that stabilize the dimer structure (between Arg-108 and Glu-100 in wHsp16.9). In the R116C αA-crystallin and R120G αB-crystallin mutants, the positively charged Arg was replaced with uncharged Cys or Gly, and interactions that stabilized dimer structure were disrupted, which will affect the trimer and oligomer structures as observed in conformational studies ^{8 9 10 11 12} as well as protein–protein interactions observed in the current study. The nature of conformational change is different for the R116C mutation of αA-crystallin and the R120G mutation of αB-crystallin, as reported by a circular dichroism (CD) study, ¹⁰ and it is therefore not surprising that changes in protein–protein interactions are different for these two mutations. 

A particularly interesting observation is that interactions between the R116C αA-crystallin mutant and WT αB-crystallin or Hsp27 increased five- to sixfold over those between WT αA-and αB-crystallin or Hsp27. αB-crystallin and Hsp27 are sHsps. Their expression increases under stress. Intuitively, we may attribute the increased interactions to increased expression of αB-crystallin and Hsp27, because introducing the R116C mutant gene may exert stress on the cells. However, that effect was not seen for either the R120G or the T5P mutant gene. Furthermore, Western blot analysis indicated that expression levels of αB-crystallin and Hsp27 were almost identical with those in the control. The more likely explanation is that conformational change in the R116C mutant either exposed additional or increased the interaction sites with αB-crystallin and Hsp27, but not with αA-crystallin. 

Our previous two-hybrid system study indicates that genes of the N- and C-terminal fragments (αAn or αBn and αAc or αBc) of αA- and αB-crystallin contribute differently to oligomerization. ³³ Both αAn and αAc contribute to oligomerization of αA-crystallin, but αAn is more important than αAc. In αB-crystallin, only the αBc fragment makes an important contribution to oligomerization, which may explain the decrease in protein–protein interactions observed between αΒ-crystallin and the R120G αB-crystallin mutant and, in contrast, the lack of change in the interactions between WT αA-crystallin and the R116C αA-crystallin mutant. The exact domain or domains that contribute to oligomerization may be determined with site-specific mutations. The increased aggregation for either R116C αA-crystallin or R120G αB-crystallin reported in prior studies ^{9 10 12} appears not to be a dominant factor in the observed changes of protein interactions, because changes were not the same for the two mutations. Another point that should be emphasized is that changes in protein–protein interactions by mutation are relative to the interactions of WT crystallins. Therefore, although protein–protein interactions involving γC-crystallins are very low compared with those involving αA- or αB-crystallin or Hsp27, the percentages of decrease by mutations are very high. 

Our recent two-hybrid study indicates the presence of heterogeneous interactions among crystallins, such as αA and βB2, αA and γC, and βB2 and γC. ³³ Our speculation that these interactions are important in the maintenance of protein solubility in the lens is further strengthened by the recent report that significant amounts of β- and γ-crystallins are present in the inclusion body in the αA-crystallin knockout mouse lens (Horwitz J, ARVO Abstract 1921, 2002). This finding indicates that β- and γ-crystallins must interact with α-crystallin, especially the major component αA-crystallin, to maintain their solubility in vivo. A similar finding of αB-crystallin in the inclusion body in the αA-crystallin knockout mouse lens was reported earlier. ⁴¹ The decreased solubility of T5P γC-crystallin reported previously ¹³ may also result from the decreased protein interactions observed in the current study. 

The present study may enhance our understanding of human age-related cataract (ARC). The current perception of the etiology of ARC is mainly that it is related to aging and environmental risk factors and not to genetic factors. However, there is some speculation that there are ARC-related genes, although no such genes have yet been identified. ^{42 43 44} It is reasonable to assume that there is a link between individual genetic backgrounds and environmental risk factors in ARC. The search for such ARC-related genes is highly crucial in cataract research. The study of inherited cataract genes may pave the way for future studies of ARC genes. 

 Supported by Grant EY05803 from the National Eye Institute, and a grant from Massachusetts Lions Eye Research Fund, Northborough, Massachusetts.

 Submitted for publication September 26, 2002; revised October 21, 2002; accepted October 31, 2002.

 Disclosure: L. Fu, None; J.J.-N. Liang, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

CAT activities for detection of protein–protein interactions involving various crystallin mutants. (A) T5P γC-crystallin, (B) R116C αΑ-crystallin, and (C) R120G αΒ-crystallin. The activity is expressed as the multiple of increase in activation compared with the basal control (vectors without inserts) and normalized for total protein concentrations. Data are the mean ± SD of results in three independent experiments. Both X-AD and Y-BD fusions gave the same results as Y-AD and X-BD fusions.

Figure 2.

 

Coimmunoprecipitation and Western blot analysis. Cell lysates were first immunoprecipitated with polyclonal antibody specific to either (A) αA- or (B) αB-crystallin. The complexes were separated by SDS-PAGE and immunoblotted with monoclonal antibody specific to GAL4 DNA-BD (top). Immunoblot analysis was performed directly on lysates, using polyclonal antibody specific to either (A) αA- or (B) αB-crystallin (bottom).

Figure 3.

 

Immunoblot analysis with (A) monoclonal antibody specific to GAL4 DNA-BD and (B) polyclonal antibody specific to Hsp27. Lysates from cells cotransfected with AD-Hsp27 and (lane 1) BD-αA, (lane 2) BD-R116C αA, (lane 3) BD-αB, and (lane 4) BD-R120G αB.