Introduction

The two homologous, vertebrate proteins αA-crystallin and αB-crystallin are members of the small heat shock protein (sHSP) family. Both proteins represent a major component of the vertebrate eye lens, and together with β-crystallins and γ-crystallins, they constitute approximately 90% of all proteins in the lens (Bassnett et al. 2011; Bloemendal et al. 2004; Clark et al. 2012). Additionally, αB-crystallin is constitutively expressed in a wide variety of tissues such as the brain, cardiac and skeletal muscles, and kidney (Bhat and Nagineni 1989; Dubin et al. 1989; Iwaki et al. 1990; Iwaki et al. 1989; Kato et al. 1991b). In contrast αA-crystallin is found mainly in the eye lens (Robinson and Overbeek 1996) and only in minor amounts in other tissues like the spleen, heart, and thymus (Kato et al. 1991a; Srinivasan et al. 1992). Functionally, the ensemble of both α-crystallins is involved in maintaining eye lens transparency (Bassnett et al. 2011; Bloemendal et al. 2004; Clark et al. 2012; Delaye and Tardieu 1983; Jaenicke and Slingsby 2001; Tardieu 1988). The eye lens represents an environment with an exceptionally high protein concentration of up to ∼450 mg/mL (Fagerholm et al. 1981; Siezen et al. 1988) and little or no protein turnover (Hoehenwarter et al. 2006; Wistow and Piatigorsky 1988). Throughout lens development and aging, lens proteins go through covalent and non-covalent modifications and become progressively susceptible for aggregation and unfolding. α-Crystallins are thought to bind these fragile, non-native proteins and protect them from irreversible aggregation by means of their chaperone activity, ensuring proper light refraction and visual acuity. Both α-crystallins were found in vitro to inhibit thermal aggregation (Horwitz 1992; Horwitz 1993; Jakob et al. 1993), aggregation induced by the reduction of disulfide bonds (Bhattacharyya and Das 1998; Raman et al. 1995), and high-energy electromagnetic radiation like UV light (Bhattacharyya and Das 1998; Raman and Rao 1994; Reddy et al. 2001). Dysfunctions caused by mutations, degradation, or unspecific post-translational modifications lead to lens opacification, clinically termed cataract (Clark et al. 2012). One characteristic feature of sHsps and α-crystallins is to form ensembles of large homo-oligomeric and hetero-oligomeric complexes that are polydisperse and dynamic. Subunit exchange between different sHSPs in vitro and in vivo further contributes to ensemble complexity and variation of properties (Bova et al. 1997; Haslbeck et al. 2016; Kundu et al. 2007). The ability to form these dynamic ensembles also is closely correlated to the chaperone activity of sHsps and α-crystallins. In this context, a variety of co-translational and post-translational modifications (PTMs), with phosphorylation being one of the most prominent, have been shown to influence the ensemble composition and its activity (Hoehenwarter et al. 2006; Peschek et al. 2013). Structurally, α-crystallins exhibit a tripartite domain organization with a central conserved α-crystallin domain (ACD) flanked by less conserved N-terminal (NTS) and C-terminal sequences (CTS). The ACD represents the signature motif of all sHSPs and comprises ∼80 amino acids in human α-crystallins. It contains predominantly β-sheet structures forming an immunoglobulin-like β sandwich (Laganowsky et al. 2010; Laganowsky and Eisenberg 2010). The NTS of human αA-crystallin and αB-crystallin consists of 63 and 67 amino acids, respectively. While it remains elusive which structural elements are present in the NTS of αA-crystallin, for αB-crystallin, the first five residues were proposed to be highly flexible, and based on ssNMR and secondary structure prediction, it was proposed that the NTS includes three short α-helices and a short β-strand connected by long unstructured linkers (Braun et al. 2011; Carver et al. 1992; Jehle et al. 2011). Interestingly, the NTS of both α-crystallins displays high hydrophobicity and conformational mobility and is essential for higher order oligomer assembly (Eifert et al. 2005; Haslbeck et al. 2005; Jehle et al. 2011). N-terminally truncated mutants of bovine αB-crystallin form more heterogeneous ensembles of oligomers which contain overall a lower number of subunits than the oligomers formed by the full-length protein. NTS-depleted rat (Merck et al. 1992; Merck et al. 1993), bovine (Eifert et al. 2005; Laganowsky et al. 2010; Salerno et al. 2003), and human (Berengian et al. 1999; Bova et al. 2000; Kundu et al. 2007) αA-crystallin forms dimers and preferentially tetramers. Furthermore, the chaperone activity of the NTS-depleted mutants is lower than that of the full-length protein (Kundu et al. 2007; Laganowsky et al. 2010; Mainz et al. 2015; Merck et al. 1993).

Intriguingly, alternative RNA splicing of α-crystallin as a co-transcriptional and post-transcriptional process has not yet been investigated in detail. The two human genes CRYAA and CRYAB are the result of a gene duplication event of an ancestral α-crystallin gene leading to isoforms with 173 and 175 amino acids, respectively. For αA-crystallin, in rodents (Cohen et al. 1978a; Cohen et al. 1978b; de Jong et al. 1980) and certain mammals like hedgehog, bat, and pika (Hendriks et al. 1988), an alternative splice variant termed αAins with decreased chaperone activity (Derham et al. 2001) was identified. This variant is identical to αA-crystallin but has an additional “insert exon” located in the first intron, resulting in 23 additional amino acids between the NTS and ACD (King and Piatigorsky 1983; van den Heuvel et al. 1985). In primates however, the analogous exon probably developed to a pseudo-exon 30 to 40 myr ago and is in the early stages of being extinguished (Jaworski and Piatigorsky 1989).

In humans, the canonical isoform 1 of both αA-crystallin and αB-crystallin is encoded by three exons. The last two exons correspond to the ACD and the CTS. The first exon codes for the respective NTS (Graw 2009). For αB-crystallin, the Ensemble database (Release 86, October 2016) annotates several differential transcripts some of which are truncated at the 3′ or 5′ ends which prevent annotation of the complete coding sequence (CDS). Nevertheless, three transcripts are proposed to be protein coding. One is the canonical isoform with 175 aa. The others result in proteins with 155 aa (Transcript ID ENST00000533971.1) and 108 aa (Transcript ID ENST00000525823.1). The first variant originates from skipping of the splice site behind the second exon, therefore extending the second exon and terminating at a new stop codon. This would be a C-terminally shortened version of αB-crystallin changing the sequence of half of the ACD and the CTS. The second variant would lead to a skipping of the first exon, being equivalent to a truncation of the complete NTS.

Interestingly, for human αA-crystallin, only splice variants leading to substitutions of the NTS are predicted. Here, we describe the presence of an alternative splice variant of αA-crystallin with an alternative NTS in the lens tissue. We set out to analyze the structural and functional implications of the exchange of the NTS in this variant. Additionally, we show that the alternatively spliced αA-crystallin variant interacts with canonical human α-crystallins and modulates their oligomeric size and chaperone activity.

Materials and methods

Cloning, expression, and purification of α-crystallins

The synthetic gene (Thermo Fisher Scientific) for αA-crystallin isoform 1 was cloned in a pE-SUMOpro (Life Sensors) expression vector via BsaI and XbaI with an N-terminal 6× His-SUMO tag. For purification, the plasmid containing αA-crystallin isoform 1 was introduced into the Escherichia coli strain BL21 cod+. Cultures were grown in LB medium at 37 °C until an A 600 of ∼0.6 was reached. Subsequently, 1 mM IPTG was added for induction of the protein expression. After 4 h of incubation at 37 °C, cells were harvested, washed, and resuspended in buffer A (40 mM NaH2PO4/Na2HPO4 pH 7.4, 500 mM NaCl, 20 mM imidazole, 1 mM DTT). Cell lysis was accomplished by passing the cell suspension through a Basic Z cell disruption system (Constant Systems) at 2.3 kbar. After centrifugation for 30 min at 48,000×g, urea was added to 3.5 M to the supernatant. The protein was purified via metal chelate chromatography on a 5-mL HisTrap FF column (GE Healthcare). The column was pre-equilibrated and after sample application washed with buffer Aurea (40 mM NaH2PO4/Na2HPO4 pH 7.4, 500 mM NaCl, 20 mM imidazole, 3.5 M urea, 1 mM DTT) and buffer Burea (40 mM NaH2PO4/Na2HPO4 pH 7.4, 500 mM NaCl, 30 mM imidazole, 3.5 M urea, 1 mM DTT). The protein was eluted with buffer Curea (40 mM NaH2PO4/Na2HPO4 pH 7.4, 500 mM NaCl, 300 mM imidazole, 3.5 M urea, 1 mM DTT). Fractions containing protein were pooled and dialysed into low salt buffer (50 mM MES pH 6, 3.5 M urea, 2 mM EDTA, 1 mM DTT). For further purification, the protein was loaded onto a RESOURCE S cation exchange column (GE Healthcare) and eluted with a gradient spanning 20 column volumes (CV) to high salt buffer (50 mM MES pH 6, 500 mM NaCl, 3.5 M urea, 2 mM EDTA, 1 mM DTT). The protein was dialysed into buffer A, and the His-SUMO tag was cleaved off with His-tagged SUMO protease (Bepperling et al. 2012). Following a second metal affinity chromatography to separate from the SUMO protease and the tag, the protein was dialysed into PBS containing 2 mM EDTA and 1 mM DTT and after concentration loaded onto a Superdex 200 pg gel filtration column (GE Healthcare). The αA-crystallin containing fractions were pooled, concentrated, frozen in liquid nitrogen, and stored at −80 °C. αA-crystallin isoform 2 was synthesized (Thermo Fisher Scientific) and cloned into a modified pET-28b expression vector with an N-terminal 6× His-SUMO tag (Bepperling et al. 2012) and expressed in BL21 cod+ in LB medium. The cells were grown at 37 °C until an A 600 of ∼0.6 was reached. Expression was induced with 0.1 mM IPTG, and the cells were incubated at 30 °C for 16 h after induction. Further purification steps were equivalent to isoform 1 except that no urea was used throughout the protein preparation. The eluted protein was dialysed into buffer A, and the His-SUMO tag was cleaved off with SUMO protease. The protein was loaded onto a nickel column, the flow through dialysed into 40 mM NaH2PO4/Na2HPO4 pH 6.8, 40 mM NaCl, 1 mM DTT, and purified with a RESOURCE S cation exchange column (GE Healthcare) and a 20 CV gradient to 40 mM NaH2PO4/Na2HPO4 pH 6.8, 500 mM NaCl, 1 mM DTT as elution buffer. The eluted protein was dialysed into PBS with 2 mM EDTA and 1 mM DTT, concentrated, frozen in liquid nitrogen, and stored at −80 °C. αB-crystallin was purified as described elsewhere (Peschek et al. 2009).

Reverse transcription and PCR analysis of HLEpiC total RNA isolate

First-strand synthesis of cDNA of HLEpiC total RNA isolate (ScienCell) was accomplished with M-MLV RT (Promega) using oligo dT primers following the manufacturer’s instructions. PCR amplification and sequencing (Eurofins Genomics) were performed with the primers listed in Table 1. Positions are indicated in Fig. 1a. GoTaq G2 polymerase (Promega) was used for PCR analysis.

Table 1 Primers used for nested RT-PCR measurements to verify the presence of different splice variants in HLEpiC total RNA isolate
Full size table
Fig. 1
figure 1

Verification of an alternative splice variant of αA-crystallin in primary human lens epithelial cells. a The structure of the human CRYAA gene with the resulting transcripts and potential protein segmentation. Numbered boxes represent exons and the lines in between introns. Narrow gray boxes represent untranslated regions (UTRs). N-terminal sequence (NTS) (blue, orange, and green), α-crystallin domain (ACD) and C-terminal sequence (CTS) (brown). Green and red triangles indicate start and stop codons. Half headed and numbered arrows indicate primer positions. b Nested PCR analysis of cDNA preparations from HLEpiC cells

Full size image

Far-UV CD spectroscopy

Far-UV circular dichroism (CD) spectra of α-crystallins were recorded at room temperature using a Chirascan-plus CD spectrometer (Applied Photophysics) flushed with dry nitrogen. Spectra were collected from 260 to 195 nm in a 106-QS cuvette (Hellma) with 0.2-mm path length. The measurement time per data point was set to 0.7 s. Proteins were diluted in PBS with 2 mM EDTA and 1 mM DTT to a final concentration of 0.5 mg/mL and incubated at 37 °C for at least 1 h prior to analysis. The depicted spectra are the average of 10 individual scans. Savitsky–Golay smoothing was applied to spectra with a polynomial order of 3 and a smoothing window of 10 points. Spectra were deconvoluted using CDNN 2.1 (Bohm et al. 1992).

Analytical ultracentrifugation

Sedimentation velocity experiments were carried out with a ProteomLab XL-I (Beckman) equipped with absorbance optics. Cells with quartz windows and 12-mm path length were centrifuged at 42,000 rpm and 20 °C using an An-50 Ti rotor (Beckman). αA-crystallin isoform 2 was preincubated in PBS including 1 mM TCEP at 37 °C for at least 1 h prior to analysis. The signal was monitored at 230 nm. Data analysis was carried out in continuous c(s) distribution mode of SEDFIT (Schuck 2000; Schuck 2003). For each concentration, a best fit frictional ratio was calculated and a consistent average value was used for calculation of c(s) distributions. The area under each curve was normalized to 1.

Analytical size exclusion chromatography and precipitation

The apparent molecular mass of the individual α-crystallins and mixtures was measured by size exclusion chromatography. A Superdex 200 10/300 GL (GE Healthcare) analytical gel filtration column was applied at a constant flow rate of 0.5 mL/min at 20 °C. The column was equilibrated with PBS including 2 mM EDTA and 1 mM DTT. Proteins were preincubated at 37 °C for at least 1 h prior to analysis. For measurements of αA-crystallin isoform 2 alone, 5.2 nmol protein was loaded, while for measurements in mixtures with αA-crystallin and αB-crystallin isoform 1, 2.3 nmol of each protein was used. Since αA-crystallin isoform 2 does not contain a tryptophane residue, fluorescence at 303 nm with an excitation at 273 nm was recorded. The gel filtration standard (Bio-Rad) was used for calibration. Proteins eluting from the column were fractionated, and the protein content was precipitated. In a 1:1000 ratio, a 10% sodium deoxycholic acid (NaDOC) solution was added, and the mixture was incubated on ice for 30 min. Trichloroacetic acid (TCA) was added to a final concentration of 10%, and the mixture was further incubated at 4 °C overnight. After centrifugation at 17,000×g, the pellet was resuspended in 2× SDS-PAGE loading dye and neutralized with 1 M Tris solution. Samples were applied to SDS-PAGE followed by western blotting if applicable. Immunodetection was conducted with a polyclonal rabbit antibody raised against αA-crystallin isoform 1 (Pineda Antikörper-Service) in a 1:10,000 dilution and peroxidase coupled anti-rabbit IgG (Sigma) in a 1:20,000 dilution. WesternBright ECL Spray (Advansta) was used as substrate solution for chemiluminescence detection with an ImageQuant LAS 4000 (GE Healthcare). Signal bands were quantified with the 1D gel analysis function of the ImageQuant TL software (GE Healthcare). Where indicated, fluorescence signals and densities of western blot bands were normalized to the highest signal.

Multi-angle light scattering analysis

The molar masses (M) of the proteins were determined using a multi-angle light scattering (MALS) detector and Astra V software (Wyatt Technology). Prior to MALS analysis, protein samples were separated on a Superdex 200 5/150 GL (GE Healthcare) size exclusion column connected to an HPLC (Shimadzu) equipped with UV and RID detectors. PBS containing 2 mM EDTA and 1 mM DTT was used as running buffer, and the flow rate was set to 0.5 mL/min at 20 °C. The signal intensities of depicted differential refractive index peaks were normalized to one. Proteins were preincubated at 37 °C for at least 1 h prior to analysis. Each sample consisted of 30 μL with a total protein concentration of 50 μM. Mixtures of α-crystallins were in equimolar concentrations.

Chaperone activity

Chaperone activity of α-crystallins in terms of suppression of thermal aggregation of model substrates was determined using bovine l-malate dehydrogenase from pig heart (MDH, mitochondrial, Roche) at 42 °C and rhodanese from bovine liver (Sigma) at 45 °C as substrates. The substrate concentration was 2 μM for MDH and 2 μM for rhodanese, while α-crystallins were used in different ratios to substrates. Mixtures of α-crystallins were in equimolar concentrations. Absorbance was measured in at least triplicates at 360 nm in 105.201-QS cuvettes (Hellma) using an Ultrospec 3100 pro UV-Vis spectrometer (GE Healthcare). The assay buffer consisted of PBS with 2 mM EDTA and 1 mM DTT (Mymrikov et al. 2016), and proteins were preincubated at 37 °C for at least 1 h. Absorbance signals were normalized to the aggregation signals of the substrates in the absence of chaperone, setting the maximum A 360 value achieved to 1. All measurements were repeated at least three times, and the mean values and standard deviations (SD) were calculated.

Results and discussion

Alternative splicing of the primary transcript results in differentially spliced variant of αA-crystallin

According to the Ensemble database (Release 86, October 2016), the gene CRYAA is predicted to produce up to five different splice products. One transcript does not contain an open reading frame, and one retains an intron, which renders it unlikely to be further processed and translated. The other three splice products are predicted to be protein coding. All these potential protein-coding products consist of three exons after splicing, whereby the last two exons are identical in all splice products (Fig. 1a), indicating that five exons represent the building blocks to produce the potentially protein-coding splice variants. The first exon, however, is variable, and alternative use of three different versions of this exon might produce three different, protein-coding splice variants. Interestingly, the first exon codes for the complete NTS, respectively, the first 63 amino acids in isoform 1 which generate the common wild-type form of αA-crystallin. Exon 4 and exon 5 (Fig. 1a) of the variants encode roughly the first half of the ACD and the second half of the ACD including the CTS, respectively. Attachment of exon 1 leads to isoform 1 of αA-crystallin and the alternatively attached exons 2 and 3 to shorter isoforms with completely different NTS. The linkage of exon 3 leads to the shortest splice variant referred to as isoform 2 by NCBI (acc. no. NP_001307648; Supplementary Fig. 1), which was already identified in adult human retina and fetal eye tissue on the basis of expressed sequence tag analysis (Bonaldo et al. 1996; Graw 2009; Wistow et al. 2002). This indicates that isoform 2 has an expression profile similar to that of isoform 1. To verify these results, we first tested if the three potential protein-coding splice variants of αA-crystallin are present in the human eye lens.

In cultivated human lens epithelial cells, the expression of αA-crystallin decreases throughout passaging (Fleming et al. 1998; Lenstra et al. 1982). Therefore, we investigated the presence of all splice variants in primary epithelial cells. The total RNA isolate of these HLEpiC cells was reverse transcribed, and nested PCRs with varying primers were conducted (Fig. 1b, Table 1, and Supplementary Fig. 2). Using primers complementary to the UTR before exon 1 (I, II) and after exon 5 (IX; X) resulted in a strong band at ∼600 bp corresponding to the isoform 1 of αA-crystallin, while targeting sequences close to or within exon 2 (III, IV, V) did not show any bands (Fig. 1b and Supplementary Fig. 2). With primers close to exon 3 (VI, VII), several bands developed. The bands were isolated and reamplified with primers binding closer to (VII) or inside of exon 3 (VIII) (Supplementary Fig. 2). Isolation and DNA sequence analysis of these fragments confirmed the band in lane 1 to be equivalent to the fused exons 1, 4, and 5, which corresponds to isoform 1. The upper band at ∼500 bp comprises exons 3, 4, and 5 including the intron between exons 3 and 4. This demonstrates the presence of a splice product, which retains an intron matching the predicted CRYAA-005 (Transcript ID ENST00000468016.1) from the Ensemble database, which additionally includes several premature stop codons disabling synthesis of the respective αA-crystallin isoform on the protein level. On the other hand, the lower band at ∼400 bp originates from the fused exons 3, 4, and 5 representing αA-crystallin isoform 2. Aligning several human sHsps with this variant, we found that there are no highly conserved sites in the NTS of αA-crystallin isoform 2, but secondary structure prediction tools showed that potential α-helices are located in the NTS at positions similar to αA-crystallin and αB-crystallin isoform 1 (Supplementary Fig. 1). In addition, the hydrophobic properties in this segment turned out to be comparable (Supplementary Fig. 3a and b). Taken together, our results substantiate the presence of isoform 2 as an alternative splice variant in human lens epithelial cells, whereas we were not able to find evidence for the existence of CRYAA-003.

The splice variant αA-crystallin isoform 2 is a soluble and structured protein with high similarity to isoform 1

To compare the structural properties of αA-crystallin isoforms 1 and 2, we expressed and purified both variants and conducted far-UV CD measurements of both αA-crystallin isoforms and αB-crystallin isoform 1 (Fig. 2). The CD spectra show that αA-crystallin isoform 1 and isoform 2 represent folded proteins after purification. For both proteins, the spectra indicate a high β-sheet content with small amounts of α-helices which is characteristic for α-crystallins and other sHsps (Haslbeck et al. 2005; Liang et al. 1985). Deconvolution of the spectra revealed that isoform 1 has a slightly lower content of β-sheet (38% isoform 1 vs. 43% isoform 2) but a higher α-helical content (9% isoform 1 vs. 6% isoform 2) (Table 2). This indicates that the stable α-crystallin domain does not seem to be affected by the changes in the NTS. Additionally, the results are in agreement with previous results on NTS truncation mutants of αA-crystallin (Bova et al. 2000; Kundu et al. 2007; Merck et al. 1993) (Table 3).

Fig. 2
figure 2

Far-UV CD sprectra of α-crystallins. The splice variant αA-crystallin isoform 2 (green line) is a soluble and structured protein with high similarity to isoform 1 of αA-crystallin (blue line) and αB-crystallin (red line). Addition of 5 M guanidine hydrochloride to αA-crystallin isoform 2 results in the loss of the secondary structure (green, dashed line)

Full size image
Table 2 Deconvolution of far-UV CD spectra of αA-crystallin isoforms 1 and 2 and αB-crystallin isoform 1 using CDNN
Full size table
Table 3 Calculated apparent molecular mass and number of subunits according to SEC and MALS measurements
Full size table

αA-crystallin isoform 2 forms a monomer–dimer equilibrium

An important general feature of α-crystallins and other sHsps is that they are able to form polydisperse high molecular weight oligomers. In this context, it was proposed that the NTS includes one of the interaction sites necessary for the formation of higher oligomers (Delbecq and Klevit 2013; Haslbeck et al. 2016). Therefore, we addressed the question whether the exchange to an alternative NTS in αA-crystallin isoform 2 affects its ability to form higher oligomers. To monitor the apparent molecular mass of the splice variant, we utilized size exclusion chromatography which revealed that the main peak eluting from the column is indicative for a monomer while a slightly earlier eluting side peak belongs to a dimer (Fig. 3a). Western blot analysis confirmed that the two peaks correspond to αA-crystallin isoform 2 (Fig. 3a and Supplementary Fig. 4), indicating an equilibrium between both states. To validate these results, we applied analytical ultracentrifugation (aUC) sedimentation velocity measurements. Interestingly, at low concentrations, only a monomer peak at 1.5 S was detected, but with increasing concentrations, dimers were observed at 2.3 S, which shows that αA-crystallin isoform 2 is in a monomer–dimer equilibrium. This observed trend of forming smaller oligomers after exchanging the NTS by a shorter, alternative version correlates to previous reports on truncations or deletions of the NTS of αA-crystallin which lead to dimer and tetramer formation in rat αA-crystallin (Merck et al. 1992). Similarly, N-terminally truncated bovine αA-crystallin was identified as a predominant tetramer with minor dimer content (Eifert et al. 2005; Laganowsky et al. 2010; Salerno et al. 2003). In addition, the formation of higher oligomeric species was reported to be concentration dependent also for full-length αA-crystallin and some truncated variants (Laganowsky et al. 2010). Furthermore, removal of more than 20 residues from the NTS drastically reduces the oligomeric size of human αA-crystallin, and complete removal of the NTS resulted in the formation of a tetramer (Kundu et al. 2007). In the case of the splice variant, the further reduction of the complex size might be caused by the significant change in the charge state of the NTS. Assuming that the histidines are uncharged, the NTS of αA-crystallin isoform 1 has a predicted net charge of −2, where isoform 2 has a total predicted charge of its shorter NTS of +4 (Supplementary Fig. 5). The charge variation in combination with the loss of the predicted α-helices, the alternative NTS seems to hinder αA-crystallin isoform 2 to form tetramers out of dimers and thus interferes with the assembly of higher order oligomers.

Fig. 3
figure 3

αA-crystallin isoform 2 forms preferentially monomer and dimeric complexes. a SEC elution profile (black line) is overlaid with the relative density of the respective quantified western blot bands of precipitated fractions (gray bars). b Sedimentation velocity analysis of αA-crystallin isoform 2 at 3 μM (red line), 5 μM (orange line), 7.5 μM (yellow line), 10 μM (green line), 12.5 μM (light blue line), 15 μM (dark blue line), and 20 μM (purple line)

Full size image

αA-crystallin isoform 2 shows decreased chaperone activity

Oligomer formation of sHsps commonly correlates directly with their chaperone activity, and the NTS is thought to be involved in the interaction with the substrate proteins. Hence, we next investigated if the alternative NTS of αA-crystalline isoform 2 modulates its activity. To test this, we measured the influence of αA-crystallin isoform 2 on the thermal denaturation of malate dehydrogenase (MDH) and rhodanese as model substrates. For MDH, no significant inhibition of the aggregation could be observed at substoichiometric ratios of αA-crystallin isoform 2/MDH (Fig. 4a, c). Chaperone activity was only observed when a molar excess of αA-crystallin isoform 2 was used, and complete inhibition was only reached at a tenfold excess (Fig. 4b). For rhodanese, no suppression of the thermal aggregation by αA-crystallin isoform 2 could be achieved. These observations are in accordance with N-terminal and C-terminal truncations of bovine αA-crystallin which were found to show diminished aggregation prevention towards different target proteins (Laganowsky et al. 2010). The same applies for human αA-crystallin isoform 1 after deletion of more than 50 residues (Kundu et al. 2007). The shortened alternative splice variant αA-crystallin isoform 2 seems to follow a similar trend.

Fig. 4
figure 4

αA-crystallin isoform 2 has a weak chaperone activity towards thermally denatured MDH and no chaperone activity towards rhodanese at close to equimolar concentrations. a MDH aggregation assay with αA-crystallin isoform 2 at 0 μM (black line), 0.125 μM (blue line), 0.25 μM (cyan line), 0.375 μM (magenta line), and 0.5 μM (yellow line). b MDH aggregation assay with αA-crystallin isoform 2 at 0, 2, 5, 8, 10, 20, and 40 μM ranging from black to light green. c Rhodanese aggregation assay with αA-crystallin isoform 2 at 0.125 μM (blue line), 0.25 μM (cyan line), 0.5 μM (yellow line), 0.75 μM (brown line), 1 μM (orange line), and 2 μM (gray line)

Full size image

αA-crystallin isoform 2 modulates the high oligomeric complexes of canonical αA-crystallin and αB-crystallin

Human α-crystallins assemble into large polydisperse oligomers that are able to exchange subunits. In a further set of experiments, we investigated the influence of the alternative splice variant of αA-crystallin on the apparent molecular size of αA-crystallin isoform 1 and αB-crystallin oligomers using size exclusion chromatography (Fig. 5a). Because the cellular ratio between the canonical isoform and αA-crystallin isoform 2 cannot be concluded from our experiments and requires further investigations, we always used equimolar concentrations of α-crystallins in mixtures to study the effect of the alternative splice variant. αA-crystallin and αB-crystallin formed large oligomers with an average of ∼24 subunits which is in accordance with the literature (Kundu et al. 2007). Interestingly, the splice variant αA-crystallin isoform 2 was able to form hetero-oligomeric complexes with αA-crystallin isoform 1 and αB-crystallin. The mean oligomeric size of the hetero-oligomeric complexes decreased. Interestingly, the change in the apparent molecular mass cannot be solely explained by the smaller protein size of αA-crystallin isoform 2, since the amount of isoform 2 in the eluting peak is much lower than would be expected by calculation assuming a 24mer (Fig. 5b). This indicates that the mean amount of subunits in the complex is altered as well. To verify that αA-crystallin isoform 2 indeed is incorporated in the high molecular weight complexes of αA-crystallin isoform 1 and αB-crystallin, we analyzed the peaks by SDS-PAGE (Fig. 5b). The majority of αA-crystallin isoform 2 was found to be present in the early eluting large hetero-oligomeric complexes. Interestingly, the modulation of the αB-crystallin oligomer by incorporation of αA-crystallin isoform 2 seems to be less pronounced, but still a distinct shift in the apparent molecular mass was detected. SEC-MALS measurements to determine the apparent molecular mass of the hetero-oligomeric complexes in the presence of the splice variant of αA-crystallin were conducted (Fig. 5c). Again, a decrease in the apparent molecular mass for αA-crystallin isoform 1 and αB-crystallin in case of the hetero-oligomer with the splice variant was detected.

Fig. 5
figure 5

Analysis of oligomer formation of α-crystallins. a Analytical SEC of αA-crystallin isoform 1 (blue line), αB-crystallin isoform 1 (red line), an equimolar mixture of αA-crystallin and αB-crystallin isoform 1 (green line), and the respective mixtures with αA-crystallin isoform 2 in the same colors as dashed lines. b Polyacrylamid gels of NaDOC/TCA precipitated fractions of the SEC runs. c SEC-MALS measurements of αA-crystallin isoform 1 (left panel), αB-crystallin (middle panel), and an equimolar mixture of αA-crystallin and αB-crystallin isoform 1 (right panel) showing the normalized differential refractive index (dRI) solid line and molecular mass (g/mol) as filled squares. Dashed lines and empty squares represent the elution profiles in the presence of equimolar concentrations of αA-crystallin isoform 2

Full size image

The presence of αA-crystallin isoform 2 enhances the chaperone activity of the canonical αA-crystallins and αB-crystallins

Since αA-crystallin isoform 2 is largely devoid of chaperone activity, we were interested in determining the chaperone activity of the respective hetero-oligomers. Both canonical α-crystallins have a strong chaperone activity towards MDH and rhodanese as substrates (Mymrikov et al. 2016) and are by far more active than αA-crystallin isoform 2 (Figs. 4 and 6). At equimolar ratios of sHsp to substrate, αA-crystallin isoform 2 alone showed no protective effect on the aggregation of the substrate. However, when αA-crystallin isoform 1 and αA-crystallin isoform 2 are present together, the aggregation of MDH and rhodanese was suppressed to a higher extent than for the αA-crystallin isoform 1 alone (Fig. 6a, d). This effect is especially pronounced at low concentrations of αA-crystallin and attenuates with increasing concentration of isoform 1. Interestingly, αB-crystallin is influenced in a similar manner by αA-crystallin isoform 2 as αA-crystallin isoform 1 (Fig. 6b, d). Moreover, the splice variant of αA-crystallin is even able to modulate the chaperone activity of the hetero-complex of the canonical αA-crystallins and αB-crystallins and enhances their potential to suppress the aggregation of MDH and rhodanese (Fig. 6c, d). A similar trend was found for other human sHsps like HspB1 and HspB6 (Heirbaut et al. 2016). The large homo-complex forming HspB1 together with the dimer forming HspB6 predominantly generated hetero-oligomeric entities smaller than HspB1 alone, but with a higher apparent molecular weight than HspB6 alone. Additionally, the hetero-oligomers had an increased chaperone activity compared to the components alone (Heirbaut et al. 2016).

Fig. 6
figure 6

Analysis of the modulating influence of αA-crystallin isoform 2 on the chaperone activity of the canonical α-crystallins. The suppression of the aggregation of MDH (a–c) and rhodanese (d) as substrates was analyzed. The spontaneous aggregation of the respective substrate alone is depicted in black. a αA-crystallin isoform 1 alone (solid lines) and in equimolar mixtures with αA-crystallin isoform 2 (dashed lines) at 0.125 μM (blue), 0.375 μM (magenta), and 0.5 μM (yellow). b αB-crystallin isoform 1 alone (solid lines) and in equimolar mixtures with αA-crystallin isoform 2 (dashed lines) at 0.05 μM (red), 0.125 μM (blue), and 0.375 μM (magenta). c Equimolar mixtures of αA-crystallin and αB-crystallin isoform 1 (solid lines) as well as with further, equimolar addition of αA-crystallin isoform 2 at 0.05 μM (red), 0.0625 μM (green), and 0.25 μM (cyan). d Aggreagtion of rodanese alone (black) and in the presence of different canonical α-crystallins (solid lines) as well as with further, equimolar addition of αA-crystallin isoform 2 (dashed lines). αA-crystallin isoform 1 (blue, 0.25 μM), αB-crystallin (red, 0.25 μM), or equimolar mixtures of αA-crystallin and αB-crystallin isoform 1 (green, each 0.25 μM)

Full size image

Conclusions

Taken together, our results verified the presence of an alternatively spliced variant of human αA-crystallin in human lens tissue. This alternative splice variant, annotated as isoform 2 in the NCBI database, has a substituted NTS that is shorter and has a high predicted total positive charge in comparison to the NTS of the canonical isoform 1 but is similar in the composition and distribution of hydrophobic amino acids. αA-crystallin isoform 2 turned out to be a soluble protein with a secondary structure typical for sHsp. According to secondary structure prediction, isoform 2 loses the two initial N-terminal helices but still carries one N-terminal helix at a similar position as the canonical αA-crystallin and αB-crystallin. In contrast to αA-crystallin isoform 1 which preferentially forms large oligomers, the alternative splice variant was found to be in a monomer–dimer equilibrium. On its own, it exhibited only low aggregation prevention activity towards heat-induced aggregation of model substrates, which the canonical crystallins protect with ease, even at low concentrations. Nevertheless, αA-crystallin isoform 2 integrated into the high oligomeric complexes of the canonical α-crystallins and led to a decrease in the subunit number of both αA-crystallin and αB-crystallin isoform 1 oligomers. Interestingly, the hetero-oligomers of the splice variant and the canonical α-crystallins show enhanced chaperone activity. Hence, in addition to post-translational modifications, alternative mRNA splicing yields at least one additional, formerly not characterized α-crystallin species. Furthermore, this additional α-crystallin variant modulates the ensemble of oligomers. Hetero-oligomers of canonical α-crystallins and the splice variant show enhanced chaperone activity. Overall, the observed functional implications of this splice variant are likely to add a further layer of complexity on the regulation of the α-crystallin ensemble in the human eye lens.