Loss of αB-crystallin function in zebrafish reveals critical roles in the development of the lens and stress resistance of the heart

Genetic mutations in the human small heat shock protein αB-crystallin have been implicated in autosomal cataracts and skeletal myopathies, including heart muscle diseases (cardiomyopathy). Although these mutations lead to modulation of their chaperone activity in vitro, the in vivo functions of αB-crystallin in the maintenance of both lens transparency and muscle integrity remain unclear. This lack of information has hindered a mechanistic understanding of these diseases. To better define the functional roles of αB-crystallin, we generated loss-of-function zebrafish mutant lines by utilizing the CRISPR/Cas9 system to specifically disrupt the two αB-crystallin genes, αBa and αBb. We observed lens abnormalities in the mutant lines of both genes, and the penetrance of the lens phenotype was higher in αBa than αBb mutants. This finding is in contrast with the lack of a phenotype previously reported in αB-crystallin knock-out mice and suggests that the elevated chaperone activity of the two zebrafish orthologs is critical for lens development. Besides its key role in the lens, we uncovered another critical role for αB-crystallin in providing stress tolerance to the heart. The αB-crystallin mutants exhibited hypersusceptibility to develop pericardial edema when challenged by crowding stress or exposed to elevated cortisol stress, both of which activate glucocorticoid receptor signaling. Our work illuminates the involvement of αB-crystallin in stress tolerance of the heart presumably through the proteostasis network and reinforces the critical role of the chaperone activity of αB-crystallin in the maintenance of lens transparency.

Although a role of ␣B-crystallin in the proteostasis network is well established and its involvement in the physiology and homeostasis of multiple tissues is supported by direct evidence, a detailed understanding of the molecular mechanisms underlying these roles has yet to emerge. An attempt to generate a knock-out mouse line was confounded by the inadvertent knock-out of HspB2, another sHSP with a presumed role in the development of muscle tissues (34). Although the knock-out mice developed progressive myopathy into adulthood (34), disentangling the contributions of the two sHSPs has proven challenging (35,36).
Here, we capitalize on the intrinsic advantages of the zebrafish as a vertebrate model system to carry out a detailed investigation of the consequences of ␣B-crystallin loss of function on lens and heart developments. We find that although disruption of the two ␣B-crystallin genes leads to lens defects, they are not essential for cardiac development under normal rearing conditions. Rather, a novel role of ␣B-crystallins is uncovered under stress conditions, specifically in response to glucocorticoid receptor (GR) signaling. Zebrafish lines, deficient of either ␣B-crystallin, show the development of cardiac edema when embryos are subjected to crowding conditions or challenged with GR agonists. Together, our results provide novel insights on the physiological roles of vertebrate ␣B-crystallins.

Generation of ␣B-crystallin loss-of-function alleles by CRISPR/ Cas9 system
With the goal of defining the physiological roles of ␣B-crystallin and linking them to its well-understood chaperone mechanism, we have taken advantage of the CRISPR/Cas9 technology (37)(38)(39) to generate zebrafish mutant lines where the two ␣B-crystallin orthologs, cryaba and cryabb, have been disrupted (hereafter, we use the nomenclature "␣Ba" and "␣Bb" for simplicity). We successfully generated multiple alleles carrying various deletions or insertions for each gene (summarized in Table 1; see details under "Experimental procedures") and a few alleles were subsequently propagated after confirmation of the mutations by DNA sequencing (Fig. 1A). For the rest of this study, we focus only on the ␣Ba 35bpDEL and ␣Bb 10bpDEL mutant alleles (Fig. 1B), given that both are predicted to encode extremely truncated polypeptides and are expected to function as null alleles. Henceforth, these deletion mutations, ␣Ba 35bpDEL and ␣Bb 10bpDEL , are referred to as ␣Ba Ϫ/Ϫ and ␣Bb Ϫ/Ϫ .
To confirm the loss of the ␣B proteins in ␣B-crystallin mutant lines, we utilized iTRAQ-based proteomics (40) to quantify the relative changes in the abundance of ␣B proteins in the adult lens of ␣Ba Ϫ/Ϫ and ␣Bb Ϫ/Ϫ . In both water-soluble (WSF) and -insoluble (WIF) fractions, we observed that compared with WT, the relative abundance of native ␣B peptides in each respective mutant was reduced significantly to a marginally-detectable level (Fig. 1D). Additionally, we generated polyclonal antibodies against purified zebrafish ␣Aand ␣Ba-crys-   Figure 1. Generation of knock-out alleles of ␣B-crystallin genes by CRISPR/Cas9. A, alignment of the DNA sequences of the mutant alleles of ␣Ba and ␣Bb. B, electrophoresis of the genomic DNA of the adult zebrafish amplified by PCR: ␣Ba 35bpDEL on 3% agarose gel and ␣Bb 10bpDEL on 4% agarose gel. The genotyping of ␣Bb 10bpDEL allele was confirmed by digestion with restriction enzyme BseLI. C, Western immunoblot of the water-soluble and water in-soluble protein fraction of the excised lenses of the adult zebrafish. After transfer on nitrocellulose membrane, proteins were probed with polyclonal antibody against zebrafish Cryaa (␣A) or zebrafish Cryaba (␣Ba). D, ␣B-crystallin iTRAQ ratios in the lens of both ␣Ba Ϫ/Ϫ and ␣Bb Ϫ/Ϫ adult fishes. The left panel shows iTRAQ ratios of ␣Ba and ␣Bb in mutant animals compared with wild-type animals in the lens WSF, and the right panel shows the same from the WIF.

␣B-crystallin in stress response of zebrafish lens and heart
tallin proteins, and we probed crude protein extracts from the lens of mutant fish lines. Although the level of ␣A-crystallin was not affected by the loss of either ␣Ba or ␣Bb (Fig. 1C, Anti-␣A), we confirmed the loss of ␣Ba-crystallin in the lens of both ␣Ba Ϫ/Ϫ and the double mutant, ␣Ba Ϫ/Ϫ ;␣Bb Ϫ/Ϫ (both WSF and WIF; Fig. 1C). Interestingly, using the ␣A Western blotting signal to normalize expression levels in the ␣Bb Ϫ/Ϫ lens, we observed a slight increase of ␣Ba protein in the WSF of the ␣Bb Ϫ/Ϫ lens, which was in agreement with our iTRAQ data (Fig. 1D), suggesting a possible compensatory response to the loss of ␣Bb protein. Moreover, we observed mobility shifts of the ␣Ba band in the WSF (but not in WIF) of the lens (anti-␣Ba; in WT and ␣Bb Ϫ/Ϫ ) that likely represent post-translational modifications, such as phosphorylation or acetylation (32,41,42). These supershifted bands were particularly noticeable in the ␣Bb Ϫ/Ϫ lens, which might suggest an increase in the amount of post-translational modification of ␣Ba-crystallin. We are currently pursuing the molecular nature of these shifted species and trying to determine whether they are phosphatase-sensitive.

␣B-crystallins are expressed in both the lens and heart during zebrafish embryogenesis
Previously, we showed that both ␣Ba and ␣Bb are expressed in zebrafish embryos from as early as 2 dpf (43). To determine their spatial expression pattern, we performed whole-mount RNA in situ hybridization (WMISH) for ␣Ba and ␣Bb on 2-dpf embryos according to the established protocol (44). As expected, we observed the expression of ␣Ba and ␣Bb in the lens of 2-dpf embryos, as well as in the hearts, albeit with lower staining signal ( Fig. 2A, black arrows). In addition, both genes were broadly expressed in the brain region and showed particularly enriched expression in the otic vesicles ( Fig. 2A, white  arrowheads).
We explored whether a compensatory transcriptional regulation exists between the two paralogs of ␣B genes in the reciprocal mutants, which is often found in duplicated genes (45). However, we observed no significant change (Ͻ1.6 cycle difference (46)) of ␣Bb expression in ␣Ba mutant embryos and vice versa (Fig. 2B). In contrast, we did not observe nonsense-mediated mRNA decay (NMD) for ␣Ba or ␣Bb mRNA transcripts that were derived from their respective mutant alleles, even though they are expected to behave as protein null (Fig. 2B). Nonetheless, as shown in our Western blot analysis, the retained ␣Ba mRNA did not produce a detectable amount of ␣Ba protein (Fig. 1C), which further supported that the ␣Ba mutation is a loss-of-function allele.

Loss-of-function of ␣B-crystallin causes lens defects in zebrafish
Consistent with our previous demonstration that knockdown of ␣Ba and ␣Bb by morpholinos can perturb lens transparency in zebrafish (43), the loss of ␣B-crystallin function, either ␣Ba or ␣Bb, led to apparent abnormalities in the embryonic lens starting at 3 dpf which became evident at 4 dpf (Fig.  3A), while the overall morphology of the embryos remained normal. The nature of these lens defects was similar to those previously described for ␣B-crystallin morphants or ␣A-crystallin knock-out lines (␣A Ϫ/Ϫ ) (43). Following the same classification system previously established, lens defects were categorized into three classes based on the severity: "Major" (Fig.  3A, panels b and c), "Minor" (data not shown), and "WT-like" class" (Fig. 3A, panel a) (43). Initially, we noted that the incidence of lens defects in the offspring derived from heterozygote incrosses, both minor and major classes, occurred in a frequency inconsistent with the lens phenotype being strictly recessive (Ͼ25%). Genotyping of phenotypic embryos revealed that both the minor and major phenotypes occur in heterozygous and homozygous mutants (data not shown). Similar to ␣A Ϫ/Ϫ embryos, both ␣Ba and ␣Bb homozygous mutants (␣Ba Ϫ/Ϫ and ␣Bb Ϫ/Ϫ ) showed lens defects with a range of ␣B-crystallin in stress response of zebrafish lens and heart severity. Greater penetrance and severity of lens defects were observed in ␣Ba Ϫ/Ϫ embryos. For ␣Ba Ϫ/Ϫ , ϳ75% of the total embryos presented lens defects, and 38% were of the major class, whereas ϳ50% of the ␣Bb Ϫ/Ϫ embryos showed lens abnormalities and 25% were of the major class (Fig. 3B). These results suggested that ␣Ba might play a more important role than ␣Bb in the maintenance of zebrafish lens transparency. Interestingly, a gene dosage effect was observed in ␣Ba and ␣Bb genes. When we crossed homozygous ␣Ba mutant adult with WT fish (␣Ba Ϫ/Ϫ ϫ WT), the resulting heterozygous ␣Ba mutant embryos (␣Ba ϩ/Ϫ ) exhibited reduced lens defects (ϳ48%), relative to ␣Ba Ϫ/Ϫ embryos. A similar effect was observed for ␣Bb Ϫ/Ϫ embryos, in which ϳ37% of them showed lens defects (Fig. 3B).
Because both ␣Ba and ␣Bb mutants were adult-viable, we generated ␣Ba/␣Bb double mutants (␣Ba Ϫ/Ϫ ;␣Bb Ϫ/Ϫ ) to determine whether a higher penetrance of the lens phenotypes would be observed by complete elimination of ␣B-crystallin. Compared with the single mutant, particularly ␣Ba Ϫ/Ϫ , the frequency of major lens defects increased moderately in ␣Ba/␣Bb double mutants (by about 10%), but the overall penetrance was not significantly changed, consistent with the conclusion that ␣Ba plays a more prominent role in lens development than ␣Bb (Fig. 3B).
Although both ␣A and ␣Ba appear critical for proper lens formation, the penetrance and severity of the phenotypes vary in the single knock-out lines. Therefore, we generated a double loss-of-function line by crossing ␣Ba Ϫ/Ϫ fish with the ␣A Ϫ/Ϫ line generated in our earlier work (43), and we examined the occurrence of the lens defects. To avoid the protective effect of maternal ␣A expression, ␣A Ϫ/Ϫ ;␣Ba Ϫ/Ϫ embryos were derived from ␣A Ϫ/Ϫ ;␣Bb Ϫ/Ϫ adult incross, and as expected, they exhibited an almost complete penetrance (ϳ99%), with ϳ90% of lens defects belonging to the major class (Fig. 3B).

␣B-crystallin mutants showed stress-induced cardiac phenotypes
In addition to its role in the lens, ␣B-crystallin has been implicated in the cellular integrity of the cardiomyocytes through interactions with Titin and Desmin (11,17,47). Therefore, we examined whether ␣Ba mutants exhibited abnormalities in heart development by incrossing either ␣Ba ϩ/Ϫ or ␣Ba Ϫ/Ϫ . The resulting embryos from both crosses displayed morphologically normal hearts (Fig. 4A, panels a and b) under routine collection and regular embryo-rearing conditions (28.5°C, in 35-mm Petri dish, up to 5-6 dpf). Similar results were observed in ␣Bb mutants (both ␣Bb ϩ/Ϫ and ␣Bb Ϫ/Ϫ ). These results contradict a previous study reporting that knocking down ␣Ba and ␣Bb genes by morpholinos would lead to heart failure and skeletal muscle defects in zebrafish embryos (48). Given the prevalent non-specific effects and the reported unreliability of using morpholino to interfere with gene functions (49), we conclude that ␣B-crystallin genes are largely dispensable for muscle and heart development during embryogenesis. Indeed, ␣B-crystallin double mutant fishes (␣Ba Ϫ/Ϫ ; ␣Bb Ϫ/Ϫ ) are viable and fertile as adults. Raising the progenies derived from an ␣Ba ϩ/Ϫ or ␣Bb ϩ/Ϫ adult incross, we found that the ratio of either ␣Ba Ϫ/Ϫ or ␣Bb Ϫ/Ϫ progeny that survived to adulthood (Ͼ3 months old) was in accordance with Mendelian ratio (Table 3). Moreover, the ␣Ba Ϫ/Ϫ and ␣Bb Ϫ/Ϫ adults did not exhibit lower survival rate compared with siblings under normal housing conditions for at least 10 months (Table 3). These data further suggested that neither ␣B-crystallin gene is essential for zebrafish overall viability.
Although ␣B-crystallin-deficient embryos showed no apparent cardiac phenotype under the normal rearing conditions described above, we examined whether ␣B-crystallin is involved in the stress tolerance of the zebrafish developing heart. As a member of sHSPs, ␣B-crystallin has been shown to be inducible by various cellular stresses, including oxidative, genotoxic, and heat shock stresses (33, 50 -52). Therefore, we raised 1 dpf ␣B-crystallin mutant (␣Ba Ϫ/Ϫ and ␣Bb Ϫ/Ϫ ) embryos in high density (75 embryos per well of a 6-well plate; see "Experimental procedures" for detail) to simulate a "stressed" condition or a non-permissive condition induced by crowding, whereas controls from the same clutch were stocked at normal/non-stressed density (15 embryos per well of a 6-well plate). The gross morphology and heart development of the embryos were monitored daily until 5 dpf. Compared with the control embryos, the ␣B-crystallin mutant embryos subjected to crowding stress progressively developed pericardial edema with moderate penetrance (Fig. 4, A and B; ␣Bb Ϫ/Ϫ shown here but similar phenotypes were observed in ␣Ba Ϫ/Ϫ ), which becomes more evident at 4 dpf (Fig. 4A, panels c and d). In contrast, no such cardiac phenotype was detectable in the control (non-stressed) ␣B-crystallin mutant embryos (Fig. 4A, panels a and b). These cardiac phenotypes were very similar to what has been described previously in embryonic lethal mutants that developed cardiomyopathy, such as titin mutants (53). The stress-induced pericardial edema in ␣B-crystallin mutants was not reversible, as we were unable to raise those ␣Ba Ϫ/Ϫ and ␣Bb Ϫ/Ϫ embryos showing cardiac phenotypes to adulthood (none survived before reaching the juvenile stage; data not shown).
Interestingly, in addition to cardiac phenotypes, the lens defects shown in both ␣Ba Ϫ/Ϫ and ␣Bb Ϫ/Ϫ embryos were also influenced by crowding stress (Fig. 4C). Compared with the non-stressed condition (control), we observed higher penetrance of overall lens defects in ␣Ba Ϫ/Ϫ (ϳ20% increase) and ␣Bb Ϫ/Ϫ (ϳ12% increase) under the crowded rearing conditions.

Sensitivity of ␣B-crystallin mutants to crowding stress is mediated by glucocorticoid levels
Based on the observation that crowding causes phenotypic enhancement in two distinct tissues, we hypothesized that the underlying stress may be mediated by a systemic and intrinsic factor. Crowding stress of adult zebrafish has been shown to increase the body cortisol level (54,55). Indeed, we examined the changes of the expression of known GR signaling downstream genes (fkbp5, pxr, and pepck) (56) in embryos raised under crowding conditions for 2 days, and we found that although all three genes showed trends of increased expression, only pepck was significantly up-regulated (Fig. 4D), which was consistent with the finding that pepck is a more sensitive GR signaling reporter gene during embryonic stages (56).

␣B-crystallin in stress response of zebrafish lens and heart
Earlier reports have shown that the cardiac performance of zebrafish embryos declined when exposed to elevated cortisol levels (57). Therefore, we tested whether the ␣B-crystallin mutants lower the tolerance of the heart to cortisol, which is the main glucocorticoid stress hormone. For this purpose, the ␣B-crystallin mutant embryos were challenged with two different synthetic glucocorticoids, dexamethasone and hydrocortisone, to mimic the elevation of stress hormone level and overactivation of GR signaling. The ␣B-crystallin mutants (␣Ba Ϫ/Ϫ ; ␣Bb Ϫ/Ϫ and ␣Ba Ϫ/Ϫ ;␣Bb Ϫ/Ϫ ), treated with both GR agonists from 1 to 4 dpf, developed pericardial edema and dilated cardiac chambers (Fig. 5, A and C, panels e and f), similar to the phenotypes under the crowding stress (Fig. 4A). In contrast, non-treated ␣B-crystallin mutants (DMSO control) or WT embryos subjected to the same regimen of synthetic glucocorticoids mostly remained normal (Fig. 5B). As expected, the ␣A-crystallin mutants (43) did not exhibit cardiac abnormality when subjected to the crowding stress and dexamethasone treatment (Fig. 6D), which emphasize the distinct involvement of ␣B-crystallin in the maintenance of cardiac function. Together, these results clearly suggested that without the pres-ence of ␣B-crystallin (i.e. loss-of-function), the embryonic heart is more vulnerable to increase in GR signaling activity.

Early cardiac development is unperturbed but cardiac functions appear compromised in ␣B-crystallin mutants when challenged with stress
We examined whether the heart phenotypes were a result of failures of cell fate specification by immunostaining of the ventricle and atrium. Compared with WT, both non-stressed ␣Bb Ϫ/Ϫ embryos (and ␣Ba Ϫ/Ϫ ) at 2 dpf showed largely normal specification and partition of the cardiac chambers (Fig. 5C, panels a and b; data not shown), consistent with the notion that ␣B-crystallin genes are not required for early cardiac development. Because cell death/apoptosis has been implicated in the ␣B-crystallin-associated cardiomyopathy, we performed TUNEL assay on both ␣Bb Ϫ/Ϫ and ␣Ba Ϫ/Ϫ embryos at 2 dpf and found no evidence of increased cell death in the heart, which suggested that apoptosis is not involved in the stressinduced heart phenotypes described above (Fig. 5C, panels c  and d). in ␣B-crystallin mutants (␣Ba Ϫ/Ϫ and ␣Bb Ϫ/Ϫ ) was influenced by the crowding stress. D, relative expression changes of GR signaling target genes (pepck, pxr, and fkbp5) in 3-dpf WT embryos after being raised in crowding conditions for 2 days (from 1 to 3 dpf), as measured by qRT-PCR.

␣B-crystallin in stress response of zebrafish lens and heart
Even though the early heart development appeared to be unaffected in ␣B-crystallin mutants, taking advantage of the embryos' optical transparency, we performed time-lapse imaging to monitor the heart rate and contraction dynamics of ␣Bb mutants, without stress. We found that, compared with the WT, ␣Bb mutants exhibited a slower heart rate (ϳ10% decrease) at 1 dpf and the heart rate of ␣Bb mutants remained lower at 4 dpf (Fig. 5D, upper panel). The ventricular shortening fraction, which can be used as an estimate of myocardial contractility, was not affected at 2 dpf (Fig. 5D (lower panel)). However, at 4 dpf, a small reduction of shortening fraction was observed in ␣Bb mutants, albeit not significant, compared with WT ( Fig. 5D (lower panel)). This marginally affected cardiac performance maybe offset by compensatory mechanisms, given the fact that the ␣B-crystallin mutants survive to adulthood.
In contrast, treatment with dexamethasone significantly reduced (ϳ50%) the ventricular shortening fraction of the ␣Bb mutants when compared with non-treated ␣Bb mutants (Fig. 5D (lower panel)), whereas the heart rate of ␣Bb mutants was not further affected (Fig. 5D). Also, compared with non-treated ␣Bb mutants (Movie S1), the apparent abnormality in cardiac contraction could be observed in dexamethasone-treated ␣Bb mutants (Movie S2). This result strongly highlights the contribution of ␣B-crystallin to the stress tolerance of the heart, in this case, for cortisol stress. Further detailed experiments are needed to examine the molecular/cellular mechanisms underlying the hypersusceptibility of cardiac tissues toward GR signaling activation in ␣B-crystallin mutants.

Phenotypes of ␣B-crystallin mutants result from haploinsufficiency instead of a toxic gain-of-function effect
The lack of NMD and the persistent mRNA (Fig. 2B) described above raised a concern that the observed lens and heart defects in ␣B-crystallin mutants could be due to the dominant (or toxic) effects (i.e. gain-of-function) of residual N-terminal truncated ␣B-crystallin protein possibly resulting from an alternative translation even though such a possible N-terminal truncated (⌬1-43 amino acids) mutant ␣Ba protein should be detectable by our anti-␣Ba antibody (Fig. 6A). To further exclude this possibility, we injected a morpholino specifically targeting a possible alternative initiation site into ␣Ba mutants and examined whether the phenotypes could be suppressed. As shown in Fig. 6, C and D, there was no suppression of lens and heart phenotypes by morpholino knockdown, which argued against the ␣Ba mutant allele functioning as dominant-negative alleles. Finally, lens-specific expression of zebrafish ␣Bb (Tg[cryabb]) also partially alleviated the lens defects of ␣Bb mutant embryos (Fig. 6B), which further supports the loss-offunction and haploinsufficient nature of our ␣B-crystallin mutant alleles.

␣B-crystallin loss-of-function mutants reveal important physiological roles
To our knowledge, this study reports the first ␣B-crystallinspecific loss-of-function lines in vertebrate animals. The previously described ␣B-crystallin knock-out mouse was in fact a compounded mutant line that also contained disruptions in the adjacent Hspb2 gene (34) precluding a direct assessment of an exclusively ␣B-crystallin loss-of-function phenotype. The availability of the zebrafish model yielded novel insights into the role of ␣B-crystallin in the lens and cardiac muscle. Although there has been evidence of transcriptional control by the glucocorticoid receptor (58 -60), our results tie this observation to a prominent role of ␣B-crystallin in the heart resistance to stress.
Unexpectedly, results from qRT-PCR suggested that the mRNAs encoded by the ␣B mutant alleles did not undergo significant NMD despite premature stop codons introduced in the first exons. This phenomenon of escaping NMD has been observed previously in first exon mutations (61,62), which underscores the importance of selecting mutation target sites and confirming the mutations given the explosion of CRISPR/ Cas9 genome-editing techniques (63). Several lines of evidence support our contention that the ␣B alleles described in this study function as null alleles or at least strong hypomorphic alleles, including the lack of stably expressed ␣Ba protein as revealed by Western blot analysis (Fig. 1C) and quantitative proteomics by mass spectrometry (Fig. 1D) of homozygous mutants (␣Ba Ϫ/Ϫ , ␣Bb Ϫ/Ϫ , and ␣Ba Ϫ/Ϫ ;␣Bb Ϫ/Ϫ ) lenses. Although it remains theoretically possible that an N-terminal truncated Cryab protein encoded by alternative downstream initiation sites contributes to the phenotype and functions as antimorphic (or neomorphic) dominant alleles (i.e. gain-offunction), we confidently discount this possibility not only because the N terminus of ␣B-crystallin has been demonstrated critical for the oligomerization and its chaperone activity (23, 64 -66), but also based on the data from the alternative-translation blocking morpholino (Fig. 6, C and D), as well as the rescue of lens defects by lens-specific expression of ␣Bb (Fig.  6B). Moreover, by Western analysis we were able to detect the purified N-terminal truncated mutant ␣Ba protein that expressed in the bacterial system (Fig. 6A), which suggests that in ␣Ba mutants the amount of the alternatively translated ␣Ba protein, if does exist, would be negligible (Fig. 1C).
Unlike the lens in the mouse model (34), proper formation of embryonic lens in zebrafish shows a dosage dependence on ␣B-crystallin. Homozygous mutants of ␣Ba and/or ␣Bb develop lens abnormalities in the majority of the embryos, whereas heterozygous ␣Ba and ␣Bb mutants also showed lens defects albeit to a lesser extent (Fig. 3). Furthermore, consistent with an elevated in vitro chaperone-like activity (67), we found that ␣Ba plays a more critical role in the maintenance of lens

␣B-crystallin in stress response of zebrafish lens and heart
transparency compared with ␣Bb (Fig. 3). Similar to what we observed in ␣A-crystallin knock-out mutants (43), this gene dosage effect (or haploinsufficiency) in ␣B-crystallin mutants may likely reflect the low abundance of overall ␣-crystallin proteins in the zebrafish embryonic lens, where a small change in the expression could cause a strong response in the proteostatic network. Although not frequently reported, cataract formation could be caused by haploinsufficiency, such as mutations in SLC16A12, PITX3, and PAX6 genes (68 -70).
In contrast to previous reports (71-73), we detected lowlevel expression of ␣Ba and ␣Bb in the lens and the heart during zebrafish embryogenesis (Fig. 2) (43, 74). Furthermore, we found no evidence of compromised skeletal muscle functions in both ␣B-crystallin mutants, unlike the severe muscular phenotypes shown in a previous study using morpholino knockdown strategies (48). In addition to the notorious off-target effects of morpholino usage, it is also likely that other members of sHSP genes expressed in the myotomes, such as hspb1, hspb7, hspb8, and hspb9 (73,75), can compensate for the loss of both ␣B-crystallin genes and maintain a functional proteostasis network in the muscle.

Cardiac phenotypes in zebrafish ␣B-crystallin mutant embryos
Given the convincing literature supporting the contribution of ␣B-crystallin to cardiac function, particularly the genetic evidence linking an R120G mutation in ␣B-crystallin (CRYAB R120G) to human desmin-related cardiomyopathy (17) as well as a novel cardiomyopathy-associated CRYAB mutation, D109H (76), the lack of apparent phenotypes in the developing hearts of ␣B-crystallin homozygous mutants, including ␣Ba Ϫ/Ϫ ; ␣Bb Ϫ/Ϫ double mutants, was unexpected. However, the full effect of ␣B-crystallin deficiency may be masked or compensated by several sHSP chaperones that are robustly expressed in the developing heart tubes, such as the hspb1, hspb7, and hspb12 (73,75).
A major finding of this paper is the emergent role of ␣B-crystallin under stress conditions. ␣B-crystallin mutants raised under crowding stress developed severe pericardial edema in a fraction of mutant embryos, even though the penetrance was low compared with the lens phenotypes (about 10 -20%; Fig.  4B). Interestingly, a difference in sensitivity to loss of ␣Ba and ␣Bb was also observed in the heart that is opposite to the lens phenotype (Fig. 3B). This may be a consequence of the higher level of ␣Bb compared with ␣Ba in the cardiomyocytes at 4-dpf embryos (77). Although we did detect expression of both ␣B-crystallin genes in the heart of 2-dpf embryos, the nonquantitative nature of the WMISH procedure precluded us from comparing the relative expression level between ␣Ba and ␣Bb genes. Additionally, differential substrate specificity or different post-translational modification of the two ␣B-crystallin paralogs may contribute to the variation of the gene dosage requirement in these tissues.
Similar to the lens phenotype, the cardiac phenotype does not appear to be due to a failure of specification of the cardiac tissue as both atrial and ventricular markers are normally expressed. In the non-stressed ␣B-crystallin mutants, we did observe a small decline of cardiac function such as slower heart rates (Fig. 5D). Given the known association of ␣B-crystallin with Desmin or Titin (10 -12), subtle changes may exist at the cellular level such as sarcomere integrity, which requires further detailed analyses. Our current hypothesis is that loss of ␣B-crystallin would only slightly weaken the myocardium but not enough to cause severe damages (i.e. become sensitized); however, exposure of excess stress (i.e. GR agonists here) would exacerbate the phenotypes and lead to cardiac dysfunction, which is strongly supported by our observation (Fig. 5D).

Stress-induced phenotypes in ␣B-crystallin mutants uncover a novel intersection between steroid signaling and proteostasis
Following up on the established link between crowding stress and activation of glucocorticoid signaling (54,55), we determined that the ␣B-crystallin mutants were highly sensitive to the treatment of GR signaling agonists, dexamethasone and hydrocortisone, and prone to develop pericardial edema (Fig.  5A). This synergism may be explained by the transcriptional regulation by the glucocorticoid receptor of a particular proteostatic target that leads to phenotype enhancement. Although GR-mediated transcriptional activation of sHSPs, including ␣B-crystallin in cell culture systems, has been demonstrated (58 -60), we did not detect changes in expression of ␣Ba or ␣Bb (by qRT-PCR; data not shown). In fact, the action of GR signaling agonists here is unlikely to be mediated through direct transcriptional activation of the ␣B genes as the ␣B protein levels in homozygote mutants would not be affected by increased ␣B transcripts.
Although the importance of GR signaling in cardiac function and maintenance has been repeatedly demonstrated, the exact roles remain controversial due to its intricate regulations and interactions (78 -82). Here, we provide additional wrinkles to this complex signaling system in which the proteostatic state of the cells (i.e. loss of ␣B-crystallin), in this case the cardiomyocytes, may modulate the GR signaling activities and influence the cellular responses. Thus, further experiments are needed to delineate the interactions between GR signaling and ␣B-crystallin.
It has been well established that prolonged corticosteroid therapy is a major risk factor for the formation of steroid-induced posterior subcapsular cataracts (PSCs), which account for ϳ15% of all cataracts (83)(84)(85)(86)(87). Although the pathoetiology and the mechanism of glucocorticoid action in the lens remain unresolved (88), two main underlying mechanisms have been examined and debated: glucocorticoid-induced gene transcription events and glucocorticoid-lens protein adduct formation (88) Only steroids with glucocorticoid activity are associated with steroid-induced PSC (89), strongly suggesting involvement of the GR (90).
Our result suggests that a compromised proteostatic system in the lens may contribute to the occurrence of steroid cataracts. Indeed, the formation of PSC is quite heterogeneous, and its dose dependence on steroid use is also controversial (91)(92)(93). Thus, we propose that individual susceptibility, which reflects the difference in genetic predisposition (such as polymorphisms in ␣B-crystallin gene) that impacts the lens proteostatic capacity, plays a critical role in the development of PSC. Under this assumption, instead of a "determining factor," the steroid use acts as an "environmental factor" that would facili-

␣B-crystallin in stress response of zebrafish lens and heart
tate the PSC formation via interactions with ␣B-crystallin (or other resident proteostatic factors).
Given that the physiological responses of GR signaling activation are notoriously diverse and often exhibiting profound variability in specificity or sensitivity between different tissues and individuals, partly due to its heterogeneity and complex regulation (94), the novel intersection uncovered in this study between GR signaling and proteostasis in the lens and the heart is likely tissue-or context-dependent. With gene profiling studies revealing a vast array of downstream target genes of GR signaling, further experiments utilizing genetic tools in zebrafish are set up to decipher this new interaction with potentially significant clinical relevance.

Generation of Cryab mutants with CRISPR/Cas9 system
To establish zebrafish mutant lines of cryaba and cryabb, gRNAs targeting the first exons of cryaba and cryabb genes were designed and screened to induce indels by injection with Cas9 mRNA in zebrafish. Several mutant lines were established using four designed gRNAs for each gene (see Table 1). Used gRNA target sequences are presented in Table S1 and the detailed protocol of mutation generation are available upon request. Each Cryab mutant allele was outcrossed to AB at least two generations and then in-crossed to generate homozygous mutants, which were then analyzed for their phenotypes. Every clutch was examined carefully for peculiar phenotypes as we are aware of the possibility of off-target effects. The primers used for screening and genotyping for Cryaba and Cryabb mutant alleles are listed in Tables 2 and 3.

Zebrafish transgenesis
To establish the transgenic zebrafish expressing zebrafish cryabb gene specifically in the lens, transgenic construct of Tg(cryaa:cryabb,myl7:TagRFP) was constructed by MultiSite Gateway (Invitrogen) assembly reactions using protocols established previously (Tol2kit (95)). Specifically, entry vectors, p5E-Cryaa (96) and p3E-poly(A) were assembled with pME-Cryabb and recombined into pDestTol2CR2 vector for microinjections. Tol2-mediated transgenesis was performed as described previously (Zou et al. (43)). At least two founder lines (F0) for each construct were screened and out-crossed to established stable F1 generations. Each F1 line was propagated and crossed with desired mutant lines.

Crowding and drug treatments
For crowding stress, 1-dpf embryos were manually dechorined, and 75 embryos were placed in one well of a 6-well plate (polystyrene, tissue culture grade) with 5 ml of 0.3ϫ Danieau water. For control, 15 embryos per well were used as nonstressed condition. To mimic the effect of glucocorticoid stress, we used pharmacological stimulation by treating embryos with 50 M dexamethasone (Sigma, D1756) and 50 M hydrocortisone (Sigma, H2270) diluted in 0.3ϫ Danieau water (24-well plate, 10 embryos/well) from 1 until 4 dpf to examine the cardiac phenotypes. The optimal dosages of dexamethasone and hydrocortisone were determined after dose-response analysis.

Western analysis
Both lenses of each adult zebrafish (WT and Cryab mutants) were excised, washed, and homogenized in lysis buffer (20 mM Tris-HCl, pH 7.6, 100 mM NaCl, 1 mM NaN 3 , 1 mM EDTA) (4) and supplemented with 1 mM PMSF and COmplete protease mixture (Roche Applied Science). The homogenate was centrifuged at 15,000 ϫ g at 4°C to separate water-soluble and water-

␣B-crystallin in stress response of zebrafish lens and heart
insoluble fractions. Both fractions from each lens were heated in a water bath for 5 min after adding 2ϫ Laemmli Sample Buffer supplemented with DTT and spun at benchtop to clear the solutions. Protein concentrations were measured by RcDc protein assay kit (Bio-Rad). 10 g of total proteins separated by 12% SDS-PAGE were transferred to nitrocellulose membrane in Towbin buffer and probed with custom Anti-␣A and Anti-␣Ba polyclonal antibodies (Vanderbilt Antibody and Protein Resource Core), which were generated by injecting full-length ␣A and ␣Ba proteins, purified from bacterial expression system, as described previously (67), in rabbits. Antisera from two rabbits each, for both proteins, were tested for titer, and after booster shots the antibodies were affinity-purified from the final bleeds using purified proteins as antigens. The blots were visualized by anti-HRP secondary antibody (Promega) using enhanced chemiluminescence (HyGLO TM Quick Spray Chemiluminescent; Denville Scientific Inc.). The cDNA construct encoding the N-terminally truncated ␣Ba protein (⌬1-43 amino acids) was generated by PCR cloning and cloned into pET-20b(ϩ) bacterial expression vector (Novagen).

WMISH
Linearized full-length coding sequence of cryaba and cryabb, with added T7 polymerase site, was PCR-amplified from the cDNA clones (obtained from Mason Posner (Ashland University, Ashland, OH). The digoxygenin-labeled ISH probes were generated by transcribing with T7 RNA polymerase (New England Biolabs) and DIG RNA Labeling Mix (Roche Applied Science). Subsequently, RNA probes were purified after TRIzol precipitation using Direct-zol miniprep kit (Zymo Research) and confirmed for purity and length (507 bp for cryaba and 498 bp for cryabb) by agarose gel electrophoresis. The wholemount in situ hybridization on 2-dpf embryos was performed according to established protocol (44).

Quantitative RT-PCR
Total RNAs were isolated from embryos at desired stages (2and 4-dpf) by using TRIzol reagent (Invitrogen), followed by DNase treatment (Ambion). The iScript cDNA synthesis kit (Bio-Rad) was used for reverse transcriptions, and subsequent qRT-PCRs (SsoAdvanced Universal SYBR Green Supermix, Bio-Rad) were performed on Bio-Rad CFX96 Real-Time PCR Detection System according to manufacturer's protocols (the cycling parameters were as follows: 95°C for 10 min, 40 cycles of 95°C for 15 s, and 60°C for 30 s). The PCR products were analyzed by gel electrophoresis to confirm that they were of the expected size. The CFX Manager software provided with the thermocycler (Bio-Rad) was used to determine Ct values. Expression differences between samples were calculated by the Ϫ⌬⌬Ct (comparative Ct) method and reported without log2 conversion to fold changes. Three pools of 10 embryos from separate clutches of each transgenic line were collected, and each sample was analyzed in triplicate. The primers (including ␤-actin as an internal control) used in this study are listed in Table S4.

Cell death assays
Embryos were fixed overnight at 4°C in 4% paraformaldehyde in PBS, dehydrated with 100% methanol, and stored in Ϫ20°C. The procedures of TUNEL staining were carried out following the manufacturer's suggested protocol (In Situ Cell Death Detection Kit, TMR red; Sigma catalog no. 12156792910).

Heart imaging and cardiac function measurements
Videos taken of the zebrafish hearts in vivo (Zeiss Axio-ZoomV16 microscope) were used to calculate shortening fraction (calculate the shortening fraction (%) for the ventricle as follows: 100 ϫ (width at diastole Ϫ width at systole)/(width at diastole) and for heart rate (count the number of beats in 15 s and then multiply the number of beats counted as beats/min), following the established protocol (98).

Microscopy and image processing
Lenses of live embryos in 0.3ϫ Daneau water with PTU/ tricaine were analyzed by bright field microscopy (Zeiss Axiovert 200) at 4 dpf and graded into three classes depending on the severity of lens defects as defined in our previous study (43). Briefly, the phenotypic features appeared as either round, shiny crystal-like droplets spread across the lens that were classified as minor defects or large irregular protuberances located in the center of the lens classified as major defects. Fluorescence images were taken with Zeiss AxioZoom.V16 microscope.

iTRAQ quantification of protein changes
To quantify protein expression changes in zebrafish lenses, one lens from 6-month-old fish of each genotype (WT, cryaba Ϫ/Ϫ , cryabb Ϫ/Ϫ and cryaba Ϫ/Ϫ ;cryabb Ϫ/Ϫ ) were homogenized in 50 ml of homogenizing buffer (for lens: 25 mM Tris, 150 mM NaCl, 1 mM PMSF, 5 mM EDTA, 10 mM NaF, pH 7.5). The samples were centrifuged at 20,000 ϫ g for 20 min, and pellets were washed with another 50 ml of homogenizing buffer ␣B-crystallin in stress response of zebrafish lens and heart followed by centrifugation at 20,000 ϫ g. The supernatants were pooled together as the WSF. Protein concentrations for WSF fractions were measured by Bradford assay (Thermo Fisher Scientific, Rockford, IL). The pellets were considered as the WIF. WIF was washed with water twice and suspended in 100 ml of water, and an aliquot was mixed with equal volume of 5% SDS for a BCA assay (Thermo Fisher Scientific). 80 mg of WSF from each lens (and 50 mg from each heart) was reduced with 50 mM tris-(2-carboxyethyl)phosphine at 60°C for 1 h, alkylated with 200 mM methyl methanethiosulfonate (MMTS) at room temperature for 10 min, and digested with sequencinggrade trypsin overnight. 20 g of each WSF fraction was then labeled with iTRAQ reagents according to the manufacturer's instructions (AB Sciex, Foster City, CA) (114 for WT, 115 for cryaba Ϫ/Ϫ , 116 for cryabb Ϫ/Ϫ ,and 117 for cryaba Ϫ/Ϫ ;cryabb Ϫ/Ϫ ). For WIF fractions, the entire reconstituted pellet for each WIF sample was digested with trypsin, and 18 g was labeled with iTRAQ reagents. Reagents were reconstituted in ethanol such that each protein sample was iTRAQ-labeled at a final concentration of 90% ethanol, and labeling was performed for 2 h. The iTRAQ-labeled samples were mixed, acidified with TFA, and subsequently desalted by a modified Stage-tip method prior to LC-MS/MS analysis. iTRAQ-labeled samples were analyzed using MudPIT analysis with 13 salt pulse steps (0, 25,50,75,100,150,200,250, 300, and 500 mM and 1 and 2 M ammonium acetate). Peptides were introduced via nano-electrospray into a Q Exactive HF mass spectrometer (Thermo Fisher Scientific, San Jose, CA). The Q Exactive was operated in data-dependent mode acquiring HCD MS/MS scans (R ϭ 15,000) after each MS1 scan (R ϭ 60,000) on the 15 most abundant ions using an MS1 ion target of 3 ϫ 10 6 ions and an MS2 target of 1 ϫ 10 5 ions. The HCD-normalized collision energy was set to 30, dynamic exclusion was set to 30 s, and peptide match and isotope exclusion were enabled. For iTRAQ data analysis, mass spectra were processed using the Spectrum Mill software package (version B.04.00, Agilent Technologies). MS/MS spectra acquired on the same precursor m/z (Ϯ0.01 m/z) within Ϯ1 s in retention time were merged. MS/MS spectra of poor quality, which failed the quality filter by not having a sequence tag length of Ͼ1, were excluded from searching. A minimum matched peak intensity requirement was set to 50%. For peptide identification, MS/MS spectra were searched against a Uniprot zebrafish database (June 21, 2012). Additional search parameters included the following: trypsin enzyme specificity with a maximum of three missed cleavages; Ϯ20 ppm precursor mass tolerance; Ϯ20 ppm (HCD) product mass tolerance; and fixed modifications including MMTS alkylation of cysteines and iTRAQ labeling of lysines and peptide N termini. Oxidation of methionine was allowed as a variable modification. Autovalidation was performed such that peptide assignments to mass spectra were designated as valid following an automated procedure during which score thresholds were optimized separately for each precursor charge state, and the maximum target-decoy-based false-discovery rate was set to 1.0% (99).

Statistics
Differences among groups were analyzed by Student's t test. Data are shown as means Ϯ S.E. Statistical significance was accepted when p Ͻ 0.05.
For statistical analysis of iTRAQ protein ratios, log 2 protein ratios were fit to normal distribution using non-linear (least squares) regression. The mean and standard deviation values derived from the Gaussian fit were used to calculate p values, using Z score statistics. A given log 2 iTRAQ protein ratio (x), with the calculated mean () and standard deviation of the fitted data (), was transformed to a standard normal variable (z ϭ (x Ϫ /). Calculated p values were subsequently corrected for multiple comparisons using the Benjamini-Hochberg method (100).
Author contributions-S. M. and S.-Y. W. contributed to conception of the research, designed and performed the research, analyzed the data, and wrote the paper. A. W. F. performed the research and analyzed the data. Z. W., K. L. R., and K. L. S. performed the research, analyzed the data, and wrote part of the paper. H. S. M. contributed to conception of the research, designed the research and wrote the paper. All authors reviewed the results and approved the final version of the manuscript.