Heat shock proteins (HSPs) were originally discovered based on their up-regulation in response to sudden temperature elevation. HSPs have since been found to be upregulated by a variety of other cellular stressors, including heavy metals, ischemia, alcohol, and ultraviolet light exposure. Homologs of the HSPs that are not upregulated by cellular stressors are properly termed heat shock cognates (HSCs), but are frequently referred to as HSPs and are members of the HSP superfamily. Small heat shock proteins (sHSPs) are a subclass of HSPs characterized by their small molecular size and presence of a conserved domain of approximately 80 amino acid residues known as the α-crystallin domain (Narberhaus,2002). Ten sHSPs exist in human (Fontaine et al.,2003; Kappe et al.,2003), and thirteen have been reported in teleost fish (Franck et al.,2004), including zebrafish (Elicker and Hutson,2007). Several are known by more than one name (Table 1), but all will be referred to here according to the human gene nomenclature.
Table 1. Small Heat Shock Proteins in Vertebratesa
Based on data from Franck et al. (2004) and Elicker and Hutson (2007).
Mammals, teleosts, amphibians, avians
Mammals, teleosts, amphibians, avians
Mammals, teleosts, amphibians, avians
Mammals, teleosts, amphibians, avians
Mammals, teleosts, amphibians, avians (two in zebrafish)
Mammals, teleosts, amphibians, avians
Mammals, teleosts, amphibians, avians
HSP22, E2IG1, H11
Mammals, teleosts, amphibians, avians
Teleosts, amphibians, avians
Like HSPs generally, many sHSPs act as molecular chaperones, preventing aggregation of newly synthesized or denatured proteins (Narberhaus,2002), but some have additional cellular functions. For example, HspB1 (HSP27) can cap or stabilize actin filaments (reviewed by Liang and MacRae,1997), regulates several aspects of cell motility (reviewed by Liang and MacRae,1997; Mounier and Arrigo,2002), regulates intermediate filament assembly (Perng et al.,1999), directly inhibits apoptosis signaling, and regulates cellular redox states (Arrigo,2007); HspB2 binds to myotonic dystrophy protein kinase, activating it and protecting it from heat inactivation (Suzuki et al.,1998); HspB4 (αA-crystallin) and HspB5 (αB-crystallin) are best known for their roles in maintaining the transparency of the lens (Litt et al.,1998; Posner et al.,1999; Berry et al.,2001; Runkle et al.,2002), although HspB5 has some additional activities in common with HspB1, specifically actin capping (Wieske et al.,2001) and stabilization (Wang and Spector,1996) and direct inhibition of apoptosis (Kamradt et al.,2002); HspB6 (Hsp20) may regulate relaxation of vascular smooth muscle (Flynn et al.,2003) or contraction of cardiac myocytes (Pipkin et al.,2003); HspB8 (Hsp22, E2IG1, H11) reportedly has kinase activity (Smith et al.,2000); and HspB10 (ODF1) is a structural component of the outer dense fiber of spermatozoan tail (Burfeind and Hoyer-Fender,1991).
In mammals, most sHSP genes have been reported to be expressed in heart and skeletal muscle (Bhat and Nagineni,1989; Dubin et al.,1989; Gernold et al.,1993; Lutsch et al.,1997; Krief et al.,1999; Benndorf et al.,2001; Kappe et al.,2001), with the exceptions of HSPB4, which is expressed specifically in lens (Posner et al.,1999; Runkle et al.,2002) and HSPB9 and HSPB10, both of which appear to be testis-specific and/or sperm-specific proteins (Kappe et al.,2001; Fontaine et al.,2003; Kappe et al.,2003). Like HSPB4, HSPB5 is expressed in lens, and both HSPB1 (Dubin et al.,1989; Gernold et al.,1993; Plumier et al.,1997) and HSPB5 (Dubin et al.,1989) are expressed in the nervous system. Several of the sHSPs are also expressed in other tissues, including kidney, liver, pancreas, lung, ovary, testis, prostate, skin, and placenta. Several sHSPs have been reported to be expressed during development in many of the same tissues in which they are expressed in adults (Lang et al.,1999; Loones et al.,2000; Tallot et al.,2003; Verschuure et al.,2003; Mao and Shelden,2006; Tuttle et al.,2006). We have previously screened the relative expression levels of all of the sHSPs between the 16-cell stage and 5 days postfertilization using quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) (Elicker and Hutson,2007). However, no systematic analysis of the spatial expression patterns of the entire family in a single organism has been performed to date. We have used whole-mount in situ hybridization to examine the developmental expression patterns of all zebrafish sHSPs and to determine which are upregulated by heat shock at 12, 24, and 48 hours postfertilization (hpf). Many are expressed in tissue-specific patterns, and five are upregulated by heat shock at one or more of these stages.
Only 2 of the 13 sHSPs, hspb1 and hspb8, are detectable by qRT-PCR at approximately 10 somites, or 12 hpf (Elicker and Hutson,2007). In addition, hspb4, while expressed at extremely low levels under normal circumstances, was found to be up-regulated by heat shock at this stage (Elicker and Hutson,2007). We, therefore, used whole-mount in situ hybridization to analyze the spatial expression of these three genes at approximately 10 somites as well as their heat shock-induced upregulation. Our previous work indicated that most of the sHSPs are detectable by qRT-PCR by 48 hpf (Elicker and Hutson,2007). We therefore performed whole-mount in situ hybridization for all 13 sHSPs on embryos at 24 hpf and 48 hpf. In order that the staining intensity for a given expression level be roughly comparable between genes, in situ hybridization for all genes was performed in parallel using probes of similar length and of the same concentration (Table 2). Furthermore, development of the chromogenic substrate was allowed to proceed for the same length of time for all probes for each developmental stage. To minimize background, we used extremely low concentrations of probe, and to demonstrate the absence of background using these parameters, we hybridized at least one probe for a gene that was not detected by qRT-PCR for each stage examined. The results are summarized below and in Table 3.
At 10 somites, hspb1 was found to be expressed throughout the embryo, including the yolk syncytial layer (Fig. 1A), consistent with a previous report (Thisse et al.,2001). hspb1 staining in heat shocked embryos was also slightly stronger than that of non–heat shocked embryos when these were examined early in the development reaction, suggesting transcriptional upregulation. These results are consistent with our previous findings (Elicker and Hutson,2007).
At 24 hpf, hspb1 was expressed in somites, mid–hindbrain boundary, heart, lens, in small patches in the dorsal and ventral forebrain, and in the yolk syncytial layer, particularly in the region of the yolk extension (Fig. 2A). Expression of this gene has been characterized previously, and the patches of expression in the head presumably correspond to differentiating jaw and ocular muscles, which express hspb1 at very high levels at later stages (Thisse et al.,2001). In addition, a prior study showed that it is expressed throughout the somite at this stage (Mao and Shelden,2006). hspb1 was strongly and ubiquitously upregulated by heat shock at 24 hpf (Fig. 2B), consistent with previous reports (Mao and Shelden,2006; Elicker and Hutson,2007).
By 48 hpf, hspb1 expression was substantially diminished in the somites and yolk syncytial layer, but expression was maintained in the mid–hindbrain boundary, heart, lens, head, and was now seen in caudal spinal cord, trunk, and pharyngeal arches (Fig. 3A). Expression in the head and trunk (Fig. 4A) corresponds to ocular, jaw, hypaxial, pectoral fin muscles (Thisse et al.,2001). It appeared to be expressed specifically in the ventricle of the heart and is concentrated in the rostral ventricle/outflow tract and the vicinity of the atrioventricular constriction (Fig. 4A) similar to the expression pattern of bmp2 at this stage (Milan et al.,2006), although hspb1 is expressed somewhat more broadly. Its expression in somites correlates roughly with the location of slow-twitch muscle fibers, which lie at the lateralmost extent of the muscle, superficial to the fast-twitch fibers (Fig. 5A). hspb1 was also expressed in notochord, although at extremely low levels compared with hspb8, hspb9, and hspb11 (Fig. 5). hspb1 was broadly upregulated by heat shock at this stage, although it appeared to be somewhat less so at this stage than at 24 hpf, consistent with earlier findings (Elicker and Hutson,2007).
hspb2 and hspb3
At 24 and 48 hpf, both hspb2 and hspb3 were expressed ubiquitously at levels so low as to be indistinguishable from background (data not shown). These results were not surprising, as the results of qRT-PCR indicate that they are expressed at extremely low levels at 24 and 48 hpf (Elicker and Hutson,2007). Neither gene was up-regulated by heat shock at either stage (data not shown), also consistent with the results of RT-PCR.
hspb4 was found to be expressed ubiquitously at 10 somites (Fig. 1B) and did not appear to be upregulated by heat shock (data not shown). This finding is not consistent with our previous finding, where hspb4 was not detected normally but was detectable after heat shock (Elicker and Hutson,2007). Our inability to see a difference in expression between unstressed and heat shocked embryos using in situ hybridization is likely due to variability in the heat shock response combined with the difficulty of seeing subtle differences in expression levels using this technique. To rule out the possibility that the apparent expression we observed here was due to background staining, we performed in situ hybridization in parallel using a probe for hspb5a, which we know to be absent at this stage. As the hspb5a embryos are completely unstained (Fig. 1C), we believe that the hspb4 staining is likely to be real.
At 24 hpf, hspb4 was expressed at low levels specifically in the lens (Fig. 2C). In embryos just a few hours younger (∼22 hpf), no expression was detected whatsoever (data not shown), suggesting that its onset of expression is just before 24 hpf. Of interest, its lens expression at this stage increased in response to heat shock (Fig. 2D). Although it has been reported to be selectively transcriptionally upregulated by Cu2+ in cultured lens epithelial cells (Hawse et al.,2003), hspb4 has not been reported to be upregulated by heat shock in any other system. However, this finding is consistent with the results of qRT-PCR, where hspb4 expression was found to increase approximately eight-fold in response to heat shock (Elicker and Hutson,2007).
By 48 hpf, hspb4 was expressed at high levels in the lens (Fig. 3C), consistent with previous reports (Runkle et al.,2002). After heat shock hspb4 continued to be expressed specifically in lens (Fig. 3D). From direct examination of embryos at earlier stages of the development reaction, it appeared that hspb4 was expressed at higher levels in response to heat shock at this stage – again, a somewhat surprising result, but one that is consistent with our previous findings (Elicker and Hutson,2007).
hspb5a and hspb5b (αB-crystallin)
We were unable to detect expression of either hspb5a or hspb5b at any stage examined (Fig. 1C, and data not shown), and neither gene was upregulated by heat shock at either stage (data not shown). This finding is consistent with our previous findings that, by qRT-PCR, both genes are either undetectable or barely detectable during this period (Elicker and Hutson,2007) and with previous report that hspb5a is not expressed during development (Posner et al.,1999). This finding is interesting in light of the fact that hspb5 is one of few sHSPs that is upregulated by heat shock in other systems. However, the heat shock-responsiveness of genes can change during development (Davis and King,1989; Heikkila,2004), possibly due to epigenetic effects or changes in the cellular composition of signaling molecules.
We were unable to detect normal or heat shock-induced expression of hspb6 at 24 hpf (data not shown), consistent with our previous finding that it is expressed at extremely low levels at this time (Elicker and Hutson,2007). At 48 hpf, hspb6 was expressed ubiquitously at low levels and did not appear to be upregulated by heat shock (data not shown).
hspb7 was expressed in heart and at low levels in the somites and notochord at 24 hpf (Fig. 2E). The expression in heart appeared to be down-regulated somewhat by heat shock at this stage (Fig. 2F), although the effect was quite variable, occurring in approximately 50% of embryos. As this variation was observed within individual batches of embryos, it suggests that the heat shock response may vary somewhat between embryos. While unusual, this result is consistent with a model in which mRNAs that fail to splice owing to heat shock (Yost and Lindquist,1988) are actively degraded through nonsense-mediated decay.
At 48 hpf, hspb7 continued to be expressed in heart, was expressed at extremely low levels in notochord, and was absent in somites (Fig. 3E). Like hspb1, hspb7 was restricted to the ventricle of the heart, rostral ventricle/outflow tract, and the vicinity of the atrioventricular canal (Fig. 5B). hspb7 expression remained essentially unchanged with heat shock at this stage (Fig. 3F), consistent with the results of qRT-PCR (Elicker and Hutson,2007).
At 10 somites, hspb8 was expressed in the notochord and in a patch in the ventral portion of the mid–hindbrain boundary. It was also expressed at low levels in the yolk syncytial layer (Fig. 1D).
By 24 hpf, hspb8 was no longer expressed in the mid–hindbrain boundary, and its expression was restricted to the somites and the yolk syncytial layer in the yolk extension, and it was expressed at very low levels in the notochord (Fig. 2G). In response to heat shock, it was upregulated throughout the embryo, although the extent of up-regulation depended on the tissue. It was most strongly upregulated in the eye, mid–hindbrain boundary, and otic vesicle (Fig. 2H).
At 48 hpf, hspb8 continued to be expressed broadly in the somites and was now expressed in specific structures in the head (Fig. 3G). Based on the striking similarity between its expression and that of hspb1 at this stage (Fig. 4C), the head expression likely corresponds to ocular and jaw muscles, although it did not appear to be expressed in the medial rectus. We also found that it was expressed in neurons in the lateral spinal cord, in the notochord, and in paired structures on either side of the dorsal aorta in what is likely the pronephric ducts (Fig. 5B). At 48 hpf, hspb8 was up-regulated by heat shock in all of its normal expression domains and was also upregulated strongly in the epiphysis, heart, ventral hindbrain, and somites, and weakly throughout the rest of the embryo (Fig. 3H). These findings are consistent with the results of qRT-PCR (Elicker and Hutson,2007).
At 24 hpf, hspb9 was expressed in specifically in the somites at slightly lower levels than hspb8 (Fig. 2I), and it was selectively upregulated by heat shock, again, specifically in the somites (Fig. 2J). At 48 hpf, hspb9 continued to be expressed in the somites, although its expression was reduced relative to 24 hpf, and expression was somewhat patchy, particularly in the caudal somites (Fig. 3I). While it appeared to be expressed throughout the somite, it was generally most highly expressed in the dorsal half of each (Fig. 5C). It was also expressed in notochord at this stage (Fig. 5C). hspb9 was also upregulated by heat shock at 48 hpf, again mostly in the somites but also in the dorsal mid- and hindbrains as well as throughout the rest of the embryo (Fig. 3J). The level of expression in the dorsal mid- and hindbrains after heat shock varied between embryos, showing anywhere from modest expression levels (example in Fig. 3J) to barely above background. Like the variation seen with hspb7 staining, this difference was observed within batches, suggesting that the amplitude of the heat shock response is very sensitive to even slight differences in ages of the embryos or other experimental variation. These findings are consistent with the results of qRT-PCR (Elicker and Hutson,2007).
Like hspb9, hspb11 was expressed specifically in somites at 24 hpf (Fig. 2K). However, it appeared to be expressed at even lower levels than hspb9 and in only the rostral-most somites. Also like hspb9, hspb11 was upregulated by heat shock only in somites (Fig. 2L).
At 48 hpf hspb11 continued to be expressed in the rostral-most somites and was also seen in the notochord at this stage (Fig. 3K). The somite expression was restricted to the vicinity of the horizontal myoseptum (Fig. 5D). At 48 hpf, hspb11 was upregulated in response to heat shock throughout the somites, in heart, and in the dorsal mid- and hindbrains (Fig. 3L). However, the extent of regulation in the mid- and hindbrains varied with approximately 30% of embryos exhibiting no upregulation. These findings are consistent with the results of qRT-PCR (Elicker and Hutson,2007).
At 24 hpf, hspb12 was expressed specifically in heart (Fig. 2M). In slightly younger embryos (∼22 hpf), hspb12 was similarly expressed but at substantially higher levels (data not shown), suggesting that its peak of expression may occur before 24 hpf. Like hspb7, its expression at 24 hpf appeared to decrease somewhat in response to heat shock (Fig. 2N), consistent with results of qRT-PCR (Elicker and Hutson,2007).
At 48 hpf hspb12 continued to be expressed in the heart (Fig. 3M). As with hspb1 and hspb7, hspb12 expression was restricted to the ventricle and was found to be concentrated primarily in the rostral ventricle/outflow tract and the vicinity of the atrioventricular canal (Fig. 5D). Its expression did not appear to change in response to heat shock at this stage (Fig. 3N), consistent with the results of qRT-PCR (Elicker and Hutson,2007).
This is the first systematic description of the developmental and heat shock-induced spatial expression patterns of the entire sHSP family in any organism. Many of the 13 zebrafish sHSPs are expressed in tissue-specific patterns, suggesting novel roles in development and/or protection from stressors. hspb1, hspb8, hspb9, and hspb11 are expressed in many of the same tissues (most notably the somites and, to some extent, notochord), suggesting the possibility that these genes may have similar evolutionary origins. Their expression patterns also suggest that these four genes may function redundantly in the development of axial muscle, cartilage, and or bone. Given that hspb1 is involved in actin dynamics (Liang and MacRae,1997; Mounier and Arrigo,2002) and protecting actin filaments from the catastrophic effects of heat shock (Lavoie et al.,1993), it is reasonable to consider the possibility that these sHSPs are important for assembly or stabilization of the skeletal muscle sarcomere specifically through regulation of actin dynamics and/or stability. Alternatively, given the recent evidence that hspb1 may protect cells from the deleterious effects of oxidation (Arrigo,2007), their expression in developing skeletal muscle (as well as heart) may reflect these metabolically active tissues' need for extra protection from oxidative stress. It is also interesting to note that, while heat shock disrupts segmentation of the somites (Kimmel et al.,1988; Roy et al.,1999), only hspb1 and hspb4 are expressed in somites during segmentation, with hspb4 being expressed at quite low levels. This finding suggests that hspb1 expression alone is not sufficient to protect the process of somitogenesis from the effects of heat shock. However, it does not rule out the possibility that, without hspb1, the effects of heat shock could be even more dramatic.
Mutations in human HSPB1 (HSP27) and HSPB8 (HSP22) cause Charcot-Marie-Tooth disease and Distal Hereditary Motor Neuropathy (Evgrafov et al.,2004; Irobi et al.,2004; Kijima et al.,2005; Tang et al.,2005a,b), two closely-related diseases in which the axons of motor neurons are affected. Therefore, we were particularly interested in the expression patterns of hspb1 and hspb8. hspb1 is expressed throughout the embryo at 12 hpf, hspb8 is expressed in spinal cord neurons at 48 hpf, and hspb1 and hspb8 are upregulated by heat shock in the central nervous system at 24 and 48 hpf. However, neither gene is expressed at particularly high levels in motor neurons, the neurons affected in these two diseases, in unstressed embryos at any stage. Therefore, the diseases are caused either by the mutations acting indirectly through another cell type (somite-derived muscle comes to mind), disruption of the stress response, or disruption of a later function of each gene.
Expression of sHSPs in the heart may also be clinically important, as cardiovascular malformations are the most common type of birth defect. Two of the sHSPs, hspb7 and hspb12, are expressed only in the heart, primarily the ventricle, the atrioventricular canal, and the vicinity of the presumptive outflow tract. In the case of hspb7, this result is not surprising, as it was originally identified in humans as a heart-specific transcript (Krief et al.,1999). The expression of hspb7, hspb12, as well as hspb1, in the unstressed atrioventricular canal and outflow tract suggest that these sHSPs may be involved in the morphogenesis of these structures. On the other hand, endocardial cushion and subsequent valve formation depend on normal myocardial contraction (Bartman et al.,2004). The force imparted by this contraction on the narrow walls of the atrioventricular canal and outflow tract may result in increased pressure, and possibly shear stress, on the myocardial wall at these points. The sHSPs may be important for maintaining tissue integrity in the face of this stress.
Not surprisingly, hspb4 is expressed specifically in lens by 24 hpf, reflecting its key structural role in lens and maintenance of clarity (Litt et al.,1998). While hspb5a and hspb5b are known to be expressed in adult lens (Posner et al.,1999; Smith et al.,2006), and mutations in human HSPB5, like mutations in HSPB4, lead to cataract (Berry et al.,2001), neither hspb5a nor hspb5b is expressed at the stages examined, highlighting the uniquely important role of hspb4 in early lens development.
In summary, several of the sHSPs are known to be important for disease prevention in humans. However, every disease known to be caused by mutations in sHSPs are diseases of the mature organism. Here, we find that many of the sHSPs have compelling spatial expression patterns during development, suggesting roles in tissue morphogenesis and protection from the effects of stressors during embryogenesis. The overlapping expression patterns of several of these genes indicate the likelihood of substantial functional redundancy within the family. Combined, these features of their expression indicate that sHSPs likely have, or have had, key roles in vertebrate development and evolution.
Wild-type fish were maintained and bred using standard procedures and in accordance with Williams College animal welfare protocol (USDA assurance certificate A3133-01). Embryos were staged according to Kimmel et al. (1995) as equivalent hours postfertilization (hpf) at 28.5°C.
A plasmid containing the complete coding sequence of hspb1 was obtained from the laboratory of Igor Dawid; hspb2 and hspb3 expressed sequence tags (ESTs) were obtained from Jinrong Peng of the Institute of Molecular and Cell Biology, Singapore; plasmids containing hspb4 and hspb5a were obtained from Mason Posner of Ashland University, Ashland, Ohio; hspb5b, hspb6, hspb9, and hspb11 ESTs were obtained from Open Biosystems; and hspb7, hspb8, hspb12, and hspb15 ESTs were obtained from the American Type Culture Collection. All plasmids and ESTs were sequenced for confirmation. hspb5b, hspb7, hspb8, hspb9, hspb12, and hspb15 subclones were created by PCR subcloning into pCRII. All subclones were sequenced completely in both directions for confirmation. Parameters for probe synthesis are found in Table 2.
In Situ Hybridization
Embryos from crosses of wild-type mating pairs were raised at 28.5°C for 22 or 46 hr. Each clutch was then divided in half, and half was heat shocked for 1 hr at 37°C and allowed to recover for 1 hr at 28.5°C. Both heat shocked and non–heat shocked embryos were then manually dechorionated, anesthetized with 0.016% tricaine, and fixed for 2 hr at room temperature in 4% fresh formaldehyde (Ted Pella) in PBS. Whole-mount in situ hybridization was performed on batches of embryos according to a publicly available protocol (http://zfin.org/ZFIN/Methods/ThisseProtocol.html) with the following exceptions: 24 hpf embryos were Proteinase K-treated for 5 min, 48 hpf embryos for 25 min; probe concentration was 200 ng/ml for all hybridization; and BM Purple (Roche) was used for color development. Photomicrographs of embryos were taken using an Insight firewire digital camera mounted on a Nikon SMZ1500 stereomicroscope. Images were cropped and resized in Adobe Photoshop. Light levels were adjusted slightly in cases where it was essential to improve visualization of stained structures.
We thank Maria Recco for expert technical assistance; Jonathan Gillig for fish care; and Drs. Jinrong Peng, Mason Posner, and Igor Dawid for their generous contributions of clones.