The posterior lateral line (pLL) of zebrafish is composed of clusters of sensory organs known as neuromasts that contain several mechanosensory hair cells surrounded by populations of internal and peripheral supporting cells. The pLL is formed during embryonic development by the work of the migrating pLL primordium (pLLp), which deposits protoneuromasts at regular intervals along the horizontal myoseptum beginning at about 22 hr post-fertilization (hpf; Ghysen and Dambly-Chaudière, 2007). The leading zone of the pLLp is composed primarily of undifferentiated progenitors that give rise to cells that organize into rosettes in the trailing zone. These rosettes are deposited as protoneuromasts, and the cells therein differentiate into the hair cells and supporting cells of the neuromast (Laguerre et al., 2005; Nechiporuk and Raible, 2008; Ma and Raible, 2009; Chitnis et al., 2011). This small population of progenitors in the leading zone maintains an elevated level of proliferation in order to generate new cells in the trailing zone following each deposition of a protoneuromast (reviewed in Ma and Raible, 2009).
Several studies suggest that the Wnt/β-catenin signaling pathway is critically involved in maintaining the elevated levels of proliferation in the undifferentiated progenitors of the pLLp leading zone. In the absence of Wnt signal, the main effector of the canonical Wnt pathway, β-catenin, is phosphorylated by a complex containing glycogen synthase-kinase 3β (GSK3β), axin, and APC and is subsequently marked for ubiquitin-mediated destruction (reviewed in Nusse, 2008). In the presence of Wnt, however, the recruitment of axin to the cell membrane prevents the formation of the GSK3β phosphorylation complex and allows β-catenin to translocate into the nucleus and bind to transcription factors in the TCF/LEF family, which promotes the transcription of various genes critical to cell-cycle progression and re-entry such as cyclin D1 and cMyc (reviewed in Nusse, 2008). This pathway has been implicated in regulating proliferation during PLLP migration given that the expression of the downstream transcription factor lef1 is restricted to the leading zone of the pLLp (Aman and Piotrowski, 2008). Additionally, inactivation of the Wnt pathway with lef1 morpholino knock-down or genetic knock-out causes a significant reduction in the number of proliferating cells in the leading zone of the primordium and failure to deposit posterior neuromasts (Gamba et al., 2010; McGraw et al., 2011; Valdivia et al., 2011). In contrast, when Wnt signaling is held constitutively active in mutant fish that express a non-functional form of APC, a protein essential to the formation of the β-catenin phosphorylation/degradation complex, proliferation is dramatically increased not only in the leading zone but also throughout the entire migrating pLLp (Aman et al., 2011). While the role of these signals in controlling migration of the primordium and deposition in neuromasts has been well studied, there has been little work examining the signaling that mediates the proliferation and differentiation of the hair cell and supporting cell progenitors within the neuromast following deposition.
Canonical Wnt/β-catenin signaling has been shown to play a central role in regulating the proliferation of retinal progenitors in zebrafish during development, on-going growth, and retinal regeneration following injury (Yamaguchi et al., 2005; Stephens et al., 2010; Ramachandran et al., 2011; Meyers et al., 2012). In particular, pharmacological activation of Wnt signaling biases retinal progenitors to remain in a proliferative state at the expense of differentiated progeny in the initial retinal progenitors, the retinal stem cells that reside in the ciliary marginal zone, and in the Müller glia that dedifferentiate and proliferate to regenerate lost photoreceptors (Yamaguchi et al., 2005; Ramachandran et al., 2011; Meyers et al., 2012).
In lateral line neuromasts, the internal and peripheral (mantle) supporting cells surrounding the sensory hair cells remain proliferative throughout life and serve as the progenitor population, both as a source for new cells within the neuromast and as a source for the budding off of new neuromasts (Ledent, 2002; Grant et al., 2005; Ghysen and Dambly-Chaudière, 2007; Nuñez et al., 2009; Wada et al., 2010). In response to loss of the sensory hair cells, the mantle cells and supporting cells re-enter the cell cycle and quickly replace the lost cells via regenerative proliferation (Jones and Corwin, 1993; Harris et al., 2003; Hernández et al., 2007; Ma et al., 2008). Thus, like the retina, the lateral line neuromasts have initial progenitors that produce all of the differentiated cell types, cells that remain proliferative for ongoing growth of the fish, and cells that can be stimulated to return to the cell cycle in response to injury. Given the importance of Wnt signaling in control of proliferation at the leading edge of the pLLp, and its role in retinal development, growth, and regeneration, we hypothesize that Wnt signaling may play similar key roles in controlling the proliferation of lateral line progenitors in neuromasts following their deposition.
In order to test the role of Wnt signaling on proliferation of lateral line neuromasts post deposition, we began by stimulating Wnt/β-catenin signaling with the pharmacological GSK3β-inhibitor 1-azakenpaullone (Az). This leads to an upregulation of Wnt/β-catenin target genes such as lef1 throughout the primordium and in the protoneuromasts. Neuromasts in fish treated with Az during or immediately after neuromast deposition are significantly larger than those of controls, and we show that this is due to both increased size of the protoneuromasts being deposited and proliferative growth in the neuromast following deposition. This increase in proliferation results in a significant increase in the number of hair cells that differentiate within Az-treated neuromasts compared with controls. We next asked whether similar increases in proliferation would be seen if Wnt signaling was activated in mature, largely quiescent neuromasts. Again, stimulation of Wnt signaling with Az triggers a significant increase in proliferation of supporting cells. Similarly, we found that Az stimulates a significant increase in proliferation of supporting cells following aminoglycoside lesioning of the hair cells. Finally, we show that inhibition of Wnt signaling by overexpression of dkk1b suppresses proliferation during both development and regeneration. Together these data suggest that Wnt/β-catenin signaling plays a central role in controlling supporting cell proliferation during initial development, ongoing growth, and regeneration.
1-Azakenpaullone Increases Expression of the Wnt Target dkk1b
Constitutive activation of Wnt/β-catenin signaling in mutant zebrafish expressing a non-functional form of APC has been shown to induce a dramatic increase in the level of proliferation in both leading-zone progenitors and throughout migrating primordia (Aman and Piotrowski, 2008; Aman et al., 2011). However, the effects of Wnt/β-catenin activation on proliferation in neuromasts following their deposition by the primordium have yet to be determined. In order to temporally control the activation of Wnt signaling, we use the glycogen synthase kinase 3β (GSK3β) inhibitor 1-azakenpaullone (Az), which prevents the phosphorylation and degradation of β-catenin (Kunick et al., 2004), and which we have shown is able to exogenously activate Wnt signaling in the retina of zebrafish embryos and larvae (Meyers et al., 2012). To confirm that Az was leading to activation of the canonical Wnt pathway in lateral line neuromasts, we treated zebrafish embryos with 2.5 μM Az at 24 hpf, fixed them at 36 hpf or 48 hpf, and measured the expression of dkk1b and lef1, which are downstream Wnt targets (Niida et al., 2004; Niehrs, 2006; Gamba et al., 2010; Stephens et al., 2010; McGraw et al., 2011; Valdivia et al., 2011), via in situ hybridization. In DMSO-treated controls at 48 hpf, dkk1b expression was restricted to a narrow middle domain of the lateral line primordium, which had migrated to nearly the tail of the fish (Fig. 1A, arrow). Consistent with prior reports, in embryos treated with Az, primordium migration is slowed and stalls out midway along the trunk (Fig. 1B, arrow). However, in these fish dkk1b expression was strong throughout the primordium, in the rosettes about to be deposited, and was also found in the neuromasts that have already been deposited (arrowheads in Fig. 1B,C). lef1 expression at 36 hpf is restricted to the leading edge of the primordium in DMSO-treated controls (Fig. 1D), but in fish treated with Az, lef1 is found throughout the primordium and in the protoneuromast about to be deposited (Fig. 1E), This suggests that in response to Az treatment, the cells of the developing lateral line upregulate dkk1b and lef1, as expected for hyperactivation of Wnt signaling. Further, this suggests that Az can affect cells both prior to and following deposition from the primordium.
1-Azakenpaullone Stimulates Proliferative Growth of Neuromasts Following Their Deposition
We next examined the effect of treatment with Az on lateral line development, using the Tg(cldnb:gfp) line (Haas and Gilmour, 2006) to visualize the primordium and deposited neuromasts. Similar to what has been reported in Wnt-activating apcmcr mutants (Aman and Piotrowski, 2008; Aman et al., 2011), treatment with Az limited how far the primordium migrated but had little effect on the number of neuromasts deposited. The primordium migrates almost to the tip of the tail within 48 hpf in DMSO-treated controls, depositing 7.8 ± 0.4 neuromasts along the length (N = 7; Fig. 2A,E,F). In contrast, in fish treated with Az at 24 hpf, the primordium migrated less than half the normal distance, stalling out partway through the yolk extension (Fig. 2B,E), but depositing 7.0 ± 0.4 neuromasts within that distance (N = 7; Fig. 2B,F). Delaying the onset of Az treatment increased the distance the primordium migrated, but had little effect on the number of deposited neuromasts (Fig. 2C,D). The differences in distance traveled were statistically significant in all pairwise comparisons (P < 0.05; ANOVA with Tukey post-hoc analysis). This confirms that treatment with Az phenocopies the effects of activating Wnt mutations, leading to the expected changes in migration of the lateral line primordium. It also shows that pharmacological manipulation can be used to control the specific onset of the induction of Wnt signaling in the lateral line.
In addition to the distance the primordium migrated, we noted that neuromasts in all of the Az-treated embryos appeared much larger than those in DMSO-treated controls (Fig. 2A–D, insets). To examine this more specifically, we treated cldnb:gfp zebrafish with either 0.025% DMSO or 2.5 μM Az beginning at 24 hpf and fixed the fish at 72 hpf. When we measured the area of the second neuromast deposited in 7 fish, Az-treated neuromasts were significantly larger at 72 hpf than the DMSO-treated controls (Fig. 3A–C; 1,936.9 ± 81.8 μm2 vs. 830.7 ± 37.9 μm2; P < 0.001; t-test). To test whether this increase was due to changes in cell shape/size or cell number, we quantified the number of cells in the widest cross-section of each neuromast. DMSO-treated controls had 42.3 ± 1.8 cells in a single cross-section through the neuromast while Az-treated fish had 107.3 ± 7.0 cells (Fig. 3A',B',C; P < 0.0001; t-test). Notably, the ratio of cells to cross-sectional area shows that each cell in a DMSO-treated control takes up approximately 19.7 ± 0.6 μm2 while each cell in an Az-treated neuromast takes up 18.3 ± 0.7 μm2, which was not a significant difference in cell size (P > 0.1).
This increase in neuromast size and cell number stimulated by Az could be explained either by an increased number of cells in the protoneuromast as it is being deposited, or by stimulation of proliferative growth following deposition. To help distinguish between these two hypotheses, we conducted time-lapse microscopy of neuromasts from DMSO- or Az-treated zebrafish measuring the size of neuromasts shortly after deposition at 36 hpf and 36 hr after deposition at 72 hpf. Az-treated neuromasts were significantly larger than DMSO-treated controls shortly after deposition (Fig. 3D,F,H; 989.3 ± 128.2 μm2 vs. 707.5 ± 48.4 μm2; P < 0.001), suggesting that the activation of the Wnt/β-catenin pathway was triggering an increase in the number of cells being deposited within one protoneuromast. Furthermore, while DMSO-treated controls decreased in size slightly over the next 36 hr as the neuromast condensed into its round shape (to 654.8 ± 7.2 μm2), Az-treated neuromasts had increased in size by more than a third (Fig. 3D–H; 1,366.2 ± 105.4 μm2; P < 0.001). Thus, Az treatment leads to both an increase in the size of the protoneuromast being deposited and in growth of the neuromast post-deposition.
In order to test more directly whether Az stimulation was leading to growth of the neuromast following its deposition, we treated fish with DMSO or Az beginning at 24 hpf, then gave them a 10-min pulse of BrdU at either 48 or 96 hpf, fixing them 2 hr after BrdU-treatment. Az led to a more than doubling in the number of BrdU+ cells per neuromast when the BrdU pulse was given at 48 hpf (Fig. 4C; 31.75 ± 0.6 BrdU+ cells per neuromast for Az compared with 11.1 ± 0.5 BrdU+ cells per neuromasts for DMSO-treated controls, N = 20; P < 0.001). A BrdU pulse at 96 hpf, when the cells of the control neuromasts have become largely quiescent (Harris et al., 2003; Laguerre et al., 2005), labels only 5.6 ± 0.4 BrdU+ cells per neuromast, but Az-treated fish still maintained more than twice as many proliferating cells (12.5 ± 0.6 BrdU+ cells per neuromast; P < 0.001 compared to DMSO at 96 hpf; Fig. 4A–C). Thus, Az is capable of maintaining lateral line progenitors in a proliferative state long after deposition of the neuromast from the primordium.
As GSK3β is a component of many signaling pathways, we next wanted to confirm that the Az-mediated induction of proliferation was due to its effects on the canonical Wnt/β-catenin signal transduction cascade. We used fish expressing dominant negative TCF, which lacks a β-catenin binding site, under control of the heat shock promoter Tg(hsp70:dnTCF-GFP); Lewis et al., 2004) to test whether we could suppress the Az-stimulated augmentation of proliferation by blocking the output of the canonical Wnt pathway. We heat shocked transgenic and wild-type sibling embryos at 48 hpf for 1 hr, then treated them with DMSO or Az until 72 hpf followed by pulse labeling with BrdU and fixation. While wild-type siblings showed a similar Az-mediated increase in the number of proliferating cells in each neuromast compared to DMSO-treated controls (Fig. 4D,F), this effect was completely blocked in transgenic fish expressing dominant negative TCF (Fig. 4E,F).
As a method for blocking upstream activation of the Wnt cascade, we also used Tg(hsp70:dkk1b-GFP) zebrafish (Stoick-Cooper et al., 2007), which express the secreted inhibitor of the Wnt ligand, dkk1b, following heat shock. We allowed both wildtype and dkk1b transgenic fish to develop to 48 hpf, then gave them a 2-hr heat shock to induce expression of the transgene. Fish were reared until 60 hpf, given a 10-min pulse of BrdU, then fixed 3 hr after BrdU labeling. While wildtype siblings had 4.6 ± 0.8 BrdU-positive nuclei per neuromast (range 2–10 BrdU+ nuclei), transgenic fish expressing the dkk1b transgene had only 0.8 ± 0.4 BrdU-positive nuclei per neuromast (Fig. 4G–I; range 0–3 BrdU+ nuclei; P < 0.001). Thus, expression of dkk1b to block activation of the Wnt cascade significantly suppresses proliferation in the lateral line neuromasts just after their deposition.
Treatment With 1-Azakenpaullone Leads to Overproduction of Hair Cells
Prior work by Aman and Piotrowski (2008) suggested that constitutive activation of Wnt signaling in mutant embryos expressing a non-functional APC during lateral line development does not prevent the differentiation of precursor cells into mature hair cells. In contrast, pharmacological stimulation of Wnt signaling in the retina of fish at similar ages maintains progenitors in a proliferative fate at the expense of production of differentiated progeny (Yamaguchi et al., 2005; Meyers et al., 2012). To test whether the maintenance of lateral line progenitors in a proliferative state by Az treatment blocks differentiation, we examined the number of hair cells differentiated at 96 hpf in fish where Az was applied beginning at 2 dpf, such that all effects of Az were after deposition of the primary neuromasts. At 4 dpf, DMSO-treated fish had 10.3 ± 0.5 HCS-1-positive hair cells per neuromast in the two anterior-most neuromasts, while fish treated with Az had 16.3 ± 0.9 hair cells per neuromast in the two anterior-most neuromasts, a statistically significant increase (Fig. 5C; N = 12; P < 0.001). Thus, the initial maintenance of progenitors in a proliferative state leads to increased production of differentiating hair cells in Az-treated fish. To test whether this increased production could be maintained over longer incubations with Az, we counted HCS-1-positive hair cells in fish treated with DMSO or Az from 2–7 dpf. The two anterior-most neuromasts in DMSO-treated controls had 16.1 ± 1.6 hair cells per neuromast (a 56% increase from 4 dpf), while those treated with Az had 34.6 ± 2.3 hair cells per neuromast at 7 dpf (a 112% increase from 4 dpf; Fig. 5A–C). We were also able to observe the increase in hair cell production using the Tg(pou4f3:gap43-GFP) line of fish, which express membrane-bound GFP under control of the Brn3c promoter and thus have constitutively fluorescent hair cells (Xiao et al., 2005; Fig. 5D,E). Thus, Az-treatment not only led to an increase in the number of hair cells that were produced, but led to a sustained overproduction of hair cells, leading to a more than doubling of the normal number by 7 dpf. This demonstrates that the increase in proliferation triggered by Az does not come at the expense of hair cell differentiation, consistent with the work of Aman and Piotrowski (2008). In fact, these data suggest that Az is able to sustain an increased rate of hair cell production long after normal production has slowed.
Treatment With 1-Azakenpaullone Stimulates Proliferation in Mature Neuromasts
The findings reported above demonstrate that exogenous activation of Wnt signaling stimulates progenitor proliferation in protoneuromasts and recently deposited neuromasts, but those experiments all initiated the stimulation of Wnt signaling when the progenitors/precursors were still proliferative, either as part of the proliferative primordium or shortly after deposition. However, the initially high levels of proliferation decrease during cell differentiation between 48–96 hpf, leading to quiescence of most cells in the neuromast, such that there are only 1–2 proliferative cells at any one time per neuromast in uninjured fish by 4–5 dpf (Williams and Holder, 2000; Harris et al., 2003; Laguerre et al., 2005; Ma et al., 2008). To test whether stimulation of Wnt signaling by Az would be able to trigger proliferation in these normally quiescent supporting and mantle cells, we added 2.5 μM Az, or DMSO as a vehicle control, to fish from 6–8 dpf, with BrdU present throughout to label all cells that pass through S-phase during the incubation period. To determine whether there was any distinction between the stimulation of peripheral supporting cells or internal supporting cells, we counted proliferative cells based on their position in the neuromast according to the definitions of Ma et al. (2008): the outermost row of cells and cells 2–3 layers past the neuromast border were defined as peripheral and cells inside that boundary were defined as internal. We counted the number of BrdU-positive cells for two anterior neuromasts in each of 6 fish per condition. Az led to a more than doubling of the number of proliferative cells during those two days, both for peripheral and internal supporting cells, compared to control, a statistically significant increase (Fig. 6A–C; P < 0.001). Thus, activation of Wnt-signaling by Az is sufficient to stimulate normally quiescent supporting cells in mature neuromasts to return to the cell cycle.
Wnt-Activation Increases Proliferation During Hair Cell Regeneration
Although uninjured neuromasts have a relatively low rate of ongoing proliferation, quiescent supporting cells readily return to the cell cycle to regenerate lost hair cells (Harris et al., 2003; Hernández et al., 2007; Ma et al., 2008). Since Az was sufficient to increase the number of normally quiescent supporting cells that proliferate in mature, uninjured neuromasts, we next asked whether it would also increase the number of cells that proliferate in response to hair cell lesion. We lesioned lateral line hair cells in zebrafish at 5 dpf with a 1-hr incubation in 500 μM neomycin (Harris et al., 2003), then incubated them in either DMSO or 2.5 μM Az for 2 days following the lesion with BrdU present continuously in the media. We quantified the number of BrdU+ cells in three anterior neuromasts for each of 13 fish. Az again stimulated a more than doubling in the number of BrdU-labeled supporting cells per neuromast from 16.8 ± 1.8 BrdU+ cells per neuromast in DMSO controls to 36.7 ±1.8 BrdU+ cells in fish treated with Az (Fig. 7A–C; P < 0.001). When we distinguished between internal and peripheral supporting cells, Az led to a doubling for each population (Fig. 7C), suggesting that, as seen above in unlesioned neuromasts, it is not selectively affecting only internal or peripheral supporting cells, but triggers both populations to proliferate.
As our data from the developmental studies suggest that stimulating Wnt/β-catenin signaling can maintain progenitors in a proliferative state for several days and over several rounds of division, it is possible that the increase in proliferating cells we see over the 48 hr following a lesion is not due to an increase in the number of supporting cells recruited to proliferate in response to the lesion, but merely maintenance of a limited number of cells in a proliferative state. To test whether Az actually stimulated an increase in the number of supporting cells returning to the cell cycle, we examined the number of BrdU+ cells in two anterior neuromasts 18 hr after lesion, a point when the number of cells returning to the cell cycle is peaking but without enough time for multiple rounds of mitosis. Az-treated fish had 13.0 ± 0.6 BrdU+ cells per neuromast (8.0 ± 0.3 internal supporting cells and 5.0 ± 0.5 peripheral supporting cells), a significant increase compared with only 5.5 ± 0.5 BrdU+ cells in DMSO-treated controls (Fig. 7D; 2.4 ± 0.3 internal supporting cells and 3.1 ± 0.4 peripheral supporting cells; P < 0.001 for each). Thus, Az stimulates a significant increase in the number of supporting cells that have returned to the cell cycle within 18 hr of the lesion, suggesting that Wnt/β-catenin signaling is sufficient to trigger supporting cells to re-enter the cell cycle.
In order to determine whether the canonical Wnt pathway was being activated in response to neomycin-induced hair cell loss, we examined expression of the TOP:gfp reporter transgene as a readout for active Wnt signaling (Dorsky et al., 2002). Control fish at 6 dpf that were not treated with neomycin show little expression of GFP within the neuromast (Fig. 7E). In contrast, 12 hr after 500 μM neomycin exposure to lesion the hair cells, the TOP:gfp transgene became strongly expressed in the neuromast (Fig. 7F). This suggests that the canonical Wnt pathway is being activated in neuromasts at the same time that the cells are beginning to proliferate in response to hair cell loss.
To test whether Wnt signaling was required for regenerative proliferation, we again used the Tg(hsp70:dkk1b-GFP) transgenic fish line to suppress activation of the Wnt pathway. The fish were heat-shocked at 39.5°C for 2 hr at 5 dpf, then placed in 500 μM neomycin for 1 hr to lesion the hair cells. Fish were maintained in BrdU for 18 hr, then fixed for staining. Wild-type siblings had 8.8 ± 1.5 BrdU+ cells per neuromast while transgenic fish expressing dkk1b had 3.9 ± 1.0 BrdU+ cells per neuromast, a significant decrease (Fig. 7G–I; P < 0.05; t-test).
1-Azakenpaullone Increases the Number of Regenerated Hair Cells
Finally, we asked whether the Wnt-stimulated increase in the number of supporting cells proliferating during hair cell regeneration led to a functional increase in the number of hair cells regenerated. We counted the number of GFP-positive hair cells from the Brn3c:gfp line for 12 fish per condition 72 hr after neomycin lesioning of the hair cells. While DMSO-treated controls had 8.7 ± 0.6 hair cells per neuromast, fish treated with Az had 14.1 ± 1.1 hair cells per neuromast, a statistically significant increase (Fig. 8A–C; P < 0.001). Thus, as during development, the stimulation of proliferation by Az results in an increased number of differentiating hair cells during regeneration.
Together these data demonstrate a novel role for Wnt/β-catenin signaling in controlling the proliferative capacity of lateral line progenitors during the development of the neuromasts and also of supporting cells in more mature neuromasts during their normal growth and regeneration.
We have used pharmacological and genetic manipulation of Wnt/β-catenin signaling to examine the role of Wnt signaling on proliferation during development, ongoing growth, and regeneration of the zebrafish lateral line. While Wnt signaling has been previously shown to control proliferation at the leading edge of the migratory primordium (Aman and Piotrowski, 2008; Aman et al., 2011; McGraw et al., 2011; Valdivia et al., 2011), this study represents the first analysis of the role of Wnt signaling in neuromasts following their deposition and maturation. We find that activation of the canonical Wnt/β-catenin pathway stimulates proliferation of progenitors in newly deposited neuromasts and in supporting cells of mature neuromasts, and that this stimulation results in the overproduction of hair cells during development and regeneration. Additionally, we find that having active Wnt-signaling is required for proliferation during neuromast differentiation and regeneration.
Wnt Signaling During Lateral Line Development
Wnt signaling is well established in controlling proliferation within the migratory lateral line primordium (Aman and Piotrowski, 2008; Aman et al., 2011; McGraw et al., 2011; Valdivia et al., 2011). Genetic activation of Wnt/β-catenin signaling results in slowing of the migratory primordium and expansion of proliferation into the trailing zone (Aman and Piotrowski, 2008; Aman et al., 2011; Valdivia et al., 2011), while inhibition does not affect migration of the primordium but limits organization of the rosettes and deposition of protoneuromasts (Aman and Piotrowski, 2008; McGraw et al., 2011). We show that pharmacological activation of Wnt/β-catenin signaling phenocopies the genetic activation, slowing migration of the primordium. We also show that this leads to deposition of larger neuromasts, though approximately the same number of neuromasts are deposited whether or not Wnt signaling is stimulated (Fig. 2). Additionally, whether we activate Wnt/β-catenin signaling during primordium migration (e.g., 24–30 hpf) or shortly after all neuromasts have been deposited, we find that activation stimulates a persistent increase in proliferation of the progenitors in each neuromast. This results in continued growth of the neuromast following its deposition from the primordium (Figs. 34), and expansion of the number of hair cells that form within each neuromast (Fig. 5). Although we have not quantified supporting cell numbers, the significant increase in the total number of cells in Az-treated neuromasts at 72 hpf (Fig. 3) is much larger than the increase in hair cell number and thus suggests that both hair cells and supporting cells are being overproduced. Alterations in notch signaling, such as the classic mind bomb mutation, also produce excess hair cells in the lateral line and inner ear (Haddon et al., 1998, 1999; Itoh and Chitnis, 2001). However, in the case of mind bomb, the overproduction of hair cells comes at the expense of supporting cells due to a failure of lateral inhibition (Haddon et al., 1999; Itoh and Chitnis, 2001), whereas with hyperactivation of Wnt signaling, the overproduction of hair cells appears to be due to the continual proliferation of progenitors past the normal point of quiescence, resulting in continued growth of the neuromast rather than conversion of cell fate.
It is possible that Wnt signaling, in addition to stimulating proliferation, directly affects hair cell differentiation. Wnt-dependent fibroblast growth factor (FGF) signaling drives the specification of HC precursors in the trailing zone of the migrating primordium (Itoh and Chitnis, 2001; Matsuda and Chitnis, 2010). Therefore, overactivation of Wnt signaling in neuromasts following their initial deposition may result in an increase in the secretion of FGF ligands that would in turn lead to an increase in the specification of HC precursors and eventually to an increase in the number of differentiated hair cells. dkk1b expression, although canonically considered a Wnt target and negative feedback inhibitor (Niida et al., 2004; Chamorro et al., 2005), may be more reflective of FGFR1 activation in the zebrafish lateral line (Aman and Piotrowski, 2008; Chitnis et al., 2011). Thus, the expanded dkk1b expression we observe in Az-stimulated primordia and deposited neuromasts may reflect increased FGF signaling secondary to Wnt signaling, potentially consistent with Wnt inducing an increase in specification of the HC precursors. Further experimentation will be necessary to better characterize the specific effect that increased Wnt signaling has on lateral line progenitors, and the potential involvement of additional signaling cascades such as FGF.
Lateral line progenitors are highly proliferative upon deposition of the neuromast, but rapidly become quiescent as they differentiate (Laguerre et al., 2005), until only 1–2 supporting cells are proliferating in any one neuromast by 96 hpf (Williams and Holder, 2000; Harris et al., 2003; Laguerre et al., 2005; Ma et al., 2008). Although stimulation of Wnt signaling maintains an elevated number of proliferating cells in each neuromast at 48 and 96 hpf compared to controls, the number of proliferative cells does decline between these time-points, suggesting that cells are leaving the cell cycle. This may reflect upregulation of negative feedback components of the Wnt pathway that are triggered by the pharmacological overactivation, or may reflect the contribution of other signaling pathways that normally aid in promoting differentiation of the progenitors into supporting cells.
Wnt Signaling in Ongoing Growth and Regeneration of the Lateral Line
In addition to maintaining developing neuromast progenitors in a proliferative state, we find that activation of Wnt signaling with Az also stimulates a significant increase in the number of proliferating supporting cells in uninjured, more mature neuromasts. In normal, uninjured neuromasts from 5–6-day-old fish, only 1–2 supporting cells are typically dividing at any one time, while we find that over a 48-hr period following Az treatment more than twice as many cells have accumulated BrdU (Fig. 6) as in DMSO-treated controls. This suggests that activation of Wnt/β-catenin signaling is sufficient to override the quiescence that develops as lateral line supporting cells mature.
When there is injury, such as aminoglycoside-induced hair cell loss, lateral line supporting cells can rapidly re-enter the cell cycle to produce replacement hair cells (Jones and Corwin, 1993, 1996; Harris et al., 2003; Ma et al., 2008; Namdaran et al., 2012). We again find that stimulation of Wnt signaling is sufficient to significantly increase both the number of proliferative cells and the production of replacement hair cells. Furthermore, overexpression of the inhibitory dkk1b transgene can suppress the increase in proliferation following hair cell lesion. This suggests that Wnt signaling may be both necessary and sufficient to trigger supporting cell re-entry into the cell cycle, similar to what is proposed for Müller glia in the retina (Ramachandran et al., 2011; Meyers et al., 2012).
Notably, in mammals, there is a decrease in proliferation of supporting cells during postnatal maturation of vestibular organs, though this change to a quiescent state is largely permanent in mammals rather than readily reversible as it is in fish and birds (Hume et al., 2003; Davies et al., 2007; Gu et al., 2007; Meyers and Corwin, 2007; Collado et al., 2011). The inability of supporting cells to return to the cell cycle following this maturational quiescence is believed to be one of the major limitations that prevents significant hair cell regeneration following injury in mammals. Notably, in the organ of Corti of mice, the common hair cell/supporting cell progenitor cell is responsive to Wnt activation, and produces additional hair cells in vitro when Wnt signaling is activated (Shi et al., 2012). Further, GSK3β, the target of Az (Kunick et al., 2004), plays a significant role in the development of supporting cell quiescence in the vestibular system of mammals, and inhibition of GSK3β maintains supporting cell proliferation past when they would normally become permanently quiescent (Lu and Corwin, 2008). These studies, paired with our similar findings in the development of quiescence in zebrafish lateral line supporting cells, strongly suggests that the central role for Wnt/β-catenin in control of supporting cell proliferation may be highly conserved across vertebrates from fish to mammals. The fact that exogenous activation of this pathway is sufficient to trigger significant numbers of supporting cells to re-enter the cell cycle in zebrafish lateral line regeneration suggests that further examination of this pathway in control of the limited proliferation seen in mammalian hair cell regeneration may be warranted.
We demonstrate a novel ability of activated Wnt/β-catenin signaling to stimulate the proliferation of supporting cells in neuromasts of the pLL both following its initial development and later in more mature larval neuromasts during the regeneration of mechanosensory hair cells. These findings suggest that activated Wnt signaling is sufficient to maintain supporting cells in a proliferative state and to induce quiescent supporting cell populations to return to the cell cycle. This stimulation of proliferation results in production of excess hair cells both during development and regeneration, suggesting that Wnt signaling expands the population of progenitors and differentiating cells. The significant stimulatory effect that this has on both developmental production and regeneration of sensory hair cells in the zebrafish, paired with work suggesting that Wnt signaling plays a crucial role in regulating progenitor proliferation in mammals, suggests that better understanding the role of this signaling pathway in control of hair cell progenitor fate may have significant clinical impacts on recovery from hearing loss and vestibular dysfunction.
All zebrafish were maintained at 28.5°C on a 14hr light:10hr dark cycle. The following zebrafish strains were used: TL as wild-type fish for general experimentation (obtained from ZIRC, Eugene, OR); the Tg(−8.0cldnb:lynGFP)zf106 line, which express membrane-bound green fluorescent protein (GFP) in all cells of the PLL (Haas and Gilmour, 2006, a kind gift of D. Gilmour); Tg(hsp70l:tcf7l1a-GFP) w26, which express a dominant negative TCF to inhibit Wnt signaling, under control of the heat shock promoter (Lewis et al., 2004; obtained from ZIRC, Eugene, OR); Tg(hsp70:dkk1b-GFP)w32, which express the Wnt-inhibitory dkk1b gene under control of the heat shock promoter (Stoick-Cooper et al., 2007; obtained from ZIRC); Tg(TOP:GFP)w25, which express GFP under a synthetic series of lef1 binding sites to provide a readout of canonical Wnt activation (Dorsky et al., 2002; obtained from ZIRC); and Tg(pou4f3:gap43-GFP)s356t line of fish, which express membrane-bound GFP under control of the Brn3c promoter (Xiao et al., 2005; a kind gift of H. Baier and A. Chitnis). Embryos were reared in E3 media (Westerfield, 2007) in a 28.5°C incubator. All experiments with vertebrate animals were approved and monitored by the Animal Care and Use Committee at Colgate University.
To constitutively activate canonical Wnt/β-catenin signaling, we used the glycogen-synthase kinase-3β (GSK3β) inhibitor 1-Azakenpaullone (US Biological, Cleveland, OH), which allows for cytoplasmic accumulation of the Wnt signaling effector β-catenin (Leost et al., 2000; Kunick et al., 2004). Stock solutions were made in DMSO, and diluted to 2.5 μM in E2 media; 2.5 μM Az was previously shown to be the minimal concentration needed to see effects on retinal development and regeneration (Meyers et al., 2012), and was also the optimal dose for arresting lateral line primordium migration in a dose-response study (data not shown). Controls were treated with 0.025% DMSO in E2 media as a vehicle control. For drug treatment to block activation of Wnt signaling, we incubated the fish in 10 μM XAV939 (Millipore, Billerica, MA). Drugs were applied to fish at various times, and fish were returned to incubation until fixation in 4% paraformaldehyde (PFA). Heat shocks were done with mixed populations of wildtype siblings and transgenic fish. Fish were placed in mesh baskets, then transferred to pre-heated media at 39.5°C for the duration of the heat shock, then immediately transferred to media at 28.5°C for incubation.
To lesion hair cells, the protocol of Harris et al. (2003) was followed. Briefly, 5–6-day-old larvae were incubated in 500 μM neomycin (in E2 medium) for 1 hr at 28.5°C, then washed three times in E2 media, and cultured as appropriate. To confirm that the neomycin ablated hair cells, we examined Tg(Brn3c:gfp) zebrafish that express GFP in their hair cells to confirm loss of hair cells 4 hr after lesion, and also confirmed loss at 4 hr after lesion with the HCS-1 antibody.
To directly monitor the growth of the neuromasts, we conducted time-lapse microscopy using the Tg(−8.0cldnb:lynGFP) line to measure the size of neuromasts during and following their deposition from the lateral line primordium. We followed the general protocol of Haas and Gilmour (2006). Briefly, we anesthetized zebrafish at approximately 30 hpf in 0.003% tricaine, embedded them in 0.75% low-melt agarose containing either 2.5 μM Az or 0.025% DMSO as a control, and incubated them in E2 medium containing either 2.5 μM Az or 0.025% DMSO as a control at 28.5°C on a Zeiss 710 laser scanning confocal microscope. Images were acquired every 15 min for approximately 56 hr.
Measurements of NM size were taken from maximum intensity projections of z-stacks taken during time-lapse microscopy of the most anterior NM that was visible throughout the duration of the time lapse. The size of the deposited neuromast was measured immediately after deposition, when the deposited NM condensed into a round shape, and at the end of the time-lapse recording. Area of the neuromast was determined by the extent of the GFP expression.
Cell Proliferation Assay
Dividing cells were labeled by incorporation of the thymidine analog 5-bromo-2-deoxyuridine (BrdU; MPBio). For labeling in embryos of 96 hpf or younger, fish were labeled with a 10-min pulse of 10 mM BrdU in 15% DMSO 2–4 hr prior to fixation. For larvae 96 hpf or older, fish were either labeled with a pulse of BrdU as above, or with continuous BrdU (5 mM) in HEPES-buffered E2 medium for the duration of the experiment.
For whole-mount labeling of cells that had incorporated BrdU, fish were processed generally following the protocols of Ma et al. (2008). Briefly, fixed fish were washed into MeOH at −20°C for at least 1 hr, rehydrated, antigen retrieved with 10 μg/ml proteinase K for 20 min, re-fixed in 4% PFA, then treated with 2N hydrochloric acid for 1 hr. Fish were then blocked in 10% normal goat serum (NGS), incubated in mouse anti-BrdU (1:50; BD) overnight at 4°C, incubated in Dy568-conjugated anti-mouse secondary antibodies (1:200, Jackson Immunoresearch, West Grove, PA) for 5 hr, and counterstained with DAPI prior to imaging. To label differentiated hair cells in whole-mounts, fish were blocked in 10% normal goat serum (NGS), incubated in mouse anti-HCS1 (Goodyear et al., 2010; 1:200, a kind gift of J. Corwin) overnight at 4°C, incubated in Dy568-conjugated anti-mouse secondary antibodies (1:200, Jackson Immunoresearch) for 5 hr, and counterstained with DAPI prior to imaging.
Quantification of BrdU and hair cell number was performed on z-stacks and not maximum intensity projections in order to minimize overlapping cells in the Z-plane. Unless otherwise stated, neuromasts imaged were primary neuromasts around the end of the yolk extension (approximately the third or fourth neuromast along the body). Central and peripheral supporting cells were defined as in Ma et al. (2008): cells in the two outermost bands of nuclei in each neuromast were defined as peripheral and cells in the center of that were defined as central. Normality of residuals was tested with Kolmogorov-Smirnov test, and comparison of means was done using a Student's t-test or ANOVA as appropriate. All data is presented as mean ± standard error.
In Situ Hybridization
In situ hybridization was done following the protocol of Thisse et al. (2001) and Thisse and Thisse (2004). Digoxygenin-labeled antisense RNA probes against dkk1b and lef1 (both kind gifts of A. Chitnis) were hybridized to fish fixed at 36 or 48 hpf. Detection of hybridization was via alkaline phosphatase-conjugated anti-digoxygenin (Roche, Indianapolis, IN), followed by NBT/BCIP color reaction (BioRad, Hercules, CA).
We thank Jeff Corwin and Ajay Chitnis for gifts of reagents and helpful discussions about the project. The Zeiss 710 Confocal LSM was funded by NSF DBI-0923310 to J.R.M.