Proliferative regeneration of zebrafish lateral line hair cells after different ototoxic insults

Abstract

Sensory hair cells in the zebrafish lateral line regenerate rapidly and completely after damage. Previous studies have used a variety of ototoxins to kill lateral line hair cells to study different phenomena including mechanisms of hair cell death and regeneration. We sought to directly compare these ototoxins to determine if they differentially affected the rate and amount of hair cell replacement. In addition, previous studies have found evidence of proliferative hair cell regeneration in zebrafish, but both proliferation and non-mitotic direct transdifferentiation have been observed during hair cell regeneration in the sensory epithelia of birds and amphibians. We sought to test whether a similar combination of regenerative mechanisms exist in the fish. We analyzed the time course of regeneration after treatment with different ototoxic compounds and also labeled dividing hair cell progenitors. Certain treatments, including cisplatin and higher concentrations of dissolved copper, significantly delayed regeneration by one or more days. However, cisplatin did not block all regeneration as observed previously in the chick basilar papilla. The particular ototoxin did not appear to affect the mechanism of regeneration, as we observed evidence of recent proliferation in the majority of new hair cells in all cases. Inhibiting proliferation with flubendazole blocked the production of new hair cells and prevented the accumulation of additional precursors, indicating that proliferation has a dominant role during regeneration of lateral line hair cells.

Figure 1. Rapid hair cell loss after treatment with water-borne copper.

Wild type larvae were treated at 5 dpf with serial dilutions of copper(II) sulfate for a period of 30 min to 8 h. Hair cell death was rapid in most cases, although 0.3 μM copper had a minimal effect. N = 8 fish per group. Error bars are +/− SD.

Figure 2. Copper may impair hair cell regeneration.

(A, B) Wild type larvae were treated at 5 dpf with serial dilutions of neomycin or copper for 30 min. Hair cell regeneration was scored using FM 1-43FX and normalized to controls. Regeneration was incomplete in groups treated with 3 or 10 μM copper, but other groups recovered fully. N = 7 fish per group. (C) Individual counts of hair cells labeled with antibodies against parvalbumin confirmed incomplete regeneration in larvae treated for 1 h with copper. N = 8 fish per group. Error bars are +/− SD.

Figure 3. Hair cells are derived from proliferating progenitors.

(A) Wild type larvae were treated at 5 dpf with neomycin or copper for 1 h followed by recovery in 5 mM BrdU until 24 hpt followed by fresh embryo medium. Greater numbers of BrdU-positive hair cells were observed in all ototoxin-treated groups at 72 hpt. (B) One-way ANOVA followed by a Tukey post-hoc analysis of the proportion of BrdU-positive hair cells revealed significant increases in all ototoxin-treated groups compared to control (p<0.001). No significant difference was observed between 1 and 10 μM copper that would suggest a dose-dependent effect on proliferative regeneration. A small but significant difference was observed between neomycin and 1 μM copper (*, p<0.05). N = 8 fish per group. Error bars are + SD. (C) Example neuromast at 72 hpt from a fish treated with 1 μM copper. Arrows mark BrdU-positive hair cells.

Figure 4. Hair cell regeneration is blocked by inhibition of mitosis.

ET4:GFP larvae were treated at 5 dpf with neomycin or copper for 1 h and incubated in flubendazole for 48 h. (A) Those not treated with flubendazole continued to add mature hair cells. Two-way ANOVA found significantly more cells expressed GFP than parvalbumin (p<0.001), indicating the presence of hair cell precursors. Bonferroni post-hoc analysis confirmed a significant difference in control and ototoxin-treated groups between GFP- and parvalbumin-positive hair cell counts at 48 hpt (**, p<0.01; ***, p<0.001). (B) Those treated with flubendazole exhibited little or no increase in mature hair cells. There were at most 1 or 2 parvalbumin-negative precursors present per neuromast, indicating that undifferentiated precursors did not accumulate. N = 8 fish per group. Error bars are +/− SD.

Figure 5. Flubendazole impairs division of hair cell progenitors.

ET4:GFP larvae were treated at 5 dpf with neomycin or copper for 1 h and incubated in flubendazole for 24 h. Immunohistochemistry was performed for GFP and PHH3, which is upregulated during mitosis. (A) Two-way ANOVA followed by Bonferroni post-hoc analysis found significant increases in the proportion of PHH3-positive cells in neomycin- and copper-treated larvae when treated with flubendazole (*, p<0.05; ***, p<0.001), demonstrating that cell division was arrested in GFP-positive hair cell precursors. N = 8 fish per group. Error bars are + SD. (B) Example neuromast at 24 hpt from a fish treated with neomycin and flubendazole. Arrows mark PHH3-positive nuclei.

Figure 6. Regeneration is obscured by growth of the immature lateral line.

(A, B) Wild type embryos were treated at 3 dpf with serial dilutions of neomycin or copper for 30 min. Hair cell regeneration was scored using FM 1-43FX and normalized to controls. Complete regeneration was observed in all groups. N = 14 fish per group. (C) Individual counts of hair cells labeled with antibodies against parvalbumin confirmed complete regeneration after 1 h treatment with copper or neomycin. However, all ototoxin-treated groups closely matched control embryos, which had few hair cells at the time of treatment. Instead of regeneration, most hair cell addition appeared related to early development of the lateral line. N = 8 fish per group. Error bars are +/− SD.

Figure 7. Proliferation is required for regeneration after treatment with gentamycin.

(A) Wild type larvae were treated at 5 dpf with 200 μM gentamicin for 30 min (acute) or 50 μM gentamycin for 6 h (chronic) and allowed to recover for 96 h from the beginning of treatment. Gentamicin-induced hair cell death was delayed compared to neomycin. There was no dose-dependent effect on regeneration, which was complete by 96 hpt. (B) When fish were incubated in flubendazole during recovery, minimal regeneration was observed, suggesting that proliferation was required. N = 8 fish per group. Error bars are +/− SD.

Figure 8. Proliferation is required for regeneration after treatment with cisplatin.

(A) Wild type larvae were treated at 5 dpf with 50 μM cisplatin for 24 h and allowed to recover for 72 h (96 h from the beginning of treatment). Regeneration was considerable yet remained incomplete. (B) When fish were incubated in flubendazole during recovery, minimal regeneration was observed, suggesting that proliferation was required. N = 8 fish per group. Error bars are +/− SD.