(A) Double whole-mount immunostainings for phosphorylated and total TAU of 32-hour-old transgenic zebrafish embryos expressing hTAU-P301L and DsRed. TAU is phosphorylated in spinal cord neurons at residues Thr231/Ser235 (AT180), Thr181 (AT270), Ser262/Ser356 (12E8), Ser396/Ser404 (PHF1), Ser422 (422), and Ser202/Thr205 (AT8). (B) WBs of total and phosphorylated TAU of 48-hour-old transgenic zebrafish embryos expressing hTAU-P301L and DsRed or DsRed alone or nontransgenic siblings. Phosphorylated TAU is only detected in TAU/DsRed transgenic embryos. No cross-reacting bands are detectable in controls at the same molecular weight. In addition to the specific band above 64 kDa (arrow) in TAU-positive embryos, antibodies PHF1 and 422 detect a nonspecific band at lower molecular weight in all embryos (asterisk). (C) TAU changes its conformation to a pathologic state, as shown in whole-mount immunostainings, by using the conformation-specific antibody MC1, in most neurons of 32-hour-old embryos expressing the TAU transgene. (D) Side views of 5-week-old living zebrafish in bright field and DsRed illumination. The fish are still rather translucent, and the transgene-expressing cells can be detected by their red fluorescence (arrowheads). (E) Immunohistochemical stainings of spinal cord paraffin sections of the same 5-week-old TAU transgenic zebrafish for total TAU (antibody DA9) and pathologically phosphorylated TAU (AT8) (arrowheads). (F) In addition, tangles are observed by Gallyas silver staining in sections of the same 5-week-old TAU transgenic zebrafish (arrowheads). Scale bars: 20 μm.

(A–D) Side views of 6-day-old living zebrafish larvae expressing TAU/DsRed (A, see single acridine orange–positive neuron in inset, depicted by arrowhead), only DsRed (B), or no transgene (siblings, C) stained with acridine orange. TAU-expressing fish show substantial cell death while DsRed-expressing and nontransgenic fish show only a low, basal amount of dying cells (quantified in D). AO, acridine orange. Data represent mean ± SD. ***P < 0.01. Scale bars: 100 μm; 10 μm (insets). (E) Still images of several time points of a time-lapse video, showing a close-up of TAU-expressing neurons in the spinal cord of transgenic zebrafish, which were stained with acridine orange. An intact neuron (white arrowhead) with an axon first changes its shape and rounds up. Subsequently, fragmentation and uptake of acridine orange is observed (yellow arrowhead; see also Supplemental Video 1). Scale bars: 10 μm.

(A) Left panel shows surface representation of the x-ray structure of GSK3β with inhibitor AR-164 in the active site. Polar areas are colored blue and lipophilic areas red. The methylpiperazine sulfone amide extends out toward the solvent area. Right panel shows top view of the active site of GSK3β. Dotted yellow lines represent hydrogen bonds between the protein backbone and AR-164. The 6-membered aromatic pyrazine moiety binds together with its anilino function to the backbone of the kinase. The pyridine ring binds to the conserved salt bridge formed by Lys85 and Asp200 (x-ray resolution: 2.47 υ). AR-534 binds in a similar way (not shown). (B and C) Chemical structure of AR-164 and AR-534 and compound characterization values including Ki (mean from 3 independent experiments performed in duplicate), selectivity over CDK2, solubility and permeability coefficient Pe (cm/min) over cell membranes (Caco2 cells), and BBB. (D) AR-164 and AR-534 inhibit pSer396 TAU phosphorylation in 3T3 fibroblasts harvested at 4 hours after treatment in comparison with total TAU, as analyzed by quantitative WB. (E) Graphical representation with IC₅₀ values of AR-164 and AR-534 effects on inhibition of TAU phosphorylation compared with SB-415286 and LiCl. (F) Alignment of protein sequences of human and zebrafish GSK3β. Over 90% of the amino acids are identical; the residues in the active site of the enzyme, which interact with the inhibitors, are completely conserved.

(A) The Driver construct contains the neuronal zebrafish promoter HuC driving the expression of Gal4-VP16, which binds to the UAS on the Responder construct. Here, it activates the bidirectional expression of hTAU-P301L and DsRed via the minimal promoters. UAS-dependent gene expression of TAU and DsRed is indicated in living fish by DsRed fluorescence. Driver and Responder constructs are flanked by Tol2 transposon sites. (B) To generate transgenic fish, the Driver and Responder constructs were mixed and injected together with Tol2 mRNA. The mRNA is translated to active transposase, which detects the flanking Tol2 elements and catalyzes random integration into the zebrafish genome in a subset of embryonic cells for a short time period, generating mosaic founder embryos. Mosaic DsRed-positive larvae were raised and outcrossed with wild-type fish. A subset of the offspring will be transgenic and can be easily identified and sorted by DsRed-positive neurons. Scale bar: 1 mm. (C) Double immunostainings for total TAU (T46 antibody) and DsRed of 32-hpf transgenic zebrafish embryos expressing hTAU-P301L and DsRed. Transgenic embryos express both hTAU-P301L and DsRed in spinal cord neurons, showing effective bidirectional expression from the Responder construct. Lateral views of the trunk above the end of the yolk extension, anterior to the left. Scale bar: 20 μm.

(A) Double whole-mount immunostainings for TAU phosphorylated at the AT8 epitope (a late marker of pathology) and the AT270 epitope (an early marker) together with staining by antibody K9JA, which shows expression of total TAU in 24-, 32-, and 48-hpf and 7-day-old transgenic zebrafish embryos. There are very few AT8-positive neurons in the spinal cord of 24-hour-old embryos, while this number increased significantly in 32-hour-old embryos and rose even further in 48-hour-old embryos. Finally, all TAU-expressing neurons contained the AT8 epitope in 7-day-old larvae. (B) In contrast, immunoreactivity of the early marker AT270 was already strong in many TAU-expressing cells at 24 hpf and became only slightly stronger at older stages. The level of total TAU detected in the same embryos by double staining with the K9JA antibody was similar between 24 hpf and 32 hpf. After 48 hours, a substantial part of TAU was transferred to the neuronal projections. In contrast, AT8-positive TAU remained mainly in the cell bodies. The AT8-positive somata combined with a lack of strongly AT8-positive neuronal projections is consistent with the pathological accumulation of modified TAU in neuronal cell bodies of patients. Lateral views of the trunk above the end of the yolk extension, anterior to the left. Scale bars: 50 μm.

(A–G) Whole-mount immunostainings for znp1, which labels synaptotagmin in the extending axons of primary motoneurons. Expression of TAU causes a dramatically reduced extension of znp1-positive motoneurons already at 28 hpf (A and B). The difference in motoneuron length is quantified (C) by measuring the length of the first 4 znp1-stained motoneuron projections (marked 1, 2, 3, 4 in A and B) before the end of the yolk extension (marked by dotted line). Triangles and diamonds represent values from individual motoneurons; colored horizontal lines represent mean ± SD. ***P < 0.01. In 48-hour-old embryos, the motoneurons have grown further in both TAU fish and controls, but the motoneuron extensions are still reduced (D and E). 5 days later, the motoneurons have grown around the muscle in both TAU fish and controls (F and G). The fine projections of motoneurons (see enlarged insets, arrowheads), are still highly reduced in TAU transgenic fish. Lateral views of the trunk above the end of the yolk extension, anterior to the left. Scale bars: 50 μm. (H–J) The stereotypic escape response, which is normal in DsRed-expressing (H) and nontransgenic larvae at 48 hpf (data not shown), is highly reduced or absent in TAU-expressing larvae (I; see also Supplemental Video 2). The phenotype was quantified in groups of 50 TAU/DsRed versus DsRed transgenic larvae, which were pooled from several clutches and selected for strong and comparable DsRed expression (J). Error bars represent mean ± SD; ***P < 0.01.

(A) GSK3 inhibitors have specific effects on early wild-type zebrafish development when treatment occurs between 4 and 24 hpf. LiCl treatment caused strongly reduced eye formation (black arrowheads), while SB-216763 and SB-415286 also perturbed formation of brain, somites, and tail (gray arrowheads). In addition to these phenotypes, AR-534 also caused multiple ear formation (arrowheads in enlargement). In contrast, AR-164, which has in vitro activity comparable to that of AR-534, did not cause any detectable change in phenotype even at 4-fold concentration. Embryos injected with dN-GSK3β mRNA to suppress the function of endogenous GSK3β phenocopied alterations of compound-treated fish; however, only AR-534 was potent enough to phenocopy the ear multiplication seen in embryos expressing high levels of dN-GSK3β (arrowheads in enlargement). Embryos injected with wild-type GSK3β did not display a phenotype at comparable mRNA concentrations. Scale bars: 500 μm; 100 μm (insets). (B) GSK3 inhibitors reduce pathologic TAU phosphorylation in vivo when applied to TAU transgenic fish between 20 and 100 hpf. Phosphorylated TAU was detected by WB using antibody PHF1 and normalized to total TAU (K9JA). The band intensities were compared with a DMSO control on the same gel. The upper specific PHF1-band (arrow) was quantified. As already shown in Figure 2, the lower PHF1 band was unspecific (asterisk). (C) Quantification of WB band intensities of phospho-specific TAU as a percentage of the total amount of TAU detected by K9JA shown in B.