https://orcid.org/0000-0003-3444-1077
Abstract
Disrupted circadian rhythms are a prominent feature of multiple neurodegenerative diseases. Yet mechanisms linking Tau to rhythmic behavior remain unclear. Here, we find that expression of a phosphomimetic human Tau mutant (TauE14) in Drosophila circadian pacemaker neurons disrupts free-running rhythmicity. While cell number and oscillations of the core clock protein PERIOD are unaffected in the small LNv (sLNv) neurons important for free running rhythms, we observe a near complete loss of the major LNv neuropeptide pigment dispersing factor (PDF) in the dorsal axonal projections of the sLNvs. This was accompanied by a ~50% reduction in the area of the dorsal terminals and a modest decrease in cell body PDF levels. Expression of wild-type Tau also reduced axonal PDF levels but to a lesser extent than TauE14. TauE14 also induces a complete loss of mitochondria from these sLNv projections. However, mitochondria were increased in sLNv cell bodies in TauE14 flies. These results suggest that TauE14 disrupts axonal transport of neuropeptides and mitochondria in circadian pacemaker neurons, providing a mechanism by which Tau can disrupt circadian behavior prior to cell loss.
Introduction
Disrupted circadian sleep–wake cycles are a prominent feature of multiple neurodegenerative tauopathies (1), including traumatic brain injury (2,3), frontotemporal dementia (FTD) (4) and Alzheimer’s disease (ad) (5–10). Disruption of core body temperature rhythms (6,11) and diurnal melatonin rhythms was observed in ad patients, even before the onset of clinical symptoms (12), and the cell loss in the suprachiasmatic nucleus (SCN) was correlated with disrupted circadian activity (13). The microtubule-associated protein tau (Tau) is implicated in the pathogenesis of ad and other tauopathies. Yet how pathogenic Tau leads to perturbed circadian behavior remains unclear.
To study this question, we are using the fruit fly Drosophila melanogaster, a model system well established for studies on circadian rhythms. There are ~150 clock neurons in the fly brain that dictate circadian rhythms (14,15). All of these neurons contain a molecular clock, consisting of the CLOCK-CYCLE transcription factor complex, which drives the expression of its own repressor complex: period (PER) and timeless (TIM) (16), resulting in 24 h oscillations in the expression of clock components. This core clock signals to locomotor output pathways that dictate free-running ~24 h rhythms in constant dark (DD) conditions. In 12-h light:12-h dark (LD) conditions, wild-type flies exhibit a bimodal pattern of locomotor activity, increasing their activity preceding lights on and lights off, termed morning and evening anticipation, respectively.
The neuropeptide pigment-dispersing factor (PDF) is required for robust behavioral rhythms in Drosophila (17,18). PDF is expressed in a subset of ventral lateral circadian neurons (LNvs), which can be further divided into subgroups based on their size and axonal projection location. There are four large PDF-expressing (PDF+) LNvs (lLNvs) per hemisphere that contain ventral projections and four small PDF+ LNvs (sLNvs) that send their axonal projections dorsally. PDF+ sLNvs have been shown to be crucial in driving rhythmic behavior in constant conditions (19–21).
Distinct subsets of clock neurons also likely play specific roles in regulating morning and evening anticipation behavior. The sLNv and posterior dorsal neuron (DN1p) groups have been implicated in promoting morning anticipation (20–24), with a direct sLNv- > DN1p excitatory circuit likely contributing (25). Evening anticipation is controlled by a subset of non-PDF-expressing clock neurons, collectively termed ‘evening cells’: the fifth sLNv, a subset of dorsal lateral neurons (LNds) and a subset of DN1p (20,23,24,26–28).
To address the link between Tau and circadian rhythms, we are employing fly models of human tauopathies that recapitulate many disease phenotypes, including reduced lifespan, increased neurodegeneration, disrupted learning/memory and decreased climbing ability when broadly expressed in neurons (29–32). Previous studies have examined the effect of human Tau on circadian behavior and sleep in Drosophila with mixed results. One study showed that the expression of wild-type Tau [Tau-WT (29)] has modest effects on Drosophila circadian behavior when expressed in all clock neurons, including a 40% decrease in rhythmicity and decrease in total sleep time of ~200 min (33), while other studies did not observe any effects on circadian rhythms (34) and/or sleep (35,36). Moreover, the molecular and/or morphological basis for the sleep and circadian phenotypes observed was not determined.
Phosphorylation of Tau is important to its toxic effects (37), including aggregation propensity, strength of microtubule binding (38,39) and synaptic vesicle trafficking (40). Similarly, the phosphomimetic TauE14 construct has been shown to exacerbate neurodegenerative phenotypes observed with Tau-WT (41,42), such as retinal degeneration (30), cell death (43) and lethality (39). Here, we express either human wild-type Tau or a phosphomimetic mutant TauE14, where 14 Serine/Threonine sites commonly phosphorylated in human disease have been mutated to glutamate to mimic permanent phosphorylation (30,41,42), selectively in clock neurons to assess their effects on circadian behavior. We find that Tau can suppress circadian rhythms and dramatically reduce the axonal distribution of the PDF neuropeptide and mitochondria consistent with disrupted microtubule transport. Our studies provide a pre-degenerative in vivo model to mechanistically explore the link between pathogenic Tau and circadian behavior.
Results
Expression of a phosphomimetic Tau mutant, but not wild-type Tau, in circadian clock neurons suppresses the strength of free running rhythms
To further investigate the effect of human Tau on circadian behavior, we backcrossed the relevant reagents for 6 generations into the iso31 background to rigorously control for genetic background. We assessed the effects of Tau-WT expression in all post-mitotic neurons using elavGAL4. We find that pan-neuronal expression of Tau-WT disrupts sleep, with a 200 min decrease in total daily sleep time compared with background controls (Supplementary Material, Table S1). However, we find that expression of background-controlled Tau-WT in all clock neurons using timGAL4 or in PDF clock neurons using pdfGAL4 did not decrease DD locomotor rhythmicity (Fig. 1, Table 1) or change sleep (Supplementary Material, Table S1).
TauE14 expression in circadian neurons disrupts free-running rhythms. Representative double plotted actograms for individual flies are shown over 4 days of 12 h light: 12 h dark conditions (LD) followed by 7 days of DD conditions. Average ± SEM of rhythmic power of locomotor behavior over 7 days constant darkness (DD) are shown above the actogram. ‘*’ indicates significant difference compared with both UAS/+ and GAL4/+ controls using one-way ANOVA followed by Bonferroni’s comparison test. See Table 1 for more details.
TauE14 expression in circadian neurons disrupts behavioral rhythms
| Genotype | Period | Rhythmic Power | N (DD) | LD Morning Index | LD Evening Index | DD Morning Index | N (LD) |
|---|---|---|---|---|---|---|---|
| UAS-Tau-WT/+ | 23.9 ± 0.0 | 82 ± 6 | 51 | 0.21 ± 0.02 | 0.28 ± 0.01 | 0.20 ± 0.03 | 51 |
| UAS-Tau14/+ | 23.7 ± 0.1 | 72 ± 6 | 42 | 0.16 ± 0.02 | 0.35 ± 0.01 | 0.15 ± 0.03 | 45 |
| tim-GAL4/+ (Tau-WT iso31 control) | 24.6 ± 0.1 | 71 ± 8 | 27 | 0.12 ± 0.02 | 0.25 ± 0.02 | 0.10 ± 0.04 | 28 |
| tim-GAL4/+ (TauE14 iso31 control) | 24.6 ± 0.1 | 76 ± 7 | 36 | 0.12 ± 0.02 | 0.33 ± 0.01 | 0.15 ± 0.02 | 38 |
| tim-GAL4/UAS-Tau-WT | 24.8 ± 0.1 | 75 ± 8 | 22 | 0.10 ± 0.02 | 0.27 ± 0.01 | 0.11 ± 0.04 | 25 |
| tim-GAL4/UAS-TauE14 | 24.02 ± 0.1 | 44 ± 5* | 31 | −0.01 ± 0.02 | 0.29 ± 0.02 | 0.04 ± 0.02* | 33 |
| Pdf-GAL4/+ (Tau-WT iso31 control) | 24.4 ± 0.1 | 73 ± 9 | 21 | 0.14 ± 0.01 | 0.22 ± 0.02 | 0.16 ± 0.02 | 24 |
| Pdf-GAL4/+ (TauE14 control) | 24.1 ± 0.0 | 80 ± 7 | 43 | 0.13 ± 0.01 | 0.27 ± 0.01 | 0.18 ± 0.02 | 46 |
| Pdf-GAL4/UAS Tau-WT | 24.7 ± 0.1 | 91 ± 7 | 26 | 0.15 ± 0.02 | 0.18 ± 0.01 | 0.18 ± 0.03 | 28 |
| Pdf-GAL4/UAS TauE14 | 23.8 ± 0.1 | 34 ± 4* | 37 | 0.09 ± 0.01 | 0.29 ± 0.01 | 0.14 ± 0.02 | 41 |
| Genotype | Period | Rhythmic Power | N (DD) | LD Morning Index | LD Evening Index | DD Morning Index | N (LD) |
|---|---|---|---|---|---|---|---|
| UAS-Tau-WT/+ | 23.9 ± 0.0 | 82 ± 6 | 51 | 0.21 ± 0.02 | 0.28 ± 0.01 | 0.20 ± 0.03 | 51 |
| UAS-Tau14/+ | 23.7 ± 0.1 | 72 ± 6 | 42 | 0.16 ± 0.02 | 0.35 ± 0.01 | 0.15 ± 0.03 | 45 |
| tim-GAL4/+ (Tau-WT iso31 control) | 24.6 ± 0.1 | 71 ± 8 | 27 | 0.12 ± 0.02 | 0.25 ± 0.02 | 0.10 ± 0.04 | 28 |
| tim-GAL4/+ (TauE14 iso31 control) | 24.6 ± 0.1 | 76 ± 7 | 36 | 0.12 ± 0.02 | 0.33 ± 0.01 | 0.15 ± 0.02 | 38 |
| tim-GAL4/UAS-Tau-WT | 24.8 ± 0.1 | 75 ± 8 | 22 | 0.10 ± 0.02 | 0.27 ± 0.01 | 0.11 ± 0.04 | 25 |
| tim-GAL4/UAS-TauE14 | 24.02 ± 0.1 | 44 ± 5* | 31 | −0.01 ± 0.02 | 0.29 ± 0.02 | 0.04 ± 0.02* | 33 |
| Pdf-GAL4/+ (Tau-WT iso31 control) | 24.4 ± 0.1 | 73 ± 9 | 21 | 0.14 ± 0.01 | 0.22 ± 0.02 | 0.16 ± 0.02 | 24 |
| Pdf-GAL4/+ (TauE14 control) | 24.1 ± 0.0 | 80 ± 7 | 43 | 0.13 ± 0.01 | 0.27 ± 0.01 | 0.18 ± 0.02 | 46 |
| Pdf-GAL4/UAS Tau-WT | 24.7 ± 0.1 | 91 ± 7 | 26 | 0.15 ± 0.02 | 0.18 ± 0.01 | 0.18 ± 0.03 | 28 |
| Pdf-GAL4/UAS TauE14 | 23.8 ± 0.1 | 34 ± 4* | 37 | 0.09 ± 0.01 | 0.29 ± 0.01 | 0.14 ± 0.02 | 41 |
Average ± SEM of period, locomotor rhythmicity (DD) and anticipation index (4 days of LD and first day of DD) from the genotypes indicated. ‘+’ indicates wild-type. UAS-TauWT/+ and UAS-TauE14/+ (lines 1 and 2) are controls in which there are no GAL4s to drive expression. Tim-GAL4/+ and Pdf-GAL4/+ controls (lines 3, 4, 7 and 8) are GAL4 only background controls for either Tau-WT or TauE14.
*Significant difference compared with both UAS/+ and GAL4/+ controls using one-way ANOVA followed by Bonferroni’s comparison test.
TauE14 expression in circadian neurons disrupts behavioral rhythms
| Genotype | Period | Rhythmic Power | N (DD) | LD Morning Index | LD Evening Index | DD Morning Index | N (LD) |
|---|---|---|---|---|---|---|---|
| UAS-Tau-WT/+ | 23.9 ± 0.0 | 82 ± 6 | 51 | 0.21 ± 0.02 | 0.28 ± 0.01 | 0.20 ± 0.03 | 51 |
| UAS-Tau14/+ | 23.7 ± 0.1 | 72 ± 6 | 42 | 0.16 ± 0.02 | 0.35 ± 0.01 | 0.15 ± 0.03 | 45 |
| tim-GAL4/+ (Tau-WT iso31 control) | 24.6 ± 0.1 | 71 ± 8 | 27 | 0.12 ± 0.02 | 0.25 ± 0.02 | 0.10 ± 0.04 | 28 |
| tim-GAL4/+ (TauE14 iso31 control) | 24.6 ± 0.1 | 76 ± 7 | 36 | 0.12 ± 0.02 | 0.33 ± 0.01 | 0.15 ± 0.02 | 38 |
| tim-GAL4/UAS-Tau-WT | 24.8 ± 0.1 | 75 ± 8 | 22 | 0.10 ± 0.02 | 0.27 ± 0.01 | 0.11 ± 0.04 | 25 |
| tim-GAL4/UAS-TauE14 | 24.02 ± 0.1 | 44 ± 5* | 31 | −0.01 ± 0.02 | 0.29 ± 0.02 | 0.04 ± 0.02* | 33 |
| Pdf-GAL4/+ (Tau-WT iso31 control) | 24.4 ± 0.1 | 73 ± 9 | 21 | 0.14 ± 0.01 | 0.22 ± 0.02 | 0.16 ± 0.02 | 24 |
| Pdf-GAL4/+ (TauE14 control) | 24.1 ± 0.0 | 80 ± 7 | 43 | 0.13 ± 0.01 | 0.27 ± 0.01 | 0.18 ± 0.02 | 46 |
| Pdf-GAL4/UAS Tau-WT | 24.7 ± 0.1 | 91 ± 7 | 26 | 0.15 ± 0.02 | 0.18 ± 0.01 | 0.18 ± 0.03 | 28 |
| Pdf-GAL4/UAS TauE14 | 23.8 ± 0.1 | 34 ± 4* | 37 | 0.09 ± 0.01 | 0.29 ± 0.01 | 0.14 ± 0.02 | 41 |
| Genotype | Period | Rhythmic Power | N (DD) | LD Morning Index | LD Evening Index | DD Morning Index | N (LD) |
|---|---|---|---|---|---|---|---|
| UAS-Tau-WT/+ | 23.9 ± 0.0 | 82 ± 6 | 51 | 0.21 ± 0.02 | 0.28 ± 0.01 | 0.20 ± 0.03 | 51 |
| UAS-Tau14/+ | 23.7 ± 0.1 | 72 ± 6 | 42 | 0.16 ± 0.02 | 0.35 ± 0.01 | 0.15 ± 0.03 | 45 |
| tim-GAL4/+ (Tau-WT iso31 control) | 24.6 ± 0.1 | 71 ± 8 | 27 | 0.12 ± 0.02 | 0.25 ± 0.02 | 0.10 ± 0.04 | 28 |
| tim-GAL4/+ (TauE14 iso31 control) | 24.6 ± 0.1 | 76 ± 7 | 36 | 0.12 ± 0.02 | 0.33 ± 0.01 | 0.15 ± 0.02 | 38 |
| tim-GAL4/UAS-Tau-WT | 24.8 ± 0.1 | 75 ± 8 | 22 | 0.10 ± 0.02 | 0.27 ± 0.01 | 0.11 ± 0.04 | 25 |
| tim-GAL4/UAS-TauE14 | 24.02 ± 0.1 | 44 ± 5* | 31 | −0.01 ± 0.02 | 0.29 ± 0.02 | 0.04 ± 0.02* | 33 |
| Pdf-GAL4/+ (Tau-WT iso31 control) | 24.4 ± 0.1 | 73 ± 9 | 21 | 0.14 ± 0.01 | 0.22 ± 0.02 | 0.16 ± 0.02 | 24 |
| Pdf-GAL4/+ (TauE14 control) | 24.1 ± 0.0 | 80 ± 7 | 43 | 0.13 ± 0.01 | 0.27 ± 0.01 | 0.18 ± 0.02 | 46 |
| Pdf-GAL4/UAS Tau-WT | 24.7 ± 0.1 | 91 ± 7 | 26 | 0.15 ± 0.02 | 0.18 ± 0.01 | 0.18 ± 0.03 | 28 |
| Pdf-GAL4/UAS TauE14 | 23.8 ± 0.1 | 34 ± 4* | 37 | 0.09 ± 0.01 | 0.29 ± 0.01 | 0.14 ± 0.02 | 41 |
Average ± SEM of period, locomotor rhythmicity (DD) and anticipation index (4 days of LD and first day of DD) from the genotypes indicated. ‘+’ indicates wild-type. UAS-TauWT/+ and UAS-TauE14/+ (lines 1 and 2) are controls in which there are no GAL4s to drive expression. Tim-GAL4/+ and Pdf-GAL4/+ controls (lines 3, 4, 7 and 8) are GAL4 only background controls for either Tau-WT or TauE14.
*Significant difference compared with both UAS/+ and GAL4/+ controls using one-way ANOVA followed by Bonferroni’s comparison test.
To address whether phosphorylated Tau may have stronger effects, we expressed the phosphomimetic TauE14 in circadian neurons using yeast galactose-induced transcription factor/upstream activating sequence (GAL4/UAS). TauE14 induces larger neurodegenerative phenotypes than those observed with Tau-WT expression in Drosophila (39,43). We find that expression of UAS-TauE14 either in all clock neurons (timGAL4) or only in the PDF-expressing LNv subset (PdfGAL4) substantially reduces free-running locomotor rhythmicity compared with controls in the same genetic background (iso31) (Fig. 1A and B, Table 1).
Phosphomimetic Tau expression in clock neurons reduces morning anticipation amplitude
We also assessed the effects of Tau-WT and TauE14 on morning and evening anticipation in light:dark conditions (LD). Consistent with effects observed in DD, neither expression of Tau-WT using timGAL4 nor PdfGAL4 was sufficient to change morning or evening anticipation. We find that phosphomimetic TauE14 expression using timGAL4 strongly reduces morning anticipatory behavior (Fig. 2A and B, left panels; Table 1). Loss of morning anticipation cannot be explained by a delay in the morning peak masked by lights-on as morning behavior is also not evident on the first day of DD (Fig. 2B right panels and Fig. 2C). Interestingly, PdfGAL4 driven expression of TauE14 does not induce a significant change in morning anticipation, suggesting that TauE14 may be working in other subsets of circadian neurons, such as the DN1p (22–25), to disrupt morning behavior.
TauE14 expression in all circadian neurons disrupts morning anticipation. (A) LD morning anticipation index for flies expressing UAS-TauE14 with tim-G4 (purple) or Pdf-G4 (pink) drivers compared with GAL4 only (black) and UAS only (gray) controls. n = 35–46. Statistical significance determined using one-way ANOVA followed by Bonferroni’s comparison test. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0005. Error bars show SEM. (B) Normalized activity profiles for the genotypes indicated, averaged over 4 days LD. Light phase is indicated by white bars; while dark phase is indicated by black bars. n = 31–43. Error bars show SEM. Arrows indicate approximate time of morning anticipation behavior with corresponding anticipation index value. (C) Morning anticipation index on DD Day 1 is indicated for male flies expressing U-TauE14 under tim-G4 (purple) or Pdf-G4 (pink) drivers compared with GAL4 only (black) and UAS only (gray) controls. n = 33–46. Statistical significance determined using one-way ANOVA followed by Bonferroni’s comparison test. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0005. Error bars show SEM.
Phosphomimetic Tau nearly eliminates PDF in the sLNv dorsal terminals without reducing sLNv number
Given the effects of TauE14 expression on behavioral rhythms, we asked whether the PDF+ LNv neurons remain intact. We co-expressed TauE14 and membrane Green Fluorescent Protein (GFP) (UAS-mGFP) using PdfGAL4 (Fig. 3A) to determine LNv cell number. We find that TauE14 expression does not change the number of GFP-labeled small or large LNvs (sLNv, lLNv) (Fig. 3B and C), indicating that the changes in circadian behavior are not due to cell death.
TauE14 expression disrupts morphology and induces a complete loss of PDF signal in sLNv dorsal projections. (A) Representative max projection images of whole mount Drosophila brains from Pdf-GAL4 UAS-mGFP/+ (PdfG4/+; top panels) and Pdf-GAL4 UAS-mGFP/UAS-TauE14 (PdfG4 > TauE14; bottom panels) labeled with anti-PDF (magenta), anti-GFP (green), anti-mammalian TAU (red) and GFP + PDF merged. Scale bar indicates 10 μm. For TAU labeling, non-specific labeling is detected in the optic lobe of PdfGAL4/+ control brains (no human Tau expression) but non-specific signal is not evident in LNv cell bodies or dorsal projections. (B and C) Number of GFP+ cells detected per hemisphere in 7-day-old Pdf-GAL4, UAS-mGFP (PdfG4/+) and UAS-TauE14; Pdf-GAL4, UAS-mGFP (pdf > TauE14) flies. (B) sLNv cell count, (C) large LNv (lLNv) cell count, n = 18–21 hemispheres. (D and E) Normalized mean intensity of PDF (D) and GFP (E) in the dorsal projections of Pdf-GAL4 UAS-mGFP control versus TauE14 flies. (F) GFP area of the dorsal sLNv projections (DP) of Pdf-GAL4 UAS-mGFP control versus TauE14 flies. n = 17–19, *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0005. Error bars show SEM. (G) Normalized mean intensity of PDF levels in sLNv cell bodies in PdfGAL4/+ (PdfG4/+) controls versus PdfGAL4/UAS-TauE14 (PdfG4- > TauE14) flies. (n = 36–43 cell bodies) *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0005. Error bars show SEM.
Next, we asked if TauE14 expression affects PDF neuropeptide expression, in particular in the dorsal projections of the sLNv, known to be critical for free-running rhythmicity (19,21,44). We observe a striking loss of PDF from the dorsal projections of the sLNv in PdfGAL4 UAS-TauE14 (Pdf > TauE14) flies, while the lLNv contralateral projections remain largely intact (Fig. 3A and D). Despite the near-complete loss of PDF signal in sLNv dorsal projections in Pdf > TauE14 flies, mGFP labeling demonstrates that these projections are still present (Fig. 3A, second panel) although with decreased intensity and area compared with control flies (Fig. 3E and F). In contrast to the prominent effect on PDF in sLNv projections, we observe a more modest decrease in PDF levels in both sLNV and lLNV cell bodies (Fig. 3G). Taken together these data indicate that TauE14 strongly reduces PDF neuropeptide levels in the sLNv dorsal projections with more modest effects on sLNv axonal morphology and cell body levels (Fig. 3A, E and F).
We next examined the effects of Tau-WT expression in Pdf-expressing neurons and found a 77% decrease in PDF levels in the sLNv dorsal projections, similar to but less severe than TauE14 (Fig. 4A and B). Unlike TauE14, we did not see a change in the area of the dorsal projection terminals using mGFP labeling (Fig. 4A, C and D) or a change in PDF and GFP levels in the LNv cell bodies (Fig. 4E and F). Taken together, our data indicate stronger behavioral and molecular effects of TauE14 expression compared with Tau-WT, consistent with previous studies (30,32,41,42). The phenotypes we observe are also quite distinct from those reported in Drosophila Tau (dTau) amorphic mutants, which retain free-running rhythmicity and PDF axonal expression, indicating that the effects of TauE14 and Tau-WT expression cannot be attributed solely to the disruption of endogenous dTau function (45). Nonetheless, the effects of Tau-WT are similar, though less severe, than TauE14.
Tau-WT expression decreases PDF signal in sLNv dorsal projections without changing terminal morphology. (A) Representative max projection images of whole mount Drosophila brains from Pdf-GAL4 UAS-mGFP/+ (PdfG4/+; top panels) and Pdf-GAL4 UAS-mGFP/UAS-Tau-WT (PdfG4 > TauWT; bottom panels) labeled with anti-PDF (magenta), anti-GFP (green) and GFP + PDF merged. Scale bar indicates 10 μm. (B and C) Normalized mean intensity of PDF (B) and GFP (C) in the dorsal projections of Pdf-GAL4 UAS-mGFP control versus Tau-WT flies. (D) GFP area of the dorsal sLNv projections (DP) of Pdf-GAL4 UAS-mGFP control versus TauE14 flies. n = 20–27 (E and F) Normalized mean intensity of PDF (E) and GFP (F) levels in sLNv cell bodies of Pdf-GAL4 UAS-mGFP control versus Tau-WT flies. (n = 17–38 cell bodies) *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0005. Error bars show SEM.
PER oscillations are intact in the small and large LNvs of phosphomimetic Tau flies
Given the effects of TauE14 expression on behavioral rhythms, we asked whether the core clock was disrupted in Pdf > TauE14 flies by assessing PER levels at peak circadian time (CT0) and trough (CT12) timepoints on DD Day 1 (Fig. 5A). We find comparable PER staining profiles in Pdf > TauE14 flies and controls, with robust differences between CT0 and CT12 PER levels in both Tau and control flies in both small and large LNvs (Fig. 5B and C). Thus, the molecular clock of LNv neurons is intact in Pdf > TauE14 flies despite disrupted rhythms.
No decrease in PER expression or oscillation upon TauE14 expression. (A) Representative whole mount Drosophila brains from Pdf-GAL4 UAS-mGFP/+ (PdfG4/+; top panels) and Pdf-GAL4 UAS-mGFP/UAS-TauE14 (PdfG4 > TauE14; bottom panels) on the first day of DD at CT0 and CT12 labeled with anti-PER (red). Normalized mean intensity of PER in the small (B) and large (C) LNv cell bodies of PdfGAL4/+ (PdfG4/+) controls versus PdfGAL4/UAS-TauE14 (PdfG4- > TauE14) flies. n = 57–64, *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0005. Error bars show SEM.
Phosphomimetic Tau reduces mitochondria in the sLNv dorsal terminals but increases them in cell bodies
Previous work has shown that mitochondrial structure and function are disrupted in Alzherimer’s disease (46–52). To determine the effects of TauE14 expression on mitochondria, we co-expressed the mitochondrial marker, mitoGFP, with TauE14 using PdfGAL4. MitoGFP is a GFP construct fused to the import sequence of the mitochondrial subunit VIII, which allows us to visualize mitochondrial localization in vivo (53). We find that mitochondria are undetectable in sLNv dorsal projections but are intact in lLNv contralateral projections (Fig. 6A), similar to our findings with PDF (Fig. 3A). Notably, we observe an increased mitoGFP signal in both sLNv and lLNv cell bodies (Fig. 6B–D). This suggests an impairment of mitochondrial transport from cell bodies into the axons, leading to mislocalization.
TauE14 expression induces loss of mitochondria in sLNv dorsal projections. (A) Representative max projection images taken at 20X magnification of whole mount Drosophila brains from Pdf-GAL4 UAS-mitoGFP/+ (PdfG4/+; top panels) and Pdf-GAL4 UAS-mitoGFP/UAS-TauE14 (PdfG4 > TauE14; bottom panels) labeled with anti-GFP (green) and anti-PDF (magenta). Note that non-specific mitoGFP staining is detected in both control and Tau brains, but in a different Z-plane than that of the projections. Insets display max projection mitoGFP images of the Z-planes with the dorsal projections only, without non-specific staining. Scale bar indicates 10 μm. (B) Representative 60X images of sLNv and lLNv cell bodies from the right hemisphere of panel A. (C and D) Normalized mean intensity of GFP in the small (C) and large (D) LNV cell bodies of Pdf-GAL4 UAS-mitoGFP flies with or without UAS-TauE14. (n = 70–75 cell bodies) *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0005. Error bars show SEM.
Discussion
Using a Drosophila model for human tauopathy, we find that Tau-dependent changes in circadian behavior are accompanied by profound reductions in the axonal localization of a key neuropeptide output in specific pacemaker neurons while sparing the core molecular clock. In addition, we find that phosphomimetic Tau also strongly impairs the axonal localization of mitochondria, consistent with the established role for Tau in microtubule-dependent transport. Taken together, our findings demonstrate in vivo molecular, cellular and behavioral effects of phospho-Tau in Drosophila that are observed in human tauopathies, validating the fly model and opening up the possibility of using high throughput genetic modifier screens to uncover novel mechanisms of Tau toxicity.
Using a novel tauopathy model, we show that expressing a phosphomimetic Tau construct (TauE14) in the PDF-positive LNv clock neurons results in a near-complete loss of PDF neuropeptide from axonal projections of the sLNv neurons. To our knowledge, such defects in neuropeptide localization have not previously been observed with Tau overexpression. Pdf > TauE14 flies also exhibit a substantial reduction in free running rhythmicity, yet LNv PER cycling remains intact in DD conditions. These findings suggest that the behavioral defects in PDF > TauE14 flies are due to the disruption in sLNv output as opposed to a core clock effect. Notably, in our hands, expression of Tau-WT in either all or the PDF-subset of clock neurons does not impair rhythms, yet Tau-WT expression in LNv neurons also strongly reduces axonal PDF levels (~75% reduction). This provides independent evidence for a role of Tau in PDF localization. These results are also consistent with other observations that flies with strong but partial disruptions of PDF and/or sLNv axonal morphology can still retain robust rhythmicity (17,19,54). These data indicate that the circadian clock network is robust to sizable axonal PDF reductions.
We observe a complete loss of mitochondrial marker (mitoGFP) expression from the small, but not large, LNv projections in TauE14 flies, similar to the pattern of PDF loss. mitoGFP expression is also increased in cell bodies of Pdf > TauE14 flies, consistent with a defect in mitochondrial transport (55). Overexpression of Tau in rodent and fly models has previously been shown to disrupt kinesin-dependent trafficking of vesicles and mitochondria, limiting their presence in the neuronal projections (56–59). Ectopic expression of human Tau in Drosophila results in decreased microtubule density and increased microtubule fragmentation (60). The loss of both PDF and mitochondria in axons upon TauE14 expression suggests a generalized defect in axonal transport in the sLNv. Neurons with mutant mitochondrial transport proteins show deficits in synaptic transmission and dysregulation of Ca2+ homeostasis (61–63). Further experiments are needed to determine if the mitochondrial loss in axons is contributing to the changes in rhythmic behavior and/or the decrease in PDF expression.
Our results suggest that TauE14 expression in clock neurons reduces morning anticipation. We find that TauE14 expression in all clock neurons (timGAL4), but not solely in the PDF-expressing LNv subset (PdfGAL4), decreases morning anticipatory behavior. DN1p clock neurons are known to play a role in promoting morning anticipation (20–24) and thus may contribute to this phenotype. The observation that TauE14 expression in all clock neurons disrupts morning, but not evening, anticipation behavior suggests selective vulnerability within different clock neuron groups, as functional disruptions of most or all clock neurons are associated with loss of both morning and evening behavior (20,26). Future experiments limiting TauE14 expression to specific clock neuron subsets would allow us to pinpoint the neurons most susceptible to Tau toxicity in the context of circadian behavioral disruption.
TauE14 appears to act differently than other neurotoxic proteins in the LNv such as mutant Huntingtin protein (Htt-Q128) (64,65). Expression of Htt-Q128 results in a substantial loss of PDF+ sLNv cell bodies (64,66), while these cells are intact in TauE14 brains. In contrast, Pdf > Htt-Q128 flies retain PDF signaling in the sLNv dorsal projections (67), whereas TauE14 causes a profound loss of PDF from dorsal terminals. Finally, Htt-Q128 results in decreased PER in cell bodies (65,67), but PER oscillations remain intact in the sLNv cell bodies upon TauE14 expression.
We find that TauE14 expression using Pdf-GAL4 disproportionally targets the sLNv, even though the GAL4 driver promotes expression in both small and large LNvs. This disproportionate effect is evident in the striking defects in PDF and mitochondrial-GFP distribution in sLNv axons (dorsal projections), whereas expression of these axonal proteins in the large LNv axons (medial/optic lobes) appears relatively intact (see Figs 3 and 6). Notably, selective effects on the sLNv have been observed previously upon expression of mutant Huntingtin protein using Pdf-GAL4 (64,66). Notably, substantial differences in gene expression have been reported between small and large LNvs, and this may contribute to the differential sensitivity of these groups to neurotoxic protein expression (69). However, we cannot exclude that the selective effects of TauE14 may reflect differential timing of Pdf-GAL4 expression during development of the small versus large LNv (68). The use of alternative GAL4 drivers and/or adult-induced expression of TauE14 will be required to distinguish between these possibilities.
Here, we have established a new model for Tau-dependent disruption of neuropeptide and mitochondrial localization, in which phosphomimetic TauE14 expression targeted to 16 LNvs is sufficient to disrupt in vivo behavioral rhythms, similar to disrupted rhythms observed in ad, FTD and other tauopathies. We propose that disrupted axonal transport of the neuropeptide PDF and mitochondria in sLNv neurons provides a molecular and cellular link between Tau overexpression and disruption in behavioral rhythms. Our findings reveal an important pre-death mechanism of cellular disruption in a neurodegenerative disease model, providing a window to screen for novel genes and potential therapeutic targets involved in disease pathogenesis.
Materials and Methods
Fly stocks
All fly stocks and crosses were maintained on standard food: 81.4 g cornmeal, 19.2 g yeast, 11.1 g soy flour, 6.42 g agar, 42.8 mL corn syrup, 42.8 mL molasses, 5.35 mL propionic acid and 13.4 mL ethanol and 1.43 g methylparaben per 1 L of food. Stocks were kept at room temperature, while crosses were kept in 25°C incubators under a 12 h light: 12 h dark cycle (LD) as previously described (66). UAS-Tau-WT(III) and UAS-TauE14(X) were obtained from Dr Mel Feany (29,30). UAS-mGFP and UAS-mitoGFP were obtained from the Bloomington Stock Center (BDSC# 5137, 8442, respectively). Pdf-Gal4(II) (18), tim-Gal4(II) (14) and elav-Gal4 c155 (70) have been described previously. UAS-Tau-WT, UAS-TauE14, Pdf-Gal4 and tim-Gal4 were backcrossed to the iso31 genetic background at least 6 times. Male flies were used for all experiments.
Locomotor activity recording and behavior data analysis
Adult male progeny 1–3 days post-eclosion (dpe) were loaded into glass tubes containing standard 5% sucrose 2% agar behavior food. Flies were monitored using the Drosophila activity monitoring system (Trikinetics) in an incubator running a 12 h light: 12 h dark cycle (LD) for 5 days, followed by 7 days of constant dark (DD). Sleep and anticipation data were collected in LD, discarding the first day after loading. Sleep data were analyzed with a custom MATLAB script (Sisobhan and Allada, unpublished) using 1 min bins, where the fly was considered asleep if there was no activity for 5 or more minutes. The total amount of sleep per 24 h, average bout length and number of sleep bouts were derived using the activity trace (number of infrared beam crossings per minute) for each fly, similar to (71). Sleep latency is calculated as the number of minutes until first sleep bout after lights turn off. Free-running rhythmicity data were collected over 7 days DD. Data collection, processing and circadian analyses were done as previously described (71–73). Rhythmic power was calculated from chi-squared periodogram data as power minus significance (P-S) as calculated by Clocklab (Actimetrics, Wilmette, IL). Activity actograms were plotted using ClockLab. Normalized group activity and sleep profiles were generated from light/dark (LD) conditions using a custom MATLAB-based analysis program (Sisobhan and Allada, unpublished). Morning and evening anticipation index was calculated similar to previous reports (25). The sample sizes indicated in the figures and tables represent the number of flies analyzed for each genotype combined from at least two independent experiments.
Whole mount immunostaining
Fly crosses and progeny were kept under 12:12 LD cycles at 25°C. Adult male progeny were used at 7–9 dpe. Brains were dissected at CT0-1 and CT12-13 for PER staining experiments, and within a Zeitgeber time (ZT) 0–2 time window for all other experiments. Staining was performed as previously described (66). The following primary antibodies were used in various combinations depending on experiment: mouse anti-PDF (1:800, Developmental Studies Hybridoma Bank, University of Iowa, Iowa City IA), rabbit anti-GFP (1:800, ThermoFisher Scientific, Waltham, MA), chicken anti-mammalian TAU (1:500, Neuromics, Edina, MN) and rabbit anti-PER [1:8000, Rosbash lab, Waltham, MA (74)]. The corresponding secondary antibodies were used as follows: donkey anti-mouse Alexa647 (1:1000, Invitrogen, Carlsbad, CA), goat anti-rabbit Alexa488 (1:1000, Invitrogen), goat anti-chicken Alexa594 (1:1000, Invitrogen) and donkey anti-rabbit Alexa594 (1:800, Invitrogen).
Confocal imaging and data quantification
Confocal imaging was performed as previously described (66). For cell body quantifications of GFP, PDF and PER, 60× magnification was used for imaging, and the intermediate stack of each cell was chosen for measuring the mean intensity; 40× magnification was used for imaging the projections. For quantification of dorsal projections, a threshold was applied to each channel in ImageJ (GFP threshold ranged from 20 to 30; PDF threshold ranged from 20 to 50). The same threshold was applied to all brains within the same experiment. The region of interest was drawn using the rectangular selection tool that included only the tip of the projection after the bend. All area measurements were taken in thresholded images. For GFP, mean intensity was calculated as the thresholded mean minus non-thresholded background mean. For PDF, mean intensity was calculated as the thresholded PDF mean divided by the thresholded GFP area minus the non-thresholded background mean in the PDF channel.
Statistical analysis
All statistical analyses were performed using GraphPad Prism [https://www.graphpad.com]. For comparisons with two groups, two-tailed t-tests were performed. For multiple comparisons, one-way ANOVAs were used with Bonferroni’s post-hoc correction.
Acknowledgements
We would like to thanks Dr Bob Vassar for comments on the manuscript.
Conflict of Interest statement. We have nothing to disclose.
Funding
National Institutes of Health (W81XWH2010211 to R.A., 2T32HL007909- 21 to M.Z., 5T32GM008152-33 to M.Z.); U.S. Army Medical Research and Materiel Command (1R21 NS110420-01 to R.A.); Alzheimer's Association (AARG-17- 53626 to R.A.).
