Brain preparation

For medulla imaging, brains were prepared in PBS according to Sugie et al. (2015, 2017) and fixed for 50 min in 4% paraformaldehyde. Brains were stained with mouse anti-Chaoptin (1:25; catalog #24B10, Developmental Studies Hybridoma Bank) and incubated overnight at 4°C in PBS plus 1% Triton X-100 (PBT) containing 0.1% BSA. After washing, secondary antibody (Ab) was incubated for 2 h at room temperature (Alexa 568 conjugated anti-mouse Ab, 1:400; Life Technologies). Brains were then mounted in Vectashield (Vector Laboratories), with the posterior side up, after a second washing step. Insect pins of 0.1 mm diameter (Entomoravia) were used as spacers to keep the brain in its original shape. To visualize the axons in their extension in the medulla (Fig. 1B,D; see Fig. 10), brains were oriented ventral side up and imaged in a ventral to dorsal orientation (Sugie et al., 2017). Confocal microscopy was performed with an Olympus FV3000 (see Figs. 7–10) or with a Zeiss LSM 780 (Fig. 1; see Figs. 3, ​,5,5, ​,6).6). Images were processed using Imaris software (Bitplane) and Fiji software (Schindelin et al., 2012).

Open in a separate window

Figure 1.

Quantitative analysis of sporadic axonal degeneration in Drosophila R7 photoreceptors. A, Scheme of the adult fly visual system. The retinal layer (blue) contains the cell bodies of the PRCs, which are organized in ~750–800 columnar units (ommatidia) all containing one PRC of each type (R1–R8). R1–R6 (gray columns) project their axons (data not shown) to the lamina (orange). R7 and R8 PRCs (green) project their axons to the medulla (green); R7 termini (black) are found in medulla layer M6, whereas R8 terminates in M3 (data not shown; Fischbach and Dittrich, 1989). B, C, Three-dimensional image stacks of the adult fly brain (gray) depicting the positions of the medulla (green) and the lobula and lobula plate (yellow) in a top (B) or posterior (C) view (left part of the brain) as well as a z projection of R7 axons in a top view (B) or the R7 termini in the posterior view (C; right part of the brain). D, The z-projection of tubulinGFP-labeled R7 and R8 axons imaged in a top brain view. R7 axons terminate in M6 (arrowhead), whereas R8 axons terminate in M3 (arrowhead). E, TubulinGFP-labeled R7 termini in a posterior brain view after extraction from a 3D brain reconstruction. The genotypes in this and all the rest of the figures are included in Table 1. F–H, Identification of R7 axon termini in the M6 medulla layer. Three-dimensional projection of tubulinGFP-labeled R7 and R8 axons in the medulla in a posterior brain view before extraction of R7 termini (F), 3D reconstruction of created mask (red surface) to extract R7 termini in M6 (G), and R7 termini after extraction (H). H', Close-up of R7 axonal termini in the medulla of flies expressing UAS-tubulinGFP driven by GMR-Gal4. The organization of medullar R7 termini in rows (lines) facilitates the detection of missing axons upon prolonged light exposure (arrowheads). I, I', A spot is assigned to each R7 axonal terminus (yellow). The total number of dots represents the amount of R7 axon termini in the medulla. Scale bars: F–I, 10 µm; H', I', 5 µm.

Open in a separate window

Figure 5.

Reversibility of light-induced axonal damage in R7 photoreceptors. A–K, Close-ups of medullas showing axonal termini of R7 in flies expressing UAS-tubulinGFP (green) driven by GMR-Gal4 exposed to various light and dark cycles (L). Arrowheads point at missing termini. Lines are drawn between rows of axonal termini. Scale bar, A–K, 10 µm. A'–K', Corresponding full medullas are depicted. Squares highlight regions of individual close-ups. Scale bar, A'–K', 50 µm. We considered 1dLD as a nondegenerated condition (A, A'), whereas 11dLL was chosen as a condition of strong axonal loss (B, B'). L, Scheme of the various illumination conditions used in this experiment. Development proceeded in LD. At eclosion, adult flies were exposed to LL for various time periods (white). A fraction of flies was put back to DD (black) or in an LD cycle (black/white). M, TubulinGFP-positive R7 termini numbers in the medulla of flies exposed to various light and dark cycles. Error bars indicate mean ± SD (ns, not significant, **** P < 0.0001).

Open in a separate window

Figure 6.

Apoptosis is not associated with axonal degeneration. A, B, Eyes of flies kept for 13 d in an LD cycle (A) or for 13 d in LL (B). Scale bar, 250 µm. C–E', Optical cross-sections through the retina of adult flies exposed to 13 d of either control (13dLD in C–C'') or constant light conditions (13dLL in D–D”). Eyes were stained with FITC-Phalloidin to highlight the actin-enriched rhabdomeres (green) and DAPI to mark the nuclei (magenta). TUNEL labeling (white) showed no signal except in the positive control, in which 13dLD eyes (E–E”) were submitted to a DNase treatment before TUNEL detection. Scale bar, 10 µm. F–K, Optical cross-sections through the retina of GMRwhite RNAi/w;GMR-Gal4/+;UAS-tubulinGFP/+ adult flies exposed to control (LD) or LL conditions for 5 (F, G) or 7 d (I, J) or exposed to 5 or 7 d of constant light followed by 2 d of constant darkness (H, K). Eyes were stained with FITC-Phalloidin to highlight the actin-enriched rhabdomeres (green). Arrowheads point to R7 (G, J). Scale bar, 10 µm.

Open in a separate window

Figure 7.

Expression of the caspase inhibitor p35 does not suppress axonal degeneration. A–J, Axonal termini of R7 in the medulla of control (A–E) or of flies expressing p35 (F–J) exposed for 1, 5, 9, and 13 d to LL or to control 13dLD. R7 axonal termini are revealed by UAS-tubulinGFP expression (green). Scale bar, 50 µm. K, TubulinGFP-positive R7 termini counts in the medulla of flies expressing p35 or of control flies exposed for various times to LL or to control 13dLD. Error bars indicate mean ± SD (ns, not significant, *** P < 0.001, **** P < 0.0001).

Open in a separate window

Figure 10.

Axonal integrity of R7 is preserved upon loss of its synaptic function. A–A”, B–B”, R7 axons in the medulla imaged in a dorsal to ventral orientation showing the presence of GRASP-positive synapses (green in A, A,”; arrowheads in A) and intact axons labeled with chaoptin (24B10) in magenta (A', A”) in a 7dLD control condition. Upon constant light exposure, the GRASP signal disappears in axonal termini after 7 d (green in B, B”), whereas intact axons are still present (24B10, magenta in B', B”). Scale bars, 5 µm.

Axonal termini counts

To quantify the number of R7 axonal termini in medullas, confocal z-stacks of ~100 µm in depth were acquired with a 40× objective (Zeiss LSM 780) or a 60× objective (Olympus FV3000). The z-stack interval was fixed to 1 µm, and the following settings were used: 1 A.U., 1024 frame with LSM 780 or 512 frame with FV 3000, 8 bit with LSM 780 or 16 bit with FV3000. We then reconstructed the stacks in 3D using the Imaris 9.7.2 software. Axonal termini of R7 PRCs project into the M6 medulla layer, which can be masked by using the Imaris surface function. To create a perfect mask, we navigated slide by slide through the 3D-reconstructed medulla in Imaris and manually marked the M6 region containing only R7 axonal termini in every third slide of the reconstructed z-stack. We then used the surface function to recreate a compact 3D mask of R7 termini based on the manually drawn regions. In an accompanying study, we developed a software enabling automated mask creation, which was partially used in this study and is described in detail in Nitta et al. (2021). After mask creation, we isolated the signal of the mask using the mask channel function. In this step, we could visualize only the R7 termini by selecting the masked channel in the display adjustment tool. We subsequently aimed to count the number of individual axonal termini by using the spot function. To do so, we added four image processing steps into our protocol. We selected gamma correction values to 1.4, and threshold cutoff values were set up individually for each image, as half of the automatically determined peak of intensity. In addition, we chose the background subtraction function with values suggested by the program, and the Gaussian filter was selected as 0.6 µm. These processing steps allowed us to visualize axonal termini in better quality, and thus facilitate the detection of termini by the spot function.

By using the estimated XY diameter function, we set the size of a terminus to 2 µm (based on healthy axonal size in LD). Additionally, we aimed at avoiding the presence of off-targets and thus used the background subtraction function while keeping the quality thresholds manual. For images obtained with the LSM 780 confocal microscope, the quality threshold was determined to be 2 for tubulinGFP signals and 1 for 24B10-positive axons; for images obtained with the Olympus confocal setup, the quality thresholds were adapted to 300 for 24B10 and 150 for the GFP signal. To provide highly accurate counts and detect early axonal loss, which consists of only 5–10% of the total amount of axonal termini (see Fig. 3J,K), we used the highly organized structure of medulla and took advantage of its axons organized in lines and rows to verify that the assigned dots belonged to axonal termini (Fig. 1H',I'). Off-targets were removed manually from the final count.

Retinal staining/TUNEL assay

Eyes were removed from heads of GMRwhite RNAi/w;GMR-Gal4/+;UAS-tubulinGFP/+ female flies in PBS and fixed in 4% paraformaldehyde overnight at 4°C. After washing in 0.1% PBT, eyes were incubated in 50 mg/ml sodium borohydride (Sigma-Aldrich) in PBS for 20 min to remove pigments and subsequently washed in PBS and PBT. The cornea was removed and retinas were stained overnight at 4°C with Phalloidin-iFluor 488 (1:1000; Biomol) in 0.1% PBT BSA. On the next day, retinas were washed with PBT and PBS and embedded directly in Fluoromount (SouthernBiotech) or further processed for TUNEL labeling. TUNEL staining was performed according to manufacturer instructions (In Situ Cell Death Detection Kit TMR red, Roche). For the positive control, retinas were incubated with 300 U/ml DNase I in Tris, pH 7.5, and 1 mg/ml BSA for 10 min at 25°C before TUNEL labeling. Retinas were washed in PBS and mounted with Fluoromount containing DAPI (SouthernBiotech).

Eye pigment measurements

For brown (ommochrome) eye pigment measurements, we used a protocol adapted from Mackenzie et al. (1999). Fifty heads of 7–10-d-old female flies were homogenized in 150 μl of 2 m HCl and 0.66% sodium metabisulfite (w/v, Sigma-Aldrich). Two hundred microliters of 1-butanol was added, and the mixture was placed on an orbital shaker at 150 rpm for 30 min before being centrifuged at 9000 × g for 5 min. The organic layer was removed and washed with 150 μl of 0.66% sodium metabisulfite in dH₂O and placed back on the orbital shaker for a further 30 min. The organic layer was removed and measured for absorbance at 492 nm. Absorbance was determined with a Nanodrop One C spectrophotometer (Thermo Fischer Scientific). Adult eye pictures were obtained with a Canon EOS 700D camera mounted on a Leica S8AP0 binocular and processed in Fiji (Schindelin et al., 2012).

Circadian rhythm and sleep behavioral assay

The behavioral trials were conducted using the procedure described previously (Fogg et al., 2014; Yoshii et al., 2015), with some modifications. A DAM5 Drosophila activity monitor system (TriKinetics) was used to record locomotor activity in 1 min bins. Individual adult male flies of the genotype GMRwhite RNAi/Y;GMR-Gal4 UAS-tubulinGFP/+ from 1 to 4 d old were transferred to recording tubes containing 5% sucrose in 0.9% agar on one end. For LL experiments, male and female flies were entrained to a 12 h LD cycle (light, 4000 lux) at 25°C for 3 d. Subsequently, test flies were subjected to LL conditions for 10 d and control flies to LD cycles for 10 d. Activity recordings were analyzed using ActogramJ software (Schmid et al., 2011). The sleep was defined as previously described (Huber et al., 2004).

Reversibility of light-dependent axonal damage in R7 photoreceptors

We sought to identify the time point at which axonal degeneration was initiated. Five days of constant light exposure (5dLL) did not lead to R7 axonal loss in the medulla, but R7 axons underwent morphologic changes, becoming enlarged (Fig. 3D–D',J, K, M–M”). We wondered whether axons at 5dLL are already damaged and whether a fraction of them is potentially committed to initiate degenerative processes. We thus placed those flies back into the dark (5dLL + 6dDD) or exposed them to a light/dark cycle (5dLL + 6dLD) and examined them at day 11 (Fig. 5L,C–E',M). Counts of tubulinGFP-positive termini in the medullas of 5dLL, 5dLL + 6dDD, or 5dLL + 6dLD flies were not distinguishable from those of control animals (Fig. 5M). This is in strong contrast to the widespread axonal degeneration observed in 11dLL flies (Fig. 5B,B',M). Thus, the light-induced changes that had taken place in the R7 axons at 5dLL are reversible, and the axons are not yet terminally committed to degenerate. Interestingly, the swelling presented by the axon termini in flies exposed to 5dLL (Fig. 5C,C') was reversed in the 5dLL + 6dDD condition, implying that axon swelling is a reversible modification at this stage (Fig. 5D,D'). Axon swelling, however, was still present in the medulla of 5dLL + 6dLD, indicating that prolonged D phases are needed for the swelling to disappear (Fig. 5E,E').

At day 7 of continuous light exposure (7dLL), first individual axons started to lose the sensitive tubulinGFP signal (Fig. 5F,F',M). This situation was not worsened 4 d later if the flies were kept in the dark (7dLL+ 4dDD), suggesting that no additional axons had been terminally committed to degenerate at this stage (Fig. 5G,G',M). Axon swelling of remaining axons was also reduced in this condition. Nonetheless, if the 7dLL flies were placed in a light/dark cycle, progressive axonal degeneration became apparent (Fig. 5H,H',M). The final outcome was intermediate between that of 9dLL exposure and 11dLL exposure (Fig. 5M). This observation suggests that at 7dLL some axons have accumulated reversible damage and that further light exposure is necessary to drive these axons into degeneration.

After 9 d of continuous light exposure (9dLL), the medulla was obviously damaged, with many missing axons (Fig. 5I,I',M). Putting the flies back in the dark for 2 d (9dLL+ 2dDD) failed to slow down the progressive axonal degeneration, and medullas of these flies were undistinguishable from those of flies exposed to 11dLL (Fig. 5J,J',B,B',M) or to those of flies exposed for 9dLL+ 2dLD (Fig. 5K,K',M). Thus, at day 9 of light exposure, axon damage may be irreversible, and a fraction of axons might be already committed to degeneration.

Together, these data suggest that R7 axons slowly accumulate reversible damage as a result of constant activation. Axons individually reach a level at which the accumulated damage initiates degeneration, a point of no return.

Apoptosis is not associated with axonal degeneration

Axonal degeneration is an early event in NDs and precedes cell death (Cavanagh, 1979; Adalbert and Coleman, 2013; Neukomm and Freeman, 2014; Tagliaferro and Burke, 2016). To monitor whether axonal degeneration of R7 photoreceptor cells was a consequence of cell death or was preceding it in our model system, we first monitored the cornea. Eyes become rough in conditions in which individual PRCs are missing or when the number, arrangement, or identity of PRCs is modified (Tomlinson et al., 1987, 1988; Basler et al., 1991; Van Vactor et al., 1991). Flies exposed to 13 d of constant light (13dLL) did not show any detectable defects in corneal morphology (Fig. 6B), indicating that widespread cell death was not taking place at this stage. To address specifically whether individual PRCs might be undergoing apoptosis, we exposed flies to LD or LL treatments for various periods of time and performed TUNEL staining to detect apoptotic nuclei within the retinas (Fig. 6C–E”). Even after 13 d in constant light, we did not detect TUNEL-positive PRC nuclei (Fig. 6D”), suggesting that apoptotic cell death had not started at a time point at which axonal degeneration was at full pace (Fig. 3H,H',J,K; Nitta et al., 2021). Nuclear morphology was similar between controls and light-exposed flies (Fig. 6C',D'). We observed a reported remodeling of the actin-rich rhabdomeres starting after 5 d of light exposure (Fig. 6G). Interestingly, R7 rhabdomeres seemed to be less affected than those of R1–R6 by this light-dependent actin remodeling after 5 or 7 d in constant light (Fig. 6G,J). This type of actin redistribution in photoreceptors was already observed after short-term light exposure and shown to be reversible in darkness (Kosloff et al., 2003). We thus tested whether we could reverse this phenotype by putting light-exposed flies (5dLL or 7dLL) back in darkness. A large majority of rhabdomeres recovered after 2 d in complete darkness (Fig. 6H,K), thus suggesting that these changes at the rhabdomeres are not indicative of a permanent cell damage.

To independently validate that apoptosis was not involved in light-induced axonal degeneration during the time frame of our experiments, we expressed in PRCs the baculovirus caspase inhibitor p35 that suppresses apoptosis and analyzed axonal termini numbers after 1, 5, 9, and 13 d in LL (Fig. 7A–D, F–I; Zhou et al., 1997). Expression of p35 did not modify termini numbers after 13 d in a light/dark control condition (Fig. 7J,K). Importantly, it did not rescue axonal degeneration at 9 or 13 d in constant light (Fig. 7H,I,K), indicating that apoptosis is not associated to the initial steps of axonal degeneration.

Neurotransmission is involved in axonal degeneration

Recent evidence indicates that synaptic failure in AD is at least partly induced by neuronal hyperactivity in the early stages of the disease, and this mechanism could also be involved in developing a late-onset autosomal dominant inherited form of PD (Nuriel et al., 2017; Hector and Brouillette, 2020; Lucumi Moreno et al., 2021). We thus tested whether activity of PRCs is linked to the axonal degeneration described in our model. First, we tested whether we could induce neurodegeneration in complete darkness by acutely activating R7 using the heat-sensitive Drosophila TrpA1 channel (Hamada et al., 2008; Pulver et al., 2009). Indeed, activation of photoreceptors by TrpA1 at 29°C was sufficient to induce axonal degeneration in the dark (Fig. 8B), starting at 11 d after shifting to the restrictive temperature (Fig. 8C). Although TrpA1 might activate R7 at nonphysiological levels compared with light, these data indicate that activation of PRCs can induce axonal degeneration. As a next step, we expressed in R7 the temperature-sensitive transgene UAS-shibire^ts (UAS-shi^ts), which prevents endocytosis at temperatures exceeding 29°C, thus rapidly stopping synaptic vesicle recycling and neurotransmitter release (Kitamoto, 2001). To avoid developmental defects caused by the leaky expression of UAS-shi^ts, we included a tubGal80^ts transgene in the experimental genotype (McGuire et al., 2003). Flies thus developed at the permissive temperature and were shifted to the restrictive temperature at eclosion, allowing for UAS-shi^ts expression and dominant negative activity only at adult stages (see above, Materials and Methods). In contrast to the degeneration observed in control flies (Fig. 8D,F), flies expressing UAS-shi^ts in their PRCs and shifted to the restrictive temperature did not show axonal degeneration during prolonged light exposure, even after 13 d of constant light exposure (Fig. 8E,F).

Because shi^ts impairs endocytic function, we sought to pinpoint more specifically the involvement of neurotransmission and to rule out potential additional effects of hindering endocytosis by Shi^ts. We thus blocked the postsynaptic response to PRC activation, leaving PRC activity intact. Fly photoreceptors are inhibitory and release histamine, which activates the Ort histamine chloride channel promoting hyperpolarization of R7 postsynaptic neurons in the medulla (Gisselmann et al., 2002; Gao et al., 2008; Pantazis et al., 2008; Liu and Wilson, 2013; Schnaitmann et al., 2018). However, we did not succeed in establishing a line containing an ort mutant (Gengs et al., 2002) in the GMRwhite RNAi background. As an alternative, we therefore examined a transallelic ort¹/ort⁵ combination in white mutant flies. Exposure of adult white flies to constant light gave rise to a very strong degeneration after only 5 d of exposure, as detected by 24B10 staining (Fig. 8G,I). With this experiment, we thus confirmed previous results obtained in the retina, showing that pigments protect photoreceptor cell bodies from excessive exposure to light (Lee and Montell, 2004; Bulgakova et al., 2010; Ferreiro et al., 2017). The presence of ort¹/ort⁵ protected flies against axonal degeneration, in fact decreasing the loss of axons after 5 and 7 d of constant light exposure (Fig. 8H,I). A similar protection was obtained in homozygous ort⁵ mutant flies (24B10-positive termini at 5dLL in ort⁵ 626.00 ± 47.58, n = 19 vs white 282.40 ± 68.91, n = 19, unpaired t test with Welch's correction, ****p < 0.0001). It is conceivable that in the white mutant background the mechanisms of degeneration might involve additional components than those acting in pigmented eyes, and additional experiments will be required to clarify this. Nonetheless, together with the shi^ts rescue, this result suggests that involvement of postsynaptic partners is required to initiate R7 axon degeneration. It further suggests that prolonged hyperpolarization of the postsynaptic neurons represents an important signal for the initiation of axonal degeneration in R7.