To live-image RGC mitochondria within axons, we first intravitreally injected Mitotracker Deep Red into Nieuwkoop and Faber, 1994 stage 48 (10 days post-fertilization) X. laevis tadpoles, animals whose RGC axons were readily visible due to expression of an RGC-specific Tg(Isl2b:GFP) transgene (Figure 1A). After allowing the dye to spread down the axons, the optic nerves of anesthetized and immobilized animals were imaged using a spinning disc confocal microscope (Figure 1B, and Figure 1—video 1, top) at 1 Hz for 1 min, to measure axonal mitochondria movement through kymographs (Figure 1C, D). Since the morphology and movement of mitochondria labeled by Mitotracker were similar to those labeled by an RGC-specific Tg(Tom20-mCherry) transgene (Figure 1—video 1, bottom), intravitreal Mitotracker injection was deemed suitable for labeling RGC mitochondria in the optic nerve. To determine the stability of Mitotracker labeling over time, optic nerves were imaged 3.5 or 18 hr after intravitreal Mitotracker injection. Since the distribution (Figure 1—figure supplement 1A) and velocity (Figure 1—figure supplement 1B) of stationary, anterogradely moving, and retrogradely moving mitochondria were similar at both time points, all subsequent imaging experiments were carried out 15–18 hr after Mitotracker intravitreal injection. The kymographs of axonal RGC mitochondria were used to determine the fraction of mitochondria that were stationary, defined here as moving less than 0.1 μm/s, as opposed to moving anterogradely and retrogradely (Figure 1C, D). In an analysis of over 500 axonal mitochondria derived from 6 animals, we found that stationary mitochondria account for nearly half of the total axonal mitochondria, whereas both anterogradely and retrogradely transported mitochondria each contribute a quarter of the total (Figure 1E), consistent with other studies of axonal mitochondria (Zheng et al., 2019; Fellows et al., 2020; Suh et al., 2021). Among the motile mitochondria, the average speed was 0.63 and 0.78 μm/s for anterogradely and retrogradely moving mitochondria, respectively (Figure 1F), which is within the range of the values previously measured by others (Morris and Hollenbeck, 1993; Ligon and Steward, 2000; De Vos et al., 2003; Miller and Sheetz, 2004; Jiménez-Mateos et al., 2006; Pilling et al., 2006; Mironov, 2007; Misgeld et al., 2007; Kang et al., 2008; Wang and Schwarz, 2009; MacAskill and Kittler, 2010; Chang et al., 2011; Chen et al., 2016; Niescier et al., 2016). Thus, the X. laevis tadpole optic nerve was deemed a suitable model to study RGC axonal mitochondria.

Next, to test whether the Mitotracker intravitreal injection itself might affect mitochondria movement within the optic nerve, we crossed Tg(Isl2b:GFP) and Tg(Isl2b:Tom20-mCherry) animals and obtained doubly transgenic animals which express both a mitochondria-targeted mCherry fluorescent reporter as well as a cytoplasmic GFP transgene, both specifically in RGCs. Mitotracker Deep Red was injected into Tg(Isl2b:GFP) expressing tadpoles at NF stage 48, into both animals with or without expression of the Isl2b:Tom20-mCherry transgene, and then the optic nerves were live-imaged as described above within 1 day of the intravitreal Mitotracker injection. The Mitotracker-based quantifications showed no significant difference in mitochondria movement between the two different Mitotracker-injected groups (with and without presence of the mitochondria-targeted transgene), whereas the Tom20-based quantifications showed that the Mitotracker intravitreal injection does have a small but significant effect on slowing retrograde transport and increasing the fraction of stopped mitochondria at the expense of anterograde moving mitochondria (Figure 1—figure supplement 1F, E, respectively). Because of this effect, in all subsequent experiments, animals with or without intravitreal injection were compared only to other animals similarly treated.

Mutations in OPTN can lead to impaired or aberrant mitophagy and the accumulation of defective mitochondria (Wong and Holzbaur, 2014a; Lazarou et al., 2015; Shim et al., 2016; Evans and Holzbaur, 2020). To test whether OPTN mutations might affect the behavior of RGC mitochondria and components of the mitophagic machinery, OPTN and LC3b, specifically within axons, we generated transgenic X. laevis in which fluorescently tagged LC3b and variants of OPTN could be expressed selectively in RGCs by induction with doxycycline (Figure 2A, B). In these large transgene constructs, regulatory regions from the zebrafish Isl2b gene (Pittman et al., 2008) which drive RGC-specific expression in the X. laevis retina (Watson et al., 2012), were here used to drive a version of the reverse tetracycline trans-activator, rtTA2, that is highly efficient in X. laevis (Das and Brown, 2004; Mills et al., 2015) in plasmids that also contain a bidirectional tetracycline operator that expresses EGFP-LC3b in one direction and in the other mCherry-tagged versions of the following human OPTN variants: Wt, glaucoma-associated mutations E50K, M98K, and H486R, ALS-associated mutation E478G, and two synthetic mutations, F178A and D474N, previously shown to be defective in the OPTN interaction with LC3b and ubiquitinated mitochondria, respectively (Wild et al., 2011; Korac et al., 2013; Wong and Holzbaur, 2014a; Sirohi et al., 2015; Shim et al., 2016; Li et al., 2018; Liu et al., 2018; Chernyshova et al., 2019; Padman et al., 2019). The optic nerves of 4–6 transgenic tadpoles per transgene were imaged using the spinning-disc confocal first as a 1-min single plane t-series acquired at 1 Hz, and then as a z-scan imaging the full thickness of the optic nerve at 1 µm steps. In the case of animals expressing E50K and Wt OPTNs, analyses were done with and without previous labeling of RGC mitochondria through intravitreal injection of Mitotracker (Figure 2C, D); in the case of all other OPTN variants, analyses of OPTN and LC3b were carried only without Mitotracker labeling. Kymograph analyses of the t-series were used to quantify the percentage of each movement class (stationary, anterograde, and retrograde) and determine the average speed of anterograde and retrograde movement. For the live-imaging and kymograph analyses of OPTN and LC3b, in the absence of Mitotracker intravitreal injection, analyses were carried out in both primary transgenic F0 animals (Figure 2E, E’, Figure 2—figure supplement 1C, D) as well as in their F1 progeny (Figure 2—figure supplement 1A, B, E, F), to minimize the possibilities of transgene copy number or integration position effects confounding the interpretation. Such analyses in F0s are possible in X. laevis because the restriction enzyme-mediated integration transgenesis method used to generate the animals (Kroll and Amaya, 1996) involves transgene integration prior to the first cell division, and thus provides fully transgenic rather than mosaic animals. The results from F0 transgenic founders and F1 progeny were similar and showed that relative to Wt OPTN, all OPTN mutants with the exception of the LC3b-binding-defective F178A increased the fraction of stopped OPTN or LC3b in either F0 or F1 experiments. However, it was only the glaucoma-associated E50K and M98K mutations that increased stopped OPTN in both F0 and F1 analyses, and only the E50K mutation that increased both OPTN and LC3b in both F0 and F1 analyses. Analyses of anterograde and retrograde velocity for OPTN and LC3b in both F0s (Figure 2—figure supplement 1C, D) and F1 animals (Figure 2—figure supplement 1E, F) were generally consistent with one another. Overall, there were no consistent significant changes in velocity in both F0 and F1 analyses, suggesting the increases in stopped OPTN and LC3b were specific effects and not secondary to making the RGCs or their axons unhealthy. To determine whether the observed increases in stopped OPTN and LC3b might represent an increase in axonal mitophagy, the studies of OPTN and LC3b movement were repeated, but now 15–18 hr after intravitreal injection of Mitotracker (Figure 2F); these studies were carried just in F1 animals expressing Wt or E50K OPTN, as well as in animals carrying no transgenes at all. Compared to the non-transgenic control animals (only Mitotracker-injected), those with expression of Wt OPTN had no alteration in the fraction of stopped mitochondria, similar to what had been previously observed after expression of the Tom20-mCherry transgene (Figure 1—figure supplement 1E), though they did have a small effect on the velocity of retrogradely transported mitochondria (Figure 2—figure supplement 1G); this effect on retrograde velocity was similar after both Wt and E50K OPTN expression. In contrast, expression of E50K OPTN uniquely resulted in a significant increase in the fraction of stopped mitochondria, above the already high baseline amount (Figure 2F), without affecting retrograde transport any further (Figure 2—figure supplement 1G). Of note, as was the case for OPTN and LC3b in the absence of intravitreal Mitotracker injection, expression of neither Wt nor E50K OPTN affected the relative balance of retrograde versus anterograde movement of mitochondria, OPTN, or LC3b but rather only affected the fraction of the stopped populations. Consistent with the previous results showing that intravitreal injection of Mitotracker had a small but significant effect on mitochondria movement (Figure 1—figure supplement 1E, F), here too we found that the intravitreal injection of Mitotracker affected the behavior of axonal OPTN, including an increase in the fraction of stopped OPTN; however, so did animals receiving an injection of just the biological salts solvent, showing that it is the eye injection procedure rather than the Mitotracker itself that modestly perturbs the system (Figure 2—figure supplement 1H).

Next, to determine whether the increases in stopped mitochondria and stopped OPTN might be linked, the degree of mitochondria and OPTN co-localization was examined. In the nerves labeled by Mitotracker, visual inspection of the raw images (Figure 2C) and the derived kymographs (Figure 2D) was suggestive that OPTN and the Mitotracker-labeled mitochondria might co-localize, particularly in the stopped populations, and more so in the animals expressing E50K OPTN. To examine the degree of mitochondria and OPTN co-localization quantitatively, we re-analyzed the previous kymographs of mitochondria and OPTN and separately quantified mitochondria that co-localized, with OPTN (mito-OPTN) versus mitochondria not associated with OPTN (mito-ONLY) (Figure 2G, Figure 2—figure supplement 1I). Strikingly, while mitochondria not co-localizing with OPTN had relatively balanced populations of stopped, and anterogradely and retrogradely moving, the majority of mitochondria co-localizing with OPTN were stopped, with the remainder being largely engaged in retrograde transport; the same was true in animals expressing Wt OPTN (Figure 2G) and E50K OPTN (Figure 2—figure supplement 1I). The glaucoma-associated E50K OPTN also significantly increased the fraction of mitochondria co-localizing with OPTN, and this increased co-localization was observed not only in the stationary but also the two moving populations (Figure 2H). In summary, expression of OPTN versions carrying glaucoma and ALS-associated mutations, as well as a synthetic mutation in OPTN that perturbs the recognition of ubiquitinated mitochondria, but not a synthetic mutation that perturbs the association of OPTN with LC3b, all had or trended to increases in stopped axonal OPTN and LC3b in RGCs, and the two that had the largest and most consistent effects were two glaucoma-associated mutations, E50K and M98K, with E50K having an increased co-localization with mitochondria and leading to a greater fraction of stopped mitochondria within the optic nerve.

Since OPTN associates with damaged mitochondria when mediating mitophagy, we asked whether some of the stopped mitochondria co-localizing with OPTN measured in our kymographs might be mitochondria undergoing mitophagy. If so, inhibiting autophagosome-lysosome fusion with chloroquine would be expected to affect the number of these stopped mitochondria. Surprisingly, as measured by the Tom20-mCherry mitochondria reporter, neither the total number (Figure 3—figure supplement 1A) nor the proportions of mitochondria stopped or moving in either direction (Figure 3—figure supplement 1B) were affected by 24 hr of chloroquine treatment. These negative results lead us to confirm the efficacy of chloroquine in our system by using transgenic animals expressing the autophagosome/autophagolysosome reporter LC3b-EGFP-mCherry, expressed in astrocytes through a promoter from the fatty acid binding protein 7 gene (fabp7, also known as basic lipid binding protein or blbp) (Owada et al., 1996; Sharifi et al., 2011), which drives expression in frog optic nerve astrocytes (Mills et al., 2015). While discrete green or red puncta were rare, chloroquine did result in the expected increase in green puncta that report autophagosomes that have not yet fused with lysosomes (Figure 3—figure supplement 1C). Within some of those autophagosomes, there was Mitotracker signal (Figure 3—figure supplement 1D), suggesting that at least some axonal mitochondria might be degraded by astrocyte autophagolysosomes. To quantify the extent of RGC axonal mitochondria found in the optic nerve outside of the axons, the same chloroquine treated animals were analyzed as z-scans of the full thickness of optic nerves acquired at 1 µm steps, followed by measurements using 3D imaging software Imaris of the Tom20-mCherry signal inside or out of a mask based on a membrane GFP transgene expressed by the same axons (Figure 3—figure supplement 1E, F). Such analyses showed that 21.4% of the Tom20-mCherry signal is outside of the axons, much of it near the surface of the optic nerve, but that amount is not significantly altered by 24 hr of chloroquine treatment (Figure 3—figure supplement 1G). Note that this measure of axonal mitochondria outside axons likely relates to only the stopped mitochondria, as the z-scan imaging used frame-averaging and thus moving objects were blurred and likely underrepresented. Thus, in our system, the stopped mitochondria co-localizing with OPTN appear not to be degraded by conventional mitophagy.

Since in many of the animals some OPTN signal appeared not to be co-localized with the LC3b signal but rather to also be on the surface of the optic nerve (e.g., middle top of the OPTN E50K nerve shown in Figure 2C), similar to what was observed with the axon expressed Tom20-mCherry transgene, we then asked whether significant amounts of OPTN might also be found outside of axons, and if so, whether the OPTN mutations affected how much. Because LC3b localizes not only to autophagosomal membranes, but also is in the cytoplasm/axoplasm (Xie and Klionsky, 2007; Mizushima and Komatsu, 2011; Fu et al., 2014; Wong and Holzbaur, 2014b; Tammineni et al., 2017; Stavoe et al., 2019; Evans and Holzbaur, 2020; Boecker et al., 2021; Kuijpers et al., 2021), we used the EGFP-LC3b signal to create a mask to demark the RGC axons, so as to then be able to separately measure the OPTN (mCherry) or mitochondria (far-red) signals that co-localized or not with the axons. First, such analyses of OPTN outside axons were carried out for the same animals above analyzed for OPTN and LC3b movement metrics, but that had not been subjected to the intravitreal injections of Mitotracker, animals in which the expression of mCherry-hOPTNs (Wt, E50K, M98K, F178A, D474N, E478G, and H486R) and EGFP-LC3b was induced in RGCs through bath application of doxycycline. Measurements carried out in the 3D reconstructions showed significant increases in the fraction of OPTN outside of the RGC axons in the optic nerves of glaucoma-associated mutations, especially E50K (12.6%), when compared to Wt OPTN (0.5%), ALS-associated mutation E478G (0.2%), and the synthetic mutations F178A (3.3%) and D474N (0.1%) (Figure 3A, B). Then, to determine whether such extra-axonal OPTN might be associated with mitochondria, the same procedure was applied to the z-scan images of animals where Wt OPTN and E50K OPTN were similarly induced by doxycycline, but that also a day prior had received an intravitreal injection of Mitotracker (Figure 3C). Imaris-based 3D reconstruction and volume quantification revealed that expression of OPTN carrying the glaucoma-associated E50K mutation results in a significant increase in both the fraction of RGC axonal mitochondria outside of the LC3b-labeled RGC axons (13.4%) but also the fraction of all stopped RGC axonal mitochondria co-localizing with OPTN also outside of the LC3b-labeled RGC axons (17.9%), as compared to the much lower amounts found after expression of Wt OPTN (2.7% for mitochondria and 0.5% for mito-OPTN) (Figure 3D, E). Note that the amount of mitochondrial signal measured outside of axons based on Mitotracker labeling is less than that estimated to be outside based on the Tom20mCherry transgene. In summary, expression of E50K OPTN, but not OPTN that lacks the glaucoma-associated mutation, not only increases the amount of stopped mitochondria, OPTN and LC3b, and the degree of co-localization between mitochondria and OPTN within axons but also increases the amounts of mitochondria and the mitophagy receptor OPTN being found outside of the axons, much of which also co-localize outside of the axons.

In the experiments just described, much LC3b-negative-OPTN-positive mitochondria signal was found on what appears to be the surface of the optic nerve, which was most obvious after expression of E50K OPTN (see Figure 3A, C, and Figure 3—video 1). To determine whether additional extra-axonal OPTN and mitochondria might be found within the optic nerve parenchyma but obscured by the EGFP-LC3b mask due to the high axon density and low imaging resolution, we turned to a sparse labeling approach. Small groups of retinal cells, estimated to be 10–50 in number, were surgically transplanted from E50K OPTN transgenic progeny into progeny of a line expressing Tg(Fabp7:mTagBFP2-Ras), in which astrocyte membranes are labeled by a membrane-targeted BFP reporter. Transplants were done around NF stage 24–32, when all cells in the retina are progenitors and thus prior to RGC differentiation. A week later in such animals and 3 days after doxycycline induction, small groups of axons were readily visible in these optic nerves based on the EGFP-LC3b signal. 3D reconstruction and volume measurements of the astrocyte membrane signal derived from the host and the OPTN and LC3b signals derived from E50K OPTN expressing donor cells (Figure 4A, B, and Figure 4—video 1) showed that the fraction of OPTN signal on the surface of the optic nerve was 17.1% of the total signal, in relatively good agreement with the data previously obtained in the whole nerves (see Figure 3B). However, this represented only 23.9% of the total OPTN outside of the LC3b-labeled axons, the rest being within the optic nerve parenchyma. Furthermore, the nerves with the largest amounts of OPTN on the surface of the optic nerve were those with the greatest amount of extra-axonal OPTN (Figure 4B). These sparse labeling data thus suggested that our previous measures of OPTN and mitochondria outside of axons quantified in Figure 4 based on the whole-nerve imaging after expression of all OPTN variants likely grossly underreported the actual amounts of OPTN and mitochondria outside of the axons. Thus, for animals expressing either Wt or E50K OPTN, we repeated the same t-series and z-scan imaging and following kymograph analyses and Imaris 3D reconstructions in the context of the sparse labeled axons, but also after dye-labeling the mitochondria prior to transplantation. That is, 1 hr after injection of Mitotracker into the eye anlage, retina progenitor cells from Wt or E50K OPTN transgenic progeny were transplanted into the eyes of non-transgenic hosts (Figure 4C, Figure 4—figure supplement 1A, B). Consistent with our previous data, more stalled mitochondria, OPTN, and LC3b puncta were observed in the axons expressing E50K OPTN compared to those expressing Wt OPTN, with relatively balanced populations of anterograde and retrograde movement in both Wt and E50K OPTN (analyzed per axon in Figure 4—figure supplement 1C and analyzed per animal in Figure 4—figure supplement 1C’). Anterograde and retrograde velocities for mitochondria, OPTN, and LC3b in the sparsely labeled Wt and E50K OPTN axons also were similar to those measured in the whole nerves, and once again showed no significant differences in velocities after expression of E50K OPTN (Figure 4—figure supplement 1D). Of note, the diameter of axons based on the LC3b signal (Figure 3—figure supplement 1E) and the intensity of the LC3b signal (Figure 3—figure supplement 1F) were comparable after expression of Wt or E50K OPTN, validating the use of cytoplasmic LC3b-GFP as a suitable axon mask in both these sparse axon experiments and the previous whole nerve experiments. Consistent with the other sparse labeling results above, the amounts of mitochondria and OPTN outside of axons were higher than those previously measured. In the case of E50K OPTN, 35.8% of mitochondria and 21.8% of OPTN were determined to be outside of the axons in the sparse labeling experiments (as compared to 13.4% and 17.9%, respectively, measured in the whole nerves). The difference between whole nerve and sparsely labeled axons was far larger in the case of expression of Wt OPTN, where previously approximately 2.7% and 0.5% had been found for mitochondria and mito-OPTN, respectively (see Figure 3D, E), but now 15.2% of mitochondria and 5.7% of OPTN were shown to be outside of the RGC axons (Figure 4D, E); this order of magnitude difference in results is presumably due to nearly all the extra-axonal Mitotracker-labeled mitochondria and OPTN are found within the nerve parenchyma, not having reached high enough levels to be measurable on the surface of the optic nerve. Collectively, these data show that even in the case of expression of Wt OPTN, which we showed above does not alter the movement behavior of mitochondria within axons, there already is a substantial amount of both mitochondria and OPTN outside of the axons from which they derive, and that expression of E50K OPTN, in addition to resulting in the stalling of large number of mitochondria, OPTN, and LC3b within axons and co-localization of mitochondria and OPTN also significantly increase the amount of both OPTN and mitochondria that are found outside of the axons and further result in over an order of magnitude increase in how much mitochondria signal is found near the surface of the optic nerve.

To determine where outside axons the axonal mitochondria were degraded, we carried out correlated light EM analyses on sparsely labeled axons. First, cells from animals expressing a membrane-targeted mCherry transgene in RGCs were transplanted into animals expressing two different transgenes in astrocytes, the same membrane-associated BFP transgene used above to label all astrocyte membranes, and also an Aquaporin-4-GFP (Aqp4-GFP) fusion construct that labels principally the glial limitans at the surface of the optic nerve. One day after labeling the RGC axonal mitochondria through an intravitreal injection of Mitotracker Deep Red, the optic nerves of these animals were live-imaged in their entirety at lower x-, y-, and z-resolution for just the Aqp4-GFP transgene (Figure 5A), and then one region imaged at higher resolution in all four channels (Figure 5B). The heads of fixed animals, heavy metal processed and embedded in epoxy resin, were then analyzed by micro-CT in order to identify the same region that had been live imaged (Figure 5C) and then the full thickness of the optic nerve at that location was analyzed by serial block-face scanning electron microscopy (SBEM). After registering the live imaging and SBEM datasets based on the position of nuclei, some of the Mitotracker signal outside axons that had been live-imaged could be unambiguously identified at the level of EM. The axon-derived material on the surface of the optic nerve, in this case dye-labeled mitochondria and transgene-labeled axonal membranes, was determined to be within the soma of a transgene-labeled astrocyte (Figure 5D, E), and morphologically appeared as mitochondria and membranes at various stages of degradation. The mitochondria that live-imaging determined to be stationary within the nerve parenchyma were harder to unambiguously correlate between light and SBEM datasets, but those that could were determined to be within the fine processes of astrocytes (Figure 5F). Notably, while axonal debris was most often observed within individual fine processes, and those processes were closely apposed to fine processes of distinct astrocytes (Figure 5F). The soma of the astrocytes was located at the optic nerve periphery but had processes that extended deep into the parenchyma (Figure 5G), similar to astrocytes that surround axon bundles within larger mammalian optic nerves. Thus, it is possible that the axon-derived debris found within astrocyte soma at the surface of the optic nerve might have originated deep within the optic nerve parenchyma. However, it also may have originated near the astrocyte soma, as dystrophic axons with mitochondria undergoing early steps of mitophagy could also be observed within some axons in close contact with the astrocyte soma (Figure 5H, I).

To investigate how RGC axonal mitochondria and OPTN leave axons, we performed additional sparse labeling experiments. Live imaging of sparsely labeled axons in animals expressing both a membrane-targeted GFP (lyn-GFP) and a mitochondrial-targeted mCherry (Tom20-mCherry) transgene revealed extensive mitochondria signal outside of axons that was relatively unchanged over 10 min of imaging (Figure 6A, and Figure 6—video 1). In these nerves, 6.3 ± 0.1% (SD) of the Tom20-mCherry mitochondria signal was outside of the lyn-GFP-labeled axons, which is comparable to the measure of mitochondria outside sparse axons obtained based on Mitotracker labeling. Manual counting of Tom20-mCherry-labeled mitochondria within five discrete axon segments (100–150 µm in length) in two animals and Imaris-based semi-automatic counts within approximately 235 axon segments (75–175 µm in length) in 7 animals provided similar estimates for the frequency of stopped mitochondria, namely 3.8 ± 2.3 (SD) or 6.3 ± 5.0 (SD) per 100 µm of axon length, respectively. Among these sparsely labeled axons, there were also membranous asymmetric focal dystrophies, or protrusions, containing mitochondria similar in shape to what we previously described in the mouse optic nerve head (Figure 6B). In the same axons where mitochondria number was counted, a total of 78 of such asymmetric dystrophies were found, thus representing 0.47 per 100 µm length of axon. However, of these protrusions, only 47, or 60.3%, contained mitochondria. Thus, approximately 1 in 20 stopped mitochondria within axons is contained within such protrusions. Of note, the mitochondria that were contained within these axonal protrusions were significantly more spherical than the average axonal mitochondria (Figure 6C). Tracking of 18 of these same mitochondria containing protrusions over approximately 8 min of repeated imaging every minute found 13 of them still present at the last time point, showing that these asymmetric mitochondria containing axonal dystrophies persist on average more than 5 min.

In order to determine the localization of OPTN relative to these axonal dystrophies, we performed similar experiments where instead donors expressed LC3b-EGFP and OPTN-mCherry transgenes. Since transplants are challenging to perform and must be done before knowing whether the transplanted cells will express the transgenes, and the intermediates are short-lived, these studies were conducted mainly for axons expressing the glaucoma-associated M98K OPTN mutation, as these animals had the greatest number of stopped OPTN in both F0 and F1 analyses (Figure 2E, Figure 2—figure supplement 1A). After the repetitive z-scan live imaging of optic nerve sub-volumes, areas with focally increased LC3b and OPTN signal were digitally extracted and their position at different times registered, to measure shape and fluorescence intensity changes over time. Areas of increased LC3b and OPTN were found in four types of structures: symmetric axonal swellings with static LC3b and OPTN, asymmetric axonal dystrophies with static or changing levels of OPTN or LC3b, a small number of asymmetric axonal dystrophies that pinched off from axons during the imaging window, and structures with no connection to axons. An axonal ‘swelling’ is shown where the mitophagy machinery OPTN and LC3b are focally stopped and enriched together (Figure 6D, i). An asymmetric axon dystrophy is shown in what appears to be a ‘loading’ stage (Figure 6D, ii and Figure 6—video 2). Here, the OPTN signal increases within the dystrophy simultaneously with a decrease in the OPTN signal immediately below the dystrophy within the associated axon, suggestive of an active loading process (Figure 6E). The last example shown is a ‘pinching-off’ event where the axon dystrophy is physically detached from the axon of origin and appears to amalgamate with other areas of extra-axonal OPTN (Figure 6D, iii and Figure 6—video 3). Minimum distance measures between the outer boundaries of protrusion/evulsion and axon of origin (Figure 6F) and measures of the distance from its centroid to the nearest position along the axon (not shown) demonstrate that the detachment process is acute, occurring within a 30-s interval. The shape, size, and dynamics in these focal asymmetric axonal dystrophies lead us to conclude that at least some of the OPTN and mitochondria leave the axons by the transcellular degradation process we first described in the mouse optic nerve head (Davis et al., 2014). Finally, since the E50K OPTN mutation causes glaucoma, we then asked whether increased transcellular degradation might be associated with vision loss. For this, we used a behavior assay that reports RGC innervation of the optic tectum (Dong et al., 2009), the main brain target for RGC axons and the equivalent of the mouse superior colliculus. Animals expressing Wt or E50K OPTN showed no measurable visual impairment 3 days after inducing transgene expression with doxycycline, the time at which E50K OPTN expression results in an increase in transcellular degradation of mitochondria.

Transgenic X. laevis were generated by REMI transgenesis (Amaya and Kroll, 1999) using optimizations for larger DNA constructs that have been described previously (Mills et al., 2015). In all cases, DNAs were linearized overnight with NotI in the injection buffer, followed by 65°C 20-min inactivation of the enzyme. Embryos were kept in 0.1X Marc’s Modified Ringer’s (MMR) with 50 μg/ml of gentamicin in sterile glass Petri dishes for the first 2 days, the first day at 16°C and the second at room temperature, and then subsequently kept in 0.1X MMR in glass bowls at room temperature and under a 12/12 light/dark cycle.

Animals were treated with 100 µM chloroquine diphosphate dissolved in 0.1X MMR for either 10 hr in the case of the measurements of autophagosomes with the pCS2(xtFapb7):LC3b-EGFP-mCherry transgene, and for 24 hr in the case of measurements of mitochondria movement within axons and amount of mitochondria signal outside axons.

Approximately 10–50 eye anlage cells were surgically transferred from X. laevis homozygous or heterozygous for Isl2b:mCherry-OPTN(Wt/E50K/M98K)_EGFP-LC3b into either wild-type or Tg(xtFabp7:mTagBFP2-Ras) hosts at a developmental stage prior to retinal cell differentiation (at between NF stage 24 and 32). Surgeries were carried under Zeiss Stemi 2000 stereomicroscopes using hand-held curved tungsten micropins on animals anesthetized with 0.2 g/l of Tricaine Methanesulfonate (MS-222) in filter sterilized 0.1X MMR containing 50 μg/ml of gentamicin and with additional buffering provided by 2 mM HEPES, pH 7.4–7.6, in 35 mm Petri dishes. Animals were immobilized during surgeries in disposable clay-mold (Permoplast, non-toxic) inserts with adjacent grooves created with a pipette tip and then altered with forceps to loosely anchor donors and hosts in parallel near each other. At approximately 10 min after transplantation, the animals were released from their clay-mold enclosures and donors and hosts returned together to the same dish containing fresh sterile-filtered 0.1X MMR 2 mM HEPES with 50 μg/ml of gentamicin. Successful transplants, which were approximately half of the number of operations performed, were defined as those with donors confirmed as transgenic (since operations are done prior to onset of transgene expression), and whose hosts had eyes of normal size and only a few RGC axons labeled, as primarily assessed under a Leica MZ10F fluorescence-equipped stereomicroscope and then verified through confocal live-imaging (see details below).

Mitotracker Deep Red is a far-red fluorescent dye that chemically stains mitochondria in live cells (Audano et al., 2021; Weiss-Sadan et al., 2019). Fresh 200 µM of MitoTracker Deep Red FM (Cell Signaling Technology, #8778) dye solution was prepared by diluting a 5 mM of stock solution prepared in DMSO with filtered 0.5X MMR on the day of injection. The Mitotracker solution was front-loaded into a borosilicate glass capillary (World Precision Instruments) that had been pulled and then broken to 1–2 µm tip diameter using fine forceps, and then injected intravitreally into left eyes of anesthetized animals at around NF stage 48 (Nieuwkoop and Faber, 1994), using a pressure injector (Narishige IM-300 Microinjector); successful injections were verified by observing an acute but mild swelling of the eye at the time of injection. After injection, the animals were returned to a new dish containing 0.1X MMR at room temperature and imaged 3.5 hr after injection for Figure 1—figure supplement 1A, B and 15–18 hr after injection for all the other experiments.

In the case of doxycycline-inducible transgenes, X. laevis tadpoles were placed in 0.1X MMR with 50 μg/ml of doxycycline for 2–3 days before imaging. At around NF stage 48, X. laevis tadpoles were anesthetized in 0.1X MMR with 0.2 g/l of MS-222 and mounted into custom Sylgard 184 silicone elastomer (World Precision Instruments) molds. Molds were created using glutaraldehyde-fixed tadpoles at the same stages slightly angled using insect pin inserts so as to make left optic nerves parallel to the imaging plane, and then modified with surgical blades to provide an opening before the mouth to facilitate breathing. Tadpoles immobilized in the molds were weighed down by an 18 mm circular coverslip and imaged from beneath in a 35 mm dish modified with a 22 mm square glass coverslip as bottom. Optic nerves of the animals were imaged by spinning disk confocal microscopy (Dragonfly 503 multimodal imaging system, Andor Technology, Belfast, UK) with a ×40/1.10 (magnification/numerical aperture) HC PL APO water immersion objective, using a Leica DMi8 inverted microscope (Leica, Wetzlar, Germany), an iXon Ultra 888 EMCCD camera (Andor Technology, Belfast, UK), Fusion Software (Andor Technology, Belfast, UK) and laser lines of 100 mW 405 nm, 50 mW 488 nm, 50 mW 561 nm, and 100 mW 643 nm; first, a time-series was obtained at a single focal plane imaged for 1 min at 1 Hz, and the entire thickness of the nerve at that same location was z-scanned at 1 µm steps. Videos at 7 fps and images were created using FIJI ImageJ.

For some sparsely labeled axons, 5–10 µm thick regions of the optic nerve were subjected to repetitive z-scan live-imaging either at 30-s intervals for 5 min or at 2-min intervals for 10 min. Acquired images were saved as separate image sequence files using FIJI ImageJ and imported into IPLab software (Scanalytics). Regions of interest sampled at different time points were then registered and quantified using a macro written in IPLab. All the relevant videos (3 fps for M98K and 1 fps for E50K videos) and images were created using FIJI ImageJ.

Axonal movements of mitochondria, OPTN and LC3b: acquired time-series images were saved as separate image sequence files in FIJI ImageJ after being processed by HyperStackReg or StackReg Plugin to eliminate object drift. The aligned image sequence files were restacked using IPLab software, followed by either manual (for the sparsely labeled axons and curved axons) or semi-automatic tracing using IPLab macros so as to visualize and measure fluorescence intensity changes over time in the form of kymographs. For the semi-automatic tracings, the outer contour of the optic nerve was traced, and the approximately 30 µm diameter nerve was automatically divided into 0.9–1.8 µm parallel swaths, which contained small stretches of axons running in parallel; using such sampling distance and thickness, the same moving objects within axons were not multiply counted, but some of the larger immotile objects might have been multiply counted. Since there was high concordance between ratios of moving and static objects and the velocities of the moving objects when comparing whole-nerve semi-automatic kymographs and the more conventional single axon tracing used in the sparse labeling experiments, any error introduced by the novel semi-automatic tracing of axons was deemed minimal. Axonal movement was determined by counting the percentage of moving and stopped objects and measuring the speed of moving over the 1 min imaging period. Objects were classified as moving if their velocity was 0.1 μm/s or faster. Ratios and average speeds were calculated in Excel.

Co-localization measures: The same aligned and stacked images from the analysis of axonal movements were reprocessed and analyzed by a separate co-localization script but by the same logic through IPLab as described previously, except displaying multiple-channel fluorescence images along the merged color images, and sequentially tracing objects labeled by multiple fluorophores first, followed by those labeled by individual fluorophores. Ratios and average speeds were calculated in Excel.

Z-scan confocal images of optic nerves expressing mCherry-OPTN and EGFP-LC3b were imported into Imaris software (Bitplane). Because LC3b is localized not only in autophagosomal membrane, but also is observed more diffusely within cytosol, including axoplasm (Fu et al., 2014; Wong and Holzbaur, 2014b; Tammineni et al., 2017; Stavoe et al., 2019; Evans and Holzbaur, 2020; Kuijpers et al., 2021; Boecker et al., 2021), this EGFP-LC3b signal was used to create a mask to define the boundary of the RGC axons and to separately measure the mCherry-OPTN or far-red mitochondria signal that co-localized or not with the axons. The Imaris mask toolset was used to produce channels for OPTN, mitochondria, or OPTN–mitochondria either co-localizing or not with LC3b. The % outside axon values were calculated in Excel by comparing the volumes of objects outside of the LC3b mask relative to the total volumes.

After optic nerve live imaging, tadpoles were fixed overnight with 4% paraformaldehyde, re-imaged for confocal imaging and positioning, then fixed overnight with 2% EM grade glutaraldehyde (18426, Ted Pella Incorporated) in 0.15 M sodium cacodylate buffer (SCB) containing 2 mM CaCl₂. The tadpoles were then prepared for serial block-face imaging (SBEM) as described previously (Deerinck et al., 2010). Briefly, tadpoles were washed with SCB then incubated in 2% OsO₄ + 1.5% potassium ferrocyanide in SCB containing 2 mM CaCl₂ for 1 hr. Tadpoles were washed with double distilled water (ddH₂O) and incubated in 0.05% thiocarbohydrazide for 30 min. Tadpoles were then washed again with ddH₂O and stained with 2% aqueous OsO₄ for 30 min. Tadpoles were again washed in ddH₂O and placed in 2% aqueous uranyl acetate overnight at 4°C. Tadpoles were washed again with ddH₂O and stained with en bloc lead aspartate for 30 min at 60°C. After a last wash in ddH₂O, tadpoles were dehydrated on ice in 50%, 70%, 90%, and 100% ethanol solutions for 10 min each, followed by two 10-min incubations in dry acetone. Tadpoles were placed in 50:50 dry acetone/Durcopan resin overnight, followed by three changes of 100% Durcopan for ~12 hr each. Tadpoles were then allowed to harden in Durcopan at 60°C for 48 hr.

Tissue blocks were cut from hardened Durcopan, and the whole head of the tadpole was mounted on plastic dummy blocks, trimmed down to only the region surrounding the left optic nerve, eye, and brain. Exact location of optic nerve and the rest of the region of interest determined by X-ray microscopy imaging of tissue blocks performed on a Zeiss Xradia Versa 510 X-ray microscope (XRM) instrument (Zeiss X-Ray Microscopy, Pleasanton, CA, USA). XRM tilt series were generally collected at 120 kV and 10 W power (75 µA current) and used to create 3D reconstructions of the embedded tissue. SBEM data was collected with a 3View unit (Gatan, Inc, Pleasanton, CA, USA) installed on a Gemini field emission SEM (Carl Zeiss Microscopy, Jena, Germany). Volume was collected at 2.5 kV accelerating voltage, with a raster size of 20k × 17k and pixel dwell time of 1.0 µsec. The pixel size was 5.0 nm and section thickness was 60 nm. SBEM volumes were analyzed and annotated in 3Dmod (3dmod Version 4.11.24, Copyright (C) 1994–2021 by the Regents of the University of Colorado).

Dot avoidance behavior was carried out essentially as previously described (Fague and Marsh-Armstrong, 2023). Ten informative trials were performed for each tadpole; that is, trials where the tadpoles might have initiated or changed movement, but it is unclear whether that was in response to the oncoming dot were not counted. A 50% response rate is characteristic of normal vision.

Statistical analyses involved the comparison of means using unpaired, two-tailed Student’s t-tests or two-way ANOVAs following Tukey’s post hoc test for multiple comparisons using GraphPad Prism software. All the bar graphs represent the means ± SEM, where values were first averaged per animal and N is the number of animals. In the stacked bar graphs (except Figure 2G, Figure 1—figure supplement 1E, and Figure 2—figure supplement 1H, I) showing percentage of each movement, all statistical comparisons were performed using all the axonal movements including stationary, anterograde, and retrograde, although only the statistically significant comparisons between the stationary groups (against Wt OPTN or control group) are shown. In all the velocity graphs (except Figure 1—figure supplement 1F), all statistical comparisons were performed throughout all the transgenic lines and movements (both anterograde and retrograde movements), and only the statistically significant comparisons against the Wt OPTN or control group are shown.

All animal experiments were carried out in accordance with procedures approved by the Institutional Animal Care and Use Committee of University of California, Davis.