The ultrastructure of severely degenerated NMJ boutons in wild-type Drosophila
In our early research, we analyzed a large number of type I NMJ boutons in wild-type Drosophila flies using TEM [44]. Only type I NMJ boutons between the 6th/7th muscles were presented [45] [46], and they were divided into type Ib and type Is. In TEM, the type Ib bouton (large) (Fig. 1A) was globular, the regular dense SSR membrane circled around the axon terminal, and clear synaptic vesicles gathered in a T-bar (Fig. 1A-A’) before the presynaptic membrane with an obvious synaptic cleft. Type Is boutons are smaller, and the SSR membrane is thinner (Fig. 1B) [45, 46]. In addition, there was a compact postsynaptic area (PSA) that matched after the postsynaptic membrane in both type I boutons (Fig. 1C-C’) [47] and type Ib boutons. In the terminal of type I boutons, organelles, such as mitochondria (Fig. 1A, B), were also present.
However, through analysis of more than 20 flies, we found that degenerated NMJ boutons were rarely observed in wild-type Drosophila. Severely degenerated NMJ boutons showed extremely degenerated axon terminals and extremely severe retraction of the SSR. The degenerated terminals gathered in a small area of the outer muscle (Fig. 1D) with rare retracted SSR membranes (Fig. 1D and D’’’), and they were basically vacuolated (Fig. 1D) or contained beaded, overlapping, dark synaptic vesicles (Fig. 1D’-D’’) and residual synapses (Fig. 1G-G’’) without mitochondria or T-bars. Compared to presynaptic membranes and postsynaptic membranes in normal synapses with a typical T-bar structure (Fig. 1 A’ and C’), the residual synapse had a significantly thin presynaptic membrane and postsynaptic membrane that were stuck to each other, with almost no synaptic cleft (Fig. 1G-G’’). The residual presynaptic membranes and postsynaptic membranes were thin and dark without synaptic clefts or T-bars (Fig. 1E’’, G-G’’) and sometimes had a tendency to separate from each other (Fig. 1G-G’’), which showed the characteristics of complete synapse degeneration.
The degenerated NMJ boutons were also present inside muscle (Fig. 1E-E’’’); some were completely vacuolated (Fig. 1E), but some vacuolated boutons contained degenerated ultrafine particles and had a sparse profile of similar dark synaptic vesicles (Fig. 1E’’). Furthermore, larger degenerated NMJ boutons in the outer muscle were filled with agglomerated degenerated products (Fig. 1F) and did not contain clear and dark synaptic vesicles but dark ultrafine particles less than 3 nm in diameter (Fig. 1F’-F’’). The SSR was extremely retracted in all degenerated NMJ boutons (Fig. 1D’’’, E’’’, F’’’), and residual SSRs were rare. There was no T-bar structure or mitochondria in any of the degenerated NMJ boutons.
In severely degenerated NMJ boutons in wild-type Drosophila, the axonal terminal retracted and shrank, with retraction and degradation of presynaptic components, such as synaptic vesicles, mitochondria and the T-bar structure, and most SSR membranes were retracted from the NMJ boutons. It is worth noting that normal NMJs were singular and isolated by an SSR (Fig. 1A and B), while the severely degenerated NMJs were close together without an obvious SSR (Fig. 1D, F, F).
Gradual degeneration process of NMJ boutons in wild-type Drosophila
We observed and confirmed severely degenerated NMJ boutons with full retraction into small boutons and then identified the process of degeneration in wild-type Drosophila (Fig. 2A-C).
In the milder degeneration state (Fig. 2A-A’’’), the boutons looked similar to normal globular type Ib boutons, but in the center region of the terminal, several dark synaptic vesicles and dark ultrafine particles (Fig. 2A-A’) appeared with more clear synaptic vesicles (Fig. 2A’) and a normal T-bar (Fig. 2A’’). Furthermore, the SSR membrane swelled and withdrew (Fig. 2A’’’). In a moderately degenerating bouton (Fig. 2B-B’’’), the dark synaptic vesicles and dark ultrafine particles increased (Fig. 2B’), and clear synaptic vesicles were present around the normal T-bar (Fig. 2B’’). However, the SSR membrane further loosened and withdrew, and some swollen SSR membranes were folded inward (Fig. 2B’’’). Then, two severely degenerated type Ib boutons were observed, which were close to each other without a T-bar (Fig. 2C). The two boutons had irregular terminals (Fig. 2C-C’’’) in which there were many dark ultrafine particles (Fig. 2C’’), few dark synaptic vesicles (Fig. 2C’) and few clear synaptic vesicles (Fig. 2C’’). The SSR became disordered and collapsed, and the SSR membrane was loose and thin or even absent from some regions of severely degenerated boutons with extremely sparse residual SSR membrane (Fig. 2C’’’). It seemed that dark synaptic vesicles were the intermediates, and the dark ultrafine particles were the final product during the collapse and degeneration of clear synaptic vesicles. It is likely that severely degenerated boutons are different from synaptic footprints that contain a relatively complete SSR in TEM and postsynaptic Dlg protein but no Synapsin or Hrp [30, 31]. It is worth noting that the deformed axon terminal was not detached from the SSR of degenerated boutons but instead degenerated in situ.
Degeneration of NMJ boutons originated from SSR abnormalities in the wild-type fly. The SSR membrane, synaptic vesicles, and T-bar showed marked degeneration in boutons (Fig. 1-2), but which component is the first to become abnormal remained unknown. We observed that the SSR membrane became loose and swollen in type Ib boutons (Fig. 3A-A’’) and in type Is boutons (Fig. 3B-B’’), but the synaptic vesicles and T-bars were very typical in both types of boutons (Fig. 3A’’, B’’), and the center region of the boutons did not exhibit the degeneration that is shown in Fig. 2A-C. Moreover, there were no dark synaptic vesicles or dark ultrafine particles in type Ib boutons or type Is boutons (Fig. 3A-B). NMJ boutons undergo marked degeneration during the process of development in the early pupal stage (6 hours pupa)[48]. We found that the T-bars were typical and that most synaptic vesicles were clear and normal in both type Ib boutons (Fig. 3C, C’’) and type Is boutons (Fig. 3D, D’’), but the SSR membrane was obviously swollen, thin, loose and disordered (Fig. 3C-C’, D-D’) in the pupal stage (13 hours pupa). Furthermore, there were no dark synaptic vesicles or dark ultrafine particles in type Ib boutons or type Is boutons in the early pupal stage (Fig. 3C-D).
Therefore, degeneration of NMJ boutons originates from swelling and retraction of the SSR membrane, and then, the clear synaptic vesicles turn into dark synaptic vesicles and fragment into dark ultrafine particles along with degeneration of the T-bar structure from the presynaptic membrane.
dnrx mutation leads to degeneration of NMJ boutons
We analyzed dnrx, dnlg1, dnlg2, dnlg3 and dnlg4 single mutants and found that dnrx273, the nrx null mutant, led to degeneration of NMJ boutons in Drosophila. The degenerating NMJ boutons in dnrx273 flies (Fig. 4-5) demonstrated more significant degeneration than observed in wild-type Drosophila (Fig. 1-3). The terminal of degenerated NMJ boutons, without a T-bar structure or other organelles, was smaller than that of normal boutons (Fig. 4A) and was filled with dark ultrafine particles (Fig. 4A’-A’’) but lacked dark synaptic vesicles, whereas the adjacent type Ib bouton was filled with clear vesicles (Fig. 4A’, A’’’) and several dark synaptic vesicles (Fig. 4A). There was no SSR membrane between the degenerated bouton and the adjacent normal bouton (Fig. 4B), which also suggested that the SSR retracted severely as the NMJ bouton degenerated. The SSR membrane was sparse and loose near the degenerated bouton (Fig. 4C-C’) but was relatively normal compared with the adjacent type Ib bouton (Fig. 4C and C’’). In the seriously degenerated boutons, the axon terminals showed signs of degeneration/vacuolization, and the SSR membrane had obviously swollen and withdrawn (Fig. 4A-A’, C-C’, and D-D’). It is worth noting that the contents of degenerated terminals in dnrx273 mutants (Fig. 4A’, C’ and D-E) were much denser than those of wild-type terminals (Fig. 2C-C’’). We observed another degenerated bouton that had an irregular morphology, a seriously linearized and degenerated SSR (Fig. 4E-E’’), degenerated contents in the axon terminal, an obvious residual postsynaptic area (PSA) and a synapse with almost no synaptic cleft (Fig. 4E-E’). The residual PSA suggested that the NMJ bouton was degenerated in situ but not eliminated. The degeneration of NMJ boutons could originate from ghost boutons and developing boutons. The abnormal SSR phenotype could not be observed in ghost synapses due to their lack of an SSR. The appearance of large ghost boutons was irregular in dnrx273 mutants (Fig. 4F-G), while normal ghost boutons in wild-type flies ([44]) and in some mutants [49] were spherical and full of clear vesicles. Instead of the clear and dark synaptic vesicles, there were dense dark ultrafine particles in the ghosts (Fig. 4F-G). However, some dark ultrafine particles were sparse (Fig. 4F-G’), and other dark ultrafine particles were intensively clustered (Fig. 4G, G’’). A thin SSR membrane was occasionally visible (Fig. 4G’). The degeneration might originate from developing boutons in which there were dark ultrafine particles (Fig. 4H, H’), degraded synapses with a thin presynaptic membrane and postsynaptic membrane (Fig. 4H, H’’), and swollen and linear SSR membranes (Fig. 4H’, H’’’).
Since mutant NMJ bouton degeneration was more serious in the dnrx273 mutant, we suspected that it might be easier to observe the fine degeneration of synaptic vesicles in the mutant by utilizing TEM. As reported in the literature [45, 46], type Ib boutons had clear synaptic vesicles with a T-bar (Fig. 5 A-A’), type Is had extremely sparse dense core vesicles (Fig. 5 B-B’), type II had more dense core vesicles (Fig. 5 C-C’), type III only included dense core vesicles (data not shown), and most clear synaptic vesicles were clustered with actin filaments (Fig. 5 A’, B’, C’).
Before NMJ boutons were severely degenerated, the T-bars detached from the presynaptic membrane. In all degenerated boutons, we observed a residual synapse that only included the presynaptic membrane and postsynaptic membrane, but no presynaptic T-bar that recruits and docks synaptic vesicles was present. The T-bar (Fig. 5D) detached from the presynaptic membrane, and the shed T-bar clustered dense synaptic vesicles (Fig. 5D-D’) [50], which might hinder accumulation of synaptic vesicles near the presynaptic membrane (Fig. 5D’, Fig. 5E (shown in the black box)) and move them to the center of the NMJ bouton, but in the same type Ib bouton the peripheral synaptic vesicles gathered in another T-bar that looked as if it was about to detach from the presynaptic membrane.
Then, we observed two degeneration modes of synaptic vesicles in type Ib boutons that could avoid the interference of dense core vesicles in type Is, type II and type III, according to the electron density under electron microscopy. In the first mode, one or two dark ultrafine spots occurred on the membrane of clear synaptic vesicles near a synapse with a relatively intact T-bar (Fig. 5F), and more dark ultrafine spots developed on the clear synaptic vesicle membrane and formed a circle at a site farther from the same synapse (Fig. 5F, F’). Before another synapse without a T-bar, the dark ultrafine spots dispersed into irregular profiles with larger sizes than the other synaptic vesicles (Fig. 5 G’); therefore, we believe the irregular profiles were the result of collapse and dispersion from the degenerated synaptic vesicles with dark ultrafine spots, and two collapsed synaptic vesicles overlapped each other to form a large profile (Fig. 5 G, G’). The slightly collapsed vesicles had the appearance of an ellipsoid profile, with dark spots on the inside and outside and a size similar to clear vesicles (Fig. 5 G, lower right corner). It is worth noting that the clear synaptic vesicles away from the synapse had a tendency to detach from each other without actin filaments (Fig. 5 F’, G’). Therefore, the first mode of synaptic vesicle degeneration occurred on the membrane with ultrafine spots and showed a collapsed and dispersed irregular profile with dark ultrafine particles.
In the second mode, the clear synaptic vesicles degenerated into dense synaptic vesicles, formed irregular dark clumps, and collapsed and dispersed an irregular profile with dark ultrafine particles. The degenerating bouton, a type Ib bouton in the dnrx273 mutant with only clear synaptic vesicles [45, 46] (Fig. 5H), had five synapses without T-bars and had numerous synaptic vesicles on its periphery. In high magnification mode, the synaptic vesicles could be divided into clear vesicles and different dense vesicles both with membranes near the periphery and cortex of an axon terminal (Fig. 5H, H’), but the clear vesicles, dense vesicles, dark clumps without a membrane, and dark ultrafine particles were present in the center of the axon terminal (Fig. 5H, H’’). The dense vesicles and clear vesicles could be clustered with actin filaments (Fig. 5 I) or detached without actin filaments (Fig. 5 J), and they appeared to exhibit deepening electron density (Fig. 5 K-V). A short dark line occurred on a certain point on the membrane of clear vesicles (Fig. 5 K), and the dark line expanded along the membrane of synaptic vesicles (Fig. 5 L) until it covered the entire membrane (Fig. 5 M-N), which made the vesicle dark. The electron density in vesicles expanded inward (Fig. 5 O), and the clear region in the dark vesicle continually decreased (Fig. 5P-Q). Then, the vesicle became fully electron dense, the membrane and morphological profile of the vesicle were lost, and a dark clump without actin filaments formed (Fig. 5 R). The clump became darker (Fig. 5 S) and separated into several dark ultrafine particles at the edge of the clump (Fig. 5 T), and the number of ultrafine particles increased at the edge of the clump (Fig. 5 U) until many fragmented dark ultrafine particles were present (Fig. 5V), which could be regarded as direct evidence that the dark vesicle had broken into ultrafine particles. The dark clumps had different sizes due to the different vesicle sizes. Once many vesicles were adhered to each other with actin filaments and degenerated together, they formed a big clump of ultrafine particles (Fig. 4A’, C’, E’, G’’). Accordingly, it was easier to observe the dynamics of synaptic vesicle degeneration in dnrx273 mutants, and the speckled membrane of clear vesicles, dark vesicles, dark clumps, and dark ultrafine particles could be regarded as signs of synaptic vesicle degeneration without lysosome involvement.
dnlg1 and dnlg4 mutants exhibited NMJ bouton degeneration
According to the signs of synaptic vesicle degeneration, we found that dnlg1 and dnlg4 mutants exhibited NMJ bouton degeneration. There was significant degeneration of axon terminals in NMJ boutons in dnlg1 mutants, but the SSR remained relatively intact (Fig. 6A-A’, C-C’). In the axon terminal, there were several plaques (Fig. 6A-A’), and the degenerated synaptic vesicles were in the plaques in the form of dark synaptic vesicles (Fig. 6A’-A’’’, B-B’) and dark ultrafine particles (Fig. 6B-B’), along with the clear synaptic vesicles and T-bar (Fig. 6A’’). Therefore, plaques with dark synaptic vesicles and dark ultrafine particles could be seen as markers of synaptic degeneration.
Degeneration was accompanied by abnormal assembly of microtubules. We observed long microtubules protruding into a type Ib bouton in both directions (from above and below) (Fig. 6C-C’), and the downward pointing microtubule passed through clear vesicles (Fig. 6C’-C’’) and reached a small plaque with a dark synaptic vesicle that showed degeneration (Fig. 6C’’). Degeneration could occur in axons. White plaques occurred among microtubules in a large axon (Fig. 6D) and contained dark synaptic vesicles and dark ultrafine particles (Fig. 6D’-D’’), but clear vesicles were not present in the white plaque (Fig. 6D’, upper right corner). Furthermore, some axons of motor nerve fibers also showed degeneration in dnlg1 mutants (Fig. 6C, E). The degenerated axons contained dark ultrafine particles (Fig. 6E’-E’’’) and gathered in specific parts of fibers (Fig. 6E). In the other part of the same fiber, the axon looked intact, with clear vesicles and microtubules pointing in different directions (Fig. 6E, F-F’). Therefore, the boundary between the degenerated axons and the normal axons could be artificially drawn (Fig. 6E).
The axon terminals also degenerated with SSR retraction (Fig. G) in NMJ boutons in dnlg4 mutants, and dark vesicles (Fig. 6G-G’), dark ultrafine particles and spare SSR membrane were observed (Fig. 6G, G’’-G’’’). Interestingly, autophagy was found in degenerated axons (Fig. 6H-H’).
dnlg2 and dnlg3 coregulate synaptic footprints in Drosophila NMJs
Both dnlg2 [41] and dnlg3 [42] regulate the circulation of synaptic vesicles, but dark vesicles and dark ultrafine particles were not observed in more than 30 NMJ boutons in dnlg2 (Fig. 7A-A’) and dnlg3 (Fig. 7B-B’) single mutants or dnlg2- and dnlg3-overexpressing lines (data not shown). However, degeneration of NMJ boutons frequently occurred both in the axon terminal and SSR in dnlg2;dnlg3 double mutants. The degeneration primarily emerged in the center of the axonal terminal (Fig. 7C-C’, C’’’, E), with dark vesicles and dark ultrafine particles visible (Fig. 7C’’’), but the clear synaptic vesicles were mainly distributed around the axonal membrane with presynaptic ruffles (Fig. 7C’’). Moreover, the SSR was disordered (Fig. 7D-D’) or even retracted to form a large PSA with the T-bar (Fig. 7E-E’). In outer motor nerve fibers, the degeneration of axons mainly emerged with dark vesicles and dark ultrafine particles (Fig. 7F, F’’). However, in most axons of the same nerve fiber, the periphery of the axons was relatively intact, with clear vesicles inside (F-F’), and most axons inside the fiber (F) looked ordered, without dark vesicles or dark ultrafine particles, which suggested that the peripheral axons of nerve fibers were more susceptible to degeneration. Interestingly, we found normal microtubules along with abnormal microtubules that had a smaller diameter and dark electron density in the small axon (F’).
dnrx and dnlg3 colead degeneration in Drosophila NMJs
dnrx83 and dnrx174 are hypomorphic mutants [38], and they live to adulthood, while dnrx273 is a null mutant [39], which is lethal during the pupa stage. Under electron microscopy, dnrx273 mutants had severe degeneration in NMJ boutons (Fig. 4-5), but the dnrx83/174 mutant did not show a degeneration phenotype. In addition, SSR degeneration and obvious dark vesicles were not observed in axon terminals in dnrx83 (data not shown), dnrx174 (data not shown), dnrx83/174 (Fig. 8A-A’), and dnlg3 (Fig. 7B-B’, Fig. 8B-B’’) mutants. However, NMJ bouton degeneration occurred in both the axon terminal and SSR of the dnrx83;dnlg3 double mutant (Fig. 8C-E’). We observed dark vesicles in the axon terminal of dnrx83;dnlg3 double mutants (Fig. 8C-C’). The SSR retracted and formed a rare SSR membrane (Fig. 8C, C’’), and a portion of the SSR was disordered (Fig. 8C, C’’’). The sparse SSR membrane could form a large PSA outside of NMJs (Fig. 8C). The T-bar structure was detached from the presynaptic membrane with clustered synaptic vesicles (Fig. 8 D-D’), and several dark lysosomes were observed in the PSA that was near the postsynaptic membrane (Fig. 8 D, D’’). A degenerating type Is bouton was observed that had almost had no SSR membrane and contained a T-bar, dark vesicles (Fig. 8 E-E’), and myelin-like mitochondria that were severely damaged (Fig. 8 E, E’’). The type Ib bouton had more large-sized clear vesicles in the dnrx83/174 (Fig. 8A-A’) and dnlg3 (Fig. 7B-B’, Fig. 8B-B’’) mutants, and the large clear vesicles further increased and collapsed inwardly in the dnrx83;dnlg3 double mutant (Fig. 8C-C’, C’’’).
Synaptotagmin is not distributed in ultrafine particles in degenerated boutons
Synaptotagmin (Syt) and synapsin (Syn) are synaptic vesicular proteins and have been used as markers of synaptic vesicles in Drosophila in many studies. Based on our above results, we propose a scenario for synaptic vesicle degeneration: spherical clear synaptic vesicles collapse into dark vesicles and then fragment into 2-3 nm ultrafine particles. Syt was present in the NMJ bouton in wild-type flies under light microscopy (Fig. 9 A-A’), Syt (Fig. 9 B-C’) and Syn (Fig. 9 D-D’) were present in the synaptic vesicles in TEM, but Syt and Syn were not present in the control (Fig. 9 E-E’) evaluated with preembedding immunogold electron microscopy.
Due to the obvious degeneration of NMJ boutons and presence of ultrafine particles in the dnlg2;dnlg3 double mutants, we investigated whether Syt was present in these ultrafine particles. Syt was present in synaptic vesicles of NMJ boutons without ultrafine particles (Fig. 9F). However, Syt was not present in the ultrafine particles but was present in the synaptic vesicles in the degenerated NMJ boutons (Fig. 9G-G’’). Therefore, in the process of synaptic vesicle degeneration into ultrafine particles, synaptic vesicle-associated proteins, such as Syt, appear to be completely degraded and could not be detected by the corresponding antibodies in TEM.
Neurexin and neuroligins jointly regulate synapse degeneration at the neuromuscular junction
Syt was not observed via TEM in the ultrafine particles that were degeneration products, which indicates that the degenerated synaptic vesicles might lose the signaling of Syt and Syn proteins with the disintegration of synaptic vesicles. Next, we investigated whether the degenerated NMJ boutons could be observed via confocal microscopy with a 3D scanning function for biological samples. To facilitate the evaluation of degenerated NMJ boutons, we observed and counted type Ib boutons that had a larger size, and the synaptic vesicles were numerous and relatively dispersed in the outer layer of axon terminals with respect to the type Is boutons [45, 46].
Most type Ib boutons had strong Syt (data not shown) and Syn (Fig. 10 A-A’’) protein signals at the 6th/7th muscles in the A3 or A2 segment in wild-type lines. The Syn signals in type Ib boutons were regular, globular and covered the entire axon terminal in large pinhole mode (We adjusted the pinhole to 600 to acquire thicker optical sections with a highly sensitive GaAsP detector and used an 80 pinhole for routine observation.), and few degenerated type Ib boutons were observed (0.63 ± 0.26, N=8). After the confocal microscopy focal length was adjusted, although speckled Syn signals were present at some optical sections, in the middle optical area, 2-3 sections were always filled with Syn signals in axon terminals (Fig. 10 A-A’’). There were very weak or no Syn signals in some type Is boutons under the same microscopy parameters, including pinhole size, laser intensity and image brightness.
However, several degenerated type Ib boutons could be observed in dnrx273 (2.25 ± 0.25, N=4) (Fig. 10 B-B’’), dnlg1 (3.29 ± 0.75, N=7) (Fig. 10 C-C’’), and dnlg4 (2.63 ± 0.53, N=8) (Fig. 10 F-F’’) mutants. The criteria for judging the degenerated type Ib boutons were as follows: 1. The Syn signals were always very weak compared with those in other type Ib boutons (Fig. 10 B-B’’); 2. The Syn signals were always distributed in spots in the bouton (Fig. 10 C-C’’, F-F’’); 3. After adjustment of the confocal microscope pinhole size and laser intensity and the image brightness, the Syn signals in degenerate type Ib boutons faded (Fig. 10 B-B’’) or were distributed in small dots (Fig. 10 G-G’’) in most instances, but the Syn signals remained spherical and dense in other type Ib boutons. There were not obvious degenerate type Ib boutons in dnlg2 (1.33 ± 0.33, N=6) (Fig. 10 D-D’’), dnlg3 (0.53 ± 0.14, N=13) (Fig. 10 E-E’’), and dnrx83 (0.88 ± 0.30, N=8) mutants (Fig. 10 I-I’’), which were partial mutants of whole dnrx genes. Interestingly, the degenerate type Ib boutons were frequently found in dnlg2;dnlg3 (4.75 ± 0.85, N=4) (Fig. 10 G-G’’) and dnrx83;dnlg3 (3.13 ± 0.35, N=8) double mutants (Fig. 10 H-H’’) (Fig. 10 I-I’’). Therefore, the synaptic vesicle-associated protein Syn could not be detected with the corresponding antibodies via confocal microscopy, and the degeneration of terminals was accelerated in dnrx and dnlgs mutants.