Activity-dependent alteration of early myelin ensheathment in a developing sensory circuit

Adaptive myelination has been reported in response to experimental manipulations of neuronal activity, but the links between sensory experience, corresponding neuronal activity, and resultant alterations in myelination require investigation. To study this, we used the Xenopus laevis tadpole, which is a classic model for studies of visual system development and function because it is translucent and visually responsive throughout the formation of this retinotectal system. Here, we report the timecourse of early myelin ensheathment in the Xenopus retinotectal system using immunohistochemistry of myelin basic protein (MBP) along with third-harmonic generation (THG) microscopy, a label-free structural imaging technique. Characterization of the myelination progression revealed an appropriate developmental window to address the effects of early patterned visual experience on myelin ensheathment. To alter patterned activity, we showed tadpoles stroboscopic stimuli and measured the calcium responses of retinal ganglion cell axon terminals. We identified strobe frequencies that elicited robust versus dampened calcium responses, reared animals in these strobe conditions for 7 d, and subsequently observed differences in the amount of early myelin ensheathment at the optic chiasm. This study provides evidence that it is not just the presence but also to the specific temporal properties of sensory stimuli that are important for myelin plasticity.


29
The development and function of brain circuits relies crucially upon precise timing of neuronal 30 inputs. By insulating axons to regulate the conduction velocity of these inputs, myelination may 31 optimize temporal control over information processing, with implications for synchrony in vertebrate 32 circuits(1, 2). Effects of experience and training on biomarkers of myelination have been reported 33 with white matter imaging techniques and with cellular level investigations(3, 4). These cellular level 34 changes have been studied using extreme manipulations of axonal activity or vesicular release, 35 including sensory deprivation, chronic pharmacological treatment, genetic manipulations, and 36 electrical and optogenetic stimulation(3-5). However, the links between patterns of sensory 37 experience, corresponding neuronal activity, and myelination have yet to be fully elucidated. To 38 study how sensory patterned activity alters myelination during circuit development, we took 39 advantage of the Xenopus retinotectal system, which is amenable to imaging, shows precocious 40 visual responsiveness, and has been extensively studied in the context of the effects of patterned 41 activity on synaptic plasticity, structural remodeling, and topographic circuit refinement(6).

43
To observe myelin ensheathment, we used an antibody against myelin basic protein (MBP), which 44 showed expected band sizes for MBP isoforms (19 and 22 kDa) on Western blots of adult Xenopus 45 brain lysate, in accordance with the reported molecular weights of MBP isoforms in Xenopus (7) 46 and other species(8) (Fig 1A). The pattern of MBP immunostaining reflected the laminar 47 organization of the adult optic tectum(9) (Fig 1B), and in the hindbrain was similar to 48 immunostaining for myelin proteolipid protein reported in stage 49 tadpoles(10) (Fig 1C). We cross-49 validated immunostaining with third-harmonic generation (THG) microscopy, an emerging label-50 free technique that has been used to image the presence of myelin in the peripheral and central 51 nervous systems(11, 12). THG microscopy reveals sub-micrometer heterogeneities produced at 52 optical interfaces, allowing it to be used as a structural imaging tool in unstained samples(13).

53
Strong THG signal was observed in a subset of MBP-positive fibers and increased at later 54 developmental stages in both the hindbrain ( Fig 1D) and optic chiasm (Fig 1E), consistent with the 55 developmental progression from new ensheathment by MBP-positive processes to increasingly 56 compact myelin, giving stronger THG signal. Because MBP expression was highly specific and 57 preceded the onset of robust THG signal, we used MBP immunostaining for the rest of this study 58 to investigate effects from the onset of myelination.

59
We studied MBP expression in stage 48 to 54 tadpoles, a relevant developmental period 60 when tadpoles have just transitioned from relying on their yolk sack for nutrition to active 61 feeding(14) and show more complex sensorimotor behaviors in response to environmental cues.
Overall, MBP expression follows a caudal-to-rostral progression in the tadpole brain, highlighted 66 by comparing changes between stage 48 and stage 51 ( Fig 1G).

67
Based on our characterization of myelin ensheathment, we identified stage 48 as the 68 appropriate developmental stage in which to investigate how visual experience modulates MBP 69 expression. Strobe rearing has been previously shown to synchronize RGC firing and modulate 70 topographic map refinement in the optic tectum of Xenopus(17) and in goldfish(18). We used 71 various frequencies of stroboscopic stimuli ("strobe") to physiologically induce temporally patterned 72 activity in the retinotectal system (Fig 2A). When animals were exposed to strobe, robust calcium 73 responses in RGC axons were evoked by 0.0625 Hz ("slow") but not by 1 Hz ("fast") strobe (

79
We therefore raised animals for 7 d under slow or fast strobe conditions or dim ambient light.

80
We chose ambient light as the control since darkness is known to lead to spontaneous local

158
Upon receiving them, they were staged as per Nieuwkoop and Faber(14) and immediately fixed for 159 sectioning and immunostaining.

160
For the calcium imaging experiment, we generated bilateral mosaic tadpoles, with    stimulus. This was followed by a 5 min rest period between trials, then the stimulus sequence was repeated with the next strobe frequency being tested.

261
Calcium recordings were registered using NoRMCorre(33), and analyzed using custom Matlab 262 scripts. For each animal, the calcium signal was averaged over the neuropil region, the ΔF / F trace 263 was calculated using the 20th percentile as baseline, then the signal corresponding to the strobe 264 period was detrended with a fourth-degree polynomial before performing Fourier spectral analyses.

265
The code for analysis of calcium responses to stroboscopic visual stimulation can be found at:    indicated one outlier (from 1Hz strobe group) that was removed.