Maturation of Purkinje cell firing properties relies on granule cell neurogenesis

Preterm infants that suffer cerebellar insults often develop motor disorders and cognitive difficulty. Granule cells are especially vulnerable, and they likely instigate disease by impairing the function of Purkinje cells. Here, we use regional genetic manipulations and in vivo electrophysiology to test whether granule cells help establish the firing properties of Purkinje cells during postnatal mouse development. We generated mice that lack granule cell neurogenesis and tracked the structural and functional consequences on Purkinje cells in these agranular pups. We reveal that Purkinje cells fail to acquire their typical connectivity and morphology, and the formation of characteristic Purkinje cell firing patterns is delayed by one week. We also show that the agranular pups have impaired motor behaviors and vocal skills. These data argue that granule cell neurogenesis sets the maturation time window for Purkinje cell function and refines cerebellar-dependent behaviors.

intrinsically generated (Raman and Bean, 1999). In this context, it is intriguing that genetically

85
Mice lacking Atoh1 from the En1 domain do not form differentiated granule cells 86 To test the hypothesis that granule cell neurogenesis is essential for the functional development 87 of Purkinje cells, we first established a model of agranular mice that is not initiated by the cell-88 autonomous development of Purkinje cells. In previous models, agranular mice lack granule cells 89 due to spontaneously occurring mutations in genes with widespread expression patterns. In those 90 mice, therefore, one cannot differentiate cell-extrinsic from cell-intrinsic effects (Dusart et al.,91 2006; Gold et al., 2007). Instead, we made use of the distinct origins of Purkinje cells and      (Figure 3D and F). We tested the firing features with five parameters:     345 We have no conflicts of interest to disclose.    All images were acquired from the cerebellum of P14 mice.    100%) and then mounted using Xylene or histoclear. All steps of immunohistochemistry were 446 performed at room temperature. All mounted slides were stored at 4 °C until they were imaged.

502
We held mice on a heated surgery pad. We removed hair from skull and made an incision in the 503 skin over the anterior part of the skull. We stabilized the heads of our mice using ear bars and a 504 mouth mount when animals were large enough (most P11-P14 mice) and otherwise glued the 505 mouse skull (P7-P10 mice) to a plastic mount that was attached to ear bars on our stereotaxic 506 surgery rig to stabilize the head during recordings. Using a sharp needle or dental drill, we made 507 a craniotomy in the interparietal bone plate, ~3 mm dorsal from lambda and ~3 mm lateral from 508 the midline, with a diameter of ~3mm. We kept our surgical coordinates consistent based on the 509 distance from lambda across mice of all ages, as the skull undergoes significant growth during 510 the ages at which we measured neural activity. After making a craniotomy, we recorded neural 511 activity using tungsten electrodes (Thomas Recording, Germany) and then the digitized the 512 signals into Spike2 (CED, England). We recorded neural activity from cells that were 0-2 mm 513 below the brain surface.

514
Analysis of in vivo electrophysiological recordings: All electrophysiological recording data were 515 spike sorted in Spike2. We sorted out three types of spikes: simple spikes, complex spikes, and 516 doublets. Complex spikes were characterized by their large amplitude, and post-spike 517 depolarization and smaller wavelets. Doublets were characterized as action potentials that were 518 followed by one or more smaller action potentials within 20 ms after the initial action potential. adjacent spikes (in s). CV = stdev(ISI)/mean(ISI), and CV2 = mean(2*|ISI n -ISI n-1 |/ (ISI n +ISI n-1 )).

528
Pause Percent was the proportion of the recording time during which the ISI was longer than five 529 times the mean ISI for each independent cell, defined as followed: (sum(ISI>5*mean(ISI)))/(total 530 recording time).

531
Behavioral analyses: Righting reflex was measured on P6, P8, and P10 as followed. Mouse was 532 placed on its back in a clean cage without bedding. One finger was used to stabilize the mouse 533 on its back. The timer was set the moment the experimenter removed their finger, and time was 534 recorded until mouse righted itself up to four paws. All mice were tested twice on each time 535 point. A "failed" trial was defined when the mouse did not right itself within one minute (sixty 536 seconds).

537
At P7, we recorded pup vocalizations as described previously (Yin et al., 2018). Pups 538 were placed in an anechoic, sound-attenuating chamber (Med Associates Inc. was recorded using Fusion software (Accuscan Instruments). We analyzed the data for total 549 distance traveled, movement time, speed, and total movements during the 15-minute test period. 550 We measured tremor using our custom-built tremor monitor . Each