The β-catenin (Brault et al., 2001) and γ-catenin (Demireva et al., 2011) alleles, Olig2::Cre (Dessaud et al., 2007) and ChAT::Cre (Lowell et al., 2006) lines were generated as previously described and maintained on a mixed background. Mouse colony maintenance and handling was performed in compliance with protocols approved by the Institutional Animal Care Use Committee of Case Western Reserve University. Mice were housed in a 12-h light/dark cycle in cages containing no more than five animals at a time.

In situ hybridization and immunohistochemistry was performed as previously described (Philippidou et al., 2012; Vagnozzi et al., 2020), on tissue fixed for 2 h in 4% paraformaldehyde (PFA) and cryosectioned at 16 μm. In situ probes were generated from e12.5 cervical spinal cord cDNA libraries using PCR primers with a T7 RNA polymerase promoter sequence at the 5’ end of the reverse primer. All probes generated were 750–1000 bp in length. Wholemounts of diaphragm muscles were stained as described (Philippidou et al., 2012). The following antibodies were used: rabbit anti-cleaved Caspase 3 (1:1000; Cell Signaling, RRID:AB_2341188), mouse anti-olig2-A488-conjugated (1:500; Millipore, RRID:AB_11205039), rabbit anti-β-Catenin (1:250; Cell Signaling Technology, RRID:AB_11127203), goat anti-Scip (1:5000; Santa Cruz Biotechnology, RRID:AB_2268536), mouse anti-Islet1/2 (1:1000, DSHB, RRID:AB_2314683) (Tsuchida et al., 1994), rabbit anti-neurofilament (1:1000; Synaptic Systems, RRID:AB_887743), rabbit anti-synaptophysin (1:250, Thermo Fisher, RRID:AB_10983675), and α-bungarotoxin Alexa Fluor 555 conjugate (1:1000; Invitrogen, RRID:AB_2617152). Images were obtained with a Zeiss LSM 800 confocal microscope and analyzed with Zen Blue, ImageJ (Fiji), and Imaris (Bitplane). Phrenic MN number was quantified using the Imaris “spots” function to detect cell bodies that coexpressed high levels of Scip and Isl1/2 in a region of interest limited to the left and right sides of the ventral spinal cord.

For labeling of phrenic MNs, crystals of carbocyanine dye, DiI (Invitrogen, #D3911) were pressed onto the phrenic nerves of eviscerated embryos at e18.5, and the embryos were incubated in 4% PFA at 37°C in the dark for 4–5 weeks. Spinal cords were then dissected, embedded in 4% low melting point agarose (Invitrogen) and sectioned using a Leica VT1000S vibratome at 100–150 μm.

MN positional analysis and correlation analysis were performed as previously described (Dewitz et al., 2018, 2019). MN positions were acquired using the “spots” function of the imaging software Imaris (Bitplane) to assign x and y coordinates. Coordinates were expressed relative to the midpoint of the spinal cord midline, defined as position x = 0, y = 0. To account for experimental variations in spinal cord size, orientation, and shape, sections were normalized to a standardized spinal cord whose dimensions were empirically calculated at e13.5 (midline to the lateral edge = 390 μm). We analyzed every other section containing the entire PMC (20–30 s in total per embryo).

For the analysis of dendritic orientation, we superimposed a radial grid divided into eighths (45° per octant) centered over phrenic MN cell bodies spanning the entire length of the dendrites. We drew a circle around the cell bodies and deleted the fluorescence associated with them. Fiji (ImageJ) was used to calculate the fluorescent intensity (IntDen) in each octant which was divided by the sum of the total fluorescent intensity to calculate the percentage of dendritic intensity in each area.

Electrophysiology was performed as previously described (Vagnozzi et al., 2020). Mice were cryoanesthetized and rapid dissection was carried out in 22–26°C oxygenated Ringer’s solution. The solution was composed of 128 mM NaCl, 4 mM KCl, 21 mM NaHCO₃, 0.5 mM NaH₂PO₄, 2 mM CaCl₂, 1 mM MgCl₂, and 30 mM D-glucose and was equilibrated by bubbling in 95% O₂/5% CO₂. The hindbrain and spinal cord were exposed by ventral laminectomy, and phrenic nerves exposed and dissected free of connective tissue. A transection at the pontomedullary boundary rostral to the anterior inferior cerebellar artery was used to initiate fictive inspiration. Electrophysiology was performed under continuous perfusion of oxygenated Ringer’s solution from rostral to caudal. Suction electrodes were attached to phrenic nerves just proximal to their arrival at the diaphragm. The signal was band-pass filtered from 10 Hz to 3 kHz using AM-Systems amplifiers (Model 3000), amplified 5,000-fold, and sampled at a rate of 50 kHz with a Digidata 1440A (Molecular Devices). Data were recorded using AxoScope software (Molecular Devices) and analyzed in Spike2 (Cambridge Electronic Design). Burst duration and burst activity were computed from 4 to 5 bursts per mouse, while burst frequency was determined from 10 or more minutes of recording time per mouse. Burst activity was computed by rectifying and integrating the traces with an integration time equal to 2 s, long enough to encompass the entire burst. The maximum amplitude of the rectified and integrated signal was then measured and reported as the total burst activity.

Conscious, unrestrained P0 mice were placed in a whole body, flow-through plethysmograph (emka) attached to a differential pressure transducer (emka). We modified 10 ml syringes to use as chambers, as smaller chambers increase signal detection in younger mice. Experiments were done in room air (79% nitrogen, 21% oxygen). Mice were placed in the chamber for 30 s at a time, for a total of three to five times, and breathing parameters were recorded. Mice were directly observed to identify resting breaths. At least ten resting breaths were analyzed from every mouse. Data are presented as fold control, where the control is the average of 2 littermates in normal air.

For all experiments a minimum of three embryos per genotype, both male and female, were used for all reported results unless otherwise stated. The Shapiro–Wilk test was used to determine normality. All data, with the exception of electrophysiology burst activity data in Figures 2, 6, showed normal distribution and p-values were calculated using unpaired, two-tailed Student’s t-test. Burst activity data in Figures 2, 6 showed non-normal distribution and p-values were calculated using the Mann–Whitney test. p < 0.05 was considered to be statistically significant, where *p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001. Data are presented as box and whisker plots with each dot representing data from one mouse unless otherwise stated. Small open squares in box and whisker plots represent the mean.

Phrenic MNs express a distinct combinatorial cadherin code (Machado et al., 2014; Vagnozzi et al., 2020), but the collective contribution of these molecules to phrenic MN development, maintenance and function has not been established. We previously found that phrenic MNs express the type I N-cadherin (N-cad) and the type II cadherins Cdh6, 9, 10, 11, and 22 (Vagnozzi et al., 2020). To investigate the role of cadherin signaling in phrenic MN development, we previously eliminated cadherin signaling in MN progenitors by inactivating β- and γ-catenin using a Olig2::Cre promoter (β-catenin flox/flox; γ-catenin flox/-; Olig2::Cre). β- and γ-catenin are obligate intracellular factors required for cadherin-mediated cell adhesive function and catenin deletion enables us to interrogate the full repertoire of cadherin actions in phrenic MNs (Figure 1A). We found that loss of β- and γ-catenin led to phrenic MN disorganization, however we were unable to assess the role of catenins in respiratory behavior and function as these mice die by e14.5 (Vagnozzi et al., 2020). Therefore, we modified our genetic strategy and generated β-catenin flox/flox; γ-catenin flox/flox; Olig2::Cre mice, referred to as βγ-cat^MNΔ mice. βγ-cat^MNΔ mice show loss of β- and γ-catenin in both MN progenitors and MNs (expressing the TF Isl1/2) by e11.5 (Supplementary Figure 1A). We find that βγ-cat^MNΔ mice are born alive but die within 24 h of birth and often display severe flexion of the wrist joint (Figure 1B).

In order to assess breathing in βγ-cat^MNΔ mice, we utilized unrestrained whole body flow-through plethysmography at postnatal day (P)0 (Figure 1C). We found that βγ-cat^MNΔ mice have a 45% reduction in tidal volume (the amount of air inhaled during a normal breath), while respiratory frequency is not affected (Figures 1D, E). This results in an average 45% reduction in overall air drawn into the lungs per minute (minute ventilation, Figure 1E), indicating βγ-cat^MNΔ mice likely die from respiratory failure. Our findings indicate that catenin signaling is necessary for proper respiratory behavior and survival. To further examine respiratory circuitry intrinsic to the brainstem and spinal cord, we performed suction recordings of the phrenic nerve in isolated brainstem-spinal cord preparations (Figure 2A). These preparations display fictive inspiration after the removal of inhibitory networks in the pons via transection, and thus represent a robust model to interrogate circuit level changes. We examined whether catenin deletion impacts circuit output at embryonic day (e)18.5/P0, shortly before βγ-cat^MNΔ mice die. We observed a striking reduction in the activation of phrenic MNs in βγ-cat^MNΔ mice (Figure 2B). While bursts in control mice exhibit large peak amplitude, bursts in βγ-cat^MNΔ mice were either of very low amplitude (∼85%) or non-detectable (∼15%). After rectifying and integrating the traces, we found a nearly 70% decrease in total burst activity in βγ-cat^MNΔ mice (Figure 2C). Our data indicate that cadherin signaling is imperative for robust activation of phrenic MNs during inspiration.

What accounts for the loss of phrenic MN activity in βγ-cat^MNΔ mice? We asked whether early phrenic MN specification, migration and survival are impacted after catenin inactivation. We acquired transverse spinal cord sections through the entire PMC at e11.5 and stained for the TF Olig2 to label MN progenitors and the phrenic-specific TF Scip and MN-specific TF Isl1/2 to label all phrenic MNs (Supplementary Figure 1B). We found that MN progenitor numbers do not change in βγ-cat^MNΔ mice. However, their distribution along the dorsoventral axis is altered, indicating that catenin signaling is not required for MN progenitor generation but is necessary for progenitor restricted localization within a narrow band (Supplementary Figures 1B, C). We found a cluster of Scip+ MNs in the ventral cervical spinal cord of both control and βγ-cat^MNΔ mice, indicating that early phrenic MN specification is unperturbed (Supplementary Figure 1B). βγ-cat^MNΔ mice show a non-significant reduction of phrenic MNs at e11.5 (Supplementary Figure 1C). However, we observed a significant reduction in phrenic MN numbers by e13.5 in βγ-cat^MNΔ mice, indicating that catenin signaling is necessary for phrenic MN survival (Figures 3A, B). While we do not observe a significant increase in activated caspase 3 levels in βγ-cat^MNΔ mice at e11.5, it is possible that the relatively high levels of apoptosis at this stage mask the increase in cell death in the relatively small phrenic MN population (Supplementary Figures 1B, C).

While phrenic MNs are normally distributed along the rostrocaudal axis, we found that they sometimes show migratory defects, where several phrenic MNs remain close to the midline instead of fully migrating (Figure 3C, arrow), and also appear to shift both ventrally and medially in βγ-cat^MNΔ mice. To quantitate PMC cell body position, each phrenic MN was assigned a cartesian coordinate, with the midpoint of the spinal cord midline defined as (0,0). Bγ-cat^MNΔ mice displayed a significant shift in phrenic MN cell body position, with cell bodies shifting ventrally toward the edge of the spinal cord and toward the midline (Figures 3C–F). We quantified the average ventrodorsal and mediolateral phrenic MN position per embryo and found a significant change in phrenic MN position in both axes in βγ-cat^MNΔ mice (Figure 3G). Correlation analysis indicated that control and βγ-cat^MNΔ mice are dissimilar from each other (r = 0.47, Figure 3H), indicating that catenins establish phrenic MN coordinates during development.

In addition to changes in cell body position, we also noticed that phrenic MNs appear to lose their tight clustering in βγ-cat^MNΔ mice. PMC clustering is thought to be critical for the proper development of the respiratory system because it facilitates recruitment of motor units through electrical coupling in the embryo to compensate for weak inspiratory drive (Greer and Funk, 2005). In order to determine PMC clustering defects, we used Imaris to measure the average distance between phrenic MNs. We found that βγ-cat^MNΔ mice had a nearly 50% increase in the average distance between neighboring cells (Figure 3I), indicating that the cadherin/catenin adhesive complex contributes to the formation of a tightly clustered phrenic motor column.

Since cadherin/catenin signaling is imperative for phrenic MN organization, we also asked whether any other aspects of phrenic MN development, such as dendritic and axon growth, might rely on catenin actions. We examined dendritic orientation in control and βγ-cat^MNΔ mice by injecting the lipophilic dye diI into the phrenic nerve at e18.5 (Figure 4A). DiI diffuses along the phrenic nerve to label both PMC cell bodies and dendrites. Consistent with our earlier observations, we found that phrenic cell bodies are often scattered in βγ-cat^MNΔ mice. Interestingly, even phrenic MNs that are significantly displaced correctly project along the phrenic nerve (arrows, Figure 4A), indicating that changes in cell body position do not impact axon trajectory choice. In control mice, phrenic MN dendrites branch out in dorsolateral to ventromedial directions; in βγ-cat^MNΔ mice, however, they exhibit stunted growth, defasciculation and a failure to extend in the dorsolateral direction (Figure 4A).

To quantify these changes, we superimposed a radial grid divided into octants onto the dendrites and measured the fluorescent intensity in each octant after removal of any fluorescence associated with the cell bodies. Zero degrees was defined by a line running perpendicular from the midline through the center of cell bodies. In control mice, the majority of dendrites project in the dorsolateral direction (0–90°), representing 40–45% of the overall dendritic intensity (Figures 4B, D, E). Ventrally projecting dendrites (180–225° and 315–360°) were also prominent, giving rise to nearly 30% of the overall dendritic intensity (Figures 4B, D, E). We found that catenin deletion resulted in a striking reduction in dorsolateral dendrites, together with a significant increase in ventral dendrites (Figures 4C–E), indicating that cadherins are necessary for establishing phrenic MN dendritic orientation. These changes in phrenic dendritic topography in βγ-cat^MNΔ mice may impact their targeting by respiratory populations in the brainstem, leading to the reduction in phrenic MN activation observed.

We then asked whether catenins might also play an analogous role in phrenic axon extension and arborization. We examined diaphragm innervation in control and βγ-cat^MNΔ mice at e18.5. We found that βγ-cat^MNΔ mice display a lack of innervation in the ventral diaphragm (Figure 5A, arrow), while the parts of the diaphragm that are innervated show a reduction in terminal arborization complexity (Figure 5A, star). Quantitation of total phrenic projections revealed a significant reduction in overall diaphragm innervation (Figures 5B, C). Collectively our data point to a catenin requirement for phrenic MN topography and axon and dendrite arborization, suggesting that these changes in early phrenic MN development lead to loss of phrenic MN activity and perinatal lethality due to respiratory insufficiency.

Given the essential role for catenins in early phrenic MN development, we wanted to further understand the temporal dynamics of cadherin signaling, and asked whether sustained catenin expression is required for maintenance of the respiratory circuit. We used a ChAT::Cre promoter to specifically delete β- and γ-catenin in postmitotic MNs (β-catenin flox/flox; γ-catenin flox/flox; ChAT::Cre, referred to as βγ-cat^ChATMNΔ mice). βγ-cat^ChATMNΔ mice show MN-specific loss of catenin expression by e13.5 (Supplementary Figure 1D), survive to adulthood and do not display respiratory insufficiency or gasping at birth. We first assessed cell body position and found no changes between control and βγ-cat^ChATMNΔ mice (Figure 6A). Injecting diI into the phrenic nerve also revealed no differences in dendritic orientation between control and βγ-cat^ChATMNΔ mice (Figure 6B). To assess respiratory circuit function, we performed phrenic nerve recordings in isolated brainstem-spinal cord preparations in control and βγ-cat^ChATMNΔ mice at P4, and found that βγ-cat^ChATMNΔ mice exhibit similar phrenic MN bursting frequency, burst duration and overall activity as control mice (Figures 6C, D). Our data suggests that cadherins engage catenin signaling during a short temporal window early in development to shape respiratory motor output, but appear to be dispensable once early phrenic MN topography and morphology have been established.