Spectrin-based Membrane Mechanics Is Asymmetric and Remodels during Neural Development

Perturbation of spectrin-based membrane mechanics causes hereditary elliptocytosis and spinocerebellar ataxia, but the underlying cellular basis of pathogenesis remains unclear. Here, we introduced the conserved disease-associated spectrin mutations into the C. elegans genome and studied the contribution of spectrin to neuronal migration and dendrite formation in developing larvae. The loss of spectrin generates an ectopic actin polymerization outside of the existing front and secondary membrane protrusions, leading to defective neuronal positioning and dendrite morphology in adult animals. Spectrin accumulates in the lateral and the rear of migrating neuroblasts and redistributes from the soma into the newly formed dendrites, indicating that the spectrin-based membrane skeleton is asymmetric and remodels to regulate actin assembly and cell shape during development. We affinity-purified spectrin from C. elegans and showed that its binding partner ankyrin functions with spectrin. Asymmetry and remodeling of membrane skeleton may enable spatiotemporal modulation of membrane mechanics for distinct developmental events. Significance Statement The biomechanical regulation of neural development is largely unknown. The spectrin-based membrane skeleton is essential for the structural integrity of the plasma membrane. This study addresses the function and behavior of spectrin in neuroblast migration and dendrite formation. The loss of spectrin generates an ectopic actin polymerization outside of the existing front, leading to defective neuronal positioning and dendrite morphology. Spectrin is absent from the leading edge but accumulates in the posterior of migrating neuroblasts and redistributes from the soma into the nascent dendrites, indicating that the membrane skeleton is asymmetric and remodels. Asymmetry and remodeling of the membrane skeleton may enable spatiotemporal modulation of membrane mechanics for distinct developmental events.


Introduction
The development of the nervous system requires the correct positioning of neurons and the accurate sprouting and elongation of neurites, laying the foundation for functional neuronal connectivity. The motility of neuronal cell and the growth cone is initiated by the environmental cue-activated transmembrane receptors and followed by asymmetric actin polymerization in the leading edge and the extension of the front of the cell (1-3). Despite the wealth of information on the biochemical mechanism of neural development, ranging from the guidance cues, receptors, the protein machinery that powers motility, the physical regulation of neural development, in particular, the underlying membrane mechanics, is mostly unexplored.
The spectrin-based membrane skeleton is essential for the structural integrity of the plasma membrane and protects cells from mechanical stresses in erythrocyte and many nonerythrocyte cells (4). The functional unit of spectrin is composed of a rod-shaped tetramer that consists of two antiparallel heterodimers of α-spectrin and β-spectrin (4). Spectrin interacts with the actin cytoskeleton and associated proteins to form a static polygonal lattice structure underneath the plasma membrane. Recent super-resolution imaging studies uncovered that spectrin creates an ordered periodic longitudinal array around the circumference of axons and dendrites (3,(5)(6)(7)(8). In addition to mechanic support, the membrane-associated periodic skeleton serves as a dynamic signaling platform for G protein-coupled receptors and cell adhesion molecules-mediated receptor tyrosine kinases (9). Mutations in human spectrin genes impair spectrin-based membrane mechanics and cause various diseases. For example, the hereditary elliptocytosis (HE)-associated αspectrin L260P mutation locks spectrin in the closed dimer conformation and destabilizes the plasma membrane (10,11). The spinocerebellar ataxia type 5 (SCA5)-associated β-IIIspectrin deletion disrupts the spectrin tetramer conformation and alters the localization of the synaptosomal proteins in the cerebellum (12,13). Knockout of β-II-spectrin in mouse neural progenitors disrupted axonal growth, stability, and synaptic cargo transport (14).
These studies establish the importance of spectrin in the nervous system and reveal the association of spectrin deficiency in neurodegenerative disorders; however, the function of the spectrin-based membrane mechanics in neural development remains unclear.
To model spectrin-associated diseases in animals, we have recently introduced the conserved disease-associated mutations into the C. elegans α-and β-spectrin genes (15).
These mutations affected membrane mechanics, and caused uncoordinated animal movement, mimicking the syndromes of the patients. Using these animals, we uncovered a novel function of spectrin in cilium formation in nematodes and showed the conserved mechanism in mammals (15). Here, we chose the C. elegans Q neuroblasts as a tractable experimental system to study the function and behavior of spectrin in developing neural progenitors. The Q neuroblasts, QR on the right side of the animal and QL on the left, undergo an identical pattern of asymmetric cell divisions, directional migration, apoptosis, and neurite growth in the first larval stage (16,17). The Q cell descendants move to highly stereotypic positions along the anterior and posterior body axis but in opposite directions, with QR and progenies (abbreviated as QR.x) migrating anteriorly and QL and progenies (QL.x) migrating posteriorly ( Fig. 1A-B). We developed fluorescence time-lapse microscopy and reporters to study cell polarity and the actin cytoskeleton in migrating Q cells (16,(18)(19)(20)(21)). Using the system, we show that spectrin is crucial for neuronal migration and dendrite formation and that the spectrin-based membrane skeleton is asymmetrically assembled and remodels in developing neuroblasts.

Spectrin regulates neuronal migration by preventing ectopic actin assembly
The germline deletion of α-spectrin (spc-1) or β-spectrin (unc-70) leads to embryonic lethality (22, 23). To study the function of spectrin in postembryonic neural development, we generated weak loss-of-function alleles or conditional mutations of spectrin. Our early study introduced the disease-associated α-spectrin L260P variation or β-III-spectrin inframe deletion into the C. elegans spc-1(cas971) or unc-70(cas983) locus, respectively (15). Both alleles produced viable progenies, allowing the dissection of the role of spectrin in neural development. As a complementary approach, we used the somatic CRISPR-Cas9 method (24) in which the Cas9 endonuclease is expressed under the control of the Q cell-specific promoter Pegl-17 or the heat-shock gene hsp-16.2 promoter Phsp to create conditional mutants of spc-1 within Q cell lineages or during Q cell development. T7 endonuclease I (T7EI)-based assays demonstrated that these transgenic animals produced molecular lesions with the expected sizes at the target loci of spc-1 after heat-shock induction of Cas9 expression (Fig. 1C-D). In line with the reported phenotypes of the spc-1(cas971) animals (15), the spc-1 conditional mutant embryos but not wild-type (WT) embryos exhibited embryonic lethality and larval arrest (Fig. 1E), demonstrating the success in producing conditional mutations in the spc-1 locus. In these animals, we examined the final positions of QR.ap and QL.ap, which will become the AQR and PQR neuron after long-distance migration towards the anterior or posterior, respectively.
Compared with WT animals, AQR localized at a more posterior position, and PQR confined at a more anterior region in spc-1(cas971) or unc-70(cas983) or somatic CRISPR-Cas9-created spc-1 conditional mutants, indicating an essential role of spectrin in cell migration ( Fig. 1B and Fig. S1A). The cell migration defects from Q-cell specific spc-1 mutants suggested a cell-autonomous contribution of spectrin to cell migration.
To understand the cellular mechanism by which the cell migration is reduced in spc-1 mutants, we performed fluorescence time-lapse analysis of cell morphology and the actin cytoskeleton in migrating Q.ap cells (16). We depicted the plasma membrane and nucleus using mCherry-tagged with a myristoylation signal and a histone and the actin cytoskeleton using GFP-tagged with the actin-binding domain of moesin (GFP::moesin ABD). In WT animals, AQR extends a large lamellipodium and forms a fan-like morphology ( Fig. 1F and Fig. S1B). Strikingly, the morphology of the spc-1-mutant AQR was perturbed, including the formation of a square-shaped cell without a persistent anterior-posterior polarity during migration ( Fig. 1F and Fig. S1B). In WT migrating AQR, GFP::moesin ABD-marked F-actin accumulates in the leading edge but is absent from the lateral and the rear. The spc-1-mutant AQR cell contains GFP::moesin ABD fluorescence in the leading edge (i.e., anterior); however, an ectopic GFP fluorescence is detectable in the lateral and the rear (i.e., posterior), resulting in an even distribution of F-actin along the entire plasma membrane (Fig. 1F). The fluorescence intensity ratio between the anterior and the posterior was reduced from 2.7-fold in WT to 1.9-fold in spc-1 mutant AQR cells (Fig. 1G), indicating a defective organization of the actin cytoskeleton. In agreement with the final reduction of AQR migration in adult animals, AQR failed to persistently extend the lamellipodium and translocate the cell body towards the anterior, reducing the migration distance and persistent polarity over time ( Fig. 1H-I). These results show that spectrin is essential for F-actin organization during cell migration. Given the abnormal actin assembly outside of the front, we postulate that the spectrin-based membrane mechanics may act in the posterior to inhibit ectopic actin polymerization.

Asymmetric distribution of membrane skeleton in migrating neuroblasts
To visualize the cellular distribution of the endogenous spectrin proteins during cell migration, we constructed the knock-in (KI) animals to mark SPC-1 and UNC-70 with GFP (15). We genetically crossed the GFP-spectrin knock-in markers into a transgenic animal that expresses mCherry-labeled plasma membrane and histone within Q cell lineages. GFP- We sought to pinpoint how spectrin became asymmetrically localized in migrating Q cells.
The GFP-spectrin knock-in strains illuminate spectrin in Q cells and their neighboring tissues, the high background fluorescence of which impedes the visualization of spectrin dynamics in Q cells. To resolve this problem, we adopted a self-complementing split fluorescent protein system (15,25). We inserted seven copies of GFP11 tag that encoded the eleventh beta-strand of super-folder GFP into the spc-1 locus and expressed GFP1-10 under the control of a Q cell-specific promoter Pegl-17. After genetically crossing an SPC-1::7XGFP11 knock-in animal with a Pegl-17::GFP1-10 transgenic animal, GFP selfcomplemented and specifically illuminated spectrin with green fluorescence in Q cells.
Using the strain, we confirmed that SPC-1::7XGFP fluorescence asymmetrically localized in the lateral and the rear of the migrating Q.ap cells, the enrichment fold of which elevated from 2.7-fold in GFP knock-in animals to 5.2-fold in Q-cell specific 7XGFP knock-in animals ( Fig. 2G), possibly because of an increase of green fluorescence and elimination of background signal. Next, we followed the dynamic changes of spectrin during Q cell development. After Q cell division, spectrin accumulates at the cleavage furrow between two daughter cells (0 min, Fig. 2F, Movie S1). Spectrin then forms puncta outside of the front (23 min) and becomes enriched in the posterior (34 -89 min), indicating an asymmetric assembly of the spectrin-based membrane skeleton in migrating cells ( Fig. 2G and Fig. S1D).

Membrane skeleton remodels to safeguard dendrite formation
Our fluorescence time-lapse microscopy revealed that the timing and position of sprouting dendrites couples with cell migration. After the A/PQR neuron reaches its destination, the cell body stops, but the same leading-edge continues to elongate, transforming into the growth cone that forms the dendrite (Movie S2) (19, 26). We examined whether and how the spectrin-based membrane mechanics regulates dendrite growth. In 10~30% of the spc-1(cas971) or unc-70(cas983) mutant animals, the A/PVM and A/PQR neurons developed neurites whose positioning and extent of growth were perturbed, and the neurites were defective in the formation, guidance or outgrowth ( We performed live imaging analysis of dendrite formation in PQR to address how spectrin regulates dendrite development. In WT animals, the PQR cell body arrived at the final destination, and the leading-edge continued to elongate, forming the growth cone for dendrite formation (Fig. 3D and Movie S2). The F-actin accumulated at the leading edge, and the newly elongated dendrite became thin and stable (Fig. 3D). In spectrin-mutants, the nascent dendrite was unstable and developed the F-actin-enriched secondary growth cones that elongated and generated ectopic dendritic branches ( Fig. 3D and Movie S3).
To understand spectrin behavior during dendrite formation, we followed SPC-1::7XGFP in the growing dendrite of the WT PQR neuron: SPC-1 initially accumulated in the lateral and the rear of the migrating PQR; however, SPC-1 transited from the soma to the newly formed dendrite and became enriched in the dendrite (Fig. 3E). We did not detect SPC-1::7XGFP in the leading edge of the growth cone during dendrite outgrowth ( Fig. 3E- which is consistent with the absence of spectrin in the leading edge of migrating cells. Quantification of the SPC-1::7XGFP ratio between the dendrite and the leading edge revealed a 5.3-fold enrichment (Fig. 3F). Together, these results show that the spectrinbased membrane skeleton disassembles in the soma and assembles in the nascent dendrite, where spectrin inhibits ectopic actin polymerization and safeguards dendrite morphology.

Ankyrin functions with spectrin during neural development
To dissect the molecular mechanism of how the spectrin-based membrane skeleton regulates neuronal cell migration and neurite growth, we used affinity purification and mass spectrometry to identify interacting proteins of SPC-1. Using an anti-green fluorescent protein (GFP) antibody, we purified GFP-tagged SPC-1 with its associated proteins from the lysate of SPC-1::GFP knock-in larvae (Fig. 4A). We determined protein constituents employing liquid chromatography-tandem mass spectrometry (LC-MS) (Fig.   4B). We detected the beta-spectrin subunit UNC-70, validating our purification scheme.
Another hit was UNC-44, an ortholog of human ankyrin protein, which was a binding partner of the membrane skeleton (27). In C. elegans, UNC-44 regulates multiple neuronal events, including axon guidance, microtubule organization, and axon-dendrite sorting and transport (28, 29), but its functions in neuronal cell migration and dendrite development are less understood.
Using the unc-44(e362) allele or conditional unc-44 mutants generated by somatic CRISPR-Cas9, we showed that the loss of UNC-44 reduced Q cell migration in adult animals and disrupted the neurites in Q cell progenies as those in the spectrin mutant animals (Fig. 4H). By creating a UNC-44::GFP knock-in animal, we showed that UNC-44 was absent from the leading edge but accumulated in the lateral and the rear of migrating Q cells (Fig. 4I, and Movie S4 ), which is similar to spectrin distribution. We have previously reported the function of spectrin in ciliogenesis. In further support of the notion that UNC-44 acts with spectrin in the same processes, we found that the ciliary length and the capacity of the animal to interact with the environment (i.e., dye-filling) were defective in the loss of UNC-44 (Fig. 4E-G). Using super-resolution live imaging, we showed that UNC-44 displayed periodicity in the axon of motor neurons (Fig. 4C). Double fluorescence imaging analysis revealed that the fluorescence intensity peaks from UNC-44::GFP and SPC-1::mCherry showed intercalation in the axon (Fig. 4C), which is consistent with the current model on the organization of the spectrin-ankyrin skeleton (3,8,27). These results show that ankyrin functions together with spectrin in neuronal cell migration, the formation of the dendrite, and sensory cilia.

Discussion
We provide evidence that the spectrin-ankyrin-based membrane skeleton is essential for neuronal cell migration and dendrite formation. This work reveals that the membrane skeleton is asymmetric and remodels during neural development: When neuroblasts migrate, spectrin accumulates in the lateral and the rear; when neuroblasts use the same leading-edge to grow dendrite, spectrin transits from the soma to the nascent dendrites. In comparison to the polarized and dynamic actin and microtubule cytoskeleton, asymmetry and remodeling are not well appreciated for the membrane skeleton. The previous studies on spectrin used "snapshot" imaging or fixed samples and described the membrane skeleton as an immobile and stable structure underneath the entire plasma membrane (3,6,8,27). In addition to its rigidity, asymmetry and remodeling can be essential for the function of the spectrin-based membrane skeleton. A polarized membrane skeleton provides the locally distinct membrane mechanics in a living cell (e.g., cell migration), and the ability to remodel enables a developing cell to change its membrane mechanics at different developmental events (e.g., from cell migration to neurite outgrowth).
How can the spectrin-based membrane skeleton be asymmetric? Fluorescence microscopy of the cell from birth to migration uncovers that the membrane skeleton becomes asymmetrically assembled in the posterior (Fig. 2F) The findings that the hereditary elliptocytosis-associated α-spectrin mutation and the spinocerebellar ataxia type 5-associated β-III-spectrin deletion are defective in neuronal cell migration and neurite formation support the notion that the spectrin-based membrane mechanics plays multiple roles in neural development, providing additional insights into SCA5 pathogenesis. Our results will motivate future investigations to examine whether neuronal migration or axon and dendrite formation are abnormal in Purkinje cells and granule cell progenitors with hereditary elliptocytosis or spinocerebellar ataxia. The manipulation of mechanobiology in the nervous system may merit exploration as a potential therapeutic strategy for some types of neurodegenerative disorders.

C. elegans strains and genetics
C. elegans strains were raised on NGM plates seeded with E. coli strain OP50 at 20℃.
Tables S1-S3 summarizes the primers, plasmids, and strains used in this study.

Molecular biology
CRISPR-Cas9 targets were inserted to the pDD162 vector (Addgene #47549) by linearizing this vector with primers listed in Table S1

Mass spectrometry analysis
Unsynchronized SPC-1::GFP knock-in strains raised on 80~100 90-mm NGM plates were collected and washed for three times with M9 buffer. Lysates were made from 1~2 mL Thermo Fisher Scientific). The protein components in the list in Figure 4 were reproducibly identified from three biological repeats.
At least 100 worms of each strain were examined in two independent assays.

Quantifications and statistical analysis
ImageJ software was used to circumscribe the fluorescence field and to measure the fluorescence intensity. In all the intensity quantifications, the background was subtracted.
When the F-actin marker GFP::moesinABD or knock-in fluorescence protein signal intensity was measured along the cortex, "leading" refers to the circumference of the lamellipodium (the wide-spread area) and "rear" refers to the circumference around the rest of cell body, and the intensity ratio between the two was calculated using the corresponding mCherry fluorescence intensity as a control. The migration distances were measured as the cell body movement (in μm per hour) along the anterior-posterior body axis. The migration angle was measured between the protrusion and the anterior-posterior body axis, the anterior was defined as degree 0, and the posterior was defined as degree 180.
Quantification was represented by the mean value ± standard deviation for each group. N represents the number of animals used for the corresponding quantification, and the onesample or two-tailed Student's t-test or χ 2 analysis were employed to determine the statistical differences as indicated in figure legends.