β-III-spectrin N-terminus is required for high-affinity actin binding and SCA5 neurotoxicity

Recent structural studies of β-III-spectrin and related cytoskeletal proteins revealed N-terminal sequences that directly bind actin. These sequences are variable in structure, and immediately precede a conserved actin-binding domain composed of tandem calponin homology domains (CH1 and CH2). Here we investigated in Drosophila the significance of the β-spectrin N-terminus, and explored its functional interaction with a CH2-localized L253P mutation that underlies the neurodegenerative disease spinocerebellar ataxia type 5 (SCA5). We report that pan-neuronal expression of an N-terminally truncated β-spectrin fails to rescue lethality resulting from a β-spectrin loss-of-function allele, indicating that the N-terminus is essential to β-spectrin function in vivo. Significantly, N-terminal truncation rescues neurotoxicity and defects in dendritic arborization caused by L253P. In vitro studies show that N-terminal truncation eliminates L253P-induced high-affinity actin binding, providing a mechanistic basis for rescue. These data suggest that N-terminal sequences may be useful therapeutic targets for small molecule modulation of the aberrant actin binding associated with SCA5 β-spectrin and spectrin-related disease proteins.


Results
Delineation of the β-III-spectrin N-terminus. Prior cryo-EM analysis showed that the N-terminus of β-III-spectrin forms an extended α-helix that is tightly associated with actin, and is contiguous with helix A of CH1. To clarify the boundary between N-terminus and CH1, the β-III-spectrin sequence was aligned to other tandem-CH ABD proteins to assess amino acid conservation. The alignment shows low conservation of residues in the N-terminal region corresponding to β-III-spectrin residues 1-52, Fig. 1A. In contrast, beginning at β-III-spectrin residue D53, sequence conservation is clearly elevated. Thus, we define D53 as the first residue of the conserved CH1, similar to a prior CH1 demarcation 18 ; the novel N-terminal α-helix identified by cryo-EM is composed of N-terminal residues S37-A52, and is contiguous with CH1 helix A that begins at residue D53, Fig. 1B. The N-terminal sequence is well-conserved between human β-III-spectrin and Drosophila β-spectrin ( Fig. 1C), supporting the functional importance of the N-terminus across species. N-terminus is required for β-spectrin function and SCA5 toxicity in Drosophila. Our prior biochemical truncation studies showed that N-terminal residues 1-51 are required for binding of the β-III-spectrin to actin in vitro. To evaluate the importance of the N-terminus to β-spectrin function in vivo, new Drosophila lines were generated carrying transgenes consisting of an upstream activation sequence (UAS) fused to the Drosophila β-spectrin coding sequence, with or without the N-terminal truncation (ΔN; see Fig. 1C). The UASβspec transgenes were inserted into the attP154 genomic locus 30 for equal expression. Rescue experiments were performed to assess the ability of the βspec transgenes to support viability. Males carrying βspec transgenes were crossed to females carrying an X-chromosome containing the pan-neuronal driver, elav-Gal4, and an emsinduced mutation (em21) in the endogenous βspec gene. βspec em21 is a recessive, loss-of-function allele containing a nonsense mutation in the thirteenth spectrin-repeat domain 31 , resulting in low level expression of a truncated β-spectrin protein, Fig. S1. The βspec em21 allele causes lethality in males and homozygous females. In contrast females heterozygous for βspec em21 are healthy. Pan-neuronal expression of transgenic wild-type β-spectrin (βspec WT ) rescued adult viability in ~ 25% of male progeny receiving the βspec em21 mutant X-chromosome, Table 1. In contrast, expression of N-terminally truncated β-spectrin (βspec ∆N-WT ) failed to rescue βspec em21 lethality. This indicates that the N-terminus is required for β-spectrin function in vivo. In addition, expression of N-terminally truncated β-spectrin (βspec ∆N-WT ) in heterozygous βspec em21 females reduced viability by ~ 30%. This indicates that the N-terminal truncation mutant is mildly neurotoxic. Moreover, raising Drosophila incubation temperature from 22 to 25 °C, to increase expression of the N-terminal truncation mutant, reduced survival by ~ 95%, Table S1 This indicates that toxicity of the N-terminal truncation mutant is dose responsive, consistent with a dominant negative mechanism. Additionally, pan-neuronal expression of βspec ∆N-WT in the absence of βspec em21 also caused lethality (Table S2), indicating that βspec ∆N-WT neurotoxicity is not due to a possible functional interaction with the βspec em21 protein fragment.
Using the rescue assay, we further tested how the N-terminus functionally interacts with the CH2-localized L253P mutation. In agreement with our prior results 14 , pan-neuronal expression of β-spectrin carrying the SCA5 Figure 1. Delineation of N-terminus. (A) Alignment of N-terminal amino acid sequences preceding CH1. Tall and dark red bars indicate high conservation, while short and dark blue bars indicate low conservation. Based on increased conservation, we define the first residue of β-III-spectrin CH1 as D53. Amino acids 1-52 constitute the N-terminus. (B) Structure of β-III-spectrin CH1 (green) and N-terminal residues N35-A52 (magenta) in complex with actin (grey) (PDB: 6ANU). N-terminal residues S37-A52 form a helix that is continuous with helix A of CH1. (C) Alignment of human (Hs) and Drosophila (Dm) β-spectrin showing high conservation in N-terminal residues. Arrow indicates the site of N-terminal truncation chosen for Drosophila ΔN transgenes and binding assays.

N-terminal truncation alleviates SCA5-induced dendritic arbor defects.
To determine how N-terminal truncation impacts SCA5-induced dendritic arborization defects, βspec transgenes were expressed using the 477-Gal4 driver in Drosophila class IV da neurons. Relative to neurons expressing βspec WT , neurons expressing βspec SCA5 showed reduced arborization characterized by a loss of distal dendrites in Sholl analysis, and decreased total branch length and number of branch points (Fig. 2), consistent with our prior characterization 14 .
In contrast, neurons expressing βspec ΔN-SCA5 displayed a dendritic arbor morphology indistinguishable from control neurons expressing the wild-type transgene. This indicates that the N-terminus is required for the SCA5 mutation to induce dendritic arbor defects. Moreover, the N-terminally truncated wild-type β-spectrin (βspec ΔN-WT ) did not reduce arborization, consistent with mild neurotoxicity observed in rescue assays at 22 °C.

N-terminus modulates β-spectrin protein level in neurons.
To evaluate the potential impact of N-terminal truncation on β-spectrin protein level, an additional set of Drosophila lines were generated containing UAS-transgenes, consisting of β-spectrin, with or without N-terminus, fused at the C-terminus to GFP. The transgenes encoding the β-spectrin-GFP fusion proteins were inserted into the attP154 genomic locus, as performed for the untagged β-spectrins. In rescue assays, the β-spectrin-GFP fusions performed similar to their untagged counterparts (Table S3), indicating that the GFP tag does not interfere with function. To assess relative abundance of wild-type and N-terminally truncated β-spectrins, female βspec em21 heterozygotes expressing UASβspec GFP transgenes under the pan-neuronal elav-Gal4 driver were collected from rescue assays at 18 °C and head extracts prepared. Using a GFP antibody, western blot analysis of UAS-βspec WT-GFP extracts revealed a prominent band running over 250 kDa, consistent with the predicted size (293 kDa) of the full-length β-spectrin-GFP protein, Fig. 3. Significantly, in the soluble fraction, N-terminally truncated wild-type β-spectrin was ~ fivefold more abundant than wild-type β-spectrin with intact N-terminus. This suggests that the inability of the N-terminally truncated β-spectrin to support viability is not due to insufficient protein level. Instead, it suggests that the N-terminus is essential for β-spectrin function in neurons, presumably to enable actin binding. Moreover, the increased steady-state β-spectrin protein level potentially reflects the effect of N-terminal truncation to stabilize wild-type β-spectrin protein, as observed in thermal denaturation studies 23 . We likewise evaluated the abundance of SCA5 β-spectrin protein, with and without N-terminus. At 18 °C, a fraction of female βspec em21 heterozygotes expressing the toxic SCA5 mutant β-spectrin survive to adulthood, enabling evaluation of SCA5 mutant protein abundance in head extracts. For SCA5 β-spectrin with intact N-terminus, western blot analysis revealed that soluble SCA5 β-spectrin protein is ~ threefold more abundant than wild-type β-spectrin with intact N-terminus, Fig. 3. SCA5 β-spectrin was also more abundant than wild-type in the insoluble fraction. The increased SCA5 β-spectrin in the insoluble fraction may reflect its high affinity for actin, as the insoluble fraction is enriched in actin. Altogether, the data indicate that the SCA5 mutation, in addition to causing high-affinity actin binding, also causes the mutant β-spectrin protein to accumulate in neurons. It is possible that a high-affinity complex formed between SCA5 β-spectrin and F-actin is resistant to protein turnover. Interestingly, western blotting revealed a fraction of actin with reduced gel mobility in SCA5 β-spectrin head extracts, Fig. 3. Potentially high-affinity binding of SCA5 β-spectrin to F-actin results in posttranslational or structural modifications to actin.
Further, the average level of soluble, N-terminally truncated SCA5 β-spectrin protein was 67% lower than wild-type β-spectrin with intact N-terminus, and 89% lower than SCA5 β-spectrin, Fig. 3. In addition, the N-terminally truncated SCA5 β-spectrin did not accumulate in the insoluble fraction, nor cause reduced gel mobility of actin, as observed for SCA5 β-spectrin with intact N-terminus. The lower abundance of the N-terminally truncated SCA5 β-spectrin likely contributes to its the lack of toxicity. However, toxicity was not observed for N-terminally truncated SCA5 β-spectrin when incubation temperature was elevated to increase transgene expression, Table S1. This suggests that the lack of toxicity of N-terminally truncated SCA5 β-spectrin is not www.nature.com/scientificreports/ solely due to reduced protein level. Instead N-terminal truncation may cause loss of a toxic functional property conferred by the SCA5 mutation. We evaluated whether the SCA5 mutation or N-terminal truncation impacts the binding of β-spectrin to α-spectrin. Co-immunoprecipitation assays showed that neither the SCA5 mutation nor N-terminal truncation disrupts the co-association of the mutant β-spectrin proteins with α-spectrin in head extracts, Figs. S4 and S5. This indicates that the mutant β-spectrin proteins are not impaired in their ability to bind α-spectrin nor assemble into a spectrin heterotetramer.

L253P causes protein destabilization independent of the N-terminus.
To assess how N-terminal truncation impacts the folded state and stability of the SCA5 mutant β-spectrin, circular dichroism spectroscopy was performed. As previously reported, the β-III-spectrin L253P ABD with intact N-terminus showed a pronounced α-helical absorption profile consistent with a well-folded ABD, Fig. 4A. Further, thermal denaturation studies showed that the L253P ABD unfolded in a cooperative, two-state transition with a melting temperature (Tm) of 44.9 °C, Fig. 4B. This Tm is 14.8 °C lower than the Tm of the wild-type ABD with intact N-terminus (59.7 °C). The circular dichroism absorption spectra of the N-terminally truncated L253P ABD similarly showed an α-helical absorption profile indicating a well-folded protein, Fig. 4A. Significantly, the N-terminally truncated L253P ABD unfolded with a Tm of 46.1 °C. Thus, the N-terminally truncated L253P ABD, like the L253P ABD with intact N-terminus, is highly destabilized relative to wild-type. Notably, N-terminal truncation of the L253P ABD or wild-type ABD raised the Tm by ~ 1 °C relative to the L253P or wild-type ABD with intact N-terminus, respectively. We previously reported a 1 °C increase in Tm for the N-terminally truncated wild-type ABD 23 , and suggested that this reflects the loss of an N-terminus that is structurally disordered when not bound to actin. showing that only βspec SCA5 significantly alters total branch length relative to βspec WT . Bottom, quantitation showing that only βspec SCA5 impacts total branch points. n = 11-14 neurons for each genotype. Statistical significance was determined by One-way ANOVA followed by multiple comparisons post hoc test. P-values are reported to the right of panels. www.nature.com/scientificreports/ N-terminus is required for L253P-induced high-affinity actin binding. To determine whether N-terminal truncation impacts L253P-induced high-affinity actin binding, in vitro co-sedimentation assays were performed. Consistent with high-affinity binding previously reported 17 , the L253P ABD with intact N-terminus was entirely bound to F-actin at all actin concentrations tested (3-140 µM), Figs. 4C, S6 and S7. In contrast, the N-terminally truncated L253P ABD bound actin with very low affinity (Kd = 234 µM). Indeed, the actin-binding affinity of the N-terminally truncated L253P ABD is ~ 3000-fold lower than the affinity of the L253P ABD with intact N-terminus (Kd = 75 nM; 17 , and threefold lower than the affinity of the wild-type ABD with intact N-terminus (74 µM). The binding affinity of the N-terminally truncated L253P ABD is higher than that of the N-terminally truncated wild-type ABD, which did not bind actin with measurable affinity. To confirm the in vitro co-sedimentation data, we employed a recently developed Förster resonance electron transfer (FRET) assay that detects the binding of the β-III-spectrin ABD to F-actin in live cells 32 . In this assay, HEK293T cells are transfected with a DNA construct consisting of the ABD fused to GFP, and a second construct consisting of the actin-binding peptide Lifeact fused to mCherry. FRET between GFP and mCherry results from the formation of a ternary complex consisting of the ABD-GFP and Lifeact-mCherry bound to an actin filament. Co-expression of the wild-type ABD and Lifeact-mCherry resulted in an average FRET efficiency of 3.9%, Fig. 4D. In contrast, the L253P mutant ABD showed an increased FRET efficiency of 8.4%, indicating that the L253P mutation causes increased actin binding in live cells. Significantly, N-terminal truncation of the wild-type ABD or L253P ABD lowered the FRET efficiency to 2.1% and 2.4%, respectively. Altogether, the biochemical and live cell FRET binding assays demonstrate that the N-terminus is required for the L253P mutation to induce high-affinity actin binding.

Discussion
The binding of β-spectrin to actin is required for the formation of the plasma membrane-associated spectrin cytoskeleton that is implicated in numerous neuronal functions. Here we investigated the in vivo functional significance of the β-III-spectrin N-terminus, which flanks the conserved tandem-CH ABD, and was recently discovered by cryo-EM, to bind actin. Using Drosophila, we demonstrated that the N-terminus is essential for β-spectrin function in neurons. Moreover, we showed that the N-terminus is required for neurotoxicity and dendritic arborization defects induced by L253P. We further demonstrated that the N-terminus is required for L253P-induced high-affinity actin binding, which provides a mechanistic explanation for how N-terminal truncation rescues SCA5 neurotoxicity. Right, quantitation of β-spectrin-GFP protein levels from western blot. Head extracts were prepared a total of three times for each genotype. Statistical analysis was performed by One-way ANOVA followed by multiple comparisons post hoc test. P-values are reported to the right of panels. www.nature.com/scientificreports/ The requirement of the N-terminus for L253P-induced high-affinity actin binding emphasizes the critical role of the N-terminus in the actin-binding mechanism. The N-terminus supports β-spectrin actin binding, at least in part, through its direct interaction with actin. The N-terminus is also destabilizing, Fig. 4. Structural instability conferred by the N-terminus may promote the conformational opening of the CH1-CH2 interface that is required for CH1 to bind actin. Potentially the N-terminus supports actin-binding by additional mechanisms. The N-terminal helix is continuous with CH1 helix A (Fig. 1), which makes contacts with CH2 residues, including L253. Possibly binding of the N-terminus to actin alters CH1-CH2 contacts to facilitate opening of the CH1-CH2 interface. In addition, the N-terminus may be a site of regulation for β-spectrin actin-binding activity. Several amino acids in the β-spectrin N-terminus have been reported to be phosphorylated 33,34 . Phosphorylation of the α-actinin N-terminus reduces actin binding 28 . Potentially β-spectrin phosphorylation modulates the function of the N-terminus to promote actin binding.
Here we showed that the SCA5 L253P mutation, in addition to causing high-affinity actin binding, also causes the mutant β-spectrin protein to accumulate in neurons, Fig. 3. It is possible that this accumulation also contributes to SCA5 neurotoxicity. In contrast, the N-terminally truncated SCA5 mutant did not accumulate, and instead showed decreased abundance relative to wild-type. N-terminal truncation, by itself, is not destabilizing. www.nature.com/scientificreports/ Rather, we showed that N-terminal truncation increases protein stability (Fig. 4), and that the N-terminally truncated wild-type β-spectrin was more abundant than wild-type β-spectrin with intact N-terminus, Fig. 3.
On the other hand, the L253P mutation is strongly destabilizing, in the presence or absence of the N-terminus, Fig. 4. L253P destabilization may reflect solvent exposure of hydrophobic residues at the CH1-CH2 interface. For example, an X-ray crystal structure of filamin A showed that CH1 residue W142 is buried at the CH interface in the "closed", actin-unbound state 35 . Cryo-EM of the filamin ABD-actin complex showed that in the "open" conformation, W142 makes hydrophobic contacts with actin 24 , avoiding solvent exposure. Actin binding may stabilize the L253P ABD by burying hydrophobic residues that would otherwise be solvent exposed. In contrast, the N-terminally truncated L253P ABD has very low affinity for actin, and thus remains destabilized and prone to misfolding and protein turnover. A currently pursued therapeutic strategy for SCA5 is to identify small molecules that bind the L253P β-IIIspectrin ABD and reduce its binding to actin 32 . The critical role of the N-terminus in actin binding suggests that small molecules that target the N-terminus may effectively reduce actin binding, similar to N-terminal truncation. In addition, it is possible that the large difference in stability of L253P β-III-spectrin versus wild-type may be leveraged to selectively target mutant β-III-spectrin. We suggest that the reduced protein level of N-terminally truncated SCA5 β-spectrin, but not N-terminally truncated wild-type β-spectrin, is due to enhanced protein turnover of a destabilized SCA5 β-spectrin that cannot bind actin. Thus, a small molecule that reduces actin binding may enhance turnover of the destabilized L253P β-III-spectrin, but not the more stable wild-type protein.

Materials and methods
Drosophila studies. Drosophila stocks. All stocks were maintained on standard food. The following stocks were obtained from the Bloomington Drosophila Stock Center: 477-Gal4/CyO, ppk-CD4-tdTom (10a)/ TM6, Tb, and elav-Gal4 (C155). attP154 transgenic flies were kindly provided by Norbert Perrimon, Harvard Medical School, Boston. The βspec em21 line was obtained from Ronald Dubreuil, University of Illinois at Chicago, Chicago. We previously generated UAS-βspec WT and UAS-βspec SCA5 fly lines 14 .
Generation of UAS-βspec ∆N transgenic fly lines. The previously generated pUASTattB-β-spectrin wild-type and pUASTattB-β-spectrin SCA5 DNA constructs 14 were used as templates in PCR to delete the DNA sequence encoding the N-terminal 44 amino acids of Drosophila β-spectrin. For wild-type or SCA5 template, two PCRs were performed using the oligo set, AAA GGT ACC GCC AAG TGA AGT TCA TCC and CAC ACT CTC ACG CTC CTC GGC CAT GGC TGG GGA TCT ACG GTT , and the oligo set, AAC AGA TCC CCA GCC ATG GCC GAG GAG CGT GAG AGT GTG and AAA GCT AGC TGC TCC AGT TTC TCC TGC. The resulting two PCR products contained overlapping sequences on either side of the intended deletion. To join the two PCR products, a third PCR was performed in which the two PCR products were used as template, together with the oligo AAA GGT ACC GCC AAG TGA AGT TCA TCC, which contains a KpnI site, and the oligo AAA GCT AGC TGC TCC AGT TTC TCC TGC, which contains a NheI site. The resulting PCR product contained the 5' region of the wild-type or SCA5 β-spectrin cDNA, with the sequence encoding the N-terminal 44 amino acids deleted. The wild-type or SCA5 PCR product was then inserted KpnI-NheI into an intermediate DNA vector, pAc5.1b-FSPWT, containing the full-length Drosophila β-spectrin cDNA, digested with the same enzymes. The full-length cDNA encoding the N-terminally truncated wild-type or SCA5 β-spectrin was then subcloned from the pAc5.1b intermediate into pUASTattB using KpnI and XbaI enzymes. The final constructs, pUASTattB-β-spectrin ∆N-WT and pUASTattBβ-spectrin ∆N-SCA5 were sequence verified and inserted into the attP154 landing site using PhiC31 integrasemediated transgenesis conducted by BestGene, Inc.
Generation of UAS-βspec GFP transgenic fly lines. A multi-step protocol was followed to fuse mEGFP to the 3′ end of the fly β-spectrin coding sequence. PCR were performed to amplify a 3′ region of the fly β-spectrin coding sequence, using pUASTattB-β-spectrin wild-type as template, together with forward primer, AAA CGC CGG CGA GGG TCA CGA AGG and the reverse primer, CTC CTC GCC CTT GCT CAC CAT CTT TTT CTT TAA AGT AAA AAA CG. The reverse primer contains sequence complementary to the 3′ end of the β-spectrin coding sequence and sequence complementary to the 5′ end of mEGFP. A second PCR was performed to amplify mEGFP using the forward primer, ATG GTG AGC AAG GGC GAG GAG and the reverse primer, AAA TCT AGA CTC GTT CTT CTC TTG CTT ATG GTT GCG TTA CGG CTG TTA CTT TAC TTG TAC AGC TCG TCC ATG CC. The reverse primer includes a short β-spectrin 3'UTR sequence, attached to the 3′ end of the mEGFP coding sequence. The mEGFP DNA used as template in the PCR was synthesized by IDT DNA Technologies. A third PCR was performed to fuse the PCR products generated in the first two PCRs. In the third PCR, the PCR products from the first two PCRs were used as template, and combined with the forward primer, AAA GGT ACC CGC CGG CGA GGG TCA CGA AGG , containing KpnI and MreI restriction sites, and the reverse oligo AAA TCT AGA CTC GTT CTT CTC TTG CTT ATG GTT GCG TTA CGG CTG TTA CTT TAC TTG TAC AGC TCG TCC ATG CC, containing XbaI site. The resulting PCR product was a fusion of the 3′ coding sequence and mEGFP. The fusion DNA was inserted KpnI-XbaI into pcDNA3.1 digested with the same restriction enzymes, to generate the intermediate construct termed pcDNA3.1-fly-β-spec CT-GFP. The β-spectrin 3′ coding sequence with GFP fusion was subcloned from pcDNA3.1-fly-β-spec CT-GFP into pUASTattB-β-spectrin constructs using MreI and XbaI restriction enzymes. The final pUASTattB-β-spectrin-mEGFP constructs were sequence verified and transgenesis performed using the attP154 landing site at BestGene, Inc. www.nature.com/scientificreports/ chromosome. All classes of adult progeny were scored. Fly incubation temperature ranged from 18 to 25 °C, as indicated in the Tables containing rescue assay data.
Dendritic arbor analysis. In Fig. 2, female flies carrying 477-Gal4; ppk-CD4-tdTom/TM6 transgene were crossed to UAS-βspec males at 22 °C. Eggs were collected from grape juice plates containing yeast paste every 4 h. Third-instar female larvae 120 h after egg laying (AEL) were imaged using HC PL APO 20x/0.80 NA dry objective and and a Leica DMi8 inverted microscope equipped with X-Light V2 spinning-disc (CrestOptics), LDI laser unit (89 North), and Photometrics Prime 95B CMOS camera. Z-stack images of class IV da neurons were obtained from A3 and A4 abdominal segments (1-3 neurons per larva) and processed to generate maximum intensity projections using Metamorph software. Reconstructions of the arbors were generated using Photoshop 21.0.1. Sholl analysis, and total branch length and branch point quantitation were performed as previously described using ImageJ software 14,36 . Statistical analyses on total branch length and number of branch points were performed in Prism 8 (Graphpad) using unpaired two-tailed t test (n = 10 for βspec WT and βspec SCA5 , n = 11 for βspec ∆N-WT and βspec ∆N-SCA5 ). Co-immunoprecipitation assay. Drosophila head extracts were prepared as described above. After, homogenization and clarification by centrifugation, a Bradford assay was performed to determine lysate protein concentrations. A sample of each lysate (80-100 µg) was collected and mixed with 2 × Laemmli sample buffer. Remaining lysate supernatants were clarified by inversion with Protein A agarose resin for 1 h at 4 °C. Then ~ 500 µg total lysate protein was incubated with either 10 µg anti-Drosophila α-spectrin 3A9 (Developmental Studies Hybridoma Bank) or 10 µg normal mouse IgG (Millipore) antibodies for 1 h at 4 °C, with inversion. Protein A agarose beads equilibrated with RIPA buffer was then added to lysate/antibody mixtures and incubated with inversion overnight at 4 °C. Then, Protein A agarose beads were pelleted at 760×g for 1 min and washed 3 × with RIPA buffer. The protein samples were eluted by boiling the beads with 1X Laemmli sample buffer for 5 min at 85 °C.

Western blot analysis. Males homozygous for
Protein expression and purification. The coding sequences for L253P β-III-spectrin ABD with or without N-terminus were PCR amplified from pET-30a-ABD L253P 17 using the following forward primers for intact N-terminus AAA CAC CTG CAA AAA GGT ATG AGC AGC ACG CTG TCA CCC or truncated N-terminus AAA CAC CTG CAA AAA GGT GCA GAT GAA CGA GAA GCT GTGC, and the reverse primer AAA TCT AGA CTA CTT CAT CTT GGA GAA GTA ATG GTA GTAAG. PCR products were digested with AarI and XbaI restriction enzymes and ligated into the compatible ends in pE-SUMOpro (LifeSensors) generated by BsaI digestion. The final constructs, pE-SUMO-ABD L253P and pE-SUMO-A52-ABD L253P, together with the previously generated pE-SUMO-ABD WT and pE-SUMO-A52-ABD WT constructs 23 , were transformed into Rosetta 2 (DE3) E. coli (Novagen). Bacteria were gown in 1 L of LB broth with ampicillin (100 µg/mL) and chloramphenicol (34 µg/ mL) and SUMO-ABD protein expression induced with 0.5 mM IPTG, for 6 h, at 300 rpm and room temperature. Bacteria were pelleted at 4000 rpm and 4 °C for 30 min, and pellets stored at − 20 °C until further use. Protein extraction and purification proceeded as described previously 23 , except that a final gel filtration chromatography step was not performed due high purity of ABD protein. Buffer exchange was performed by dialysis in 10 mM Tris, pH 7.5, 150 mM NaCl, 2 mM, MgCl 2 , and 1 mM DTT using Slide-A-Lyzer, 10,000 MWCO cassettes (Thermo Scientific).
Circular dichroism spectroscopy. Purified ABD proteins were diluted to 200 ng/µL in buffer containing 10 mM Tris, pH 7.5, 150 mM NaCl, 2 mM, MgCl 2 , and 1 mM DTT. Circular dichroism absorption spectra and thermal denaturation data were collected in a Jasco J-810 spectropolarimeter with Peltier temperature controller, www.nature.com/scientificreports/ as described previously 23 . Raw absorption data was converted to mean residue ellipticity using an equation described previously 23 . For thermal denaturation absorption data measured at 222 nm, melting temperatures were calculated in Prism 8 (Graphpad) using nonlinear regression to fit a two-state transition equation previously described 23 .
Actin co-sedimentation binding assays. Actin was purified from rabbit muscle acetone powder acquired from Pel-Freez Biologicals. To extract actin, acetone powder was incubated in pre-chilled water (1 g / 20 mL) for 30 min followed by vacuum-filtration using Whatman #4 filter paper. Native-PAGE. ABD and BSA proteins were clarified by centrifugation at 40,000 RPM for 30 min at 4 °C in a TLA-100.3 rotor. ABD and BSA concentrations were determined by Bradford assay. Varying concentrations of ABD proteins (2-8 µM) were prepared in F-buffer containing 10 mM Tris, pH 7.5, 150 mM NaCl, 0.5 mM ATP, 2 mM MgCl 2 , 1 mM DTT. BSA protein was prepared at concentration of 2 µM, also in F-buffer. Samples were mixed with 2X Native Sample Buffer (Bio-Rad). Native-PAGE was performed using a 12% acrylamide gel and ice-cold Tris-glycine buffer (Bio-Rad). Gels were stained using Coomassie Brilliant Blue R-250 solution. The destained gel was imaged using the 785 nm channel of an Azure 600 imager.
Live cell FRET binding assays. Mammalian cell DNA constructs were generated to express ABD proteins fused at the C-terminus to mEGFP, to be used as FRET donors. For these constructs, the mEGFP coding sequence was PCR amplified using the forward primer CTT GGT ACC ACC ATG GTG AGC, containing a KpnI site, and the reverse primer AAA TCT AGA CTA CTT GTA CAG CTC GTC CAT GCC, containing a XbaI site, followed by digestion and ligation into pcDNA3.1, resulting in the intermediate construct pcDNA3.1-mEGFP. For PCR amplification of the ABD coding sequence, the forward primer AAA AAG CTT ACC ACC ATG AGC AGC ACG CTG TCA CCC , or AAA AAG CTT ACC ACC ATG GCA GAT GAA CGA GAA GCT GTGC, for full-length or N-terminally truncated ABD, respectively, was used with the reverse primer AAA GGT ACC CTT CAT CTT GGA GAA GTA ATG G. The amplified ABD sequences were digested with HindIII and KpnI, and ligated 5′ of mEGFP in pcDNA3.1-mEGFP. The final FRET donor constructs pcDNA3.1-β-III-spectrin ABD-mEGFP with or without the N-terminal truncation were sequence verified. The pcDNA3.1-Lifeact-mCherry FRET acceptor construct was previously described 32 . HEK293T cells were acquired from the American Tissue Culture Collection (ATTC) and were transiently transfected using Lipofectamine 3000 (ThermoFisher Scientific) with either pcDNA3.1β-III-spectrin ABD-mEGFP donor construct (D), or co-transfected with pcDNA3.1-β-III-spectrin ABD-mEGFP and pcDNA3.1-Lifeact-mCherry donor and acceptor constructs (DA), in 6-well microplates. Cells were harvested 24 h after transfection by dissociation with TrypLE (Gibco), followed by TrypLE inactivation in 1 × DMEM (4.5 g/L D-glucose and 110 mg/L sodium pyruvate) supplemented with 10% FBS (Gibco) and 1% penicillin/streptomycin (Gibco), and then washing in 1 × DPBS (Gibco). Fluorescence lifetime measurements were performed using time-correlated single photon counting method in a FS5 spectrofluorometer equipped with an EPL-472 pulse laser with 5 MHz frequency (Edinburgh Instruments). mEGFP emission signals were detected at 510 nm until the fluorescence decay reached 1000 photon counts within a 50 ns time window. The lifetime (τ) of mEGFP in D and DA samples was obtained by fitting the fluorescence decay data using Fluoracle software (Edinburgh Instruments) reconvolution analysis with instrument response function correction. The FRET efficiency was calculated using the equation: www.nature.com/scientificreports/ Sequence alignments. For Fig. 1, amino acid sequences of β-spectrin and related proteins were aligned using Clustal Omega 1.2.4 algorithm in SnapGene 5.1.7.

Data availability
Source data generated during and/or analysed during the current study are available from the corresponding author on reasonable request.