Functional Differences Between Neuronal and Non-neuronal Isoforms of Drebrin

Abstract

Actin cytoskeleton is vital for neuronal function. Drebrin is a key F-actin binding protein in neurons which is linked to the filaments’ stabilization. As mammalian brain develops, drebrin expression pattern switches from non-neuronal (drebrin E, Embryonic) to neuron-specific isoform (drebrin A, Adult), but the evolutionary need for such a switch is enigmatic. Prior in cellulo and in vivo work suggested a non-redundant role of drebrin isoforms in neuronal development and function, however, molecular level understanding of it is lacking. Here we used mutagenesis, bulk solution assays, and time-lapse TIRF microscopy to probe for functional differences between drebrin isoforms. We found that drebrin A and E are functionally distinct and differ in their ability to inhibit F-actin depolymerization. We showed that both isoforms act as permissive cappers of the barbed end of actin, however, drebrin A has a significantly stronger capping activity, compared to that of the non-neuronal drebrin E. Probing for the molecular level insights on the observed differences revealed that the adult-specific exon in neuronal drebrin A contains an actin binding interface which enhances its permissive capping activity. Strikingly, F-actin decoration by neuronal drebrin A confers significantly stronger resistance to cofilin-mediated severing compared to that of drebrin E. Our results provide novel molecular level insights on functional differences between drebrin isoforms, which deepen our understanding of cytoskeletal regulation in the neuronal context. Our results also helps interpreting the previously reported data related to the silencing or knockout of the neuronal drebrin isoform.

Introduction

Neurons express a unique subset of actin regulators and some of these proteins are not found in any other tissues. Actin binding protein drebrin is one of the key regulators of actin in neurons as well as non-neuronal cells. As mammalian brain develops, drebrin expression pattern switches from embryonic (E) to the adult-specific isoforms (A)¹ but the evolutionary need for such a switch is enigmatic. As a result, in adults, drebrin A (DrbA) is strictly neuronal and can be found in postsynaptic compartments (dendritic spines)¹ and only drebrin E (DrbE) expresses in non-neuronal cells suggesting functional differences between the isoforms which, up to date, remain unknown. Drebrin loss is a hallmark of some of the most complex neurodegenerative disorders (Alzheimer’s disease, Down syndrome, epilepsy and others).² Moreover, important roles of DrbE in non-neuronal cells such as cell–cell communication and its connection to cancer metastasis started to emerge as well.^3–6 Thus, better understanding of the functional differences between drebrin isoforms is an important biomedical goal.

Prior in cellulo and in vivo work provided evidence that DrbA and DrbE play non-redundant roles in neuronal cells.^7,8 It was shown that the depletion of DrbA (while DrbE could still express) inhibits the clustering of postsynaptic density 95 protein (PSD-95) in dendritic filopodia (precursors of dendritic spines) delaying their maturation.⁷ Moreover, silencing of DrbA blocked activity-dependent synaptic clustering of N-methyl-D-aspartate (NMDA) receptors.⁹ Furthermore, in the Drebrin A Knock Out (DAKO) mice, the loss of homeostatic plasticity of NMDA receptors, impaired long-term potentiation (LTP), and the frequency of synapses with perforated PSD complex were reported.^10,11 It was hypothesized that such phenotype may lead to memory alterations. In line with this idea, DAKO mice exhibited defects of fear learning in adulthood.⁸

Drebrin isoforms are a product of the alternative splicing of a single gene.¹¹ Protein sequence-wise, the only difference between DrbA and DrbE is the presence of 46 amino acid insert in DrbA corresponding to the adult-specific exon (ASE). Drebrins contain a N-terminal ADF/cofilin-homology domain, followed by a helical/charged motif and a unique actin-binding domain which all together compose an actin-binding core of drebrin (aa 1–300, Drb1–300).¹² F-actin is the main binding partner of drebrin. The full-length protein and its N-terminal fragments are known to bind to the sides of actin filaments.¹² It was previously documented Drb1–300 binds to F-actin with 1:3 stoichiometry¹² as opposed to the 1:5 stoichiometry reported for the full length drebrin^13,14 suggesting the existence of an unidentified F-actin binding site(s) in its C-terminal sequence (aa301–706).

On a functional level, DrbA and its N-terminal constructs (aa1–300 and 2–252) promote F-actin stabilization.¹² It was documented that these constructs can rescue actin forms with impaired self-interacting interfaces. Importantly, DrbA was more effective in such rescue compared to its N-terminal fragments.¹² Drebrin’s side-binding to F-actin has several distinct features such as positive binding cooperativity, greater filaments’ stiffness upon decoration, and increased helical pitch.¹⁴ These features are mediated via actin-binding core of drebrin (Drb1–300) since they were also observed with this construct.¹⁵ Despite the presence of the ADF homology domain, drebrin does not possess any F-actin severing activity.¹³ However, it was recently shown that neuron-specific DrbA acts as a leaky (permissive) capper of F-actin’s barbed end slowing down the elongation rates >2-fold.¹⁶ It is so far unknown whether DrbE also has a permissive capping activity.

Our understanding of debrin’s crosstalk with other actin-interacting proteins is only starting to emerge. It was previously shown that DrbA inhibits cofilin-mediated severing of F-actin but the isoform dependence of this effect remains unclear.¹⁷ Early work showed that DrbE competes with tropomyosin and some of the actin-bundling proteins (e.g., alpha-actinin and facsin) for the side-binding to F-actin.¹³ In addition, it was recently shown that neuronal DrbA can inhibit actin nucleation by formins of the diaphanous and DAAM sub-families.^16,18

Taken together, functional differences between drebrin isoforms constitute a substantial gap of knowledge which would be critical to address in order to gain better understanding of the role of drebrin in health and disease. In this study we carried out side-by-side characterization of drebrin isoforms A and E employing biochemical assays, single filaments’ imaging, computational modeling, and site-directed mutagenesis.

Results

Drebrin isoforms inhibit bulk F-actin depolymerization with different potencies

It was previously shown that drebrin inhibits F-actin depolymerization,¹² however, whether the potency of this inhibition is drebrin isoform-dependent is unknown. Thus, we monitored real time bulk depolymerization of F-actin in the presence of recombinant DrbA and DrbE using pyrene-labeled actin as a reporter and latrunculin A (LatA) as an actin sequestering agent. Pyrene’s sensitivity to its environment leads to higher fluorescence of pyrene-labeled actin in filamentous form as opposed to monomers. Thus, the drop in pyrene fluorescence can be used as a readout of bulk F-actin depolymerization while the excess of LatA suppresses actin re-polymerization. We observed that DrbA inhibited actin depolymerization significantly stronger than DrbE, as evident from the shape of the depolymerization curves from which the initial rates of actin disassembly were calculated (Figures 1a, b). Analysis of the initial rates showed that DrbA and DrbE inhibit actin depolymerisation ~26-fold and 3-fold, respectively, compared to the no-drebrin control (Figure 1b, N = 3 different preps for each drebrin isoform).

Figure 1. Drebrin isoforms inhibit bulk F-actin depolymerization with different potencies.

(a) Pyrene fluorescence assay reporting on F-actin depolymerization in the presence of drebrin isoforms and LatA. Representative set of data is shown. Conditions: [F-actin] = 2 μM (5% pyrene maleimide), [drebrin] = 0.2 μM; [LatA] = 10 μM; KEMH7 buffer. (b) Quantification of the F-actin depolymerization rates for the conditions shown in (1a). N = 3 different preps of each drebrin isoform; average initial rates of depolymerization were calculated from 4, 5, and 2 depolymerization traces. Error bars correspond to the SEM. Student’s t-test, p values are provided above the bar graphs. (c) Analysis of the LatA resistance of drebrin-decorated F-actin. Representative SDS PAGE gel of the samples incubated overnight at room temperature. S – high speed supernatant (G-actin), P – high seed pellet (F-actin). Conditions: [F-actin] = 2 μM, [drebrin] = 1 μM; [LatA] = 10 μM; KEMH7 buffer. Experiment was repeated twice with similar results. (d) Assessment of the size of the LatA-resistant of F-actin fraction with the increasing concentrations of drebrin. Black or white symbols – unlabeled actin; red or blue symbols – 5% pyrene maleimide actin (samples were taken from the fluorimeter’s plate). The data were fit to the exponential decay equation (Single, 3 parameters) using SigmaPlot v11 software. Conditions: [F-actin] = 2 μM, [drebrin] = 0.05–1.5 μM; [LatA] = 10 μM; KEMH7 buffer.

It should be noted that not only the initial rates but also the extent of actin depolymerization was different in the presence of the drebrin isoforms, compared to the F-actin-only control (Figure 1a). It was previously documented that in the presence of neuronal DrbA a fraction of actin polymers persists even after overnight incubation with LatA while undecorated filaments depolymerize almost fully.¹² We hypothesized that the differences in the fluorescence plateau levels observed in the presence of DrbA and DrbE (Figure 1a) reflect unequal amount of the LatA-resistant F-actin fractions. To test this experimentally, we used high speed co-sedimentation to determine the fractions of F-actin remaining in each sample after the overnight incubation with LatA. Strikingly, we documented that a larger fraction of the LatA-resistant F-actin (polymer) is detected in high-speed pellets in the presence of DrbA compared to the samples containing DrbE at the same concentration (Figure 1c, d). The SDS PAGE analysis revealed differences in actin distribution between the supernatants and pellets in the presence of DrbE vs. DrbA (Figure 1c). Specifically, most of the F-actin gained LatA resistance (was detected in the pellet) in the presence of DrbA but not with DrbE. We next asked if the amount of the LatA-resistant F-actin could be increased by elevating the concentration of DrbE in the assay. The titration curves show that the amount of LatA-resistant actin polymers saturate at different levels with the increasing concentrations of the two drebrin isoforms, pointing towards the functional difference between them (Figure 1d). Taken together, our data shows that DrbA is a significantly stronger inhibitor of the rate and extent of F-actin depolymerization compared to the nonneuronal DrbE.

Drebrin isoforms bind to F-actin tightly and cooperatively

To aid interpretation of F-actin depolymerization results with drebrin isoforms, we asked if we could detect any differences in the fractional saturation of actin filaments with DrbA vs. DrbE. To this end, we performed high-speed pelleting assays with the fixed concentration of F-actin (2 μM) and increasing concentrations of drebrin isoforms. Since it is known that debrin binds to F-actin with high affinity,^13,14,19 we utilized Sypro Red Protein Gel Stain to aid the protein bands’ detection (Figure 2a). We found no major differences in the fractional saturation of actin filaments by DrbA vs. DrbE under our experimental conditions. Statistical analysis of the data presented in Figure 2b showed no significant differences in the amounts of DrbA and DrbE (0.2 μM) bound to F-actin (2 μM) (p = 0.994; Mann–Whitney test, n = 4 and 5 samples for DrbA and DrbE, respectively). This indicates the same fractional saturation of F-actin with drebrin isoforms in our bulk depolymerization experiments (Figure 1a, b). Thus, the differences in the effects of DrbA and DrbE on F-actin depolymerization are unlikely to stem from the differences in the fractional saturation of actin filaments by drebrin. Note that kinetic differences in drebrin isoforms’ binding to actin filaments cannot be ruled out based on the results of the steady-state measurements presented in Figure 2.

Figure 2. Drebrin isoforms bind to F-actin tightly and cooperatively.

(a) Analysis of the drebrin isoforms’ binding to F-actin. Representative SDS PAGE gels stained with Spyro Red Protein Gel Stain are shown. S – high speed supernatant (G-actin), P – high speed pellet (F-actin); numbers above the gel panels correspond to the drebrin concentrations; A – DrbA, E – DrbE. (b) Quantification of the data shown in (2a). The binding data was pooled from 4 and 5 different pelleting experiments for DrbA and DrbE, respectively. Two independent preps of each drebrin isoform were tested. The data points were fit to the exponential rise to maximum equation (Single, 3 parameters) using SigmaPlot v11 software. Conditions: [F-actin] = 2 μM, [drebrin] = 0.05–2 μM; KEMH7 buffer. (c) Visualization of drebrin bound to actin filaments using anti-His-tag antibodies. Representative fluorescence microscopy images are shown for DrbA and DrbE. Scale bar = 15 μm. (d) Quantification of the fraction of actin filaments heavily (>50%) decorated by drebrin. Experiments were repeated 3 times with similar results. Representative repeat is shown. Each data point represents a measurement for a 265×265μm field of view (n = 20 fields per sample). Student’s t-test.

Cooperative binding to F-actin was previously reported for the N-terminal Drb1–300 construct and, therefore, should be a feature of both isoforms.¹⁵ However, considering the differences in the binding stoichiometry between Drb1–300 (1:3) and the full-length proteins (Figure 2b), the modes of the filaments’ decoration may differ for DrbA/DrbE vs. Drb1–300 due to a contribution of their C-terminal sequences to F-actin binding. To visualize the drebrin binding patterns, we took advantage of the C-terminal His-tag retained in both isoforms and utilized fluorescently labeled anti-6His-tag antibodies at 1:250 dilution (Invitrogen, 21315-D550). We formed F-actin-drebrin complexes, incubated them with the antibodies for 10 min, then diluted the complexes into the drebrin-free buffer and imaged them on the polylysine-coated surface. We predicted that under these conditions, drebrin molecules bound to F-actin’s lattice less cooperatively would dissociate faster revealing the clustering patterns of DrbA and DrbE on actin filaments. Indeed, our imaging confirmed cooperative binding of both drebrin isoforms to F-actin (Figure 2c). Specifically, we observed heavily decorated as well as bare actin filaments and ones with drebrin clusters of various sizes (Figure 2c. To quantify the potential differences in binding, we defined heavy decoration by drebrin as more than 50% filament’s coverage and measured a fraction of such population for 10–20 fields of view (265 × 265 μm) in each sample (N = 3 biological replicates per condition; 3 different preps of each drebrin isoform). A representative set of data shown in Figure 2d indicates that the fraction of heavily decorated actin filaments was significantly larger (45 vs. 16%) with DrbA compared to the DrbE-containing samples. Thus, our results are consistent with the tight binding of both isoforms to F-actin and their distinct modes of decoration which likely stems from the differences in the filaments’ binding cooperativity of DrbA and DrbE.

Drebrin E is a weaker permissive capper of actin filaments compared to the neuron-specific drebrin A

It was previously reported that DrbA is a leaky (permissive) capper of F-actin.¹⁶ Specifically, DrbA is one of the proteins which binds to actin filaments in such a way that it partially blocks the addition of new monomers to the fast-growing barbed end of F-actin (“caps” a barbed end). It is so far unknown whether DrbE possesses the same activity. At the same time, the observed differences in the rates and extent of F-actin depolymerization in the presence of drebrin isoforms could be, at least in part, explained by the different potencies of their permissive capping activities. Thus, as detailed below, we used F-actin annealing, bulk polymerization, and time-lapse Total Internal Reflection Fluorescence Microscopy (TIRFM) assays to compare the interference of DrbA and DrbE with barbed end dynamics.

It is expected that the F-actin annealing process would be impaired if a capping protein is bound to one of the polymer’s ends since it changes its structural identity or sterically blocks the end. It was previously documented that in line with its permissive capping activity, DrbA inhibits F-actin’s annealing – the ability of actin polymers to join (anneal) after fragmentation and, therefore, reform longer filaments.¹⁶ Thus, to compare the effects of DrbA and DrbE on F-actin annealing, we fragmented fluorescently labeled actin polymers (15% Alexa488SE) with a needle and then mixed them either with the drebrin constructs or corresponding buffer (Figure 3a, bottom panel). We used the actin-binding core of drebrin (Drb1–300) as a negative control since it was previously shown that it does not possess actin capping activity.¹⁶ As expected, Drb1–300 had no statistically significant effect on actin annealing (Figure 3a, top right). Side-by-side comparison of the effects of drebrin isoforms showed that the inhibitory effect of DrbE on actin annealing is significantly weaker than that of the DrbA (Figure 3a, b). These results provided the first indication that DrbA and DrbE may have different potencies as permissive cappers of actin’s barbed end.

Figure 3. Drebrin isoforms have different potencies as permissive cappers of F-actin.

(a) Probing the efficiency of F-actin annealing (90 min) in the presence of drebrin isoforms. Non-capping Drb1–300 construct was used as a negative control. Representative fluorescence microscopy images are shown. Experiments were repeated twice with similar results. Scale bar = 10 μm. Conditions: [F-actin] = 0.5 μM, [drebrin] = 0.4 μM; KEMH7 buffer. (b) Quantification of the data shown in (3a). Student’s t-test: p < 0.001 (three stars); ns – not significant. Error bars correspond to the SD. (c) Elongation of individual actin filaments in the presence and absence of drebrin isoforms (time-lapse TIRFM assays). Representative examples are shown for each condition. Conditions: [F-actin] = 0.5 μM (15% Alexa488SE), [drebrin] = 0.5 μM. (d) Quantification of F-actin elongation rates in the presence of drebrin isoforms. N = 3 independent experiments with different preps of each drebrin isoform. 10–40 actin filaments were analyzed per condition in each experiment. Error bars correspond to the SEM. Student’s t-test, p values are provided above the bar graphs. (e) Permissive capping effect of the drebrin isoforms saturates at different levels. Quantification of F-actin elongation rates in the presence of the increasing concentrations of the drebrin isoforms (time-lapse TIRFM assays). Bars: grey – actin control, blue – DrbE, red – DrbA. The dots in the graph correspond to the average elongation rates measured for each data set (N = 2, 10–20 filaments analyzed per condition in each set). For statistical analysis the data were pooled from two separate experiments (23–34 filaments total per condition). Student’s t-test: p < 0.001 (three stars). Conditions: [F-actin] = 0.5 μM (15% Alexa488SE). (f) Quantification of F-actin elongation rates in the presence of drebrin isoforms and profilin. 30 actin filaments were analyzed per condition. Mann-Whitney test (F-actin vs DrbE), Student’s t-test (DrbA vs DrbE); p values are provided above the dot graphs. Conditions: [F-actin] = 1 μM (15% Alexa488SE), [profilin] = 5 μM, [drebrin] = 0.5 μM.

Next, we compared the effects of drebrin isoforms on actin elongation in the single filament time-lapse TIRFM experiments under identical conditions. The representative traces of actin filaments growing in the presence and absence of drebrin isoforms are shown in Figure 3c. Consistent with our annealing data, we found that DrbE inhibits actin elongation to a lesser extent compared to DrbA (Figures 3c, d). Specifically, elongation rates of 0.5 μM actin in the presence of DrbE and DrbA were 4.3 ± 0.5 and 2.3 ± 0.2 subunits/s, respectively (Figure 3d, N = 3 different preps of each drebrin isoform). Under the same experimental conditions, actin alone was elongating at a rate of 5.7 ± 0.2 subunits/s which translates into a 1.3 vs 2.5-fold inhibition of the barbed end elongation in the presence of DrbE and DrbA, respectively. Importantly, the observed inhibition of filament elongation by DrbE was statistically significant compared to actin alone (p = 0.023, unpaired Student’s t-test) indicating that non-neuronal drebrin isoform also acts as a permissive capper (Figure 3d). We also performed the experiments at the fixed actin and increasing drebrin concentrations (Figure 3e). Our results indicated that the inhibitory effect of DrbA and DrbE on actin elongation saturates at different levels, suggesting that it stems from their sequence rather than F-actin binding kinetics.

We also asked whether capping activity of drebrin isoforms is retained in the presence of profilin. This question is important to address since profilin-G-actin complexes are considered to be a physiologically relevant form of monomeric actin. Moreover, drebrin contains a polyproline stretch and early work identified profilin as its biding factor.²⁰ We found that permissive F-actin capping activity of both drebrin isoforms was retained and showed the same trend in the absence (Figure 3d) and presence of profilin (Figure 3f).¹⁶ Thus, DrbE acts as a permissive capper of F-actin in a profilin-independent manner, but its potency is weaker compared to that of DrbA. Importantly, our biochemical and microscopy experiments provided the first experimental proof of differential regulation of intrinsic F-actin dynamics by drebrin isoforms.

Molecular insights on the differences in permissive capping activities of DrbA and DrbE

To gain molecular level insights into the different potencies of DrbA and DrbE as permissive cappers of actin filaments, we employed drebrin constructs schematically represented in Figure 4a.

Figure 4. Determining the sequence requirements for permissive capping of F-actin by drebrin.

(a) Schematic of the drebrin constructs used in this study. ADF – ADF homology domain (blue), HC – helical charged motif (pink), ABD – actin-binding domain (green), red bar adult specific exon (ASE). Dotted lines indicate the missing part of the drebrin sequence. (b) Quantification of F-actin elongation rates in the presence of DrbE and its phosphomimetic mutant DrbE-S142D. The dotted line indicates the average elongation rate of F-actin in the absence of drebrin. N = 3 separate experiments with each drebrin form. 15–39 actin filaments were analyzed per condition in each experiment. Error bars correspond to the standard deviation. Student’s t-test, ns – not significant. Conditions: [F-actin] = 0.5 μM (15% Alexa488SE), [drebrin] = 0.5 μM. (c) Quantification of F-actin elongation rates in the presence of the DrbA constructs with sequence deletions downstream ASE. The data were pooled from two separate experiments with each drebrin form. 11–20 actin filaments were analyzed per condition in each experiment. Mann-Whitney test, ns – not significant. Conditions: [F-actin] = 0.5 μM (15% Alexa488SE), [drebrin] = 0.5 μM. (d) Quantification of F-actin elongation rates in the presence of the increasing concentrations of the ASE peptide. The data from the two separate experiments (solids spheres and open triangulars) are shown for each ASE concentration. 10–16 actin filaments per condition were analyzed in each experiment. For statistical analysis the data were pooled from two separate experiments. Student’s t-test (ns – not significant, 3 stars – p < 0.001. Conditions: [F-actin] = 0.5 μM (15% Alexa488SE). (e) Elongation of actin filaments in the presence of the ASE peptide compared to that measured with the drebrin isoforms (time-lapse TIRFM assays). The average traces are shown for each condition; error bars correspond to the SD. Representative data sets is shown (N = 2 data sets); the average elongation rates for each condition and its SD values are displayed on the right. Conditions: [F-actin] = 0.5 μM (15% Alexa488SE), [drebrin] = 0.5 μM, [ASE peptide] = 50 μM.

It was previously reported that DrbE but not DrbA gets phosphorylated on the conserved Ser142²¹ and it was proposed that such phosphorylation leads to a conformational change in DrbE upon which it adopts an “open” (vs. “closed”) conformation.²² It is currently unknown whether the presence of the ASE in DrbA abolishes this conformational switch locking DrbA in an “open” state. However, in DrbE, such conformational change may lead to a modulation of functionality, including its effects on F-actin dynamics. Thus, we tested the permissive capping activity of DrbE vs. its previously validated phosphomimetic mutant DrbE-S142D²² using time lapse TIRFM imaging. Our results show that mimicking the phosphorylation on Ser142 does not influence DrbE-induced inhibition of F-actin elongation (Figure 4b). Thus, Ser142 phosphorylation appears to be non-essential for the permissive F-actin capping activity of drebrin.

Drebrin sequence has a high degree of disorder in its C-terminal region (aa301–706). Since Drb1–300 construct has no actin capping activity,¹⁶ it is possible that the observed capping is due to the steric hindrance contributed by the C-terminal sequence of drebrin. Since DrbA is 46 residues longer than DrbE, it is feasible that the observed differences in their permissive capping activities stem from the sheer length of the C-terminal sequence. To test this idea, we employed two drebrin constructs with large (74–106 amino acid) deletions in their C-terminal sequence downstream of the ASE (Figure 4a). If steric hindrance is a main contributor to the drebrins’ permissive capping, then we would expect to observe a weaker capping activity in the presence of these drebrin deletion constructs. In disagreement with this hypothesis, our TIRFM results revealed that F-actin elongation rates in the presence of the drebrin constructs with the large sequence deletions were the same as those observed in the presence of wild type DrbA (Figure 4c). Thus, these data support the presence of a specific actin binding/capping sequence within the ASE of DrbA. Indeed, a peptide of drebrin corresponding to the ASE sequence inhibited actin filaments’ elongation, acting as a permissive capper itself (Figure 4d, p < 0.001 actin alone vs 50–100 μM ASE). Note that a higher concentration of ASE peptide is required to saturate the inhibitory effect compared to that of the full length drebrin isoforms. This is consistent with a scenario when the side-binding of drebrin to F-actin aids local enrichment and proper orientation of the ASE sequence at the barbed ends. We found that at saturation, the inhibitory effect of the ASE peptide on actin elongation is weaker than that of the full-length DrbA (Figure 4e, red trace) but stronger than of DrbE (Figure 4e, blue trace).

The drebrin sequence corresponding to the ASE does not contain any conserved actin-binding sites. Thus, we used Alphafold Colab (version AlphaFold.ipynb, March 2024)²³ with default parameters, to model the potential interaction between the actin and ASE of drebrin. Specifically, the modeling used the rabbit skeletal actin (Uniprot ID P68135) and a peptide corresponding to the ASE sequence of the mouse drebrin (aa 322–367; Figure 5a, top). The resulting model indicated a direct interaction between actin and drebrin with the ASE peptide positioned in the hydrophobic cleft of the actin (between actin subdomains (SD) 1 and 3) (Figure 5a, b). Interestingly, this region overlaps with actin’s self-interacting interface,²⁴ thus, the binding of the ASE of DrbA in the hydrophobic cleft would be consistent with its enhanced permissive capping activity compared to that of DrbE. This interaction was predicted to be stabilized solely by the hydrogen bonds with the ASE peptide of drebrin making contacts with SD1 and SD3 of actin. Specifically, Thr348, Tyr352, Cys355, and Arg357 of drebrin were predicted to interact with the actin residues Met357, Tyr171, Gly148, Tyr145, and Ile343, respectively. Note that these actin residues correspond to the Met 355, Tyr169, Gly146, and Ile341, respectively, in the mature actin form (after the removal of the first two N-terminal residues). Only Thr348 of DrbA was predicted to form two hydrogen bonds with actin – one involving its backbone (Thr348-Tyr171) and another one involving its side chain (Thr348-Met357) (Figures 5a, b).

Figure 5. Model of the actin complex with the ASE peptide of drebrin and its validation.

(a) Alphafold-generated model of actin binding to the ASE of DrbA (right panel – zoomed view). SD1–4 – subdomains of actin. The sequence of the ASE is shown on top of the panels. Drebrin residues predicted to be involved in its interaction with actin are in red. (b) Drebrin and actin residues involved in the complex formation are highlighted in blue and pink, respectively. Note that actin residue numbering would correspond to the Met 355, Tyr169, Gly146, and Ile341 in the mature actin form (after the removal of the first two N-terminal residues). (c) Quantification of F-actin elongation rates in the presence of the DrbA and its DrbA-YC/AA mutant designed to weaken the interaction predicted by the computational model. The data were pooled from two separate experiments with each drebrin form. 28–38 actin filaments were analyzed per condition in each experiment. Student’s t-test. Conditions: [F-actin] = 0.5 μM (15% Alexa488SE), [drebrin] = 0.5 μM.

To test the actin-drebrin ASE binding model (Figures 5a, b) we employed site-directed mutagenesis aiming to weaken or abolish the predicted interaction. In 4 out of 5 cases the side chains of the drebrin’s ASE peptide were predicted to interact with the backbone of actin. Thus, we chose to introduce mutations in the drebrin’s (as opposed to actin’s) sequence. We mutated Tyr352 and Cys355 to alanines (DrbA-YC/AA), aiming to weaken the hydrogen bonding in the complex (Figure 5b, residues are indicated with red stars). In addition to our in silico prediction, the rationale for mutating these two residues was twofold – (1) these residues were predicted to be a part of a short beta-strand (Jpred 4 secondary structure prediction tool²⁵) and (2) the size and aromatic nature of Tyr352 side chain could contribute to a specificity of drebrin interaction with the hydrophobic cleft of actin. In agreement with the computational model, we found that DrbA-YC/AA mutant had weaker permissive capping activity than wild type DrbA. Specifically, F-actin elongation rates were faster in the presence of the DrbA-YC/AA mutant compared to those measured in the presence of the wild type DrbA Figure 5c). Thus, the ASE of DrbA contains an actininteracting interface which enhances its permissive capping activity.

F-actin decoration by neuronal DrbA confers stronger resistance to cofilin-mediated severing compared to that of DrbE

Drebrin cross-talk with cofilin was found to be physiologically important.^2,26 Furthermore, a previous report indicated the strong protection of actin filaments against cofilin-mediated severing by DrbA.¹⁷ Thus, we asked if F-actin decoration by DrbA and DrbE provides similar levels of protection against cofilin’s action. To this end, we employed time-lapse TIRF microscopy imaging (Figure 6a). In these experiments, we polymerized actin in the microscopy flow chambers in the absence or presence of drebrin isoforms and allowed it to “age” on the surface for 60 min before introduction of cofilin or buffer. Our data revealed that DrbA is a significantly stronger inhibitor of cofilin-induced severing, compared to DrbE, judging by the number of severing events detected within the observation time window (Figure 6a, severing events are indicated by the red arrowheads in the middle and bottom panels). Next, we quantified the length of actin filaments after five and a half minutes from the addition of cofilin and expressed it as an average length reduction (black, blue, and red bars) compared to the initial filaments’ length (grey bars) (Figure 6b). We documented that under our experimental conditions, DrbE exhibited mild protection of actin filaments from the cofilin-mediated severing (Figure 6b, blue bars). However, F-actin decorated by DrbA showed dramatic resistance against the cofilin-induced fragmentation (Figure 6b, red bars). This trend persisted in the entire range of cofilin concentrations tested (0.3–1 μM) (see Supplementary Movies 1–3 with 1 μM cofilin). To ensure that the same amounts of the residual phosphate are present in F-actin in all the samples, we changed the mixing order allowing undecorated actin to “age” on the slide surface followed by the sequential introduction of drebrin then cofilin. Strikingly, we again observed a dramatic resistance of F-actin decorated by DrbA (but not by DrbE) to cofilin-mediated fragmentation (Figure 6c).

Figure 6. Effects of drebrin isoforms on cofilin-mediated severing of F-actin.

(a) Montages of the TIRFM images from the representative movies monitoring cofilin-induced fragmentation of F-actin over time. F-actin was copolymerized with drebrin (middle and lower panels) in the microscopy chambers. New severing events are indicated by red arrowheads in the montages from the movies recorded in the presence of drebrin isoforms. Scale bar = 10 μm. Conditions: [F-actin] = 0.5 μM (15% Alexa488SE), [drebrin] = 0.5 μM, [cofilin] = 0.3 μM. (b) Quantification of the average reduction in actin filaments’ length after 300 s from the start of the recording. Imaging began 30–35 s after the cofilin addition. Average filament length at time zero was set to 1 for all the samples (grey bar). The shortening of F-actin in the no-cofilin control is represented as an average of 3 different experiments (first black bar and corresponding dots). (c) TIRFM images from the representative movies monitoring cofilin-induced fragmentation of F-actin over time. F-actin was polymerized without drebrin in the microscopy chambers followed by the sequential addition of drebrin then cofilin (or buffer in controls). Scale bar = 15 μm. Conditions: [F-actin] = 0.5 μM (15% Alexa488SE), [drebrin] = 0.5 μM (added first), [cofilin] = 0.3 μM (added second).

Discussion

In this study we carried out side-by-side biochemical characterization of neuronal (DrbA) and non-neuronal (DrbE) isoforms of the F-actin-binding protein drebrin. We found that DrbE is a weaker inhibitor of F-actin depolymerization as well as a weaker permissive capper compared to DrbA (Figures 1 and 3). Furthermore, heavier decoration of F-actin by DrbA is consistent with its greater binding cooperativity compared to that of DrbE (Figure 2c, d). Interestingly, prior mapping of the drebrin interaction interface on F-actin identified multiple modes of binding for its N-terminal fragments – Drb1–300 and the actin-binding domain (seq 233–300/317).²⁷ Thus, it is feasible that F-actin interaction interface in the C-terminal sequence of DrbA (which is missing in its N-terminal constructs and DrbE) may stabilize a mode of drebrin’s attachment which promotes its cooperative binding to F-actin.

On a sequence level, our analysis showed that adult-specific exon (present in DrbA but not DrbE) contains an actin binding/capping interface (Figures 4 and 5). As an experimental proof of it, we were able to weaken this interface via site-directed mutagenesis which reduced the permissive capping activity of DrbA (Figure 5c). It is currently unknown whether the capping-enhancing region in ASE of DrbA (identified by our modeling and mutagenesis) contributes to the differences in the modes of F-actin decoration by DrbA vs. DrbE. The same binding stoichiometry of the drebrin isoforms to F-actin suggests that actin-interacting residues present in DrbA synergize with the interface present in both isoforms as opposed to constituting an additional actin-binding site (Figure 2b). Furthermore, a dramatic resistance of F-actin decorated by DrbA to cofilin-mediated severing (which was not observed with DrbE) is well aligned with its effects on F-actin dynamics and its mode of side-binding to filaments (Figure 6).

Functional differences between drebrin isoforms have been an unanswered question in the field of regulation of neuronal cytoskeleton, thus, some in cellulo and in vivo experiments were performed to address it. Our novel results aid the interpretation of the previously reported effects. For instance, fractionation and subsequent immunostaining of the brain lysates using the drebrin isoform-specific antibodies suggested that during postnatal brain development (when both isoforms are expressed), DrbA is more stably associated with the F-actin containing fractions than DrbE.¹ At the same time, the reports from several groups as well as our results presented herein (Figure 2) indicate that the binding affinities of drebrin isoforms to F-actin are similar.^13,14,19 With G-/F-actin ratio being relatively constant on different stages of the brain development,¹ we hypothesize that DrbE could be more easily outcompeted by other actin-binding factors present in brain lysates. In line with this idea, we found that DrbE is less effective in actin filaments’ protection from cofilin-mediated severing compared to the DrbA (Figure 6). We speculate that less cooperative binding of DrbE to F-actin (compared to that of DrbA) may allow for more effective cofilin clustering on actin filaments resulting in more boundaries between cofilin-decorated and cofilin-free F-actin regions which lead to severing.^28,29

Interestingly, it appears that the emergence of DrbA during brain development coincides with the transition of DrbE into the pellet (F-actin containing) fractions of cerebral cortex and hippocampal tissue lysates (see Figure 2 in Ref. 1). We can speculate that a distinct mode of DrbA binding to F-actin (capping and/or side binding) may aid the recruitment of DrbE into filaments potentially leading to the mixed decoration of F-actin with drebrin isoforms at some of the stages of brain development. In line with this hypothesis, in the cerebellar cortex (where the expression levels of DrbA is very low compared to the cerebral cortex and hippocampus), DrbE persists in supernatant fractions for much longer and does not exhibit a sharp transition into the pellet fractions (see Figure 2 in Ref. 1). In future work it would be interesting to probe for F-actin co-decoration with DrbA and DrbE in neurons and its potential effects on the filaments’ dynamics.

At the level of neuronal substructures, actin stability and dynamics is critical for maintaining the bulbous shape of mature dendritic spines – postsynaptic terminals. In dendritic spines two pools of F-actin are distinguished – stable and dynamic.^30,31 It was previously reported that neuronal DrbA is associated with the stable pool of F-actin and is likely responsible for its resistance to disassembly.³² A recent study employed Fluorescence Recovery After Photobleaching (FRAP) to examine the dynamics of the exogenously expressed GFP-tagged DrbA and DrbE in the hippocampal neurons. The authors reported a 2-fold larger stable fraction of DrbA, compared to that of DrbE in dendritic spines.³³ Interestingly, using reconstituted system with purified proteins, we also observed ~2-fold larger LatA-resistant F-actin fraction in the presence of DrbA, compared to the DrbE-containing samples (Figure 1c, d). Our data suggest that such increased stability may (at least in part) result from the enhanced F-actin capping by DrbA and, potentially, from its unique mode of side-binding compared to that of DrbE. Importantly, we documented that DrbE is a permissive capper itself and the actin binding interface residing in ASE of DrbA is likely synergistic with the one which is present in both isoforms (Figures 3 and 7). In line with this idea, Hanamura et al.³³ reported that repositioning of the ASE to the C-termini of drebrin reduces the size of the stable drebrin fraction in dendritic spines. This could be explained by spatial separation of the drebrin residues contributing to its permissive capping activity.

Figure 7. A working model of drebrin isoforms’ binding and capping of F-actin.

Actin-binding core of drebrin (aa1–300) and its disordered C-terminal sequence are represented with a green box and a solid black line, respectively. ASE is shown as a red box. A part of an actin capping interface which is present in both isoforms is shown as a blue sphere. Note that the exact location of this sequence in drebrin molecule is unknown. ASE contains the actin binding/capping sequence which is predicted to bind in the hydrophobic cleft of actin (left panel) masking its self-interacting interface.

Taken together, experimental evidence suggests that DrbA is uniquely suited for maintaining the stable pool of F-actin in dendritic spines and its functionality is distinct from that of DrbE.

Materials and Methods

Plasmids and mutagenesis

The mouse Drebrin A (UNIPROT Q9QXS6, DrbA) gene for recombinant expression in E.coli was in the pGEX6P-2 plasmid and it was flanked by a N-terminal glutathione transferase tag (GST-tag) and C-terminal end a 6-His affinity tag (6His).¹⁶ Site-directed mutagenesis was performed using the standard approach. The following primers were used for mutagenesis:

DrbA-YC/AA

Forward: 5′-GACTCCCTTTCCCGCAATCACCGCACACCGCACCCCAA-3′.

Reverse: 5′-TTGGGGTGCGGTGTGCGGTGATTGCGGGAAAGGGAGTC-3′.

DrbAΔ368–441

Forward: 5′-GTAACAGCAGCTCCTGAGGTCTGGG CTGGCTGTGC-3′.

Reverse: 5′-GCATGGGAGGGAGGAAGAGAGGTT TGGGGTG-3′.

DrbAΔ442–547

Forward: 5′-GTGCCCACACCACCCTCAGGTGCTGAGGCCTCCCTGGCAGAG-3′.

Reverse: 5′-TTTGGCCTCCTTGCTGAGCTCTTCATCCAGGCTTGGGGTAGTCTC-3′.

Large deletions in the drebrin gene were made using inverse PCR which utilizes two phosphorylated primers oriented in the inverse direction to amplify the entire circular plasmid except the part of the gene to be deleted. PfuUltra DNA polymerase (Agilent Technologies, 600385) was added into a 50ul reaction and subjected to 18 cycles of PCR: 95 °C (45 sec), 62 °C (1 min), and 68 °C (14 min) followed by a final elongation step (10 min at 68 °C). The parental plasmid was degraded with 10U of DpnI (Promega, R6231) for 1 h at 37 °C. After the cleanup step (QIAprep Spin Miniprep Kit, QIAgen, 28104), the resulting DNA was ligated for 24 h using T4 DNA Ligase (1U) and transformed into XL10-Gold competent cells.¹⁸ To generate DrbA-YC/AA mutant, 22 cycles of PCR were performed with the following parameters: 95 °C (45 sec), 55 °C (1 min), and 68 °C (14 min) followed by a final elongation step (10 min at 68 °C). The parental template DNA was degraded with 10U of DpnI (Promega, R6231) for 1 h at 37C. DNA sequences of all the new constructs used in this study were confirmed by Sanger sequencing (Azenta).

Protein expression and purification

Rabbit skeletal actin (RSA) was extracted from acetone powder of rabbit back muscle and labeled with pyrene maleimide or Alexa488-succinimidyl ester (SE) as described.^18,34,35 Human cofilin-1 and profilin were purified as described.^36,37

DrbA, DrbE, and their mutants were expressed in Escherichia coli (E. coli) Rosetta 2 cells and purified as previously described^14,16 with minor modifications. In brief, the cells were lysed either using a microfluidizer or, alternatively, by supplementation of the lysis buffer with 1X BugBuster reagent and 4 rounds of sonication. After elution from Co²⁺ affinity column, the fractions were supplemented with 0.5 mM thesit, and 0.05% IGEPAL to aid the binding to glutathione beads. To dissociate low molecular weight contaminants, cleaved drebrin preparations were incubated in a buffer composed of 1 M urea, 4.6 mM Tris, pH 8. 89 mM KCl, 0.89 mM DTT for 5 min at 25 °C. Drebrin preparations were then dialyzed overnight (ON) against Drebrin dialysis buffer (5 mM Tris, pH8, 100 mM KCl, 1 mM DTT).¹⁸ After dialysis, proteins were centrifuged at 90 K rpm for 5 min at 4C in a TLA100 rotor and stored at −80 ° C. Drb1–300 was expressed and purified as described previously.²⁷

Drebrin ASE peptide was synthetized by GeneScript. Lyophilized peptide was soaked for 5 min at room temperature (RT) in 0.2 M ammonium bicarbonate pH 8.2 then vortexed. Reconstituted peptide was stored at −80 °C.

Pyrene fluorescence assays and high-speed pelleting

Pyrene fluorescence assays were used to monitor the kinetics of the bulk actin polymerization as described.¹⁸ Fluorescence was monitored using Varioskan Flash plate reader (Thermo Scientific), with the excitation and emission wavelength set at 365 and 407 nm, respectively. In some of the experiments testing latrunculin A (LatA) resistance of actin filaments, the samples were taken directly from the wells of a plate after the overnight incubation at RT and subjected to the high-speed pelleting (TLA-100 rotor, 95,000 rpm or 392,000 g for 20 min at 4C) which allowed us to assess the G-/F-actin distribution in the resulting samples. Supernatants and pellets were separated by SDS-PAGE and routinely stained with InstantBlue (Coomassie) dye (unless stated otherwise in the figure legend). In the experiments testing drebrin isoforms’ binding to F-actin, Sypro Red Protein Gel Stain (Invitrogen, S6653) was utilized. Actin bands’ intensities were quantified using ImageJ.

Fluorescence microscopy

For the annealing assays, monomeric actin (15% Alexa-488SE labeled) was polymerized in KMEH7 buffer (50 mM KCl, 1 mM MgCl₂, 0.2 mM EGTA, 10 mM HEPES, pH7) supplemented with 1 mM DTT and 0.2 mM ATP. then diluted to 1 μM. F-actin was fragmented 8 times with a 23G1 needle (Precision Glide) and dispensed on a glass surface. Aliquots of fragmented actin were added to the drebrin constructs of interest with a multichannel pipette (1:1, V/V ratio) and incubated at RT for 90 min. The resulting samples were diluted to 5 nM with the Imaging Buffer (IB: KMEH7 buffer supplemented with 100 mM DTT and 1 μM phalloidin) and imaged on polylysine coated surface in epi-fluorescence mode.

To visualize drebrin bound to actin filaments, anti-6His-tag monoclonal antibodies DyLight™ 550 were utilized (Thermo Fisher, Cat# MA1–21315-D550). F-actin (1.5 μM, 15% Alexa488SE labeled) and drebrin (0.4 μM) were mixed with in KMEH7 buffer containing 1 mM DTT and 0.2 mM ATP followed by the 4 h incubation at RT. Next, the samples were mixed with equal volumes of the anti-6His-tag antibody (1:250 final dilution) and incubated for 10 min at RT. Lastly, the samples were diluted to 15 nM actin concentration into the IB and imaged on a polylysine coated surface in TIRF mode. Excitation wavelengths were 488 and 561 nm for actin-Alexa488 and the anti-6His tag monoclonal antibodies DyLight™ 550, respectively.

In time-lapse TIRF microscopy experiments, untethered actin filaments were imaged on Pluronic coated surface essentially as described.^34,38 The coverslips were cleaned with 1% Hellmanex, 0.1 M KOH, and acetone with multiple water rinses in between. The coverslips were dried extensively, coated with octadecylsilane derivative (0.4% in water) (Thermo Scientific, TS-42799), and cured at 160 °C. Flow chambers (V ~ 12 μl) were treated with 2 chamber volumes (CV) of 1–5% Pluronic F127 solution (Sigma, P2443) for 3 min, then equilibrated with 2 CV of 1xTIRF imaging buffer (10 mM HEPES, 1 mM MgCl₂, 50 mM KCl, 0.2 mM EGTA (pH 7) supplemented with 50 mM DTT, 0.2 mM ATP, 20 mM glucose, 0.5% methyl cellulose). Actin monomers or G-actin-profilin mixtures (15% Alexa488-SE labeled) were incubated for 3 min at RT with ME exchange buffer (50 μM MgCl₂, 0.2 mM EGTA) then mixed with TIRF imaging buffer with or without actin-binding proteins. The final formulation of the TIRF imaging buffer also contained 0.05 mg/ml casein, 0.25 mg/ml glucose oxidase, and 0.05 mg/ml catalase to minimize protein damage due to photobleaching. To test the effects of drebrin on the cofilin-mediated severing, rectangular coverslips were functionalized as described above. Experiments were carried out using single flow chambers (~35 μl). Actin polymers were polymerized and aged on the surface for 60 min before addition of cofilin. Imaging was conducted with a Nikon Ti2-E microscope equipped with a TIRF module at 100x magnification. Images were analyzed using ImageJ (Fiji) software (NIH, Bethesda, MD). Filaments were counted manually.

Abbreviations:

DrbA

neuron-specific mouse drebrin A

DrbE

mouse drebrin isoform E

ASE

adult specific exon, seq.322–367

DrbA-YC/AA

DrbA-Y352A/C355A

DAKO

drebrin A knock out

LatA

latrunculin A

TIRFM

Total Internal Reflection Fluorescence Microscopy

NMDA

N-methyl-D-aspartate

ADF

actin depolymerizing factor

DAAM

Dishevelled-associated activator of morphogenesis

SDS-PAGE

sodium dodecyl-sulfate polyacrylamide gel electrophoresis

Drb1–300

mouse drebrin construct seq.1–300

Drb 2–252

mouse drebrin construct seq.2–252

DrbAΔ368–441

drebrin A construct missing seq.368–441

DrbAΔ442–547

drebrin A construct missing seq. 442–547

GFP

Green fluorescent protein

Appendix A. Supplementary material

Supplementary material to this article can be found online at https://doi.org/10.1016/j.jmb.2025.169015.

DATA AVAILABILITY

Data will be made available on request.