1. Introduction
Membrane remodelling is a crucial cellular process that underlies several processes, such as the formation of the intraluminal vesicles of the maturing endosome by the ESCRT machinery. The ESCRT machinery is involved in many cellular processes where membrane regions are abscised away from the cytosol [
1]. Initially, the machinery was investigated during the formation of intraluminal vesicles (ILVs) at the limiting membrane of the maturing endosome (ME). During ILV formation, the machinery comprises five complexes acting in sequence, ESCRT-0, -I, -II, -III, and an oligomeric complex of the AAA-ATPase Vps4 [
1]. The early-acting complexes concentrate cargo, which is recognised through its ubiquitin label at sites of ILV formation. ESCRT-III is the centrepiece that, together with Vps4, mediates the membrane abscission process. ESCRT-III consists of four members of the CHMP family, which assemble into a filament at the limiting membrane of the ME. Vps20 initiates the polymerisation of CHMP4B, encoded by Shrub in
Drosophila and Snf7 in yeast, into a homopolymer that is modified in some way by Vps24, Vps2, and later-acting ESCRT-III members such as CHMP5, Did2 and Ist-1.
Members of the CHMP family fold into helical hairpins and cycle between the cytosolic monomeric and polymeric membrane-bound state [
1]. It has been shown that Shrub/CHMP4B possesses two opposing electrostatic surfaces that mediate the electrostatic interactions between protomers of the filament at membranes [
2,
3]. How the assembled ESCRT-III filament performs the reverse topology membrane abscission is not understood. Recent in vitro experiments suggest that the assembled Shrub filament either serves as a template for the formation of more rigid filaments made of other CHMP members, or pinches off membranes by itself with help of the other CHMP members [
4]. The Shrub filaments appear to be constantly remodelled by the action of Vps4, which can remove and replace protomers from the filament under ATP consumption [
4,
5].
Work so far has revealed that the
Drosophila tumour suppressor lethal (2) giant discs (Lgds) and its two mammalian orthologs coiled-coil and C2 domain-containing protein 1a (Cc2d1a/Lgd2) and 1b (Cc2d1b/Lgd1) are positive regulators of the activity of the ESCRT machinery [
6,
7,
8,
9]. A consequence of loss of function of
lgd in
Drosophila is the ligand-independent activation of the Notch pathway, as well as the tissue-specific prolonged activation of the Dpp/BMP pathway [
10,
11]. In addition, it results in the defective cytokinesis of germline cells [
12]. In mammals, the loss of CC2D1a results in death shortly after birth in mice and intellectual disability and autism in humans [
13]. The loss of CC2D1b/LGD1 causes a defect in the formation of the nuclear envelope after cleavage of cell culture cells [
14]. However, the complete knock-out of Lgd1 does not affect development or vitality in mice, suggesting that either the loss is compensated by alternative pathways or Lgd2 can act in a functionally redundant manner [
15]. In favour of the second possibility, the concomitant loss of function of both CC2D1/LGD genes results in early embryonic lethality, indicating that they function in a redundant manner [
16]. Cell culture experiments addressing molecular function of the Lgd orthologs in humans reveal a variety of functions ranging from regulation of cell signalling to cytokinesis [
13].
Lgd proteins possess four tandem repeats of the family-defining
Drosophila melanogaster 14 (DM14) domain, followed by a phospho-lipid-binding C2 domain ([
17,
18,
19],
Figure 1A,A’). The DM14 domains are helical hairpins that mediate protein–protein interactions with members of the CHMP family, such as Shrub/CHMP4, CHMP2 and CHMP7 ([
8,
9,
14,
20],
Figure 1A,D,E,F). Lgd physically interacts with Shrub via its two odd-numbered DM14 domains (onDM14s) [
21]. In contrast to the two even-numbered domains, the onDM14s possess an extended positively charged surface and appear to function in a redundant manner, shown by the finding that Lgd variants with only one of the odd-numbered domains can completely rescue
lgd-null mutants if already present in one copy [
21] (
Figure 1D–F’). However, they fail to rescue a sensitised background where a copy of functional
shrub is removed in addition to
lgd (sensitised background, genotype:
lgd/
lgd;
shrub/+, see
Figure 1B–C’). This indicates that the functional redundancy among the onDM14s is required to tolerate the
lgd shrub double heterozygous situation. Whether the redundancy of the odd-numbered domains is complete or whether they also have individual functions is not known. The findings for Lgd and Shrub have been recently confirmed for the orthologs of mammals and
C. elegans [
2,
7].
The interaction of the third DM14 domain (DM14-3) with Shrub has been determined at the atomic level, revealing that the positively charged surface of DM14-3 engages in electrostatic interactions with the negatively charged surface of Shrub [
2,
3]. It also identified the amino acids (AAs) responsible for binding, which are conserved in both onDM14s of Lgd. Replacement of individual AAs of this motif, here defined as the KARRxxR motif (see
Figure 2), by Alanine (A) abolishes the function of the onDM14s. The meaning of the KARRxxR motif has also been confirmed for the mammalian LGD orthologs [
2]. Although essential, it is not clear whether the motif is sufficient for the functionality of the onDM14s. Another open question is whether the positively charged AAs of the KARRxxR motif only provide charge, or an additional function. These questions are important to answer for the understanding of the function of DM14 domains and their interaction with members of the CHMP family.
Since the binding between Shrub protomers and the binding of Shrub to an onDM14 both require the negatively charged surface of Shrub, they are mutually exclusive. Nevertheless, recent work demonstrated that Lgd is a positive regulator of Shrub function, since its loss of function strongly reduces the activity of Shrub [
6,
7,
9]. One likely possibility for how Lgd acts is that its complex formation with Shrub prevents inappropriate polymerisation of Shrub in the cytosol. The inappropriate polymerisation would reduce the function of Shrub at the endosomal membrane [
6]. Another recently suggested possibility is that Lgd is required to efficiently recruit Shrub to the LM of the ME [
7].
Here, we investigate the requirement of the DM14 domains for the function of Lgd in detail. We found that although both onDM14s can act in a functionally redundant manner, the redundancy is not complete and both contribute to the full function of Lgd. Our analysis indicates that some of the AAs that form the KARRxxR motif of the onDM14s are not exchangeable by similarly charged AAs without loss of function, indicating that they not only provide charge but also fulfil structural roles. Moreover, we analysed the importance of AAs that are conserved in all DM14 domains. Furthermore, we show that the region of Lgd between DM14-4 and the C2 domain as well as its C-terminal region to the C2 domain are important for protein stability/function. Finally, our analysis of the C. elegans ortholog of Lgd revealed that it has only one DM14 domain that is functionally equivalent to the onDM14s. Altogether, the results further the understanding of how Lgd family members regulate the ESCRT machinery.
2. Materials and Methods
2.1. Fly Stocks
Gbe+Su(H)-lacZ [
23],
wg-lacZ [
24],
lgdd7 FRT40A [
25], and
shrub4-1 FRTG13 [
26] were used in this study. Flies were raised on a standard diet and kept at room temperature. For crossings, flies were raised at 25 °C.
2.2. Generation of Constructs and Vectors
Lgd rescue constructs were expressed under the control of the endogenous Lgd promotor (
lgdP) as described before [
9]. The used vector harbours the proximal lgd genomic elements (548 bp upstream and 553 bp downstream of the lgd ORF).
Truncated Lgds were generated using Gibson Assembly with
pattB-lgdP-Lgd-HA as a template [
9]. For the Gibson Assembly, primers were used to generate gene fragments with overlaps of approximately 30 bp. Primer sequences are in the
Supplemental Information as
Supplementary Table S1. AA changes were achieved by site-directed mutagenesis.
All lgdP constructs were inserted into the genomic attP landing site 86Fb. Injection of embryos was either performed in-house or by BestGene Inc. (Chino Hills, CA 91709, USA).
2.3. Genotypes
Figure 1B: w; +/+; Gbe+Su(H)-lacZ/+; Figure 1C: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/+; Figure 2B: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-1-2-3-4-C2 86Fb; Figure 2C: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP.lgd-1-2-3-4-C2 86Fb; Figure 2D: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-3-4-C2 86Fb; Figure 2E: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP.lgd-3-4-C2 86Fb; Figure 2F: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-1-C2 86Fb; Figure 2G: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP.lgd-1-C2 86Fb; Figure 2H: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-1-1-C2 86Fb; Figure 2I: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP.lgd-1-1-C2 86Fb; Figure 3C: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-3 K387R -4-C2 86Fb; Figure 3D: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP.lgd-3 K387R -4-C2 86Fb; Figure 3E: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-3 R389K-4-C2 86Fb; Figure 3F: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-3 R390K-4-C2 86Fb; Figure 3G: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-3 R393K-4-C2 86Fb; Figure 3H: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-3 A388G-4-C2 86Fb; Figure 3I: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP.lgd-3 A388G-4-C2 86Fb; Figure 4F: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-4-C2 86Fb; Figure 4G: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-4 KARRxxR HA 86Fb; Figure 4H: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-2-2-HA 86Fb; Figure 4I: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-2KARRxxR-4 KARRxxR -C2 86Fb; Figure 4J: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd 1-2 KARRxxR-3-4 KARRxxR -C2 86Fb; Figure 5B: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-3-R368A-C2 86Fb; Figure 5C: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-3-K380A-4-C2 86Fb; Figure 5D: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-3 K396A-4-C2 86Fb; Figure 5E: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd 3G408A-4-C2 86Fb; Figure 5F: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-3 P417A-4-C2 86Fb; Figure 5G: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP.lgd-3 R368A-4-C2 86Fb; Figure 5H: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP.lgd-3 K380A-4-C2 86Fb; Figure 5I: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP.lgd-3-4-C2 86Fb; Figure 6E: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-Δ558-664 HA 86Fb; Figure 6F: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-LAP HA 86Fb; Figure 6G: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.lgd-ΔC HA 86Fb; Figure 7B: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP.C.e.Lgd 86Fb; Figure 7C: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP. C.e.Lgd 86Fb; Figure 7D: w; lgdd7 FRT40A/lgdd7 FRT40A; Gbe+Su(H)-lacZ/lgdP. Δ 1-135 HA 86Fb; Figure 7E: w; lgdd7 FRT40A/lgdd7, shrb4-1 FRTG13; Gbe+Su(H)-lacZ/lgdP.lgd-Δ 1-135 HA 86Fb.
2.4. Immunohistochemistry and Microscopy
Dissected wing imaginal discs of wandering L3 larvae were fixed with 4% paraformaldehyde in PBS (4%PFA) for 30 min and washed with PBS. Permeabilization and blocking were carried out using 0.3% Triton X-100 in PBS (PBT) and 5% normal goat serum (NGS) for 30 min at room temperature. Primary antibody incubation was performed in 5% NGS in 0.3% PBT for 2 h at room temperature followed by three washing steps with PBT. The corresponding secondary antibody was applied in 5% NGS in 0.3% PBT for 2 h at room temperature. The following antibodies were used: mouse anti-Wg 4D4 (1:10, Developmental Studies Hybridoma Bank (DSHB), Iowa City, IA, USA), rabbit anti-β-Gal (polyclonal) (1:5000, MP Biomedicals, LLC., Solon, OH, USA). Fluorophore-conjugated secondary antibodies were purchased from Invitrogen. Nuclei staining was carried out using Hoechst 33258 dye.
Images were acquired with the Zeiss Axio Imager Z1 Microscope equipped with a Zeiss Apotome or Apotome2 (Carl Zeiss Microscopy GmbH, Jena, Germany).
2.5. Western Blot Analysis
For comparison of protein levels of different expressed Lgd variants, lysates of wandering L3 larvae were analysed by Western blot. For that, the larvae were collected and washed in PBS and then transferred in lysis buffer (10% Glycerin; 50 mM HEPES (pH7.5); 150 mM NaCl; 0.5% Triton-X-100; 1.5mM MgCl2; 1mM EGTA; Protease Inhibitor Cocktail (Sigma-Aldrich, St Louis, MI, USA). Rupture of the larval tissues was then carried out by using a micro pestle. After 15 min of incubation on ice with lysis buffer, the lysates were centrifuged and the supernatant was collected in a fresh tube avoiding the transfer of excess fat. Laemmli Buffer was added and heated to 95 °C for 10 min.
Lysates were loaded on a 10% SDS-PAGE gel and blotted on a PVDF membrane. Immunostaining was performed using standard protocols. Membranes were blocked using 5% milk powder in PBS and immunostaining was carried out in 2% milk powder in PBS.
The following antibodies were used: rat anti-HA (1:3000; 3F10, Roche) and rabbit anti-actin (1:10,000, 17H19L35, Invitrogen, Waltham, MA, USA). HRS-conjugated secondary antibodies were purchased from Jackson Immuno Research (872 W Baltimore Pike, West Grove, PA, USA). Chemiluminescence was detected using WesternBright ECL HRP substrate (Advansta, 2140 Bering Dr, San Jose, CA, USA) and the Amersham ImageQuant 800 (Cytiva, Marlborough, MA, USA).
2.6. Sequence Alignments and Structure Analysis and Visualisation
Sequence alignments were carried out using CLC Genomics Workbench 20 (QIAGEN, Aarhus, Denmark). The predicted Lgd protein structure was used from the Alphafold database (ID: AFQ9VKJ9-F1). Structure analysis and visualization were achieved using PyMOL (Schrödinger Inc, New York, NY, USA). Protein Structure Source: Figure 1A: Alphafold Database, Figure 1E,F: Alphafold prediction (Colabfold), Figure 1F’,F’ Alphafold multimere prediction (Colabfold); Figure 4A–C’: Alphafold prediction (Colabfold); Figure 5J: Alphafold Database; Figure 6A,B: Alphafold Database, Figure 6C: Alphafold prediction (Colabfold).
4. Discussion
Lgd is an important regulator of the ESCRT machinery, which is required for the full activity of the ESCRT-III core component Shrub/CHMP4 [
6,
9,
27]. In its absence, the activity of Shrub is reduced by more than 50%. For its function, the direct interaction with Shrub via its two onDM14s is required [
6,
9,
27]. Previous work suggests that the onDM14s act in a redundant manner. We found here that the redundancy of the onDM14s is not complete, suggesting that both onDM14s have to cooperate to achieve the full functionality of Lgd. The mechanism which underlies this cooperativity is unclear at the moment. Formally, it is possible that the cooperativity lies simply in the increase in the affinity to Shrub. We think that this is not the case, since the rescue of the sensitised background by Lgd-1-3-C2 is complete, while it is only partial for Lgd-1-1-C2 or Lgd-3-3-C2 ([
21] and this work). These findings argue against a simple affinity reason for the requirement of the two onDM14s and favours the possibility that the onDM14s have unique functions that are cooperatively required. The nature of the unique functions should be a focus of future work, as the onDM14-mediated interaction with Shrub is also conserved in mammals and is therefore a fundamental mechanism mediated by Lgd family members [
2].
It was surprising that C.e.Lgd appears to be unique with the lack of a second functional onDM14. We have previously shown that the presence of both onDM14s is important to be able to tolerate unfavourable genetic constitutions, such as double heterozygosity for
shrub lgd [
9]. Thus, it is required for the robustness of the Lgd/Shrub interaction in the situation of reduced levels of both partners. The outstanding unique feature of the onDM14s is the KARRxxR motif. Interestingly, remnants of the KARRxxR motif can still be recognised in the corresponding C.e.DM14-3, suggesting that the functional robustness provided by the two domains in Lgd orthologs of
Drosophila and mammals was initially present, but lost during evolution of
C. elegans. It seems that the robustness is no longer required in
C. elegans. In favour of this notion is that while the loss of function of Lgd is lethal in
Drosophila or mice (concomitant loss of both Lgds), it is only conditionally lethal in nematodes, as it depends on the feeding conditions [
7,
28]. This suggests that the importance of Lgd has diminished during nematode evolution.
We found here that the Rs in the KARRxxR motif cannot be replaced by Ks without a reduction in activity. Nevertheless, the rescue of
lgd mutants by the corresponding Lgd variants with R to K substitutions is much better than that of correspondent previously analysed variants with R to A substitutions [
21]. This indicates that the positive charge is one component of the requirement for function, but not the only one. It also appears that the structure of the Rs is important to make the necessary connection to Shrub. Indeed, in the crystal structure, the Rs of the KARRxxR motif are partly engaged in bivalent interactions with AAs of Shrub [
21].
The DM14 domain is a unique feature of the Lgd family [
18,
19]. The comparison of the DM14s of Lgd identified six absolutely conserved AAs at adequate positions in all four domains. Four of these AAs are also absolutely conserved in the DM14s of LGD1 and LGD2, while the two others are conserved in most cases or replaced by similar AAs. Surprisingly, only the mutation of the absolutely conserved P at the C-terminus of each DM14 abolished the function of DM14-3, while the other AAs lead to no detectable loss of activity, or only to a weak loss of activity, detectable only in the sensitised background. Interestingly, a mutation of the corresponding P in DM14-1 of the mammalian ortholog LGD2 (CC2D1A) has been recently associated with heterotaxy and ciliary disfunction in humans [
29]. Our analysis suggests that because of the partial functional redundancy among the onDM14s, the loss of this P results in a weak loss-of-function allele of LGD2, since one of the onDM14s is inactivated. Moreover, they also suggest that the mutation might only cause disease in certain genetic conditions, such as the additional heterozygosity of orthologs of
shrub, or other mutations that weaken the activity of Shrub. Thus, our results also provide a likely explanation for the rarity of the described mutation [
29].