Reinforcing the LINC complex connection to actin filaments: the role of FHOD1 in TAN line formation and nuclear movement

Abstract

Positioning the nucleus is critical for many cellular processes including cell division, migration and differentiation. The linker of nucleoskeleton and cytoskeleton (LINC) complex spans the inner and outer nuclear membranes and has emerged as a major factor in connecting the nucleus to the cytoskeleton for movement and positioning. Recently, we discovered that the diaphanous formin family member FHOD1 interacts with the LINC complex component nesprin-2 giant (nesprin-2G) and that this interaction plays essential roles in the formation of transmembrane actin-dependent nuclear (TAN) lines and nuclear movement during cell polarization in fibroblasts. We found that FHOD1 strengthens the connection between nesprin-2G and rearward moving dorsal actin cables by providing a second site of interaction between nesprin-2G and the actin cable. These results indicate that the LINC complex connection to the actin cytoskeleton can be enhanced by cytoplasmic factors and suggest a new model for TAN line formation. We discuss how the nesprin-2G-FHOD1 interaction may be regulated and its possible functional significance for development and disease.

Keywords:

• actin filaments
• Emery-Dreifuss muscular dystrophy
• FHOD1
• formin
• LINC Complex
• nesprin
• nuclear movement
• TAN lines

Abbreviations

LINC	=	linker of nucleoskeleton and cytoskeleton
nesprin-2G	=	nesprin-2 giant
TAN lines	=	transmembrane actin-dependent nuclear lines
SR	=	spectrin repeat
KASH	=	Klarsicht
ANC-1	=	Syne homology
DRF	=	diaphanous related formin
GBD	=	GTPase binding domain
DID	=	diaphanous inhibitory domain
FH	=	formin homology
DAD	=	diaphanous autoregulatory domain
ABS	=	actin binding site
LPA	=	lysophosphatidic acid
GFP-mN2G	=	GFP-mini-nesprin-2G
CH	=	calponin homology
EDMD	=	Emery-Dreifuss muscular dystrophy

Metazoans actively control the position of the nucleus in the cell to regulate cell division, differentiation, and migration.¹ In fibroblasts and myoblasts, actin retrograde flow powers the rearward movement of the nucleus to orient the centrosome and polarize cells for directed migration (Fig. 1A).^2,3 Many nuclear movements are mediated by connections between the cytoskeleton and the nucleus established by specific LINC complexes.^1,4,5 For the actin-dependent nuclear movement in fibroblasts and myoblasts, the specific LINC complex is composed of nesprin-2G, a ˜800 kDa, actin-binding and spectrin repeat (SR)-containing outer nuclear membrane protein and SUN2, an inner nuclear membrane protein that binds the KASH (Klarsicht, ANC-1, Syne homology) domain of nesprin-2G.^3,6 The association of these proteins with dorsal actin cables results in their assembly into linear arrays termed TAN lines (Fig. 1).^3,6,7 TAN lines are anchored by association of SUN2 with A-type lamins.⁸ Two inner nuclear membrane proteins, Samp1 and emerin, are also found in TAN lines and may contribute to their anchoring.^9,10 Emerin additionally affects actin-dependent nuclear movement by interacting with myosin IIB and organizing the directionality of actin flow.¹⁰

Figure 1. TAN line dependent nuclear movement. (A) Model for TAN line dependent nuclear movement in fibroblasts and myoblasts polarizing for migration. In serum-starved wounded monolayers of fibroblasts and myoblasts, LPA stimulation induces the retrograde flow of dorsal actin cables (blue). These actin cables engage nesprin-2G on the nuclear surface to form TAN lines (red) resulting in the rearward movement of the nucleus while the centrosome stays in the center of the cell. This establishes cell polarization by orienting the centrosome toward the wound edge. (B) Immunofluorescence images of TAN lines in NIH3T3 fibroblasts expressing the nesprin-2G chimera, GFP-mini-N2G. Cells were stimulated with LPA for one hr and the formation of TAN lines visualized by staining for GFP-mini-N2G and F-actin. The boxed region in the left panel is shown at higher magnification in the panels to the right. Arrows indicate TAN lines with colocalized GFP-mini-N2G and actin cables. Scale bar, 5 μm.

TAN lines resemble other membrane adhesions, such as focal adhesions and cadherin cell-cell adhesions, in that they form by actin-dependent clustering of membrane proteins and transmit force. The large number of proteins associated with focal and cell-cell adhesions suggests that additional proteins are likely to be associated with TAN lines. In a recent study,¹¹ we identified the diaphanous related formin (DRF) FHOD1 as interaction partner of nesprin-2G and a new component of TAN lines. Like other DRFs, FHOD1 is composed of an N-terminal GTPase binding domain (GBD), followed by the diaphanous inhibitory domain (DID), the central formin homology (FH) domains FH1 and the actin binding FH2, and the C-terminal diaphanous autoregulatory domain (DAD) (Fig. 2A). Despite a similar domain organization, FHOD1 exhibits features that distinguish it from other DRFs. Importantly, It does not stimulate actin polymerization in vitro, but instead appears to cap actin filaments.¹² Also unlike other DRFs, FHOD1 bundles actin filaments both in vivo and in vitro.^12-15 This bundling activity requires dimerization provided by the FH2 domain, but also a second F-actin binding domain in the N-terminus between the DID and the FH1 domain that is unique to FHOD1 among the 15 mammalian formins (Fig. 2C). This second actin binding site (ABS) also endows FHOD1 with the ability to bind along the length of actin filaments, unlike other DRFs, which interact transiently with actin filament barbed ends. FHOD1 is regulated by autoinhibition which is accomplished by binding of the DAD to the DID. Although for most DRFs release of autoinhibition is mediated by binding of specific Rho GTPase(s) to the GBD, for FHOD1 release of autoinhibition occurs by phosphorylation of the DAD (see below).

Figure 2. Multi-site attachment model for TAN lines. (A) Schematic of the autoinhibited form of FHOD1 (shown as a monomer to emphasize domains). Individual domains are described in the text. (B) Model for the multi-site attachment of nesprin-2G (N2G) and FHOD1 to actin cables in TAN lines. The interaction of nesprin-2G with FHOD1 forms a branched connection between nesprin-2G and the actin cable with one connection provided by nesprin-2G's CH domains and the other by FHOD1s unique ABS. This branched connection is proposed to strengthen the association between the nesprin and the actin cable. In the perinuclear space between the inner (INM) and outer nuclear membrane (ONM), KASH domains of nesprin-2G interact with the SUN2 trimer. In the nucleoplasm, SUN2 is anchored by interaction with lamin A/C of the lamina and with the INM proteins Samp1 and emerin. (C) A detailed view showing FHOD1 interacting with SRs 11–12 of nesprin-2G through its N-terminal GBD and DID and to the actin cable through its ABS. The dimeric nature of FHOD1 may bring multiple nesprin-2Gs together.

The FHOD1 N-terminus (residues 1–339) is structurally different from other DRFs¹³ and we used it as a bait in a yeast 2-hybrid screen for FHOD1 interacting proteins. We found that a fragment of human nesprin-2G (residues 1340–1669) interacted with the N-terminal fragment of FHOD1.¹¹ By GST-pulldown and co-immunoprecipitation assays we confirmed the interaction of the proteins in mammalian cells. The interaction of the full length proteins was shown by co-immunoprecipitation of endogenous nesprin-2G with over-expressed FHOD1. A directed yeast 2-hybrid interaction screen with nesprin-2G fragments covering the length of nesprin-2G and additional GST-pulldown assays revealed that the only site of FHOD1 interaction on nesprin-2G was with SRs 11 and 12.

Because of the importance of nesprin-2G for centrosome orientation and nuclear movement in fibroblasts, we tested if FHOD1 was required for these processes by knocking it down with siRNA oligonucleotides in NIH3T3 fibroblasts. Similar to the phenotype of nesprin-2G knockdown cells, FHOD1-depleted cells did not orient their centrosomes due to a failure to move their nuclei rearward after stimulation with lysophosphatidic acid (LPA), a serum factor that stimulates these processes in starved NIH3T3 fibroblasts.¹⁶ Defects in centrosome orientation and nuclear movement in FHOD1-depleted cells were rescued by re-expression of wild-type FHOD1 but not by re-expression of FHOD1 lacking residues 1–339 that interact with nesprin-2G. Consistent with the importance of the interaction between FHOD1 and nesprin-2G, overexpression of the interacting fragments of nesprin-2G and FHOD1 inhibited centrosome orientation and nuclear movement.

Rearward nuclear movement requires the formation and retrograde flow of dorsal actin cables that engage the nucleus through TAN lines.^6,10 Despite its proposed role in regulating the actin cytoskeleton, FHOD1 knockdown cells did not have significant differences in total levels of F-actin, the number of dorsal actin cables above the nucleus, or the velocity of actin retrograde flow after LPA stimulation compared to control cells. However, TAN line formation was severely impaired in FHOD1 knockdown cells. Importantly, expressed RFP-FHOD1 localized to TAN lines, which were visualized by staining of either expressed GFP-mini-nesprin-2G (GFP-mN2G, a minimal functional nesprin-2G containing the N-terminal calponin homology (CH) domains and the C-terminal KASH domain⁶) or endogenous nesprin-2G.

We further dissected the role of FHOD1 in TAN line formation by expressing various forms of FHOD1 in FHOD1 depleted cells. We found that a FHOD1 construct comprising the entire N-terminus (residues 1–569) and lacking the canonical formin FH1 and FH2 domains localized in TAN lines, rescued centrosome orientation and partially rescued nuclear movement in FHOD1 deficient cells. In addition to a nesprin-2G binding site, FHOD1 1–569 contains the unique ABS of FHOD1. This consideration raised the possibility that FHOD1 bound to nesprin-2G contributes to TAN line formation by providing a second F-actin binding site in addition to the actin binding CH domains of nesprin-2G. To test this possibility, we fused the nesprin-2G interacting site of FHOD1 (residues 1–339) with the F-actin binding CH domains of α-actinin. Expression of this chimera rescued the centrosome orientation and nuclear movement defects in FHOD1 knockdown cells as well as FHOD1 1–569, implying that the FHOD1-nesprin-2G interaction provides a second linkage to actin cables in addition to that provided by the CH domains of nesprin-2G.

These findings support a new model for TAN line engagement of actin cables (Fig. 2B). In this model, nesprin-2G attaches to actin cables by 2 sites of interaction: its own CH domains and the unique ABS in the N-terminus of FHOD1. We hypothesize that this branched connection reinforces the nesprin-actin interaction and allows it to resist the high forces necessary to move such a large organelle as the nucleus. The model also implies that the orientation of nesprin-2G relative to the actin cable is likely to be nearly parallel (at least for the region between the CH domains and the SR11–12) rather than the end-on orientation that was originally envisioned.^6,7 This conclusion is based on the extended structure of SR proteins and average 6 nm length of individual SRs, yielding a distance of at least 60 nm between the CH domains and SRs11–12 of nesprin-2G.^17,18 Coupled with the predicted length of the FHOD1 N-terminal fragment that interacts with both nesprin-2G and the actin cable (<10 nm),¹³ the only way that both proteins can interact with the actin cable is if nesprin-2G lies along the axis of the actin cable. Lastly, FHOD1 is hypothesized to form a dimer through its FH2 domains.¹⁴ Thus, FHOD1 has the potential to bring two nesprin-2Gs together. In this way, FHOD1 may additionally contribute to the assembly of TAN lines by enhancing the avidity of nesprin-2G for the actin cables. Perhaps, this explains why monomeric constructs based on the N-terminus of FHOD1 did not fully rescue nuclear movement.

In muscle cells there are 2 splice variants of FHOD1 in addition to the widely expressed version in fibroblasts: one variant contains a 28 aa insert in the ABS, the other variant is a short N-terminal form that includes a unique 7 aa insert and terminates within the ABS.¹⁹ Neither of these splice variants have been functionally characterized, but it is interesting to note that the short variant is similar to the construct that we used to rescue the nuclear movement defect in FHOD1 depleted cells. Additional studies to refine the location of the actin binding site may reveal whether this short variant retains the capacity to bind actin filaments. If so, it has the potential to reinforce the nesprin-2G-actin cable interaction; if not, it may act as an endogenous inhibitor of the interaction.

The N-terminal site on FHOD1 that interacts with nesprin-2G is in the same region that is involved regulating its activity. This raises interesting questions concerning how FHOD1 interaction with nesprin-2G may be regulated and whether nesprin-2G interaction with FHOD1 may maintain it in an active state. Like other DRFs, FHOD1 is regulated by auto-inhibition through an intra-molecular interaction between its N-terminal DID and C-terminal DAD.²⁰ By co-immunoprecipitation of FHOD1 transfected 293T cells, in preliminary experiments we have observed that a nesprin-2G fragment containing SR11–12 co-immunoprecipitated the constitutively active FHOD1 ΔDAD more efficiently than it did wild-type FHOD1. This implies that the intra-molecular interaction between FHOD1's DID and DAD reduces the interaction between FHOD1 and nesprin-2G and that the release of FHOD1's autoinhibition may be a pre-requisite for efficient FHOD1-nesprin-2G interaction.

There have been several mechanistic models proposed for how the autoinhibition of FHOD1 is released. The GBD of FHOD1 interacts with Rac1-GTP, but this does not seem to activate FHOD1.^13,20,21 Phosphorylation of FHOD1 at three serine/threonine residues by ROCK1 partially releases autoinhibtion.^14,22 Additionally, phosphorylation of FHOD1 at Y99 by Src was shown to be a prerequisite for phosphorylation of the DAD by ROCK1.²³ Phosphorylation of FHOD1 by Src and ROCK1 to release autoinhibition might be relevant to activate FHOD1 for TAN line formation. However, expression of dominant negative RhoA does not inhibit centrosome orientation, suggesting that activation of FHOD1 by ROCK is not required for rearward nuclear movement.¹⁶ Perhaps for FHOD1's function in linking nesprin-2G to actin cables in TAN lines, only the phosphorylation of the N-terminus by Src is required. It will be interesting to test whether Src is required for TAN line formation and nuclear movement.

SRs11–12 are in one of 2 clusters of SRs in nesprin-2G (SRs11–13 and SRs 49–53) that are phylogenetically conserved and are predicted to interact with proteins based on their probability of surface exposure.^24,25 Interestingly, an additional protein meckelin, which is localized in plasma membrane and ER, was found to bind to the SRs 11–13.^26-28 Taking this into account, FHOD1 and meckelin binding to the SRs could work independently, cooperatively, or competitively to regulate the formation of TAN lines. While SRs 11–13 are close to the CH domains at the N-terminus of nesprin-2G, SRs 49–53 are at the other end of the molecule near the transmembrane domain.²⁴ A number of proteins bind to this region²⁹ raising the possibility that additional bridging proteins will be important for TAN line formation and nuclear movement.

Despite the importance of the FHOD1-nesprin-2G interaction for nuclear movement and polarity regulation of cells in culture, it remains to be tested whether the FHOD1-nesprin-2G interaction is involved in animal development, differentiation or pathogenesis. Nesprin-2 has numerous splice isoforms from the giant nesprin-2G to shorter isoforms that lack actin interacting CH domains or even the transmembrane KASH domain.³⁰ When the exon of Syne2 (the gene encoding nesprin-2) specifying the KASH domain is deleted, mice exhibit defects in laminar formation in the cerebral cortex and hippocampus during brain development resulting in poor learning skills.³¹ Nesprin-1G is structurally similar to nesprin-2G, but lacks the set of nesprin-2G conserved SRs that interact with FHOD1.^24,25 Unlike the KASH null nesprin-2 mouse, the KASH null nesprin-1 mouse does not have brain developmental defects.³⁰ This difference in phenotype may reflect FHOD1's specific interaction with nesprin-2G and raises the possibility that the nesprin-2G-FHOD1 interaction is important for brain development.

There is an intriguing correlation between proteins contributing to TAN line- dependent nuclear movement in fibroblasts and myoblasts and variants in their genes causing Emery-Dreifuss muscular dystrophy (EDMD). Mutations in genes encoding emerin^32-34 and A-type lamins³⁵ were the first to be associated with EDMD, but more recent studies have identified variants in nesprin-2³⁶ and SUN2.³⁷ In addition to skeletal dystrophy, EDMD frequently leads to dilated cardiomyopathy. Furthermore, mice with KASH null nesprin-1 and 2 genes in cardiomyocytes develop cardiomyopathy, and various mutations of nesprin-2 have been found in cardiomyopathy patients.³⁸ Expression of EDMD variants of lamin A and SUN2 or depletion of emerin, as occurs with many EDMD emerin variants, leads to defective nuclear movement and centrosome orientation in fibroblasts.^8,10,37 Expression of FHOD1 is found in many tissues and like nesprin-2G it is particularly high in heart, skeletal muscle, and spleen.^19,30 The coincidence of expression patterns of FHOD1 and nesprin-2G and their interaction suggest that FHOD1 like nesprin-2 plays a pivotal role in the maintenance of striated muscle, and the disruption of the interaction between nesprin-2G and FHOD1 may lead to muscular dystrophies. Although variants of FHOD1 have not been associated with disease, FHOD1's closet relative, FHOD3 is expressed in the heart and a missense variant of FHOD3 has been found in a patient case with adult-onset familial dilated cardiomyopathy.³⁹ Interestingly, FHOD1 localizes in intercalated discs and costameres in cardiomyocytes, unlike FHOD3's cross-striated pattern,^40,41 suggesting a distinct role for FHOD1 in cardiomyocyte function. It will be interesting to see if FHOD1 mutations are associated with skeletal or cardiac muscle diseases.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

Our work on nuclear movement is supported by NIH (grant GM 099481 and NS059352 to GGG) and the Deutsche Forschungsgemeinschaft (grant FA 378/ 15–1 to OTF).