The missing LINC

Abstract

Pairing of homologous chromosome is a unique event in meiosis that is essential for both haploidization of the genome and genetic recombination. Rapid chromosome movements during meiotic prophase are a key feature of the pairing process. This is usually telomere-led, and in metazoans is dependent upon microtubules and dynein. Chromosome movements culminate in the formation of a meiotic “bouquet” in which nuclear envelope-associated telomeres are clustered at the centrosomal pole of the nucleus. Bouquet formation is thought to facilitate homolog pairing. Recent studies reveal that coupling of telomeres to cytoplasmic dynein is mediated by SUN1 in the inner nuclear membrane (INM) and KASH5 a novel protein of the outer nuclear membrane (ONM). Together SUN1 and KASH5 assemble to form a transluminal LINC (linker of the nucleoskeleton and cytoskeleton) complex that spans both nuclear membranes. SUN1 forms attachment sites for telomeres at the INM while KASH5 functions as a dynein adaptor at the ONM. In mice deficient in KASH5, homologous chromosome pairing does not occur. The result is that meiosis is arrested at the leptotene/zygotene stage of meiotic prophase 1, and as a consequence both male and female mice are infertile. This study demonstrates an essential role for dynein directed telomere movement during meiotic prophase.

Introduction

The evolution and diversification of eukaryotes has, to a large extent been facilitated by sexual reproduction. The mixing and subsequent shuffling of genes between generations provides the raw material upon which natural selection can act. Sexual life cycles are underpinned by the process of meiosis, in which parental homologous chromosomes are segregated, in a reduction division, to yield haploid cells. In higher eukaryotes, meiosis is a crucial step in the development of mature germ cells, and the evolution of meiosis has been a topic of considerable debate.¹ While there is little doubt that meiosis represents a modified form of mitosis, it features several unique steps. These include the pairing of homologous chromosomes coupled with widespread recombination between non-sister chromatids. Following homolog segregation, a second mitosis-like division distributes non-replicated chromatids between daughter cells.

The first meiotic prophase (prophase 1) is characterized by the pairing of homologous chromosomes and the assembly of synaptonemal complexes.^2-5 The latter are specialized structures resembling molecular zippers extending the entire length of each chromosome and which hold the homologs together. The mechanism by which homolog pairs are established during meiotic prophase has been a topic of debate lasting more than a century. One of the significant early observations was that during prophase 1 telomeres, in association with the nuclear envelope (NE), cluster at the nuclear pole closest to the centrosome, with the chromosomes adopting a transient bouquet-like configuration (references in Scherthan, 2001⁶). It is now thought that the process of bouquet formation, which features rapid chromosome movements, might facilitate the mutual searching of the homologous chromosomes as a prelude to synapsis.^6,7 Studies in a variety of organisms now suggest that this is very likely the case and that the NE, the barrier that separates the nucleus and cytoplasm, plays a central role in the pairing process. Significantly, the NE-associated chromosomal movements that are a prelude to bouquet formation are actually driven by the cytoskeleton, and in this way NE-linked chromosome dynamics are intimately tied to meiotic progression.^7-10 The implication here is that there must be mechanisms to transmit cytoskeletal forces across the NE to individual chromosomes.

The general structure of the NE has been conserved in all eukaryotes.^11,12 Its most prominent features are the inner and outer nuclear membranes (INM and ONM) separated by a peri-nuclear space (PNS) of about 50 nm. The two nuclear membranes are periodically connected at annular junctions where they are spanned by nuclear pore complexes, massive multi-protein assemblies mediating the to-and-fro movement of macromolecules between the nucleus and cytoplasm. However, despite these connections, the INM and ONM are biochemically distinct. The INM contains a unique array of 60 or more integral membrane proteins¹³ and is associated on its nuclear face with the nuclear lamina, a thin (10–50 nm) relatively insoluble protein meshwork. The nuclear lamina is composed primarily of A- and B-type lamins,¹⁴ prototype members of the intermediate filament protein family. In addition to its interactions with the INM, which is mediated by several integral proteins, the lamina provides anchoring sites at the nuclear periphery for underlying chromatin domains and is thought to represent an important structural determinant for the NE as a whole. The ONM, on the other hand, is functionally related to the peripheral endoplasmic reticulum (ER), with which it maintains numerous connections. In this way, the ER, ONM, and INM may be considered as separate domains within a single continuous membrane system with the PNS representing a perinuclear extension of the ER lumen.

LINC Complexes and Nucleocytoplasmic Coupling

The double membrane architecture of the NE raises the question of how chromosomal regions, telomeres in particular, can be physically linked to the cytoskeleton during meiotic prophase 1. Since the NE remains intact during the process of synapsis there has to be molecular machinery in place that spans both the INM and ONM that can interact with cytoskeletal components and with chromatin. A possible solution to this puzzle was provided by studies on interphase somatic cells. These revealed that nuclear structures, including nuclear lamins, are mechanically coupled to a variety of cytoskeletal proteins.¹⁵ The nature of this coupling is partly dependent on two families of integral nuclear membrane proteins: SUN (Sad1, Unc-84) domain proteins, which reside in the INM, and KASH (Klarsicht, ANC-1, Syne Homology) domain proteins, residing in the ONM.^16,17 KASH domain proteins play important roles in nuclear migration and positioning in a variety of cell and tissue types in multiple organisms, and interact with an assortment of cytoskeletal components. For instance, both Nesprins 1 (Nesp1) and 2 (Nesp2) in mammals bind F-actin as does ANC-1 in C. elegans.^18-22 Drosophila Klarsicht and C. elegans ZYG-12 both function as ONM adaptors for cytoplasmic dynein,^23,24 while mammalian Nesprin 4 (Nesp4) represents an epithelial-specific ONM binding partner for conventional Kinesin-1.^25,26 Dynein and kinesin binding sites are also present in Nesp1 and Nesp2. Finally, mammalian Nesprin 3 (Nesp3) interacts with plectin, a versatile cytolinker molecule, that provides a connection to the intermediate filament system.^27-29 In this way, diverse ONM KASH domain proteins are able to mediate associations between the NE and all three branches of the cytoskeleton. An overview of the mammalian KASH protein family is provided in Figure 1.

Figure 1. Comparison of mammalian KASH-domain proteins.^{25,29,59,60,64,65} In all cases the N-terminal domain resides within the cytoplasm while the C-terminal KASH domain extends across the outer nuclear membrane (ONM) and into the peri-nuclear space (PNS) where it interacts directly with members of the SUN domain protein family. The largest isoforms of nesprins 1 and 2 (Nesp1 and Nesp2) feature an N-terminal actin binding domain. These proteins also possess kinesin and dynein binding functions. In the case of Nesp2, the kinesin binding region resides close to the KASH domain since a smaller Nesp2 isoform, Nesp2β, is capable of recruiting kinesin to the nuclear envelope.⁶⁶ One other vertebrate KASH domain protein, Nesp4 also functions in kinesin binding. Nesp4 is expressed largely in epithelial cells where it might have a role in nuclear positioning. Nesp3 functions as an adaptor for plectin, a versatile cytolinker molecule that provides a means of coupling nuclei to the intermediate filament system. Lymphocyte restricted membrane protein (LRMP/Jaw1) is unusual in that it doesn’t appear to interact with cytoskeletal components. Instead LRMP forms an association with the IP3 receptor (IP3R).⁶⁷ The final member of the KASH protein family, KASH5, functions as an ONM adaptor for cytoplasmic dynein in meiotic cells.

Since the nuclear membranes share numerous continuities with the peripheral ER, the question arises as to how KASH proteins are retained and concentrated within the ONM. The answer is provided by the eponymous KASH domain and its interactions. The KASH domain is a C-terminal sequence of some 50–60 amino acid residues.²² It features a single membrane-spanning region followed by a short sequence of 40 residues or less that extends in to the PNS. The KASH domain is both necessary and sufficient for targeting to the ONM. In C. elegans retention of the KASH domain protein, ANC-1, within the ONM is dependent upon the presence in the INM of UNC-84,²² a prototype SUN domain protein.

Both mouse and human genomes encode at least six SUN domain proteins, SUN1, 2, 3, 4 (SPAG4), 5 (SPAG4L), and osteopotentia.^30,31 Of these, only two, SUN1 and SUN2, are widely expressed. Both SUN1 and SUN2 (as well as SUN3–5) are type II membrane proteins. The SUN1 and SUN2 N-terminal domains (~200–400 residues) are exposed to the nucleoplasm where they may interact with a variety of nuclear components,^32-35 including members of the lamin family. The C-terminal region of SUN1 and SUN2 extends into the PNS. A membrane proximal sequence, which forms a triple helical coiled-coil^30,32-35 terminates in the globular SUN domain. It is now clear from studies in both mammals and C. elegans that SUN and KASH domains directly interact. In this way INM SUN domain proteins provide a unique function as transluminal tethers for ONM KASH proteins.^34-37 The implication of these findings is that together, SUN and KASH domain proteins form a pair of links in a molecular chain spanning both nuclear membranes that physically couples nuclear components to the cytoskeleton. These SUN-KASH pairs are termed LINC complexes (for linker of the nucleoskeleton and cytoskeleton).³⁵

Recent structural studies reveal that in mammalian SUN2, the SUN domain functions as a homo-trimer.^38,39 KASH-domain binding involves extended interactions at the interface between two adjacent SUN domains.³⁸ This association may be stabilized by the formation of an interchain disulphide bond between a cysteine within the KASH domain and an adjacent SUN domain cysteine. SUN2 (and in all likelihood, SUN1) trimerization is mediated by the membrane proximal luminal coiled-coil domain. This domain has a predicted length of 40–50 nm, sufficient to span the PNS and to bring the trimeric SUN domains close to the ONM where they can bind KASH sequences. The molecular modeling indicates that each SUN protein trimer has the capacity to bind three KASH domains. Accordingly, if individual KASH proteins are able to oligomerize, then SUN-KASH associations could promote the formation of extensive LINC complex arrays within the NE.

A number of LINC complex isoforms, defined by their SUN/KASH constituents, possess essential roles in nuclear movement and positioning in a variety of cell and tissue types. Actin-dependent nuclear repositioning during fibroblast migration⁴⁰ is dependent on Nesp1/2-containing LINC complexes. Cell cycle-dependent nuclear oscillation during neuronal and retinal differentiation utilizes LINC complexes that function as adaptors for microtubule motor proteins.⁴¹ A Nesp4-containing LINC complex is required for the microtubule- and Kinesin-1-dependent basal positioning of the nucleus in outer hair cells (OHCs) of the inner ear.^25,26 In the absence of Nesp4 OHC viability is severely compromised resulting in progressive irreversible hearing loss.²⁶ Dynein-binding LINC complexes are also employed in pronuclear migration in fertilized eggs.^24,42 In addition to these roles in directed nuclear movement and anchoring, it is now clear that SUN and KASH domain proteins may also influence intranuclear organization. During meiosis, LINC complexes have a critical role in transmitting cytoskeletal forces across both nuclear membranes to individual chromosomes.^43-46 LINC complexes therefore mediate the rapid chromosome movements that are a prelude to bouquet formation during meiotic prophase 1.^6,47

LINC Complexes in Meiosis

Studies in both fission yeast and in C. elegans and have uncovered key roles played by LINC complexes during meiotic prophase. In S. pombe, the KASH protein Kms1 serves as an ONM adaptor for cytoplasmic dynein. Kms1 is anchored at the nuclear surface through transluminal association with Sad1, a SUN domain protein of the INM.^48,49 Sad1 foci in turn constitute NE attachment sites for meiotic telomeres. The association between Sad1 and telomeres is mediated by both Taz-1 and Rap-1, a pair of telomeric proteins, as well as by the meiosis-specific proteins, Bqt-1, 2, 3, and 4.^43,50 Through this chain of interactions spanning the NE, Kms1-associated dynein drives the clustering of telomeres to form a meiotic bouquet, as well as the subsequent “horsetail”-like nuclear movements that are thought to facilitate homologous chromosome pairing. Budding yeast also utilizes LINC complexes to mediate bouquet formation.^9,51,52 However, it is unusual in that the process is actin-rather than microtubule-mediated.^8,10

Meiosis in C. elegans follows the S. pombe archetype where homolog pairing is facilitated by dynein-dependent bouquet formation. In the nematode, however, association of chromosomes with the NE is not mediated by telomeres. Instead, chromosome specific pairing centers (PCs) are used. ZYG-12, a KASH domain protein of the ONM and functional homolog of Kms1 functions as an adaptor for cytoplasmic dynein. An INM SUN domain protein, SUN-1/matefin (MTF-1), tethers ZYG-12 in the ONM and at the same time defines attachment sites for chromosomal PCs.^46,53,54 This association between SUN-1/MTF-1 and PCs, is mediated by soluble chromosome- and PC-specific proteins (ZIM-1, -2, -3, and HIM-8).^46,53,55-57

LINC complexes are also implicated in mammalian meiotic progression. Gene knockout studies in mice revealed that the INM SUN domain protein, SUN1, is essential for gametogenesis.^44,45 Both males and females deficient in SUN1 are infertile. Infertility in males is clearly due to the arrest of primary spermatocytes in meiotic prophase 1, with the complete failure of telomeres to localize appropriately to the nuclear periphery. These studies demonstrate that, as with fission yeast, telomere association with the NE is a prerequisite for bouquet formation and efficient homolog pairing. Telomere clustering always occurs at the pole of the nucleus that is closest to the centrosome or microtubule organizing center (MTOC, reviewed in ref. 6 and 58). By analogy with S. pombe and C. elegans, this is likely to be a dynein-mediated process requiring the function of a meiotic LINC complex that couples telomeres to the microtubule system. While SUN1 clearly represents the INM portion of such a LINC complex, the identity of its ONM KASH domain portion has until very recently remained uncertain.

KASH5: A New Dynein Adaptor

A candidate KASH component of a meiotic LINC complex, KASH5, was first revealed in a yeast two-hybrid screen of a mammalian testis library using the mouse cohesin protector protein, shugosin 2, as bait.⁵⁹ Although this interaction was likely an artifact, characterization of KASH5 revealed localization to the NE of primary spermatocytes in a SUN1-dependent manner. Furthermore it was found to associate with p150^Glued, a component of the dynein regulatory complex, dynactin. In complementary studies, we had searched for new proteins that displayed sequence homology with the Nesp2 KASH domain.²⁵ We were able to identify two additional KASH domain proteins, Nesp4²⁵ and lymphocyte restricted membrane protein (LRMP, also known as JAW1).⁶⁰ Both of these proteins display SUN1/2- and KASH domain-dependent localization to the ONM. In the case of Nesp4, we now know that it functions as an epithelial NE adaptor for Kinesin-1.²⁵ A second homology search using the complete LRMP sequence as probe detected a zebrafish KASH protein, Futile Cycle (Fue). Fue is required for dynein-dependent nuclear migration in zebrafish zygotes.^42,61 When aligned with mouse LRMP, Fue displays an N-terminal extension that contains its dynein-binding region. Yet a third homology search using the unique Fue sequence revealed mouse KASH5. Further in silico analysis of the complete KASH5 sequence of 578 amino acid residues highlighted a degenerate, but nevertheless recognizable C-terminal KASH domain, a central predicted coiled-coil and an N-terminal domain containing a calmodulin-like EF-hand sequence.⁶²

Consistent with the observations of Morimoto et al. (2012),⁵⁹ we confirmed that in adult mice, KASH5 expression was restricted largely to the testis.⁶⁰ However, RT-PCR also detected the presence of KASH5 transcripts in bone marrow derived cells. Although no cell lines could be found that expressed KASH5 at any level, when recombinant KASH5 was introduced in to HeLa or HEK293T cells it localized to the ONM.⁶⁰ This localization was dependent upon the integrity of its KASH domain as well as upon the presence of SUN1 or SUN2. Significantly, KASH5 was able to recruit cytoplasmic dynein and associated regulatory molecules (e.g., the dynactin complex and LIS1) to the NE. Recruitment depended on the KASH5 N-terminal domain containing the EF-hand motif. Related experiments, in cultured cells, revealed that KASH5 can self-associate through the central coiled-coil domain. Taken together, these findings indicate that KASH5 defines a new LINC complex species that couples the NE and associated nuclear structures to the microtubule system via cytoplasmic dynein. The fact that KASH5 can oligomerize also suggests that KASH5 LINC complexes could potentially form large arrays or clusters.

SUN1 is the predominant SUN domain protein expressed in spermatocytes.⁴⁴ By conventional immunofluorescence microscopy it appears to be distributed across the spermatocyte nuclear surface as discrete foci.⁴⁴ Each of these foci is coincident with a telomere attachment site. Furthermore, SUN1 and KASH5 precisely co-localize, consistent with the notion that these two molecules assemble to form a meiotic LINC complex.⁶⁰ Such a suggestion is reinforced by the finding that KASH5 is lost from the spermatocyte NE in mice deficient in SUN1.^59,60

To better define the physiological role of KASH5 we derived mice in which both KASH5 alleles were non-functional.⁶⁰ While KASH5-null mice are healthy with an apparently normal lifespan, both males and females are infertile. In the males no mature sperm cells are produced whatsoever. A more careful analysis reveals that spermatogenesis is arrested during meiotic prophase 1. Arrest was at pre-pachytene and was associated with a failure to resolve double strand DNA breaks. Loss of maturing sperm coincided with extensive apoptosis (as measured by TUNEL staining) in the seminiferous tubules. If as we suspected that KASH5-containing LINC complexes are required for telomere clustering and efficient homolog pairing then prophase arrest should be occurring prior to synaptonemal complex formation. SCs contain two major structural features. Axial elements assemble as a strip along the entire length of each homolog. During the final stages of SC formation axial elements become linked by a series of transverse elements akin to the rungs of a ladder. By following the recruitment and assembly of the axial element protein SCP3 and the transverse element protein SCP1 it is possible to follow the progress of synapsis. This analysis was greatly facilitated by the use of structured illumination microscopy (SIM),⁶³ a super-resolution technique that allowed us to resolve paired axial elements, which in synapsed chromosomes, are separated by a distance of only 100–150 nm (Fig. 2). Visualization of individual axial elements within assembled SCs is not possible using conventional light microscopy which has a diffraction limited resolution of about 200 nm (Fig. 2). These analyses revealed that in the absence of KASH5 homologous chromosomes remained unpaired. Nevertheless, SUN1 was still localized appropriately as discreet foci in the KASH5-null spermatocytes and was still capable of mediating telomere attachment to the NE. However, in the absence of KASH5, these telomeres had lost their link to dynein and to the microtubule system and therefore failed to cluster at the centrosomal pole of the nucleus. The prediction was that telomere led chromosome movements during prophase 1 should be significantly curtailed in KASH5-null spermatocytes.

Figure 2. Immunofluorescence microscopy of a wild type spermatocyte spread employing antibodies against KASH5 (green) and SCP3 (red), an axial element component of synaptonemal complexes. Individual axial elements are not resolved using conventional widefield microscopy. However, these structures can be distinguished using structured illumination microscopy (SIM).

SIM also revealed some subtle features of SUN1/KASH5 LINC complexes that were not evident with conventional light microscopy. At unpaired telomere attachment sites, the LINC complexes, visualized using either anti-SUN1 or anti-KASH5 antibodies, appeared in ring-like clusters. Following homolog pairing, rings merged in to figure-of-eight arrangements (Fig. 3). However, in KASH5-null spermatocytes, while SUN1 was still clustered to form telomere attachment sites, the ring-like organization of these clusters was lost. Although SUN1 functions as a tether for KASH5, the nanoscale distribution of SUN1 appears to be influenced by KASH5.⁶⁰

Figure 3. Structured illumination microscopy of spematocytes labeled with antibodies against KASH5 and SCP3. KASH5 is organized into ring or figure-eight-like structures.

From these results it appears that mammalian meiosis conforms to the paradigm established in S. pombe and C. elegans whereby a LINC complex couples telomeres (or PCs in the case of C. elegans) within the prophase nucleus to dynein within the cytoplasm (Fig. 4). By this route, meiotic LINC complexes promote telomere/PC clustering leading to homologous chromosome pairing (Fig. 4). It follows therefore, that KASH5 represents a functional homolog of S. pombe Kms1 and C. elegans ZYG-12. From a purely practical point of view our work has focused primarily on spermatogenesis. However, given that female KASH5-null mice are also infertile and have ovaries that are devoid of follicles, it is highly likely that KASH5 has a similar role in oogenesis.⁶⁰

Figure 4. A SUN1/KASH5 LINC complex that spans both the inner and outer nuclear membranes (INM and ONM) couples telomeres to cytoplasmic dynein, a microtubule motor protein. During the transition from leptotene to zygotene stages of meiotic prophase 1, the minus end directed motor activity of dynein will lead to clustering of telomeres (“bouquet” formation) at the nuclear pole adjacent to the centrosome.

Unanswered Questions

There are still a number of issues to be addressed concerning the function of meiotic LINC complexes in mammals. While it is clear that SUN1 defines NE attachment sites for telomeres the nature of the interaction and its regulation remain unknown. Whether the association between SUN1 and telomeres or sheltrin complex components is direct or whether there might be other meiosis-specific molecules involved that are similar in function to Bqt1–4 in S. pombe remains unknown.^43,50 Both zebrafish Fue and C. elegans ZYG-12 have essential roles in pronuclear migration in early embryos.^24,42,61 This activity is linked to the recruitment of dynein to the NE. This obviously raises the possibility that KASH5 might have a similar role following fertilization in mammals. Further studies using conditional mouse strains are being pursued to test this hypothesis as well as to explore the role of KASH5 in female germ cell development.

Horn HF, Kim DI, Wright GD, Wong ES, Stewart CL, Burke B, Roux KJ. A mammalian KASH domain protein coupling meiotic chromosomes to the cytoskeleton. J Cell Biol. 2013;202:1023–39. doi: 10.1083/jcb.201304004.

10.4161/nucl.27819