Nucleophagy at a glance

The cell nucleus is a cellular organelle whose raison d'etre is the maintenance and expression of the genome. Maintenance of nuclear structure and organization is essential for the vitality of most cell types (Mijaljica et al., 2010). Therefore, just as for other cellular organelles, we would anticipate ‘surveillance’ by processes that act to minimize nuclear damage by removal of damaged or non-functional components. Studies in both yeast and mammalian cells support the contention that selective digestion of portions of the nucleus can occur in order to maintain nuclear integrity (Roberts et al., 2003; Park et al., 2009; Mijaljica et al., 2012). By contrast, in other systems, such as the ciliated protozoan Tetrahymena thermophila and some filamentous fungi, the evidence suggests that selective digestion of whole nuclei occurs as part of a program of cellular development or mechanism of infection (Akematsu et al., 2010; Shoji et al., 2010; He et al., 2012). Autophagy is a homeostatic intracellular degradation system widely conserved in eukaryotes that is known to operate in a selective manner to remove organelles, such as mitochondria or peroxisomes, and which, as will be discussed below, can accomplish the degradation of nucleus-derived cargo or whole nuclei. Autophagy-related genes (the Atgs), which encode proteins comprising the core autophagic machinery, regulate all types of autophagy (Geng et al., 2013).

Nucleophagy is the selective removal of nuclear material from a cell by autophagy; it can occur either as a selective mode of macroautophagy or microautophagy, being referred to as macronucleophagy or micronucleophagy, respectively. In macronucleophagy, autophagosomes sequester the nucleus-derived cargo and subsequently fuse with the vacuole or lysosomes, which results in the degradation of their contents (Park et al., 2009; Liu and Yao, 2012). By contrast, micronucleophagy involves the direct engulfment of the nucleus-derived cargo by invagination, protrusion and/or septation of the vacuolar or lysosomal limiting membrane (Kvam and Goldfarb, 2007; Krick et al., 2008; Krick et al., 2009).

A number of divergent modes of nucleophagy exist and they can be differentiated principally by the mechanisms that mediate the delivery of the nucleus-derived cargo to vacuole or lysosomes, as well as by the nature of the cargo. Below, we discuss those nucleophagy processes for which some mechanistic information is available.

Piecemeal microautophagy of the nucleus (PMN) progresses in five distinct steps (see Poster) in order to remove and degrade small blebs of the Saccharomyces cerevisiae nucleus. It is initiated at nucleus–vacuole (NV) junctions, highly localized membrane–membrane contact zones at which membrane reorganization occurs (Kvam and Goldfarb, 2006; Krick et al., 2009; Dawaliby and Mayer, 2010). NV junctions are formed by interactions between Vac8, located in the vacuolar membrane, and Nvj1, in the nuclear envelope (step 1). Conditions that normally induce autophagy (e.g. a short-duration nitrogen starvation) result in the invagination of membranes that are associated with NV junctions into the vacuolar lumen. These invaginations consist of three membranes (the surrounding vacuolar membrane, which will eventually encompass the nuclear bleb that is comprised of both nuclear outer and inner membranes) and a portion of underlying nucleoplasm (step 2). Sequestration of the nuclear cargo occurs by fission of both nuclear membranes and fusion of the vacuolar membrane extensions (step 3). With the fusion of the vacuolar membrane extensions, the triple-membrane PMN vesicle is released into the vacuolar lumen (step 4), where it is eventually degraded by the resident vacuolar hydrolases (step 5) (Roberts et al., 2003; Kvam and Goldfarb, 2007; Krick et al., 2009). PMN is classified as microautophagy because the delivery of the nuclear bleb to the vacuole occurs by direct invagination and scission of the vacuole membrane, without involvement of autophagosome structures. Notably, components of the homotypic vacuole fusion machinery that are required for macroautophagy processes are not needed, emphasizing that PMN is a microautophagy process (Krick et al., 2008; Krick et al., 2009; Millen et al., 2009). As yeast has a closed mitosis (i.e. the replicated chromosomes divide within an intact nucleus), PMN provides a mechanism by which potential nuclear cargo destined for vacuolar degradation that would otherwise be enclosed – and rendered inaccessible – by the nuclear envelope can be accessed. The cargo that is sequestered by PMN includes non-essential nuclear components, such as the granular nucleolus that is enriched in pre-ribosomes, as well as the surrounding portion of nuclear membranes (Roberts et al., 2003). Nucleus-derived components, such as bulk chromatin, nuclear pore complexes and spindle pole bodies are excluded from PMN vesicles (Roberts et al., 2003; Krick et al., 2009; Millen et al., 2009).

Efficient PMN requires the core Atg genes and a set of subtype-specific Atg genes. The dependence of PMN on Atg genes seems to be restricted to the final step where the vacuolar membrane that surrounds the pinched-off piece of nucleus fuses to ‘close’ the outermost membrane (vacuolar-derived) of the PMN vesicle. In cells deleted for individual Atg genes, NV junctions and blebs still form (Krick et al., 2008; Krick et al., 2009).

Late nucleophagy (see Poster) of bulk nucleoplasm in S. cerevisiae can be detected only after prolonged periods (20–24 hours) of nitrogen starvation (Mijaljica et al., 2010; Mijaljica et al., 2012). In addition to this clear temporal difference between PMN and late nucleophagy, these two processes are also spatially separated (Mijaljica et al., 2012). Late nucleophagy is also classified as a microautophagy process because delivery of the nuclear cargo occurs without involvement of autophagosome structures. In contrast to PMN, late nucleophagy can occur in the absence of Nvj1 or Vac8, and, therefore, the formation of NV junctions. Although some components of the core Atg machinery are required for late nucleophagy, the macroautophagy-specific Atg11 is not. Intriguingly, late nucleophagy also does not require the components of the Vps34-containing PtdIns(3)P-kinase complex I (Mijaljica et al., 2012), although the mechanistic significance of this is yet to be established. Interestingly, inhibition of late nucleophagy in some autophagy-deficient mutants is accompanied by morphological alterations of the nucleus (Mijaljica et al., 2012). However, it is not clear whether these alterations are a consequence of misregulated late nucleophagy, or whether the conditions of cell growth, such as prolonged nitrogen starvation under which this phenotype is observed, leads to the depletion or mislocalization of one or more key factors (i.e. proteins or lipid) that are important for maintaining nuclear membrane morphology. At present there is no evidence that different types of cargo are delivered to the vacuole during late nucleophagy. It is possible that PMN and late nucleophagy facilitate the sequestration of different nuclear cargoes, but this remains to be established.

Diploid cells that are grown on a non-fermentable carbon source will normally enter meiosis in response to nitrogen starvation to produce four haploid nuclei, which are then packaged into spores. Under conditions of suboptimal carbon supply, cells produce only two spores and the two nuclei that do not form spores remain uncellularized in a common cytoplasm. Following spore formation, the mother cell vacuole, which is not partitioned into the daughter spores after meiosis, undergoes the final steps of disintegration. The release of the lytic contents of the vacuole into the cytoplasm leads to the destruction of the two uncellularized nuclei (see Poster) (Eastwood et al., 2012; Eastwood et al., 2013). This type of lytic dissolution of cellular contents including nuclei does not appear to be a unique feature of yeast programmed nuclear destruction (PND), as it is reminiscent of ‘mega-autophagy’, the rupture of the central vacuole in plant cells as part of programmed cell death (van Doorn and Woltering, 2005; van Doorn and Woltering, 2010). That mega-autophagy is not the conventional form of macroautophagy is emphasized by observations that show typical autophagosome structures are not formed to sequester the uncellularized nuclei (Eastwood et al., 2012; Eastwood et al., 2013).

In the filamentous fungus Aspergillus oryzae, autophagy mediates the degradation of basal cell components, such as entire nuclei, which appears to facilitate hyphal tip growth through the recycling of nutrients that are generated by nuclear degradation (Shoji et al., 2010). As the fungal cells are multinucleate, the loss of some nuclei can occur without detrimental effects. This process has been defined as macroautophagy on the basis of experiments that used EGFP-tagged Atg8 to follow autophagosomal structures (Shoji et al., 2010). These authors show that crescent-like autophagosomal precursors that are in close proximity to the targeted nuclei mature into large ring-like autophagosomes (1–2 µm in diameter) that then sequester whole nuclei. Subsequently, nuclear material is dispersed throughout neighboring vacuoles, suggesting that autophagosomes fuse with vacuoles for macroautophagy-mediated degradation of whole nuclei (see Poster). Because virtually all of the large autophagosomes contain at least one nucleus, they are likely to be specifically dedicated to degradation of nuclei (Shoji et al., 2010).

Macroautophagy-mediated degradation of the nucleus also occurs in the rice blast fungus, Magnaporthe oryzae. To be able to infect rice plants, M. oryzae forms a specialized structure, the appressorium, which generates immense intracellular turgor pressure that results in the direct rupture of the plant cuticle and, thereby, facilitates fungal entry and colonization of plant tissue (Liu et al., 2012). Appressorium development involves a number of steps, one of which is nuclear degeneration in the spore from which the appressorium develops (see Poster) (Veneault-Fourrey et al., 2006; He et al., 2012). Although, M. oryzae possesses homologs of Vac8 and Tsc13, components of a putative PMN pathway, an Nvj1 homolog has not been identified and there is no evidence for the formation of NV junctions and for PMN activity. Moreover, both M. oryzae Vac8 and Tsc13 are dispensable for nuclear breakdown during plant infection (He et al., 2012). In contrast to PMN, nuclear degradation in M. oryzae is dependent on core autophagy genes, such as M. oryzae ATG1 or ATG4 (Kershaw and Talbot, 2009). In contrast to large ring-like autophagosomes that mediate the degradation of A. oryzae nuclei, only smaller, more punctate autophagosomes surround the target nucleus during appressorium formation in M. oryzae (He et al., 2012). Moreover, the nucleus does not appear to be fully degraded, suggesting that nuclear breakdown might be performed by distinct macroautophagy-dependent processes in filamentous fungi (He et al., 2012).

T. thermophila contains two functionally distinct nuclei in one cell. The polyploid macronucleus divides by amitosis (i.e. through separation); it is a site of mRNA synthesis and governs the somatic phenotype of the cell. The diploid micronucleus is transcriptionally inactive and serves as the germ nucleus. During sexual reproduction, a complex process of nuclear differentiation and degradation takes place. The degradation of the ‘old’ macronucleus by programmed nuclear death (PNuD) shows some characteristics of mammalian apoptosis (Davis et al., 1992; Lu and Wolfe, 2001; Liu and Yao, 2012) but autophagy is also crucial (Lu and Wolfe, 2001; Akematsu et al., 2010; Liu and Yao, 2012). Two different models for the role of autophagy have been proposed (models 1 and 2, see Poster). In one model, autophagosomes do not appear to form (Akematsu et al., 2010); during degradation of the old macronucleus, the nuclear envelope undergoes significant changes and exhibits characteristics of autophagosome membranes and also exposes certain sugars and phosphatidylserine on the outer membrane surface (Akematsu et al., 2010). The authors interpreted these changes as destruction signals that facilitate the subsequent fusion of the target nucleus with lysosomes (see Poster). By contrast, the second model suggests a coordinated formation of autophagosomes that surround the target nucleus (Lu and Wolfe, 2001; Liu and Yao, 2012). In the initial study (Lu and Wolfe, 2001), the vesicles were not well characterized, but in the later study (Liu and Yao, 2012), it was observed that two orthologs of Atg8 were involved, supporting the involvement of autophagosomes (see Poster). Additional support for the role of these Atg8 proteins as bona fide components of autophagosomes comes from atg8-knockout mutant cells, which show a pronounced delay in nuclear degradation (Liu and Yao, 2012). However, further studies are required to fully understand the precise autophagic mechanism involved.

A role for autophagy in the degradation of the entire nucleus or nuclear components (see Poster), such as chromatin in mammalian cells has been suggested for many years. However, only little or no details were available with regard to the particular mode of nucleophagy involved (Mijaljica et al., 2007; Mijaljica et al., 2010).

The first definitive report of nucleophagy in mammalian cells was based on electron microscopy (EM) images that showed the presence of perinuclear vesicular structures in cells from nuclear envelopathies and emerinopathies that are caused by mutations in the genes encoding A-type lamins (LMNA) and emerin, respectively (Park et al., 2009). Consequently, mouse embryonic fibroblasts (MEFs) bearing an Lmna mutation and expressing GFP-labeled LC3 exhibit round, LC3-positive structures in very close proximity to the nucleus. These structures were identified as autophagosomes based on their incorporation of GFP-labeled LC3, the mammalian homolog of Atg8 (a marker of autophagosomes), and their colocalization with the core autophagy proteins ATG5, ATG16L and ATG9 (Park et al., 2009). In addition, the lysosomal marker protein LAMP2 was also found to colocalize to some of these structures, suggesting the fusion of autophagosomes with lysosomes to form autophagolysosomes. The colocalization of GFP–LC3 with histones H1 and γH2AX suggests the presence of chromatin within some autophagosomes. Importantly, a decreased intensity of staining by DAPI is consistent with chromatin degradation in these compartments.

Micronuclei are mainly displaced chromosomes or chromosome fragments that are enclosed by nuclear membrane (Fenech et al., 2011), and their formation can be enhanced following exposure to genotoxic compounds (Erenpreisa et al., 2012). Recently, it has been shown that autophagy is also involved in the elimination of micronuclei (Rello-Varona et al., 2012). Here, micronuclei were found to contain chromatin, and a few of these (2–5%) colocalize (adjacent to or enveloped by) to LC3-positive vesicles. Colocalization with LC3 is lost after knockdown of the core autophagy genes ATG5 or ATG7, confirming that the chromatin and LC3-positive structures are indeed autophagosomes. The autophagic LC3-positive micronuclei also contain the autophagy adaptor p62 (SQSTM1), which does not localize to non-autophagic LC3-negative micronuclei. In addition, LC3-positive micronuclei exhibit signs of nuclear envelope degradation, contain foci of γH2AX-positive DNA, a marker of DNA damage, and stain less intensively for chromatin markers. By contrast, LC3-negative micronuclei are surrounded by an intact membrane and only rarely show signs of DNA damage (Rello-Varona et al., 2012). Presumably, autophagic degradation of micronuclei might contribute to genome stabilization and protect against chromosomal instability (Boya and Codogno, 2012).

A subsequent study addressed the low occurrence of LC3-positive micronuclei and suggested that it derives from the DNA repair processes that might occur within micronuclei (Erenpreisa et al., 2012). The authors suggest that LC3-positive vesicles represent a failure of DNA repair attempts in micronuclei, so they are subsequently removed by autophagy (Erenpreisa et al., 2012). This is consistent with the reduced presence of the DNA damage marker γH2AX and chromatin staining with Hoechst 33342 in these vesicles. They also suggest that some micronuclei maintain a connection with the nucleus by means of an extension of the nuclear envelope membranes that include one or more layers of heterochromatin of a diameter of 30 nm (so-called envelope-limited chromatin sheets, ELCS). This connection allows the micronuclei to bridge to new nuclear envelope sites where their DNA content can be repaired and reincorporated into the nucleus, and might have been the LC3-negative micronuclei (Rello-Varona et al., 2012). This model further suggests that, if repair and reintegration fails, these micronuclei would ‘bud’ from the nucleus (i.e. lose their connection) and be degraded by autophagy (Erenpreisa et al., 2012). Further studies of the physical connections between micronuclei and the nucleus are clearly needed to understand any ongoing repair events.

The links of nucleophagy to human disease are only just emerging (Box 1). Although the formation of micronuclei has been observed in many disease situations, such as Bloom syndrome (Yankiwski et al., 2000) and cancer (Luzhna et al., 2013), evidence has only emerged recently that micronuclei can be degraded by autophagy (Rello-Varona et al., 2012; Erenpreisa et al., 2012) and the full significance of this process in the context of disease pathology and progression remains to be investigated.

As discussed here, nucleus-derived cargo can be subjected to degradation by autophagic processes we collectively refer to as nucleophagy. However, our knowledge of nucleophagy is incomplete and a number of unanswered questions remain (Box 2), particularly with regard to the divergent modes of nucleophagy in lower eukaryotes.

It is unlikely that a specialized process of nucleophagy, such as PMN is required in mammalian cells, because, unlike yeast cells, which undergo closed mitosis, either macroautophagy or microautophagy could be utilized during S phase when the nuclear membrane has been disassembled and nuclear material is accessible (Mijaljica et al., 2010). A specialized mode of nucleophagy has been suggested to be more important for long-lived amitotic cells, such as neurons (Kvam and Goldfarb, 2007). In these cells, the potential nuclear material that is destined for destruction is sequestered for macro- or micro-autophagy by the intact nuclear membrane, which is not dismantled for the purpose of chromosome replication and cell division. In this context, the formation of micronuclei might be a means to make nuclear material available for degradation by autophagy. To date, however, there are only very few reports for the presence of micronuclei in amitotic cells that have not been subjected to radiation or genotoxins, and they offer no clues with regard to the likely importance of micronuclei formation and degradation in physiological settings.

Despite the number of unanswered questions, the future holds the promise of significant progress in our understanding of the mechanisms and physiological significance of nucleophagy. Addressing and answering these questions in mammalian cells will contribute to a better understanding of how dysregulation of nucleophagy contributes to disease.