Class E Vps (vacuolar protein sorting) proteins are components of the ESCRTs (endosomal sorting complexes required for transport) which are required for protein sorting at the early endosome. Most of these genes have been identified and genetically characterized in yeast. Recent genetic studies in Drosophila have revealed the phenotypic consequences of loss of vps function in multicellular organisms. In the present paper, we review these studies and discuss a mechanism which may explain how loss of the human Tsg101 (tumour susceptibility gene 101), a vps23 orthologue, causes tumours.
Protein turnover of cell-surface proteins including signalling receptors, both ligand-occupied and unoccupied, occurs through endocytosis . After endocytosis, these proteins are present at the early endosome. Activated signalling receptors are able to signal from the early endosome, because their intracellular domains are still exposed to the cytosol. In fact, the endosomal location is the preferred mode of signalling for some receptors [1–5]. At the endosome, protein sorting occurs with two choices. Proteins can either be recycled back to the cell surface, or they undergo a second form of internalization at the limiting membrane of the early endosome which then matures into MVBs (multivesicular bodies) [6–13]. In MVBs, activated receptors are completely detached from the cytosol and stop signalling. The MVB matures into the late endosome which fuses with lysosomes for proteolytic degradation.
MVB formation requires four ESCRTs (endosomal sorting complexes required for transport): ESCRT-0, ESCRT-I, ESCRT-II and ESCRT-III. At least 15 class E vps (vacuolar protein sorting) genes have been identified in yeast which are the structural components of the ESCRTs . Important to this review are vps23, a component of ESCRT-I, and vps25, a component of ESCRT-II. vps genes are highly conserved from yeast to mammals, and biochemical studies have shown that mammalian Vps proteins also function in MVB formation [6,7]. However, mammalian Vps proteins also have functions besides MVB formation [14–22].
In yeast, loss of class E vps function leads to enlarged early endosomes and accumulation of ubiquitinated proteins . However, until recently, the phenotypic consequences of loss of vps genes in multicellular organisms have been unclear. Nevertheless, mutations in Tsg101 (tumour susceptibility gene 101), the human orthologue of vps23 and component of ESCRT-I, have been linked to cervical, breast, prostate and gastrointestinal cancers [14,23–26]. The reason for this oncogenic behaviour is unknown, and knockout studies in mice have been inconclusive . However, recent genetic studies in Drosophila have revealed a potential oncogenic mechanism caused by loss of vps23 and vps25 function [27–30]. Here, we review these Drosophila studies by describing the phenotypic analysis of mutants of the ESCRT-I component vps23 and the ESCRT-II component vps25.
Most findings discussed in this review were obtained in a so-called clonal analysis in Drosophila. In principle, a clonal analysis corresponds to a conditional knockout in which homozygous mutant tissue (referred to as mutant clones) is induced in heterozygous animals. These animals are also referred to as genetic mosaics because they contain tissue of different genetic origin. Usually, this analysis is performed in imaginal discs which are the larval precursors of the adult appendages such as eyes, legs and wings.
Because vps mutants in yeast are characterized by enlarged endosomes , mutants of Drosophila vps23 and vps25 were analysed for endosomal defects. Hrs (hepatocyte growth factor-regulated tyrosine kinase substrate) (Vps27), a component of ESCRT-0, has been used as a convenient marker for early endosomes . Mutant clones of vps23 and vps25 contain enlarged Hrs-positive particles [27–30], thus representing abnormal early endosomes. Therefore, similarly to yeast, loss of vps23 and vps25 in Drosophila leads to endosomal defects.
Accumulation of ubiquitinated proteins
Ligand binding triggers activation of signalling receptors, but also ubiquitination which provides the signal for endocytosis and for protein sorting into the MVBs by ESCRTs [6,32–34]. The FK1 and FK2 antibodies specifically recognize conjugated ubiquitin, but not unconjugated ubiquitin [35–37]. Using the FK antibodies, it was shown that the enlarged Hrs-positive endosomes accumulate ubiquitin-conjugated proteins in vps25 mutant clones . Because de-ubiquitination of target proteins occurs in the ESCRT-III complex before MVB formation [6,9], the accumulation of ubiquitin-conjugated proteins suggests that MVBs barely form in these mutants, leading to enlargement of endosomes.
Accumulation of Notch protein
Accumulation of ubiquitin-conjugated proteins at endosomes implies that signalling receptors accumulate at the mutant endosomes. This was indeed shown for the signalling receptors Notch, Delta, EGFR (epidermal growth factor receptor), Patched, Smoothened and Thickveins in hrs, vps23 and vps25 mutants [27–31,39]. Not all of these receptors are signalling at the mutant endosome, presumably because there are other mechanisms to inactivate them, such as, for example, dephosphorylation. However, a notable exception is the Notch receptor. Not only does Notch signal from the endosome, but it also appears that Notch becomes inappropriately activated at vps23 and vps25 mutant endosomes, perhaps even in a ligand-independent manner .
JAK (Janus kinase)/STAT (signal transducer and activator of transcription) signalling
During Drosophila development, one target of Notch signalling is the JAK/STAT pathway [40–43]. Consistently, the accumulation of Notch protein correlates with increased JAK/STAT activity in vps23 and vps25 mutant clones [27–30]. Interestingly, increased JAK/STAT activity is observed not only in vps mutant clones, but also outside the clones, i.e. non-autonomously . This non-autonomous effect is caused by Notch-dependent secretion of Unpaired, an IL (interleukin)-6-like cytokine in Drosophila, which binds to and activates Domeless, the receptor of the JAK/STAT pathway .
Non-autonomous hyperplastic overgrowth
One of the biggest surprises in ESCRT research came with the analysis of mutant animals that were mosaic for vps23 and vps25 in Drosophila. Mosaic refers to a genetic condition in which the animal is composed of homozygous mutant, heterozygous and wild-type tissues (see also the Technical note above). The eyes and heads of flies mosaic for vps23 and vps25 are larger than those of wild-type flies [27–30]. Given the potential oncogenic nature of Tsg101 in humans [14,23–26], this overgrowth is maybe not so surprising. However, what was surprising is that the overgrowth is not caused by vps23 and vps25 mutant tissue itself. Instead, the tissue abutting the vps23 and vps25 clones is overgrown [27–30]. Therefore overgrowth occurs in a non-autonomous manner. The vps23 and vps25 mutant clones appear to emit a signal which causes this overgrowth phenotype. At the time, such a non-autonomous overgrowth phenotype was unprecedented. However, two other reports have since shown that non-autonomous overgrowth also occurs in Uba1 mutants in Drosophila which encodes the E1 ubiquitin-activating enzyme in the ubiquitination pathway [37,44]. For vps23, vps25 and Uba1, the signal for non-autonomous overgrowth is provided by Notch-dependent secretion of Unpaired which stimulates JAK/STAT signalling non-autonomously in neighbouring cells (see above) [27–30,37,42].
The nature of this overgrowth phenotype is hyperplastic . Hyperplastic tissue continues to proliferate, but leaves apical–basal polarity of the cells intact and maintains a monolayer epithelium. Eventually, these cells become post-mitotic and begin to differentiate. The hyperplastic overgrowth phenotype is dependent on Notch and JAK/STAT signalling because reducing the activity of these pathways suppresses the vps23 and vps25 overgrowth phenotype [27,29,30].
Both vps23 and vps25 mutant clones are characterized by a strong apoptotic phenotype. Most mutant clones which are induced in early larval stages die during late larval stages. This corresponds to approx. 3 days of developmental time which is a relatively long time given that normal developmental cell death occurs within hours [45,46]. Thus apoptosis in vps23 and vps25 clones is rather slow. This slow apoptosis may be important for the non-autonomous overgrowth (see above) and apoptotic resistance (described below).
The cause for apoptosis of vps25 mutant cells is unknown. They up-regulate the apoptosis-inducing gene hid (head involution defective), and the stress-induced kinase JNK (c-Jun N-terminal kinase) is also active . It is unclear why these cell death pathways are activated. It is possible that the endosomal defects poison the cell so that it eventually dies. However, mutants which cause developmental defects often induce cell death . Thus it is possible that vps23 and vps25 mutant cells also die by such a ‘death by frustration’ model.
Non-autonomous apoptotic resistance
Despite the fact that vps23 and vps25 mutants are characterized by strong apoptosis, vps25 mosaics are actually resistant to apoptotic stimuli in a non-autonomous manner . Tissue adjacent to vps25 mutant clones displays enhanced resistance to apoptotic stimuli. Thus, although vps25 mutant clones themselves die, they increase the apoptotic resistance of neighbouring cells . Such an altruistic behaviour has not been described before. We showed that non-autonomous resistance against apoptosis is caused by non-autonomous increase of Diap1, the Drosophila inhibitor of apoptosis 1 . Similarly to non-autonomous proliferation, it is clear that such a phenotype must be caused by secretion of a signal from vps25 mutant cells that increases the apoptotic resistance of neighbouring cells. However, this signal is not Unpaired which accounts for non-autonomous proliferation. Overexpression of Unpaired does not or only weakly increase the apoptotic resistance of cells . Thus the hunt is on to identify the factor which controls survival in neighbouring cells.
Autonomous neoplastic overgrowth
So far, we have described phenotypes which are caused by vps23 and vps25 mutant clones in genetic mosaics. In these cases, only a small fraction of the tissue is homozygous mutant. The majority of the tissue is wild-type. Under these conditions, non-autonomous hyperplastic overgrowth and apoptotic resistance was observed (see above). However, when the entire tissue (such as the entire eye or wing in Drosophila) is mutant for a vps gene, the phenotype is different. Under these conditions, severe neoplastic overgrowth is observed. When tissue transforms to adopt neoplastic character, it loses apical–basal polarity, loses its monolayer epithelial character, fails to stop proliferating, never differentiates and becomes invasive, i.e. metastatic. Importantly also, this phenotype is autonomous, i.e. it affects the mutant vps cells. When the tissue is entirely mutant for vps23 and vps25, it displays strong neoplastic overgrowth [5,27–29,38], thus vps23 and vps25 qualify as neoplastic tumour suppressors. Neoplastic overgrowth is associated with increased Notch activity, but it is uncertain whether Notch signalling causes this phenotype, because other endocytic mutants, such as scribble, avalanche and rab5, which act upstream of the ESCRTs do not accumulate Notch activity, but yet cause strong neoplastic overgrowth .
Although we have discussed above that vps25 mutant clones in a mosaic animal induce non-autonomous hyperplastic growth, it is possible to induce neoplastic growth in mutant clones in a genetic mosaic. The reason they do not induce neoplastic growth in mutant clones is that they die by apoptosis (see above). However, if cell death is blocked by overexpression of the caspase inhibitors p35 or Diap1, or the JNK inhibitor Puckered , vps25 mutant clones survive and massively overgrow. Apoptosis-inhibited vps25 mutant cells are extremely aggressive. Upon transplantation into a wild-type fruitfly, they are invasive and display metastatic behaviour, i.e. they are neoplastic .
Vps25 as neoplastic tumour suppressor: a model for human Tsg101-induced cancer
Human ESCRT components, most notably Tsg101 (Vps23), have been implicated in tumour suppression. NIH 3T3 cells, depleted of Tsg101 by an antisense approach, formed colonies on soft agar and produced metastatic tumours in nude mice . However, the conditional Tsg101-knockout in mouse primary glands did not cause the formation of tumour cells over a period of 2 years, making the role of Tsg101 as tumour suppressor controversial . However, Tsg101 mutant cells are very sensitive to apoptotic death , implying that they die before they become harmful to the organism.
The phenotypic characterization of vps25 mutants in Drosophila provides an explanation for the failure to confirm Tsg101 as a tumour suppressor. vps25 mutant clones need to survive over extended periods of time in order to sustain Notch-dependent growth. Even though they induce non-autonomous proliferation, after they have died, Notch signalling is turned off and proliferation stops. Furthermore, the size of the adult eye of vps25 mosaics is only slightly enlarged when compared with normal flies, and does not match the strong overgrowth phenotype during larval stages. Thus, as long as vps25 clones are not resistant to their own apoptotic death, tissue repair during pupal stages may partially regress the size of the mosaic animal back to almost normal. Instead, it appears that inhibition of cell death is the triggering event for a tumorous phenotype of vps25 clones. vps25 clones expressing Diap1, p35 or Puckered are not apoptotic and make up a large fraction of the mosaic tissue, which can be five times larger than normal.
Tumorigenesis requires multiple genetic alterations that transform normal cells progressively into malignant cancer cells . Thus additional genetic hits may be necessary to inhibit apoptosis of Tsg101 mutant cells, which may then be able to induce a similar tumour phenotype to that observed for vps25. Thus, although the tumour-suppressor function of Tsg101 was not confirmed in a mouse model, the Drosophila findings still make it likely that Tsg101 and other mammalian ESCRT members have tumour-suppressor properties.
ESCRTs: from Cell Biology to Pathogenesis: Biochemical Society Focused Meeting held at Robinson College, Cambridge, U.K., 26–28 August 2008. Organized and Edited by Katherine Bowers (University College London, U.K.), Juan Martin-Serrano (King's College London, U.K.) and Paul Whitley (Bath, U.K.).
Drosophila inhibitor of apoptosis 1
endosomal sorting complex required for transport
hepatocyte growth factor-regulated tyrosine kinase substrate
c-Jun N-terminal kinase
signal transducer and activator of transcription
tumour susceptibility gene 101
vacuolar protein sorting
We thank Zhihong Chen and Clare Bolduc for excellent technical assistance.
Work in our laboratory was supported by the National Institutes of Health [grant numbers GM068016, GM074977 and GM081543] and The Robert A. Welch Foundation [grant number G-1496].