APOPTIN INDUCES TUMOR-SPECIFIC APOPTOSIS AS A GLOBULAR MULTIMER

The chicken anemia virus-derived Apoptin protein induces tumor-specific apoptosis. Here, we show that recombinant Apoptin protein spontaneously forms non-covalent globular aggregates comprising 30 to 40 subunits in vitro. This multimerization is robust and virtually irreversible, and the globular aggregates are also stable in cell extracts, suggesting that they remain intact within the cell. Furthermore, studies of Apoptin expressed in living cells confirm that Apoptin indeed exists in large complexes in vivo . We map the structural motifs responsible for multimerization in vitro and aggregation in vivo to the N-terminal half of the protein. Moreover, we show that covalently fixing the Apoptin monomers within the recombinant protein multimer by internal cross-linking does not affect the biological activity of Apoptin, as these fixed aggregates exhibit similar tumor-specific localization and apoptosis-inducing properties as non-cross-linked Apoptin. Taken together, our results imply that recombinant Apoptin protein is a multimer when inducing apoptosis, and we propose that this multimeric state is an essential feature of its ability to do so. Finally, we determine that Apoptin adopts little, if any, regular secondary structure within the aggregates. This surprising result would classify Apoptin as the first protein for which, rather than the formation of a well-defined tertiary and quaternary structure, semi-random aggregation is sufficient for activity. apoptosis the antibody ( α MBP). Nuclear morphology by DAPI staining. Detection with anti-Apoptin α VP3-C similar results. [ B ] Western blot analysis of supernatant (S) and pellets (P) of lysed Saos-2 cells, expressing full-length MBP-Apoptin and MBP-Apoptin(80-121). Cell extracts were centrifuged at 10,000 g, yielding S10 and P10 , and subsequently at 30,000 g , producing S30 and P30. ‘Pos’ is purified, recombinant MBP-Apoptin. Western blots were stained with α VP3-C. [ C ] Western blot analysis of recombinant (Rec.) MBP-Apoptin centrifuged at a concentration of 20 µ g/ml in lysis buffer alone.


INTRODUCTION
Apoptin is a protein from chicken anemia virus (CAV 1 ) that induces apoptosis in transformed chicken cells (1). It has 121 amino acids and no known functional or sequence homologues.
When the gene encoding Apoptin is transduced into cultured cells, the expressed Apoptin protein also induces apoptosis in a wide range of transformed human cell lines (2). However, non-3 transformed, normal human primary cell types are not killed. Transformed and non-transformed cells differ markedly in the subcellular localization of expressed Apoptin. In transformed cells, Apoptin migrates to the nucleus, whereas in non-transformed cells it is retained mainly within the cytoplasm. Although nuclear localization of Apoptin appears to be essential for apoptosis induction in transformed cells (3), the presence of Apoptin in the nucleus alone is not sufficient to induce apoptosis in normal cells 2 . Two clusters of basic residues (K82-R89, R111-R120) near the C-terminus of Apoptin have been implicated as a nuclear localization signal (NLS 1 ).
MBP-Apoptin fusion constructs were shown to be biologically active and retain their distinct behavior in normal and tumor cells upon microinjection 3 . Here, we addressed Apoptin's biophysical and structural properties and how these relate to its biological function, and found it to be biologically active as a higher-order multimer. However, no evidence was found for an established regular structure.
All clones were confirmed by automated fluorescent sequencing. Table 1 summarizes the protein constructs we generated.  10

Circular dichroism spectroscopy
Far-UV circular dichroism spectra were recorded on a Jasco J-715 spectrometer. Spectrometer were filtered and degassed directly prior to measurement.

Fluorescence measurements
Fluorescence emission and excitation spectra were recorded on a Perkin Elmer LS-50B. and adsorbed to a carbon-coated polioform layer grid. Both samples were incubated with concentrated streptavidin-gold conjugate (gold particle diameter: 5 nm) (KPL Inc.) and stained with 3% uranyl acetate. Electron microscopy was performed on a Philips TEM 410 transmission electron microscope.
Scanning force microscopy. 90 ng of purified MBP-Apoptin (in 10 µl 5 mM HEPES pH 7.9, 3 mM KCl, 5.5 mM MgCl 2 ) was incubated at 37 °C for 15 minutes and then deposited on a disc of freshly cleaved mica (Ted Pella Inc.). After 20 seconds, the mica was gently rinsed with HPLC water. Excess water was removed and the disc was dried with a steady flow of 0.22 µm filtered air. Images were acquired on a Nanoscope IIIa (Digital Instruments Inc.), operating in tapping mode in air with a type E scanner. Silicon tips were obtained from Digital Instruments.

Incubation of recombinant MBP-Apoptin in Saos-2 and VH10 lysates in low detergent buffer
Saos-2 cells, which are human tumor cells derived from osteosarcoma 1 , and VH10 cells, which are normal human fibroblasts 1 , were grown to around 50% confluency. The cells were Apoptin was added to 5% (w/w), which was estimated to be the average ratio of MBP-Apoptin to cytoplasmic protein in microinjected cells (7,8). Samples were incubated for 30 minutes at 30°C , in the presence of 1 mM ATP and 20 mM MgCl 2 , and for 2 and 24 hours at 4 °C, without additives. Following incubation, samples were fractionated on Superose 6 HR 10/30. Prior to fractionation, any precipitated material was pelleted by centrifugation at 29,000 g for 20 minutes, after which the pellets were washed with lysis buffer. All pellets and fractions were dot-blotted, using 10 µl per sample, as described above. Dot blots were probed with mAb 111.3.  Transient transfection. In a parallel experiment, Saos-2 cells were grown on coverslips in 9 cm plates and transfected as described above, fixed using fresh 1:1 methanol:acetone for 5 minutes and stained with a mouse monoclonal antibody that recognized MBP (clone R29.6, Abcam). An appropriate FITC-labeled secondary antibody was added and the cells were mounted in DAPI/DABCO/glycerol. Again, apoptotic cells were scored on the basis of their nuclear morphology. Detection of MBP-Apoptin with αVP3-C produced comparable results.

MBP-Apoptin
In E. coli, soluble MBP-Apoptin was expressed with a yield of up to 100 mg per litre of culture. After affinity chromatography on amylose resin and cation exchange chromatography, MBP-Apoptin migrated as a stable multimeric complex, with a molecular weight of 2.5 ± 0.3 MDa, on an analytical size exclusion chromatography column (Superose 6 HR 10/30) ( Figure 1A and Regardless of the precise nature of the two species, this result shows that the C-terminal domain of Apoptin on its own is unable to form the type of higher-order multimers such as MBP-Apoptin(1-121) does. Taken together, these results demonstrate that the N-terminal 69 residues of Apoptin contain sufficient structural elements to account for its multimerization behavior.

Apoptin has little if any ordered structure
Next, we examined whether recombinant Apoptin protein harbors any secondary or tertiary structure, using circular dichroism (CD 1 ) and fluorescence spectroscopy.

CD spectroscopy
The far-UV CD spectrum of refolded Apoptin-H 6 ( Figure 4A)  Apoptin between 210 and 230 nm was reminiscent of CD spectra reported for β-turn-orhairpin-containing peptides (14,15). However, the geometry of residues involved in β-turns and 21 -hairpins can vary considerably, making their net contribution to a CD spectrum poorly quantifiable. In general, the CD spectrum of refolded Apoptin-H 6 was very similar to that of proteins that are mostly unstructured (16)(17)(18)(19).
The CD spectrum of MBP-Apoptin confirmed these observations, being not significantly different from that of MBP on its own ( Figure 4B) (20,21). Firstly, this result shows that the Apoptin moiety did not perturb the folding of MBP. Secondly, it demonstrates that Apoptin did not contribute significantly to the secondary structure content of the fusion protein. This is consistent with the very low overall [θ] MRE of refolded Apoptin-H 6 .
We conclude that Apoptin displays very little regular secondary structure, but we cannot exclude that some residues adopt a β-strand or -turn conformation.

Intrinsic fluorescence of Tyr95
Even though Apoptin did not appear to adopt a well-defined fold, the Apoptin polypeptide might still display a certain degree of intra-or intermolecular order within the multimeric complex. Because Apoptin-H 6 does not contain any Trp residues, the intrinsic fluorescence of its single Tyr95 can act as a structural probe (22,23). Tyr95 is outside the multimerization domain defined by Apoptin's N-terminal residues. The excitation spectrum of Tyr95 in refolded Apoptin-H 6 was very similar to that of protonated L-Tyr(-OH), with excitation maxima at 280 and 275 nm, respectively ( Figure 5A). However, the fluorescence emission spectrum of Apoptin's Tyr95 at pH 6.5 closely resembled that of deprotonated, free tyrosine (L-Tyr(-O -)) at pH 12 ( Figure 5B). This apparent discrepancy can be explained by the effect that excitation has on the pK a of the Tyr side chain: the pK a of Tyr(-OH) is near 10 in the ground state, but decreases to approximately 4 upon excitation (24,25). Because the emission spectrum of 22 refolded Apoptin-H 6 was almost entirely devoid of Tyr(-OH) fluorescence, an efficient proton transfer to a nearby proton-accepting group (Glu, Asp or His) must have occurred upon excitation of Tyr95. This result indicates that Tyr95 was hydrogen bonded. The fluorescence yield of Tyr95 was increased by a factor of 4 to 5 relative to fully deprotonated and solventexposed L-Tyr(-O -) (at pH 12), which indicates that Tyr95 was at least partially protected from the solvent. The presence of an internal hydrogen-bond involving the side chain of Tyr95 suggests that at least one region of the Apoptin polypeptide was able to adopt a more ordered conformation.

Apoptin did not bind Zn 2+ despite a putative Zn 2+ -binding motif
Apoptin contains a potential metal-binding motif comprised of one His (H29) and three Cys

MBP-Apoptin expressed in tumor cells exists in an aggregated state.
Having In parallel, we determined the aggregation state of both protein constructs in cell extracts by centrifugation and found that the bulk of the extracted full-length MBP-Apoptin could be pelleted at 10,000 g, indicating that it is or is part of a very dense aggregate ( Figure 6B).
Moreover, most of the MBP-Apoptin that remained in the supernatant could be pelleted at 30,000 g. Under the same conditions, MBP-Apoptin(80-121) was fully soluble ( Figure 6A).
Moreover, we verified that recombinant MBP-Apoptin alone did not precipitate upon dilution in lysis buffer ( Figure 6C). We assume that the trace amount of full-length MBP-Apoptin remaining in the supernatant after centrifugation at 30,000 g corresponds to newly translated polypeptides that were not yet fully aggregated or had been absorbed by detergent micelles.
Taken together with the immunofluorescence data, these results imply that aggregation of Apoptin occurs in vivo and that the determinants responsible for aggregation are located in the N-terminal part of the protein.

The recombinant MBP-Apoptin protein complex is active as a multimeric species.
In a separate paper, we show that microinjected recombinant MBP-Apoptin protein induces apoptosis in tumor Saos-2 cells, but not in normal VH10 cells 3  Next, in order to ensure that MBP-Apoptin aggregates did not dissolve upon microinjection into living cells, we cross-linked the MBP-Apoptin aggregates by brief incubation with 0.05% glutaraldehyde. The majority of these cross-links are expected to be between the MBP moieties if the multimeric fusion protein, as they contain most of the Lys residues. Cross-linked complexes did not contain detectable amounts of smaller oligomers or monomers ( Figure 7A).  Figure   7B and 7C). However, the efficiency of nuclear import of cross-linked MBP-Apoptin appeared to be decreased in comparison to non-cross-linked MBP-Apoptin 3 (data not shown), which may indicate that some of its NLSs are obscured as a result of glutaraldehyde treatment. In VH10 cells, cross-linked MBP-Apoptin remained in the cytoplasm and did not induce apoptosis ( Figure   7B and 7D). Clearly, covalent cross-linking did not have any significant effect on the activity of microinjected MBP-Apoptin, implying that in vivo dissociation of the recombinant Apoptin protein multimers is not required for tumor-specific apoptosis induction.

Structure of the Apoptin complex
We demonstrated with a range of biophysical techniques that recombinant Apoptin protein which implies that there is at least some internal structure. Second, the CD spectrum of refolded Apoptin-H 6 was clearly different from that of a true random coil polypeptide, indicating that its conformational freedom was restricted (11). If Apoptin multimers present an at least partially ordered surface, this characteristic would allow them to interact selectively with cellular factors.
Such an interaction could give rise to a particular biological effect, namely the induction of apoptosis in a tumor-specific manner. Furthermore, it could be that small domains of the Apoptin multimer become ordered once recognized by cellular factors. There are precedents for such behavior; for example, high mobility group proteins (HMGA 1 ) specifically recognize DNA, yet in the absence of DNA do not adopt a regular conformation (27).
According to secondary structure prediction, Apoptin might fold as an anti-parallel β-sheet between E32 and L46 with A38 and G39 oriented in a β-turn or -hairpin. The CD spectrum of The Leu/Ile clustering of the potential amphipathic β-hairpin is reminiscent of a Rev-or PKIα−like nuclear export signal (NES 1 ) (28). Although the NES sequences of PKIα and of p53 28 have been reported to adopt an amphipathic α-helical conformation (29,30), CD spectroscopy showed that refolded Apoptin-H 6 did not contain any α-helical regions. If indeed the isoleucines and leucines of the putative NES of Apoptin are part of its multimerization motif, these residues would, because of their hydrophobic nature, be largely buried within the multimer. Thus obscured, Apoptin's putative NES would be likely to have reduced activity.

Activity of the Apoptin complex
We Apoptin was contained in an aggregate particle that was denser than the recombinant Apoptin protein multimer. Therefore, it is likely that Apoptin expressed in vivo forms a larger aggregate or co-aggregates with other cellular proteins.  (35). An example where protein aggregation seems to represent 'gain-of-function' is the human milk protein α-lactalbumin, which forms partially unfolded oligomers that are imported into the nucleus of tumor and differentiating cells, whereas the monomeric form is not.
Nuclear import of oligomeric α-lactalbumin was accompanied by induction of apoptosis (36,37). It has been suggested that the presence of globular aggregates of misfolded protein in the cell may lead to cell death if these aggregates are efficient nucleation sites for fibrous amyloid formation (33). However, the globular aggregates of recombinant Apoptin did not form recognizable fibrous amyloid deposits when microinjected into live cells 3 , and such amyloid formation by Apoptin was also not observed in vitro in the current study. Therefore, the mechanism of apoptosis induction by Apoptin is clearly different. Moreover, apoptosis induction by recombinant Apoptin multimers is not a general cytotoxic effect, as Apoptin does not appear to elicit any harmful effects when introduced into several different normal primary human cells 3 .

31
Apoptin will exert its pro-apoptotic function only when a cell has entered the pathway that leads to a transformed state (38). In that respect, Apoptin clearly differs from proteins that give rise to aggregation-linked diseases (39).
Apoptin is a viral protein encoded by chicken anemia virus. The virus has a minimal ssDNA genome and Apoptin's gene fully overlaps with that of the VP2 protein, albeit with a shift in frame (40). One possible function of Apoptin in the replication cycle of CAV is to induce apoptosis of infected chicken thymocytes in order to release them from the host cell once the viral particles have matured. Transmission of CAV may be enhanced when the virus is encased in or associated with apoptotic bodies, analogous to adenoviral vectors (41). In such a process, the potential co-transmission of Apoptin globules together with infectious viral particles may also have a biological function. In this respect it is striking that Apoptin forms a globular particle roughly the size of a virus, so it may have an evolutionary relationship with a viral coat protein.
The requirement of a compact genome in CAV has impelled the Apoptin gene to fully overlap with that of VP2 in the viral genome. The obvious price of this compactness is that the sequences of both genes are more strongly constrained. However, if a protein's function requires it merely to aggregate, rather than to adopt a well-defined quaternary conformation, the constraints on its sequence will be less stringent.
It is well possible that Apoptin's original function has demanded its multimerization, which it may have achieved through random aggregation as outlined above, rather than through the formation of a specific quaternary structure. If so, Apoptin provides an example of a novel route for the evolution of protein function. There are several possible explanations why the multimerization of Apoptin may be essential for its biological function; e.g. aggregation may stabilize Apoptin, which in its ill-defined monomeric form may be readily degradable, or the formation of globular multimers may result in cooperative binding of Apoptin moieties to certain large ligands or molecular complexes (DNA, RNA, chromatin, nuclear pores, etc.). Obviously these possibilities do not exclude one another, underscoring the likeliness that in the case of Apoptin the classical structure/function paradigm translates as an aggregation/function paradigm.