The experiments described here were undertaken to define the DNA binding domain of the adenovirus IVa2 protein, a key component of the viral genome encapsidation mechanism. While IVa2 was first described as a transcriptional activator of the MLP (Lutz and Kedinger, 1996
; Tribouley et al., 1994
) and as a minor component of the viral core (Goding and Russell, 1983
; Winter and D'Halluin, 1991
), more recent results from the Hearing and our labs have provided evidence that it has a critical role in viral genome encapsidation (Ostapchuk et al., 2011
; Zhang and Imperiale, 2003
). From an examination of the predicted structural motifs of the protein and from a previous report in which the effects of various IVa2 deletions on its ability to bind the DE1 and DE2 elements were examined (Lutz and Kedinger, 1996
), we predicted that a putative helix-turn-helix motif at the C-terminus is required for DNA binding. We produced a mutant protein lacking the ten most C-terminal amino acids and demonstrated that its biophysical properties were indistinguishable from those of the wild type protein. Similar protein levels were also seen following transfection of expression vectors into 293 cells. Thus, the ten amino acid deletion did not affect the overall structure and expression of IVa2. EMSA analysis with a packaging sequence DNA probe confirmed that the DNA binding motif was disrupted by deletion of these ten amino acids. At concentrations at which IVa2(wt) formed a single complex with the probe, IVa2(1-439) formed none; only at a thousand-fold higher concentrations, at which IVa2(wt) forms multiple specific complexes on this probe (Tyler et al., 2007
), was even a faint, single complex seen with IVa2(1-439).
While it is formally possible that the lack of DNA binding observed with IVa2(1-439) was due to the presence of inhibitory contaminants from the purification, concentrations up to at least 100 nM IVa2(1-439) have no effect on the binding of IVa2(wt) to the packaging sequence (data not shown). This result also implies that the multiple complexes observed on the packaging sequence probe as the concentration of IVa2(wt) increases depend upon interactions of the protein directly with the DNA, and not upon protein-protein interactions. It has also been reported that the C-terminus of IVa2 contains a nuclear localization sequence (Lutz et al., 1996
); however, this maps to residues 432-437, which are retained in our mutant.
Having identified a DNA binding mutant of IVa2, we examined the consequences of the loss of DNA binding on IVa2 binding to a known protein partner, L1 52/55 kDa, which also has a role in genome packaging (21, 26). Like IVa2, L1 52/55 kDa is found in assembly intermediates, but unlike IVa2, it is not found in mature virions (Hasson et al., 1992
). In contrast to the absence of DNA binding, the IVa2(1-439) protein bound L1 52/55 kDa, suggesting that this interaction is separable from the interaction with the genome. This result confirms that the ten amino acid deletion does not disrupt the overall structure of the IVa2 protein.
We conducted experiments to determine how abrogation of DNA binding affects the viral life cycle. Mutating the C-terminal DNA binding domain did not significantly affect late protein expression. While the MLP contains two binding sites for IVa2-containing complexes, DE1 and DE2, and can be activated by IVa2 in transfection assays (Tribouley et al., 1994
), viruses that cannot express IVa2 or in which both the DE1 and DE2 sites on the MLP have been mutated show only a subtle decrease in late protein expression (Ostapchuk et al., 2011
; Pardo-Mateos and Young, 2004b
; Zhang and Imperiale, 2003
). We therefore conclude that the major role for IVa2 in the viral life cycle is during assembly. Whereas IVa2(wt) was able to complement a mutant genome and produce infectious virions, IVa2(1-439) was not.
Our recent demonstration that IVa2 is found at a single vertex (Christensen et al., 2008
) led us to envisage a potential model for the role of IVa2 in adenovirus assembly in which IVa2 forms a structure at the unique vertex, alone or in concert with other viral proteins. This complex then associates with the viral genome through the A repeats of the packaging sequence and facilitates encapsidation of the DNA. This allows a role for IVa2 that is consistent with earlier studies that suggested the viral genome is inserted into a preformed empty capsid (Everitt et al., 1973
; Sundquist et al., 1973
; Tibbetts, 1977
). If IVa2 and other viral proteins form a complex at a unique vertex, it is possible they associate with a portal structure. Such portals are used by a number of other viruses with dsDNA genomes, including HSV and several bacteriophages including PRD1, which has a number of remarkable structural similarities to adenovirus (Abrescia et al., 2004
; Gowen et al., 2003
; Newcomb et al., 2001
; Wills et al., 2006
). The portals studied so far are all dodecamers (Catalano, 2005
); we have determined that there are approximately six copies of IVa2 per mature virion (Christensen et al., 2008
), making it unlikely that IVa2 is itself the portal.
IVa2 almost certainly does not act alone during assembly. One candidate for a protein partner is L4 22K/33K. We have shown that while L4 22K cannot by itself bind packaging sequence DNA, it can form complexes on the DNA in the presence of IVa2 (Ewing et al., 2007
); these same complexes are also present in infected cells (Ewing et al., 2007
; Ostapchuk et al., 2006
). It is also apparent that the presence of L4 22K enhances the binding of IVa2 to the packaging sequence (Ewing et al., 2007
), and there is recent evidence from the study of an L4 22K null virus that IVa2 is stabilized by L4 22K (Morris and Leppard, 2009
). It is interesting to note that the L4 22K null virus has a similar phenotype to a IVa2 null virus, i.e. it has normal viral gene expression and DNA replication, but it is unable to produce infectious particles (Ostapchuk et al., 2006
). It is not known whether any empty particles are produced by this mutant virus.
It also remains to be elucidated what the exact role of L1 52/55 kDa is in adenovirus capsid assembly and genome encapsidation. An L1 52/55 kDa null virus is only able to produce empty capsids (Gustin and Imperiale, 1998
) and a temperature sensitive L1 52/55 kDa mutant virus packages only a short segment of DNA, ~1,000 bp of the left end of the genome, at the restrictive temperature (Hasson et al., 1989
). Both our lab and the Hearing lab have shown that L1 52/55 kDa can interact with the packaging sequence (40, 43), although L1 52/55 kDa protein has not been shown to bind to DNA directly in the absence of IVa2 (Ostapchuk et al., 2005
; Perez-Romero et al., 2005
), which does bind DNA directly. These data suggest L1 52/55 kDa has multiple roles in the encapsidation of the adenoviral genome. There must be a role at the initiation of packaging, because without L1 52/55 kDa no DNA is packaged. The results with the temperature sensitive L1 52/55 kDa virus suggest that L1 52/55 kDa has a second function in the completion of encapsidation, as virions with only short portions of the left end of the genome are detected. These two roles may reflect the two types of L1 52/55 kDa-containing complexes: one with DNA and one with IVa2.
If the unique IVa2-containing vertex identified on the adenovirus capsid represents the site on the virion where the viral genome is recognized and translocated into the capsid, then there should be proteins present at that vertex that provide energy for translocation. The packaging motors associated with the portals of the tailed bacteriophages T4, λ and ϕ29 are ATPases (Benson et al., 1999
; Guo et al., 1986
; Guo et al., 1987
; Karhu et al., 2007
; Rao and Mitchell, 2001
; Yang et al., 1999
; Yang et al., 1997
). While a portal structure has not been defined for tailless icosahedral phage PRD1, it does contain a unique vertex (Stromsten et al., 2003
). One of the proteins associated with the unique vertex is P9, a putative ATPase. IVa2 is a compelling candidate to possess ATPase activity because it contains Walker A and Walker B motifs, characteristic of ATPases in the ABC and AAA+ families (Aravind et al., 1999
; Burroughs et al., 2007
). These motifs are absolutely conserved in all adenoviruses whose sequences have been determined, although other parts of the protein are not. While there is not yet evidence of biochemical ATPase activity, IVa2 has been shown to bind ATP with low affinity (Ostapchuk and Hearing, 2008
), and mutating a lysine residue in the Walker A motif that is predicted to be involved in ATP binding is lethal to the virus (Ostapchuk et al., 2011
; Pardo-Mateos and Young, 2004a
). Taken together, the data presented provide promising directions for further biochemical analysis of the adenovirus IVa2 protein and its partners with the goal of elucidating the mechanisms used for genome packaging in icosahedral eukaryotic viruses.