The detection of the PVAT is technically difficult because its presence at only one vertex makes it unsuitable for icosahedral reconstruction. Furthermore, it is surrounded by inconsistently arranged tegument material of roughly similar density and so it differs from a bacteriophage tail, which is not obscured by other structural components. We were able to visualize it here because the methods used were sufficiently sensitive to distinguish the set of unique signals provided by the portal vertex and use this information to direct the alignment of each subtomogram. Aligning them based on the density distribution at the portal vertex meant that the PVAT density was reinforced in the final reconstruction, while that of the surrounding pleomorphic tegument, was weakened. The MSA-guided and melon ball alignment procedures produced the same particle orientation for 151 of the 214 particles. Analysis of the other 63 particles suggested that the majority had been reliably assigned by MSA-guided alignment whereas melon ball alignment was more frequently erroneous due to systematic bias from aligning to the missing wedge. It should be emphasized that in the MSA-guided method, the assignment of the portal vertex is made on a per-particle basis. In contrast, the all-vs-all melon ball alignment simply compares all the vertices in one particle to all the vertices of another particle without any starting assumption about any of the vertices. Because the two independent and unrelated alignment methods generated the same final structure, we attach high confidence to the resulting description of the PVAT.
It is likely that the PVAT does not constitute a self-contained structure but represents a set of proteins that are consistently arranged with respect to the portal vertex and are continuous with the heterogeneous surrounding tegument. In this sense, it resembles the star-shaped densities seen at the tops of pentons, where the surface displayed in the reconstruction delineates a boundary between the constant and changeable parts of the tegument.
The PVAT occupies a position that would enable it to carry out several of the receptor binding and signalling functions that in tailed bacteriophages are performed by the tail. In these bacteriophages, the tail plays a key role in docking the virus to its host and controlling entry of the virus genome into the host cell. In herpesvirus-infected cells, the PVAT may play an equivalent role in docking the capsid to the NPC as well as in signalling the proper conditions for release of the viral DNA. Ensuring that the viral genome is released only when the capsid reaches the nucleus would require interaction between components of the virus particle and the NPC 
The composition of the PVAT cannot be directly determined from our reconstructions. The likelihood that it is DNA leaking from the capsid is small, as all existing evidence suggests that the viral genome is fully retained within capsids until its release is triggered at the nuclear pore. Furthermore, the size and shape of the PVAT are consistent with it being a complex of proteins but not with it being formed by a single DNA strand. Two proteins that have been shown to have roles in NPC recognition 
and DNA release 
, and are therefore likely to be present at the portal vertex, are pUL25 and pUL36. pUL25 together with a second protein, pUL17, has been identified as forming the capsid-vertex-specific component (CVSC) on capsids purified from infected cell nuclei 
. The CVSC occupies a position overlying triplexes Ta and Tc (as designated in 
); adjacent to, but not extending to, the pentons. This position corresponds to that of the distal portions of the star shaped densities that are seen in our map connecting the pentons to the peripentonal triplexes and hexons (). The proximal portions of our star shaped densities that are in contact with the pentons, are also present in virions that have been stripped of their envelopes and much of their tegument where they have been shown to be formed by pUL36 
. The common geometry displayed by the outer portions of the star-shaped densities at the pentonal vertices () and the flying buttresses at the portal vertex () supports a hypothesis that they are formed by these same proteins. Interestingly, the star-shaped density seen at the pentonal vertices can only account for a fraction of the mass of pUL36 (~336 kDa) 
, implying that much of it is unresolved at these locations. Therefore, reorganization of pUL36 induced by the unique environment at the portal vertex could allow it to contribute to the distinctive PVAT structure.
An interesting aspect of the PVAT is its potential role in orchestrating the arrangement of tegument and envelope around the capsid. In many of the individual virions, the distal end of the PVAT was found to lie close to the envelope. This observation, combined with the relatively fixed separation of the portal vertex and envelope evident from , appears to indicate that the PVAT is acting in the manner of a tether, connecting the envelope to the capsid (). Such behavior would suggest that formation of the virion envelope, which takes place by budding into golgi-derived vesicles, could be influenced by the orientation of the capsid around the portal vertex.
This study demonstrates the increasing power of cryo-ET and 3-D image processing techniques to uncover unexpected features, even in a virus as extensively studied as HSV-1, and illustrates the potential of such approaches to enhance our understanding of other important molecular machines.