Structural analysis of herpes simplex virus by optical super-resolution imaging

Laine, Romain F.; Albecka, Anna; van de Linde, Sebastian; Rees, Eric J.; Crump, Colin M.; Kaminski, Clemens F.

Structural information of biological nanostructures can be derived through averaging repetitive features as is commonly performed in EM. Particle averaging in SMLM images was originally shown for the nuclear pore complex 34, 47 , and more recently for HIV-1 (ref. 27 ). Here we extended this approach and developed a model-based analysis of SMLM data to extract structural parameters for specific virus proteins layers with optimal precision (1-2 nm for the shell diameter and o6 nm for the thickness). We compared different antibody-labelling methods in the context of structural analysis and found that primary labelling offers the most robust readout of protein location. We found that errors in the estimates of linker size or localization error introduce bias in the analysis and this needs to be carefully considered when secondary labelling is used: for example, we found that linker lengths varied from ca. 15 to 25 nm depending on which protein was labelled; therefore, precise estimates may be difficult in a practical situation. In contrast, we found that primary labelling resulted in a reliable and constant linker length of 10 nm. Fab fragments offer potential for even shorter linker size. However, producing Fab fragments that retained sufficient binding specificity and/or affinity proved challenging for most of the proteins we considered. In future, other labelling techniques may offer more promising routes towards short linker size, for example, nanobodies 48 or aptamers 49 . Furthermore, the effect of restricted rotational freedom is of great importance, as it affects the localization precision 50 . We show that for analysing purified virions, primary labelling offers a good compromise, introducing small to no bias and providing high localization precision. We used our model-based analysis to accurately determine the diameters of different protein layers in purified virions and were thus able to build a structural model of the HSV-1 tegument architecture (Fig. 4) . We observed significant differences in the diameter of all investigated protein layers and found that the central section of VP1/2 is located deep within the virus particle, featuring a diameter of only B122 nm. This is similar in size to the capsid itself, as established by prior EM studies 5 . This observation suggests that the central domain of VP1/2 may directly interact with the capsid, similar to the C-terminus of VP1/2 (ref. 43 ). In addition, our model suggests that the Nterminus of VP1/2 projects away from the capsid towards the envelope, as both pUL37 and VP16 interact with the N-terminal domain of VP1/2 and were located near the envelope (Fig. 4b) . To investigate this further, it would be interesting to measure particle diameters using antibodies specific to various epitopes along the length of VP1/2, in particular the N-terminus.