Mapping of herpes simplex virus-1 VP22 functional domains for inter- and subcellular protein targeting


The herpes simplex virus 1 (HSV-1) tegument protein VP22 has been utilised as a vehicle for trafficking proteins. It has a remarkable property of exiting the cell that is producing it and entering the neighbouring cells, which has been used to deliver therapeutic proteins, p53 and herpes simplex virus thymidine kinase (tk). It has a complex pattern of expression and subcellular localisation. Functions of VP22 include intercellular transport, binding to and bundling of microfilaments, inducing cytoskeleton collapse, nuclear translocation during mitosis, and binding to chromatin and nuclear membrane. The regions of VP22 which contain each of these functions have not been characterised. Finding the region carrying the property of intercellular spread would facilitate enhancement of transport function. By constructing a series of deletion constructs of VP22 tagged by the green fluorescent protein (GFP) we have mapped the functions of VP22 to specific regions in the polypeptide as follows: intercellular transport – aa 81–195; binding and reorganisation of cytoskeleton – aa 159–267; nuclear targeting, inhibition of cytoskeleton collapse – aa 81–121; and nuclear targeting and facilitation of intercellular transport – aa 267–301. Separation of VP22 functions enables focus on the mechanism of VP22-mediated transport and improve the transportation efficiency of VP22.


The herpes simplex virus-1 protein VP22 is a major component of the herpesvirus tegument, the region between the envelope and capsid of the virus particle. It has a complex pattern of expression and subcellular localisation. In expressing cells, VP22 is first localised in the cytoplasm, where it binds to microtubules and induces collapse of the cytoskeleton creating a large perinuclear lump. After the nuclear membrane has disintegrated in early mitosis, VP22 translocates into the nucleus and binds to chromatin, where it remains until after completion of mitosis and during the next interphase.1 The determinants of microtubule binding and reorganisation, nuclear translocation, and chromatin binding have not been elucidated. In addition to its functions in the expressing cells, VP22 has the capability of intercellular spread, whereby the protein exits producing cells and enters neighbouring cells, where it translocates into the nucleus.2 This property is maintained in protein fusions with VP22, which can be utilised to transfer potentially therapeutic proteins into target cells.

It has been recently demonstrated that VP22 can deliver the tumour suppressor protein p53 from expressing cells to neighbouring recipient cells where p53 retained its function in inducing cell cycle arrest and apoptosis.3 We have previously constructed a fusion gene encoding VP22-tk, ie VP22 fused to herpes simplex virus thymidine kinase (tk), in retroviral backbones. This protein was found to retain the intercellular trafficking properties of VP22, as well as tk enzyme activity. The VP22-tk protein spreads between cells in sufficient quantities to induce cell death in response to ganciclovir (GCV) treatment not only in the primary synthesising (a.k.a. synthetic) cells but also in surrounding recipient cells, thus causing a bystander phenomenon even in cells devoid of gap junctions. This effect was observed upon GCV treatment of transfected tissue culture cells and in vivo in mice injected with tumour cells transduced with VP22-tk fusion genes.4 However, the effect was only observable with a relatively high percentage (50%) of VP22-tk-synthesising cells as compared with non-VP22-tk expressing cells in the tumour cell mixture.

The green fluorescent protein (GFP) is detectable in individual cells without the use of exogenous agents, such as enzyme substrates. The localisation of GFP is essentially nonrestricted; that is, it is freely distributed in the cell cytoplasm as well as the nucleus. This makes GFP a useful reporter for studies of protein localisation. In previous studies it has been shown that a GFP-VP22 fusion protein retains the properties of the native VP22 in a recombinant HSV-1 virus,5 and allowed the characterisation in detail of the subcellular localisation of VP221 and its intercellular transfer.6,7,8,9 GFP-VP22 fusion protein has been found to retain the intercellular trafficking properties of VP22 in all the cell lines studied.

We theorised that the different functions of VP22 may be separable and localised to specific regions of the protein. To investigate the properties of VP22 and to dissect the different functions and determinants of localisation, we have made several deletion constructs of VP22 as fusions to GFP. Compared with full-length GFP-VP22, the deletion constructs showed different subcellular localisation and effects on the cytoskeleton, as well as altered intercellular transport.


The deletion constructs were made as fusions to EGFP and EYFP and are depicted in Figure 1a. The plasmids were transfected into Cos-7 cells and the cells were analysed 24 and 48 h later. To confirm the identity of the constructs, Western blotting of the transfected cell lysates was performed using antibodies to the HA tag (data not shown) and GFP (Figure 1b). Localisation of the GFP-tagged constructs in live and fixed cells was assessed by fluorescence microscopy. The localisation pattern in PFA-fixed cells was found to be identical in living cells. Cells were studied fixed to maintain their morphology at room temperature.

Figure 1

(a) Deletion constructs used in this study. Fragments of VP22 were fused to the C-terminus of GFP, or the N-terminus of EYFP (ΔC121). Restriction sites used are shown, as well as the HA tag. (b) Western blot analysis of the deletion fusion constructs. 1, GFP-VP22; 2, ΔC121; 3, ΔN81; 4, ΔN119; 5, F81–195; 6, F119–195; 7, F159–195; 8, ΔN 195; 9, ΔN 267; 10, GFP. One μg of each plasmid was transfected for 2 × 105 293 cells and detected with anti-GFP antibody.

Subcellular localisation of GFP-VP22 deletion constructs in living cells

To characterise the localisation and function of VP22 fragments, we transfected Cos-7 cells with the VP22 deletion constructs fused to GFP. We observed the cells at 24 and 48 h after transfection (Figure 2). In line with the results published by us and others,2,6 the localisation of full-length GFP-VP22 was filamentous in the early stages of transfection, ie up to about 36 h, followed by change to a bright perinuclear body and nuclear translocation of the fluorescent protein at 48 h after transfection. As a control, localisation of GFP was compared with the localisation of the constructs. GFP showed diffuse nuclear and cytoplasmic localisation at all times. The deletion constructs had each a distinct localisation pattern, and these are described below.

Figure 2

Localisation of GFP-VP22 deletion constructs in cells. Transfected cells were fixed in paraformaldehyde and photographed 24 h and 48 h after transfection. Original magnification ×40.

ΔC121 localisation was diffuse, predominantly nuclear with slight nucleolar concentration at all times. After methanol fixation ΔC121 was not retained well inside the cells, so some loss of signal could be observed. However, near complete loss of fluorescence is observed with GFP.

ΔN81 closely resembles full-length VP22 in its subcellular localisation and retains the intercellular trafficking capability. The properties of microtubule binding, cytoskeleton collapse, nuclear localisation and chromatin binding are all retained by this N-terminally deleted VP22. Before cell division, ΔN81 binds to the cytoskeleton and induces cytoskeleton collapse. During mitosis, ΔN81 binds to chromosomes (inset in Figure 2), but remains diffusely detectable in the cytoplasm as well. After cell division it remains in the cell nucleus, but also remains easily detectable in the cytoplasm. The capability to induce cytoskeleton collapse is diminished in ΔN81 compared with wt GFP-VP22. Fifty per cent of expressing cells show bundled cytoskeleton at 48 h after transfection even after cell division. In producer cells localisation does not change after fixation by PFA or methanol.

ΔN119 binds to cytoskeleton and induces its collapse at 24 h after transfection when it is not localised in the cell nucleus. At 48 h, however, when the cytoskeleton has collapsed, ΔN119 can be found as a perinuclear lump and distinct speckles on the nuclear membrane, but not inside the nucleus.

F81–195 localises diffusely in the cell cytoplasm and nucleus, with more pronounced nuclear-nucleolar predominance when the expression level is low. The localisation does not change after cell division.

F119–195 localises diffusely in the cytoplasm and nucleus when the expression level is low at 24 h. Cells expressing more protein show a collapsed cytoskeleton. At 48 h all cells show a collapsed cytoskeleton and an ‘empty’, nonfluorescent nucleus.

F159–195 remains in the cytoplasm before and after cell division. It localises as a ‘foamy’ network. No cytoskeleton collapse was observed.

ΔN195 was found in the cytoplasm. At 24 h already 90% of the ΔN195 expressing cells had a collapsed cytoskeleton, a markedly rounded shape and a strongly fluorescent perinuclear aggregate. Methanol fixation did not change this localisation.

ΔN267 localised diffusely in the cell cytoplasm and nucleus, with more pronounced nuclear–nucleolar predominance when expression levels were low. Localisation did not change after cell division. After methanol fixation cytoplasmic localisation of the fluorescent protein became more predominant due to a considerable loss of nuclear signal. No fluorescence was detectable in the surrounding cells after methanol fixation.

Intercellular spread

Figure 3 shows the transport capacity of the two deletion constructs which still allowed intercellular trafficking (ΔN81 and F81–195). Other constructs were not detectable in surrounding cells after methanol fixation (data not shown). While full-length GFP-VP22 localises primarily in the recipient cell nucleus and only a minor fraction can be found in the cytoplasm, construct ΔN81 was localised predominantly in the recipient cell cytoplasm as filamentous structures. The recipient cells nuclei were characterised by lack of fluorescence. F81–195 could be found diffusely throughout the cell, with a slight concentration in the nucleus. The ease of detection of F81–195 in non-producer cells is considerably diminished compared with full-length GFP-VP22 and ΔN81 (compare the signal in surrounding cells on Figure 3). However, the amount of F81–195 in surrounding cells may be underestimated due to its diminished capacity to bind the cytoskeleton and to chromatin, resulting in a more diffuse signal. Panels a and b depict the same fields. In panel a, only GFP fluorescence is shown, in panels b and c blue colour identifies cell nuclei. In panel c, the fusion proteins are detected by GFP antibody, followed by Cy-3-conjugated secondary antibody (red). The transfected, ie synthetic cells appear yellow due to the balance of green and red signals. Note the excess of red signal in the surrounding cells, indicating partial loss of GFP function upon transportation.

Figure 3

Intercellular spread of GFP-VP22 deletion constructs. Only the constructs showing detectable cell to cell spread are shown. Transfected Cos-7 cells were fixed with methanol and photographed 48 h after transfection. Original magnification ×40. (a) GFP fluorescence; (b) the same field: GFP (green), cell nuclei stained with bisbenzimide (blue); (c) GFP (green), cell nuclei (blue) and GFP detected by Cy-3 (red).

The properties of each of the constructs are summarised in Table 1.

Table 1 Properties of the VP22 deletion constructs


The exact mechanism of VP22 function is not known. It is likely that various dynamic interactions of VP22 with different cellular proteins are required.

The functional properties of VP22 can be broadly divided into nuclear localisation, nucleolar localisation, chromatin binding, microtubule binding, induction of cytoskeleton collapse and intercellular transport. We theorised that these functions may be separable in the VP22 polypeptide. Therefore, we constructed a series of deletion mutants as fusions to N- and C-terminus of EGFP protein. EGFP does not affect VP22 functions and has been used by us6 and others5 to tag VP22. All the deletion constructs retained some of the specific functions of VP22. As a result, we could identify the regions of VP22 required for cytoskeleton binding and reorganisation, nuclear and nucleolar localisation, nuclear membrane localisation and intercellular transport. These regions are depicted in Figure 4.

Figure 4

Localisation of the functional properties on the VP22 polypeptide.

We mapped the amino acids critical for binding to and reorganising cytoskeleton to between positions 159 and 267 (Figure 4a). Amino acids 159–195 (construct F159–195) localise in a network-like structure, but do not induce collapse. In F119–195, however, destabilisation of the cytoskeleton is more prominent, but only in cells with higher levels of F119–195 expression. ΔN195 induced a strong cytoskeleton collapse even at early times of expression. Because of the marked difference in cytoskeleton collapsing activity between the constructs ΔN81 and ΔN119, and between F81–195 and F119–195, we conclude that amino acids 81–119 contain elements inhibitory to cytoskeleton collapse (compare constructs F81–195 and F119–195 at 48 h).

ΔC121, ΔN267 and F119–195 localised to the cell nucleus independent of the cell cycle stage; mitosis was not necessary for the nuclear translocation of these constructs. No association was seen with mitotic chromatin. ΔN267 and F119–195 were concentrated in the nucleoli. ΔC121 was concentrated in the nucleoli by 24 h; later the higher level of expression resulted in a general diffuse localisation of the protein. This suggests that there are at least two separate determinants of nuclear and nucleolar localisation in the VP22 protein. These are the regions between 81–121 and 267–301 (Figure 4b). All these three constructs are of a sufficiently small size (ΔC121 = 40.6 kDa, ΔN267 = 33.2 kDa and F81–195 = 42.0 kDa) to be able to diffuse into the nucleus in the absence of a specific transport signal. However, retention in the nucleus and nucleoli, most probably results from a specific interaction. Binding to chromatin was maintained only in the N-terminally deleted ΔN81. Construct ΔN119 displayed a speckled pattern at the nuclear membrane, but did not bind to chromosomes or interphase chromatin. This suggests that binding to chromatin is a co-operative function, which requires several regions in the VP22 peptide 81–301, and aa 81–119 are necessary, but not sufficient, for the function (Figure 4c).

Nuclear membrane localisation displayed by ΔN119 is also a co-operative function within the peptide 119–301, and which requires the presence of aa 119–159.

The construct ΔC267 has been previously reported not to spread between cells,2 suggesting that the C-terminus is necessary for intercellular transport. In agreement with these previous results, ΔN81 could spread, but a different localisation pattern was observed in recipient cells. Surprisingly, the construct F81–195, which lacks both the N- and C-terminal portion, showed a reduced, but detectable intercellular spread. Therefore, we conclude that the N-terminal 80 amino acids do not contribute to intercellular spread of VP22. Instead, the N-terminal may contain a negative regulatory function which is counteracted by the C-terminal portion of VP22. The constructs ΔN195 and ΔC121 are not detectable in surrounding cells after methanol fixation. Based on this, we mapped the property to spread between cells, which can be detected visually after methanol fixation, to amino acid positions 81–195 (Figure 4e).

We propose a model for VP22 function, where each functional region either binds to a cellular partner to facilitate its function, or to another region within the VP22 protein, in which case these functions are blocked.

As a protein capable of intercellular transport, VP22 has been shown to be a potentially useful component in bystander cell targeting in gene therapy. The capacity of the fragments of VP22 described here to induce bystander cell death, in the absence of gap junctions, in HSV-TK – GCV suicide gene therapy model is currently under investigation, although it seems likely that none of the identified fragments will be as effective in inducing bystander cell death as the full-length VP22 protein.

We have dissected the functional domains of VP22 by deleting selected portions of the molecule in GFP fusion constructs. Each of the deletion constructs displayed a different subcellular localisation pattern, a different effect on the cytoskeleton, and an altered intercellular transport capability. We have therefore mapped the functions of VP22 which may be of benefit in protein delivery vector construction and the subcellular targeting of functional protein fusions.

Materials and methods


Cos-7 (rhesus monkey fibroblast) and 293 (human embryonic kidney epithelium) cell lines were obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA). Cells were grown in Dulbecco's modified Eagle's medium (4500 mg/l glucose) (Life Technologies, Paisley, UK) supplemented with 10% fetal calf serum (Life Technologies) and 2 mM L-glutamine, at 37°C 5% CO2.

Cells were seeded at 50–60% confluency the previous day and transfected with plasmids by Fugene 6 reagent (Boehringer Mannheim, Mannheim, Germany) according to the manufacturer's instructions. Briefly, DNA (0.2–1.0 μg) and Fugene 6 reagent were mixed in 100 μl serum-free cell culture medium at the mass:volume ratio 1:2 and added to cells after 15 min.

The cells were fixed 24 or 48 h after transfection with 100% methanol, or 3% paraformaldehyde (PFA) at room temperature for 15 min.


pEGFP-VP22 was created by cloning into the BamHI site of pEGFP-C1 (Clontech, Palo Alto, CA, USA) a BglII–BamHI fragment from pUL49epB7 and adding the BamHI–XbaI fragment from pINSHNES (both kind gifts from Dr P O'Hare, Marie Curie Research Institute, Oxted, UK) containing the hemaglutinin (HA) epitope. Deletion constructs fused to EGFP were created by the following strategy: pEGFP-VP22 was cleaved with ApaI and XbaI. The 5′ fragment of the VP22 gene (ApaI–ApaI) was cloned in frame to pEYFP-N1 (enchanced yellow fluorescent protein – a spectral variant of GFP) yielding the construct ΔC121. The ApaI-XbaI fragment was cloned in frame into pEGFP-C3 to yield the construct ΔN119. This in turn was cleaved with SmaI-XbaI and the resulting fragment was cloned into pEGFP-C3 to yield the construct ΔN195. To make the construct F119–194, plasmid ΔN119 was cleaved with NaeI and BamHI, the ends flushed with Klenow enzyme and religated. To make ΔN81, pUL49epB was cleaved with PpuMI and the ends filled with Klenow enzyme. Then it was cut with XbaI and cloned into pEGFP-C2. F81–195 was made by cleaving ΔN81 with NaeI and BamHI, filling the ends with Klenow and religation. To make F159–195, the SalI-SalI fragment from F119–195 was cloned into SalI site of pEGFP-C3. The SalI–XbaI fragment from ΔN195 was cloned into pEGFP-C1 to make ΔN267. All constructs were sequenced at GFP-VP22 junctions to confirm their identity.

Western blotting

Transfected 293 cell lysates were separated on a 10% SDS-PAA gel, blotted on to a nitrocellulose filter and detected with anti-GFP antibody (Clontech) followed by HRP conjugated goat anti-rabbit antibody. Blots were detected with SuperSignal ULTRA chemiluminescence detection kit from Pierce (Rockford, IL, USA).


For immunodetection, anti-GFP rabbit polyclonal antibody was used (Clontech) and secondary donkey-anti-rabbit Cy-3 conjugate antibody (Jackson Immunoresearch Laboratories, West Grove, PA, USA) both according to manufacturer's instructions. The cells were washed once in PBS, blocked for 2 h with BSA and incubated with the primary antibody in blocking solution overnight at dilution suggested by the manufacturer, at room temperature, followed by four washes with PBS. The secondary antibody was diluted in the blocking solution and incubated for 1 h at room temperature, followed by four washes with PBS. Hoechst stain (Bisbenzimide H33342; Fluka Biochemica, Buchs, Switzerland) was used to counterstain the nuclei, at 1:4000 dilution in PBS, for 5 min at room temperature followed by one PBS wash.


Photographs were taken at 24–48 h after transfection with a Leica DM RXA confocal microscope (Leica Microsystems, Heidelberg Germany) and a Spectra Source Orbis digital camera (SpectraSource Instruments, Westlake Village, CA, USA). Image processing was done using Slidebook 2.1.5 (Intelligent Imaging Innovations, Denver, CO, USA) and Adobe PhotoShop 5.0 (Adobe Systems, Seattle, WA, USA).


  1. 1

    Elliott G, O'Hare P . Cytoplasm-to-nucleus translocation of a herpesvirus tegument protein during cell division J Virol 2000 74: 2131–2141

  2. 2

    Elliott G, O'Hare P . Intercellular trafficking and protein delivery by a herpesvirus structural protein Cell 1997 88: 223–233

  3. 3

    Phelan A, Elliott G, O'Hare P . Intercellular delivery of functional p53 by the herpesvirus protein VP22 Nat Biotechnol 1998 16: 440–443

  4. 4

    Dilber MS et al. Intercellular delivery of thymidine kinase prodrug activating enzyme by the herpes simplex virus protein, VP22 Gene Therapy 1999 6: 12–21

  5. 5

    Elliott G, O'Hare P . Live-cell analysis of a green fluorescent protein-tagged herpes simplex virus infection J Virol 1999 73: 4110–4119

  6. 6

    Aints A, Dilber MS, Smith CI . Intercellular spread of GFP-VP22 J Gene Med 1999 1: 275–279

  7. 7

    Brewis N et al. Evaluation of VP22 spread in tissue culture J Virol 2000 74: 1051–1056

  8. 8

    Harms JS, Ren X, Oliveira SC, Splitter GA . Distinctions between bovine herpesvirus 1 and herpes simplex virus type 1 VP22 tegument protein subcellular associations J Virol 2000 74: 3301–3312

  9. 9

    Wybranietz WA et al. Quantification of VP22-GFP spread by direct fluorescence in 15 commonly used cell lines J Gene Med 1999 1: 265–274

Download references


This work was supported by grants from the Swedish Child Cancer Fund, the Swedish Tobias Foundation, the Swedish Cancer Fund, Åke Wibergs Stiftelse (Sweden), Alex and Eva Wallströms Stiftelse (Sweden) and Stiftelsen Clas Groschinsky Minnesfond (Sweden). We wish to thank Dr Peter O'Hare for providing the plasmids pUL49epB and pINSHNES and Dr Birger Christensson for help with confocal microscopy.

Author information

Correspondence to MS Dilber.

Rights and permissions

Reprints and Permissions

About this article


  • herpes simplex virus-1
  • VP22
  • intercellular transport
  • gene therapy
  • protein trafficking

Further reading