Dynamic ubiquitination drives herpesvirus neuroinvasion

Nicholas J. Huffmastera, Patricia J. Sollarsb, Alexsia L. Richardsa, Gary E. Pickardb,c, and Gregory A. Smitha,1

aDepartment of Microbiology–Immunology, Northwestern University Feinberg School of Medicine, Chicago, IL 60611; bSchool of Veterinary Medicine and Biomedical Sciences, University of Nebraska, Lincoln, NE 68583; and cDepartment of Ophthalmology and Visual Sciences, University of Nebraska Medical Center, Omaha, NE 68198

Edited by Bernard Roizman, The University of Chicago, Chicago, IL, and approved September 1, 2015 (received for review June 25, 2015)

Neuroinvasive herpesviruses display a remarkable propensity to enter Results the nervous system of healthy individuals in the absence of obvious Ubiquitination Promotes Retrograde Delivery of PRV Capsids to Cell trauma at the site of inoculation. We document a repurposing of Bodies of Sensory Neurons. The requirement of the pUL36 DUB cellular ubiquitin during infection to switch the virus between two during neuroinvasion may be either a protective mechanism that invasive states. The states act sequentially to defeat consecutive host stabilizes substrates by maintaining them in an unmodified state, barriers of the peripheral nervous system and together promote the or part of a regulatory mechanism that switches proteins between potent neuroinvasive phenotype. The first state directs virus access to ubiquitinated and nonubiquitinated states. To discriminate between nerve endings in peripheral tissue, whereas the second delivers virus these possibilities, ubiquitination was inhibited during PRV in- particles within nerve fibers to the neural ganglia. Mutant viruses A locked in either state remain competent to overcome the correspond- fection in a neuron culture model (Fig. 1 ). Upon infection of ing barrier but fail to invade the nervous system. The herpesvirus primary dorsal root ganglion (DRG) sensory neurons that were “ubiquitin switch” may explain the unusual ability of these viruses to pretreated with PYR41, an inhibitor of the E1 ubiquitin activating routinely enter the nervous system and, as a consequence, their prev- enzyme, the delivery of PRV capsids to the nuclear rim was sig- alence in human and veterinary hosts. nificantly reduced compared with the vehicle-treated control (Fig. 1B) (31). In contrast, interference was not observed in an epithelial

herpesvirus | PRV | HSV-1 | neuroinvasion | ubiquitin cell line nor following treatment of neurons with an inhibitor of MICROBIOLOGY the related E1 activating enzyme of the ubiquitin-like modifier, erpesviruses such as herpes simplex virus type 1 (HSV-1) NEDD8 (Fig. 1 B and C) (32). These findings indicated a specific Hand pseudorabies virus (PRV) routinely enter the peripheral role for ubiquitination in neurons during infection. Given that viral nervous system following casual contact between infected and protein pUL36 is an effector of the retrograde axonal transport that uninfected individuals. Recent findings have provided insights delivers capsids to the neural soma, we investigated whether pUL36 into this complex phenotype, but the molecular mechanisms was a target of ubiquitination during infection (2). A previous directing nervous system invasion are poorly understood (1–5). report indicated that pUL36 is ubiquitinated during transient A critical effector of this process is a conserved herpesvirus deubi- overexpression when the DUB is inactivated by a catalytic quitinase (DUB) (a cysteine protease that removes ubiquitin from mutation (23); this also proved true during infection of epi- – target posttranslationally modified proteins) (6 8). The DUB is thelial cells (Fig. 1D). In contrast, there was no evidence of housed in the amino terminus of the pUL36 protein (protein pUL36 neddylation in this context, despite the prior finding encoded by the 36th open reading frame in the unique-long seg- that the pUL36 DUB can cleave NEDD8 from substrates to ment of the viral genome), amounting to less than 10% of the total promote viral DNA replication (26). mass of this large (>250 kDa) structural component of the virion (9). pUL36 is attached directly to the icosahedral capsid shell of these enveloped viruses (10–12) and, due to its location between the Significance capsid and envelope, is referred to as a herpesvirus tegument pro- tein. Upon virus entry into a cell, pUL36 becomes exposed to the The neuroinvasive herpesviruses cause a wide range of diseases in cytosol (4, 13, 14) and directs delivery of the capsid and its DNA humans and livestock. These viruses have the unusual ability to content to the nucleus (2, 15–19). Following replication, pUL36 is routinely enter the mammalian nervous system following infection also essential for viral assembly and egress (20–22). of mucous membranes. Although infection is often asymptomatic, ThepresenceofthepUL36DUBontheexposedsurfaceofthe the consequences of the resulting lifelong nervous system in- intracellular viral particle suggests that removal of ubiquitin from fections are not fully understood. This study uncovers two cellular cytosolic proteins, or from the virus particle itself, is instru- opposing nervous system protective barriers that the virus mental for these viruses to transmit between peripheral tissues and defeats by changing its properties during the course of in- the neurons that innervate them (8). This necessarily implicates the fection. The molecular switch governing this change underlies prior addition of ubiquitin as a factor in disease progression, either as the herpesvirus neuroinvasive phenotype. part of the virus neuroinvasive program or as a cellular defense Author contributions: N.J.H., P.J.S., A.L.R., G.E.P., and G.A.S. designed research; N.J.H., mechanism that is overcome by the DUB. Although several sub- P.J.S., and A.L.R. performed research; N.J.H. produced and characterized mutant viruses, strates of the DUB have been identified, how this enzyme promotes performed immunoprecipitations and nuclear delivery assays, prepared mass spectrome- the potent neuroinvasive property of these viruses is unknown try samples, and analyzed mass spectrometry results; A.L.R. performed and analyzed viral – particle retrograde transport experiments; P.J.S. and G.E.P. performed and analyzed (23 30). To elucidate the role of the DUB, we set out to examine mouse and rat infection models; N.J.H. and A.L.R. contributed new reagents/analytic the role of ubiquitination during neuroinvasive herpesvirus in- tools; N.J.H., P.J.S., A.L.R., G.E.P., and G.A.S. analyzed data; and N.J.H., P.J.S., G.E.P., and fection. We report that ubiquitin addition and removal from a G.A.S. wrote the paper. conserved lysine residue in pUL36 switches PRV between two The authors declare no conflict of interest. invasive states that act successively to promote virus dissemina- This article is a PNAS Direct Submission. tion into the peripheral nervous system. Supporting evidence 1To whom correspondence should be addressed. Email: [email protected]. “ ” that the ubiquitin switch is conserved in HSV-1, and possibly This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. all neuroinvasive alphaherpesviruses, is provided. 1073/pnas.1512559112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1512559112 PNAS Early Edition | 1of6 Downloaded by guest on October 2, 2021 WT, K1613R, or genetically repaired K442 virus, capsids of the + A K442R mutant displayed aberrant motion in the form of frequent Axon stops and misdirected (anterograde) transport that together im- + - - + paired the long-distance delivery of capsids to the cell body (Fig. 2D + - - -- - Nuc and Movie S1). No corresponding defect was observed during ep- fluorescent extracellular virion + ithelial infection (Fig. S1). The reduction in K442R capsid targeting fluorescent intracellular capsid + to the neural soma phenocopied broad inhibition of ubiquitination +-microtubule during WT infection, implicating ubiquitin addition to pUL36 res- Soma B E B DMSO PYR41 MLN4924 D idue K442 as instrumental to neuronal infection (Figs. 1 and 2 ). The HSV-1 and PRV pUL36 DUBs Regulate the Ubiquitin State of WT Δ DUB mock DIC kDa Several Viral Proteins, Including pUL36. Mass spectrometry was used to determine whether PRV pUL36 K442 was, in fact, ubiq- 250 150 uitinated during infection. The analysis measured the amount of

pUL36 IP: GFP (pUL36) ubiquitin-like proteins (either ubiquitin, NEDD8, or ISG15) pre-

capsids sent on virus proteins isolated from HeLa cells infected for 8 h with either HSV-1 or PRV, each encoding either a wild-type DUB (WT) n.s. 125 250 or a DUB catalytic mutant (ΔDUB) (Fig. S2). Modifications were *** 100 150 detected as posttryptic di-glycine conjugates on lysine side chains Ub Δ 75 (33). A ratio of DUB:WT modification was determined for each detected virus tryptic peptide, with values greater than 1 indicating 50 250 the peptide contained a lysine that was modified more frequently in 25 150 the absence of DUB activity during infection (Dataset S1). A NEDD8 0 summary score of combined peptide ratios for individual virus est. capsids / nuclear rim DMSO PYR41 MLN4924 proteins was calculated, and a Pearson’s correlation analysis in- dicated a strong positive relationship between the two viruses (r = C 100 n.s. n.s. 0.87; Fig. 3A). The protein with the highest ΔDUB:WT value was

75 250 input pUL36 itself, which had three targeted lysines in PRV: K442, K963, pUL36 and K1453 (Fig. 3B). Although the modifications identified by mass 50 spectrometry could be from either ubiquitin, NEDD8, or ISG15, posttranslational modification at these pUL36 sites was consistent 25 150 p150 with ubiquitin given that the DUB is not active against ISG15 (6, 0 26), and NEDD8 was not detected as a pUL36 conjugate during est. capsids / nuclear rim DMSO PYR41 MLN4924 infection with the ΔDUB virus (Fig. 1D). K963 (K1294 in HSV-1) was not further examined as it was dispensable during PRV in- Fig. 1. Ubiquitination promotes capsid delivery to neuronal nuclei. (A)Diagram fection (Fig. 2 B and C) and was not enriched in the HSV-1 sample of capsid retrograde axonal transport and nuclear rim delivery. (B) Representative C images of capsids (UL25/mCherry tag) at nuclear rims of cultured DRG (Top)with (Fig. 3 ). K1453 (K1789 in HSV-1) was also not pursued, as al- estimates of capsid numbers at nuclear rims at 3–4 hpi (Bottom) following pre- though ubiquitination at this site was enriched in the ΔDUB sample treatment with DMSO (vehicle), PYR41 (10 μM), or MLN4924 (1 μM). Mean ± SD during HSV-1 infection (Fig. 3C), the lysine is not conserved in are shown (***P < 0.001 based on Tukey’s multiple-comparison test). (Scale bar: other neuroinvasive herpesviruses including HSV-2, which notably 10 μm.) (C) Estimates of capsid numbers at nuclear rims at 3–4 hpi following in- encodes arginine at the equivalent position (Fig. S3). Only K442 fection of PK15 epithelial cells. (D) pUL36 was immunoprecipitated from PK15 cells (K636 in HSV-1) is positionally conserved across the neuroinvasive either mock-infected or infected with PRV expressing a GFP-pUL36 fusion at 8 hpi alphaherpesviruses, was ubiquitinated during both HSV-1 and PRV and probed for GFP-pUL36, ubiquitin (Ub), or NEDD8. Input lysates were probed infection, and was hyperubiquitinated in HSV-1 and PRV in the for GFP-pUL36 and a cellular dynactin component (p150) as loading control. absenceofanactivepUL36DUB(Fig.3andFig. S3). These results are consistent with the noted requirement for an active ubiquiti- A Conserved Lysine in pUL36 Is a Critical Virulence Determinant. nation pathway during neuronal infection, with the critical substrate Mutational analysis was used to initially examine whether pUL36 mapping to K442 of PRV pUL36. ubiquitination promotes infection. Four lysines, the site of ubiq- uitin addition, that are invariant across the pUL36 homologs of Lysine 442 Is Necessary for Retrograde Invasion and Spread Within seven neuroinvasive herpesviruses were individually mutated to the Nervous System. To determine whether the defects associated with the K442R mutation translated to reduced neuroinvasion in arginine to prevent potential ubiquitination while maintaining the a mammalian host, variants of PRV encoding an mCherry-capsid positive charge of the amino acid side chain (Fig. 2A). The four fusion that serves as a fluorescent reporter were injected in- recombinant PRV strains propagated efficiently in epithelial cells, traocularly into the aqueous humor of the rat eye anterior but two (K442R and K1613R) exhibited moderate reductions in A B chamber (AC) (Fig. 4 ). Although not a natural route of infection, spread (Fig. 2 ). In contrast, defects associated with the K442R the AC model confines the initial infection to intraocular tissues and K1613R mutations were profound during infection of mice that are bathed in the aqueous humor, including the iris, pro- C (Fig. 2 ). K442R-infected animals did not present symptoms gressing to a well-defined invasion of the peripheral nervous system during the course of the experiment, which was terminated at 336 h (PNS) by retrograde axonal transport (34). This model was pre- postinfection (hpi). The K1613R mutant was severely attenuated viously used to document the neuroinvasive defect of PRV relative to the wild type (WT), K963R, and K1299R viruses. Ge- encoding a catalytically inactive DUB (8). At 44 hpi, both the netic repair of the K442R and K1613R mutations restored virus ΔDUB and K442R viruses had spread within the iris, approaching spread and virulence to WT levels (Fig. 2 B and C). but not equivalent to WT infection levels (Fig. 4B). Neuroinvasion To examine whether the virulence defect of the K442R mutant was monitored by imaging the innervating ipsilateral superior correlated with a reduced capacity to infect neurons, capsid ret- cervical ganglion (SCG) at 48 hpi, which was sufficient time for WT rograde axonal transport in DRG cultured ex vivo was monitored virus to infect this autonomic ganglion of the PNS (Fig. 4C). using live-cell fluorescence microscopy (Fig. 2D). In contrast to Spread to SCG neurons only occurred following PRV replication

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1512559112 Huffmaster et al. Downloaded by guest on October 2, 2021 A

B C

Fig. 2. A conserved lysine in pUL36 is a virulence determinant that sustains retrograde axonal trans- D port. (A) Illustration of PRV pUL36. Positions of lysine residues are noted with vertical bars, and the DUB is shown in light gray with the position of the active-site cysteine indicated. Four lysines conserved across seven neuroinvasive herpesviruses are shown with local se-

quence alignments (GenBank JF797219, AJ004801, MICROBIOLOGY AY665713, NC_001348, NC_004812, GU734771, and Z86099). (B) Spread of viruses in PK15 epithelial cells (n = 150 per sample; mean ± SD shown in red; ***P < 0.001 based on Tukey’s multiple-comparison test; rep, genetic repair). (C) Kaplan–Meier presentation of mouse survival following intranasal instillation of 106 E plaque-forming units (PFU) of PRV (n = 10 for K442R and K1613R; n = 5 for all others). (D) Frequency of aberrant (nonretrograde) PRV axonal transport in cultured DRG sensory neurons 30–60 min postinfection (three independent experiments; n > 100 viral particles per sample; mean values ± SEM; *P < 0.05, two-tailed unpaired t test). (E) The number of fluorescent capsids at nuclear rims of cultured DRG (Left, representative images) was estimated (Right)at3–4 hpi. (Scale bar: 10 μm.)

in innervated tissues of the eye, as a defective PRV mutant (ΔUL25) also remained neurotropic and spread within the SC, but did not that cannot spread cell to cell produced negligible infection of the transmit retrogradely within the axons of the optic nerve to the retina SCG, indicating that axon terminals in the eye are not exposed to the (Fig. 4E). The inability of the K442R mutant to transmit retrogradely aqueous humor (Fig. 4C)(35).TheΔDUB virus was incapable of from AC to SCG (indirect neuronal exposure) and from CNS to transmission to the SCG despite productive replication in the iris, retina (direct neuronal exposure) was consistent with the noted de- which is consistent with the prior observation that DUB activity is fect in retrograde axonal transport (Fig. 2 D and E) and distin- required for PRV spread from the AC to the brain, and demon- guished this neuroinvasive effector function from that of the DUB. strates that this established CNS neuroinvasive defect is due to an upstream loss of PNS neuroinvasion (7, 8). The K442R virus phe- Discussion nocopied the DUB mutant demonstrating that ubiquitination, in Neurotropic herpesviruses such as HSV-1 and PRV exhibit pro- addition to deubiquitination, at K442 is critical for invasion of the found, yet subtle, neuroinvasive properties. Infections typically PNS. Intraocular AC injection of repaired viruses resulted in in- begin in exposed mucosal tissue, with newly replicated viral parti- fection of the SCG that was indistinguishable from the WT (Fig. 4C). cles rapidly transmitting within axons of innervating neurons and To determine whether the loss of neuroinvasion with the K442R resulting in the establishment of lifelong latent infections in sensory mutant was due to a defect in epithelial-neuron transmission, virus and autonomic ganglia of the PNS. This feat is remarkable given that inoculum was injected directly into the superior colliculus (SC) of most unrelated viral infections occur one cell junction away from the brain, which receives axonal projections from retinal ganglion peripheral nerves, yet few viruses invade the nervous system. Of cells (RGCs) in the eye (Fig. 4D). In the absence of an epithelial those that do, neuroinvasion is infrequent (i.e., poliovirus, measles barrier, the ΔDUB virus replicated and spread within the SC, virus, mumps virus, West Nile virus, Japanese encephalitic virus, and reinforcing that DUB activity is dispensable for neurotropism but is Chandipura virus) or occurs following an animal bite (i.e., rabies critical for neuroinvasion (Fig. 4E)(8).Furthermore,theDUB virus). In the case of the latter, mechanical tissue disruption at the mutant spread from the SC to the retina, confirming that the virus site of inoculation bypasses intrinsic protective barriers in the skin was competent to travel retrogradely in axons. The K442R mutant and increases retrograde transport in damaged axons as part of a

Huffmaster et al. PNAS Early Edition | 3of6 Downloaded by guest on October 2, 2021 peripheral nerve repair response (36, 37). The concept that tissue 15 A trauma can increase the susceptibility of the nervous system to neurotropic viruses is best documented for poliovirus, which is rarely 10 pUL36 neuroinvasive except when provoked by tissue injury (38–40). The near impenetrability of the nervous system in healthy tissue indicates that effective intrinsic barriers to infection exist between peripheral 5 tissues and the PNS. However, these barriers are seemingly in- r = 0.87 effective against the neuroinvasive herpesviruses. We report that PRV repurposes the cellular ubiquitin pathway as di-Gly fold changes by -10 -5 5 10 15 20 HSV-1 protein ( Δ DUB/WT) a means to switch the virus between two invasive states. Each state is capable of overcoming an intrinsic barrier that normally protects -5 the PNS from infection, but neither state alone is sufficient to Sum of di-Gly fold changes by PRV protein (ΔDUB/WT) B promote neuroinvasion (Fig. 5). A combination of genetic analysis and mass spectrometry data pinpointed K442 in the capsid-bound EP0 (2) UL21 (4) UL42 (3) * * 77 tegument protein, pUL36, as the relevant site of ubiquitination and UL2 (1) UL23 (3) UL46 (6) pUL36 inset UL3 (1) UL5 (3) UL24 (1) UL47 (1) 963 deubiquitination that modulates the virus state. PRV, which is UL6 (5) UL27 (13) UL50 (7) highly neurovirulent in rodent infection models, became avirulent UL4 (1) UL7 (1) UL28 (1) UL52 (3) 1453 UL8 (4) UL30 (4) UL54 (1) and incapable of PNS invasion when encoding a single K442R co- UL40 (3) UL10 (2) UL31 (1) US1 (1) lysine position 442 don change that prevented ubiquitination at this site. By capitalizing UL11 (1) UL32 (2) US2 (2) 0246810 UL34 (2) UL13 (1) UL33 (1) US3 (2) on a well-defined PRV ocular infection model, the opposing phe- UL14 (1) UL35 (2) US4 (2) di-Gly fold changes by peptide Δ notypes of the ubiquitin switch became apparent: addition and re- UL29 (12) UL15 (5) UL37 (4) US6 (3) ( DUB/WT) UL16 (1) UL38 (1) US7 (2) moval of ubiquitin at K442 was consistent with alternating the virus UL49 (3) UL19 (13) UL39 (10) US8 (1) between an axon-targeting mode within peripheral tissues (DUB ICP4 (5) dependent) and a sustained retrograde microtubule trafficking state

UL36 (9) following entry into axons (ubiquitination/K442 dependent). The absolute requirement of the ubiquitin switch in co- 02468101214161820 ordinating the larger neuroinvasive phenotype indicates that the Sum of di-Gly fold changes by PRV protein (ΔDUB/WT) C intrinsic barriers protecting nerve endings in peripheral tissues * are diametrically opposed, and we suggest that this opposition UL2 (3) UL1 (1) UL51 (2) hinders the evolution of neuroinvasive properties, as virus ad- UL17 (3) * UL4 (2) UL53 (5) UL30 (9) aptations that overcome one barrier will be detrimental to the UL5 (1) UL54 (5) UL50 (4) UL7 (1) US1 (3) ability of a virus to overcome the opposing barrier. Given the UL27 (12) UL14 (2) US2 (2) UL10 (4) role of pUL36 in microtubule-based transport, it seems likely UL15 (1) US6 (3) UL55 (2) UL28 (1) US7 (3) that microtubule trafficking may be the source of the barrier UL34 (1) UL31 (6) US8 (1) UL3 (3) opposition (2). Preventing ubiquitination at K442 moderately UL32 (1) US10 (2) UL9 (2) UL33 (1) US11 (1) reduced viral spread in nonneuronal cells based on plaque assay, UL13 (4) UL44 (2) US12 (3) UL22 (3) but severely impaired microtubule-based retrograde delivery of UL45 (1) UL26 (4) PRV in axons in explanted sensory ganglia and eliminated ret- UL38 (2) rograde spread in vivo. The enhancement of retrograde micro- UL41 (6) UL21 (4) tubule transport that benefits PRV axonal transport is likely ICP0 (3) detrimental for spread from epithelia to nerve endings, as UL47 (5) 186 pUL36 inset Δ UL12 (7) indicated by the DUB phenotype, possibly by misdirecting vi- US3 (4) 2213 rions from sites of epithelial/axonal contact. The benefit afforded UL24 (11) 636 ICP4 (3) by the DUB in epithelial cells, why the DUB activity is not UL48 (3) 595 detrimental within axons, and identifying the E3 ubiquitin ligase

UL19 (19) lysine position UL42 (8) 1789 that targets pUL36 are important topics for future investigation. UL40 (1) 0246 Because the DUB and K442 are conserved elements of pUL36 UL49 (7) across the neuroinvasive herpesviruses, the robust neuroinvasion UL37 (6) di-Gly fold changes by peptide Δ UL25 (2) ( DUB/WT) exhibited by these Alphaherpesvirinae subfamily members are likely UL6 (9) based upon the ubiquitin switch described here with PRV. In par- UL29 (16) UL46 (5) ticular, the comparative mass spectrometry analysis of HSV-1 and UL39 (26) PRV ubiquitination included in this report indicates that the con- UL52 (10) UL36 (5) served K442 site (K636 in HSV-1) is ubiquitinated and serves as a

0 2 4 6 8 10 12 DUB substrate for both viruses. Thus, the ubiquitin switch provides Sum of di-Gly fold changes by HSV-1 protein (ΔDUB/WT) a framework for understanding how a subset of herpesviruses are capable of routinely invading the mammalian nervous system. Fig. 3. Ranking of ubiquitinated viral proteins as pUL36 DUB substrates. (A) Comparison of PRV and HSV-1 mass spectrometry datasets as detailed for Materials and Methods PRV (B) and HSV-1 (C). Tryptic peptides containing lysine di-glycine conju- Cells and Virus Strains. Vero and PK15 epithelial cell lines were maintained gates from lysates of HeLa cells infected with WT or DUB-inactivated PRV or in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% (vol/vol) HSV-1 were immunoprecipitated and examined by mass spectrometry. For each WT and ΔDUB pair, a peptide ratio (ΔDUB/WT) was determined and a value of 1 was subtracted. A summary score for each viral protein was then calculated by summing its peptide ratios. Each point on the graph represents each corresponding protein and plotted in rank order from poorest DUB one viral protein summary score (A), with greater values indicating the substrate (Top) to greatest (Bottom). Black bars indicate conserved sub- tendency of the protein to be modified in the absence of DUB activity. Only strates in PRV and HSV-1. *Proteins with summed ratios < 1 (i.e., modified proteins that had modified lysines in both PRV and HSV-1 were plotted. but not DUB substrates). The number of di-glycine modified lysines detected A Pearson correlation analysis indicated a positive relationship between the per protein is indicated in parentheses. Insets show pUL36 peptides having a two viruses (r = 0.87). (B and C) Positive peptide ratios were summed for ratio > 1 (dotted vertical lines, significance threshold of 1.5).

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1512559112 Huffmaster et al. Downloaded by guest on October 2, 2021 A D Ocular neuroinvasion model Intracranial infection model AC (eye) SCG (PNS) SC (CNS) RGC (eye) injection site 2˚ replication injection site 2˚ replication

optic iris nerve

cornea superior anterior optic cervical superior chamber nerve colliculus retinal lens retina ganglion ganglion ciliary cells body vitreous humor B iris - 44 hpi C SCG - 48 hpi E SC - 30 hpi RGC - 48 hpi TDUB (rep) WT WT Δ DUB Δ UL25 Δ DUB Δ DUB K442R K442 (rep) K442R K442R K442R (rep)

infected cells infected cells infected RGCs virus 44 hpi virus 48 hpi 120 hpi virus 30 hpi 48 hpi WT +++ ΔUL25 1.5 ± 2.1 ND ΔDUB 0 > 500 MICROBIOLOGY ΔDUB ++ ΔDUB 2.7 ± 3.1 ND K442R 0 1.0 ± 0.6 K442R ++ K442R 0 0 K442 (rep) > 500 ND WT > 5,000 ND DUB (rep) > 5,000 ND K442 (rep) > 5,000 ND

Fig. 4. The herpes DUB and lysine substrate promote serial stages of neuroinvasion. (A and D) Two models of PRV retrograde axonal transport examined in this study (AC, anterior chamber; RGC, retinal ganglion cells; SC, superior colliculus; SCG, superior cervical ganglion). (A) Nerve endings are shielded by ocular tissue (iris) and not directly exposed to viral inoculum. (B and C) Representative images of the iris (B)andSCG(C) following intraocular injection of PRV strains encoding a mCherry-capsid reporter into the AC and retrograde transport to the SCG, with amount of infection based on fluorescent cell number indicated below. (D) Neurons and axons are directly exposed to viral inoculum. (E) Representative images and quantitation following intracranial injection into the SC and retrograde transmission to RGC of the contralateral eye. Microelectrode tracks are outlined in SC injection images to distinguish inoculum from viral spread. (Scale bars: 50 μm.)

bovine growth serum. HeLa cells were freshly obtained from the ATCC and the pYEbac102 BAC (42) were transfected into Vero-Cre cells as previously de- maintained in DMEM supplemented with 10% (vol/vol) FBS. Bacterial artificial scribed (4). The resultant virus stocks were passaged once to create working stocks. chromosome (BAC) infectious clones derived from PRV pBecker3 were transfected Titers and single-step propagation kinetics were determined by plaque assays on into PK15 cells as described previously (41). HSV-1 infectious clones derived from PK15 (PRV) or Vero (HSV-1) cells as described previously (4, 43) (Tables S1 and S2).

State 1 State 2 Transmission to nerve terminals Persistent retrograde axonal transport (deubiquitinase is essential) (ubiquitin addition is essential) E1 Ub Inhibited by catalytic inactivation of the pUL36 DUB Ub Ub nucleus

Ub Inhibited by E1 antagonist or mutation of the pUL36 K442 ubiquitin acceptor site

Transmission within epithelium is state independent

Fig. 5. Ubiquitin switch model of herpesvirus neuroinvasion. The large tegument protein (pUL36) encodes a deubiquitinase (DUB) activity and a ubiquitin addition site (K442 in PRV), both of which are essential for virus invasion of the nervous system from peripheral tissues. The DUB and K442 acceptor site together constitute a molecular switch that alternates the virus between an axon targeting mode (state 1), and a sustained retrograde transport mode (state 2) that is required for the long-distance delivery of viral particles within nerve fibers to the peripheral ganglia.

Huffmaster et al. PNAS Early Edition | 5of6 Downloaded by guest on October 2, 2021 Immunoprecipitation and Western Blotting. PK15 cells were mock infected or Mass Spectrometry (UbiScan). Approximately 2.0 × 108 HeLa cells were infected infected at a multiplicity of infection (MOI) of 50 with recombinant viruses at MOI of 50 with either WT or DUB mutant PRV or HSV-1 and collected at 8 hpi encoding GFP-pUL36 fusions. Lysates were prepared at 8 hpi as described pre- for label-free UbiScan analysis (33). Detailed methods are in SI Materials and viously (44). More details can be found in SI Materials and Methods. Methods. Raw and fold-change sorted data are combined in Excel format in Dataset S1. Live-Cell Fluorescence Microscopy. Imaging of axonal transport and capsid accumulation at neuronal nuclei was performed using primary cultured neurons ACKNOWLEDGMENTS. We thank Jonathan Leis and Gina Daniel for helpful from embryonic chicken [embryonic day 8 (E8) to E10] DRGs, as previously de- discussions of the data and approach, Jenifer Klabis for assistance with scribed (13). See SI Materials and Methods for further details. recombinant virus production, Riley Junell and Anne Fischer for technical assistance, and Jeffrey Silva and Charles Farnsworth from Cell Signaling Technology for mass spectrometry. Sequencing services were performed at In Vivo Inoculation of Virus. All procedures conformed to NIH guidelines for the Northwestern University Genomics Core Facility. This work was funded by work with laboratory animals and were approved by the Institutional Animal NIH Grants R01 AI056346 (to G.A.S.) and R01 NS077003 (to G.E.P. and P.J.S.). Care and Use Committee of the University of Nebraska, Lincoln. See SI Materials N.J.H. received support from Viral Replication Training Grant T32AI060523 and Methods for further details. from the NIH.

1. Koyuncu OO, Song R, Greco TM, Cristea IM, Enquist LW (2015) The number of al- 23. Bolstad M, Abaitua F, Crump CM, O’Hare P (2011) Autocatalytic activity of the phaherpesvirus particles infecting axons and the axonal protein repertoire de- ubiquitin-specific protease domain of herpes simplex virus 1 VP1-2. J Virol 85(17): termines the outcome of neuronal infection. MBio 6(2):e00276-15. 8738–8751. 2. Zaichick SV, et al. (2013) The herpesvirus VP1/2 protein is an effector of dynein-mediated 24. Wang S, Wang K, Li J, Zheng C (2013) Herpes simplex virus 1 ubiquitin-specific pro- capsid transport and neuroinvasion. Cell Host Microbe 13(2):193–203. tease UL36 inhibits beta interferon production by deubiquitinating TRAF3. J Virol 3. Koyuncu OO, Perlman DH, Enquist LW (2013) Efficient retrograde transport of 87(21):11851–11860. pseudorabies virus within neurons requires local protein synthesis in axons. Cell Host 25. Saito S, et al. (2013) Epstein-Barr virus deubiquitinase down-regulates TRAF6-medi- Microbe 13(1):54–66. ated NF-κB signaling during productive replication. J Virol 87(7):4060–4070. 4. Antinone SE, Smith GA (2010) Retrograde axon transport of herpes simplex virus and 26. Gastaldello S, et al. (2010) A deneddylase encoded by Epstein-Barr virus promotes pseudorabies virus: A live-cell comparative analysis. J Virol 84(3):1504–1512. viral DNA replication by regulating the activity of cullin-RING ligases. Nat Cell Biol 5. Aggarwal A, et al. (2012) Ultrastructural visualization of individual tegument protein 12(4):351–361. dissociation during entry of herpes simplex virus 1 into human and rat dorsal root 27. Whitehurst CB, et al. (2009) The Epstein-Barr virus (EBV) deubiquitinating enzyme ganglion neurons. J Virol 86(11):6123–6137. BPLF1 reduces EBV ribonucleotide reductase activity. J Virol 83(9):4345–4353. 6. Kattenhorn LM, Korbel GA, Kessler BM, Spooner E, Ploegh HL (2005) A deubiquiti- 28. Inn KS, et al. (2011) Inhibition of RIG-I-mediated signaling by Kaposi’s sarcoma-as- nating enzyme encoded by HSV-1 belongs to a family of cysteine proteases that is sociated herpesvirus-encoded deubiquitinase ORF64. J Virol 85(20):10899–10904. conserved across the family Herpesviridae. Mol Cell 19(4):547–557. 29. Whitehurst CB, Vaziri C, Shackelford J, Pagano JS (2012) Epstein-Barr virus BPLF1 7. Böttcher S, et al. (2008) Mutagenesis of the active-site cysteine in the ubiquitin-spe- deubiquitinates PCNA and attenuates polymerase η recruitment to DNA damage cific protease contained in large tegument protein pUL36 of pseudorabies virus im- sites. J Virol 86(15):8097–8106. pairs viral replication in vitro and neuroinvasion in vivo. J Virol 82(12):6009–6016. 30. van Gent M, et al. (2014) Epstein-Barr virus large tegument protein BPLF1 contributes 8. Lee JI, et al. (2009) A herpesvirus encoded deubiquitinase is a novel neuroinvasive to innate immune evasion through interference with toll-like receptor signaling. PLoS determinant. PLoS Pathog 5(4):e1000387. Pathog 10(2):e1003960. 9. Schlieker C, Korbel GA, Kattenhorn LM, Ploegh HL (2005) A deubiquitinating activity 31. Yang Y, et al. (2007) Inhibitors of ubiquitin-activating enzyme (E1), a new class of is conserved in the large tegument protein of the herpesviridae. J Virol 79(24): potential cancer therapeutics. Cancer Res 67(19):9472–9481. 15582–15585. 32. Soucy TA, et al. (2009) An inhibitor of NEDD8-activating enzyme as a new approach to 10. Coller KE, Lee JI, Ueda A, Smith GA (2007) The capsid and tegument of the alpha- treat cancer. Nature 458(7239):732–736. herpesviruses are linked by an interaction between the UL25 and VP1/2 proteins. 33. Kim W, et al. (2011) Systematic and quantitative assessment of the ubiquitin-modified J Virol 81(21):11790–11797. proteome. Mol Cell 44(2):325–340. 11. Cardone G, et al. (2012) The UL36 tegument protein of herpes simplex virus 1 has a 34. Strack AM, Loewy AD (1990) Pseudorabies virus: A highly specific transneuronal cell composite binding site at the capsid vertices. J Virol 86(8):4058–4064. body marker in the sympathetic nervous system. J Neurosci 10(7):2139–2147. 12. Fan WH, et al. (2015) The large tegument protein pUL36 is essential for formation of 35. McNab AR, et al. (1998) The product of the herpes simplex virus type 1 UL25 gene is the capsid vertex-specific component at the capsid-tegument interface of herpes required for encapsidation but not for cleavage of replicated viral DNA. J Virol 72(2): simplex virus 1. J Virol 89(3):1502–1511. 1060–1070. 13. Luxton GW, et al. (2005) Targeting of herpesvirus capsid transport in axons is coupled 36. Hanz S, et al. (2003) Axoplasmic importins enable retrograde injury signaling in le- to association with specific sets of tegument proteins. Proc Natl Acad Sci USA 102(16): sioned nerve. Neuron 40(6):1095–1104. 5832–5837. 37. Curtis R, et al. (1993) Retrograde axonal transport of ciliary neurotrophic factor is 14. Granzow H, Klupp BG, Mettenleiter TC (2005) Entry of pseudorabies virus: An im- increased by peripheral nerve injury. Nature 365(6443):253–255. munogold-labeling study. J Virol 79(5):3200–3205. 38. Lancaster KZ, Pfeiffer JK (2010) Limited trafficking of a neurotropic virus through 15. Knipe DM, Batterson W, Nosal C, Roizman B, Buchan A (1981) Molecular genetics of inefficient retrograde axonal transport and the type I interferon response. PLoS herpes simplex virus. VI. Characterization of a temperature-sensitive mutant defective Pathog 6(3):e1000791. in the expression of all early viral gene products. J Virol 38(2):539–547. 39. Gromeier M, Wimmer E (1998) Mechanism of injury-provoked poliomyelitis. J Virol 16. Batterson W, Furlong D, Roizman B (1983) Molecular genetics of herpes simplex virus. 72(6):5056–5060. VIII. Further characterization of a temperature-sensitive mutant defective in release 40. Strebel PM, Ion-Nedelcu N, Baughman AL, Sutter RW, Cochi SL (1995) Intramuscular of viral DNA and in other stages of the viral reproductive cycle. J Virol 45(1):397–407. injections within 30 days of immunization with oral poliovirus vaccine—a risk factor 17. Jovasevic V, Liang L, Roizman B (2008) Proteolytic cleavage of VP1-2 is required for for vaccine-associated paralytic poliomyelitis. N Engl J Med 332(8):500–506. release of herpes simplex virus 1 DNA into the nucleus. J Virol 82(7):3311–3319. 41. Smith GA, Enquist LW (2000) A self-recombining bacterial artificial chromosome and 18. Abaitua F, Hollinshead M, Bolstad M, Crump CM, O’Hare P (2012) A Nuclear locali- its application for analysis of herpesvirus pathogenesis. Proc Natl Acad Sci USA 97(9): zation signal in herpesvirus protein VP1-2 is essential for infection via capsid routing 4873–4878. to the nuclear pore. J Virol 86(17):8998–9014. 42. Tanaka M, Kagawa H, Yamanashi Y, Sata T, Kawaguchi Y (2003) Construction of an 19. Schipke J, et al. (2012) The C-terminus of the large tegument protein pUL36 contains excisable bacterial artificial chromosome containing a full-length infectious clone of multiple capsid binding sites that function differently during assembly and cell entry herpes simplex virus type 1: Viruses reconstituted from the clone exhibit wild-type of herpes simplex virus. J Virol 86(7):3682–3700. properties in vitro and in vivo. J Virol 77(2):1382–1391. 20. Desai PJ (2000) A null mutation in the UL36 gene of herpes simplex virus type 1 results 43. Bohannon KP, Sollars PJ, Pickard GE, Smith GA (2012) Fusion of a fluorescent protein in accumulation of unenveloped DNA-filled capsids in the cytoplasm of infected cells. to the pUL25 minor capsid protein of pseudorabies virus allows live-cell capsid im- J Virol 74(24):11608–11618. aging with negligible impact on infection. J Gen Virol 93(Pt 1):124–129. 21. Fuchs W, Klupp BG, Granzow H, Mettenleiter TC (2004) Essential function of the 44. Hwang J, Kalejta RF (2011) In vivo analysis of protein sumoylation induced by a viral pseudorabies virus UL36 gene product is independent of its interaction with the UL37 protein: Detection of HCMV pp71-induced Daxx sumoylation. Methods 55(2): protein. J Virol 78(21):11879–11889. 160–165. 22. Luxton GW, Lee JI, Haverlock-Moyns S, Schober JM, Smith GA (2006) The pseudora- 45. Leelawong M, Lee JI, Smith GA (2012) Nuclear egress of pseudorabies virus capsids is bies virus VP1/2 tegument protein is required for intracellular capsid transport. J Virol enhanced by a subspecies of the large tegument protein that is lost upon cytoplasmic 80(1):201–209. maturation. J Virol 86(11):6303–6314.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1512559112 Huffmaster et al. Downloaded by guest on October 2, 2021