Alphaherpesvirus pUL21 homologues use non-canonical motifs to compete with cellular adaptors for protein phosphatase 1 binding

14 VZV , RIPPO, PIP , regulatory subunit, PPP1R15A 15 ORCIDs: 16 THB 0000-0001-6420-3665; JED 0000-0002-4863-0330; SCG 0000-0003-4547-4034 17 Running title: 18 Herpes competes for PP1 binding using novel motifs 19 Contributions: 20 Conceptualization: SCG; Data Curation: HM, DSC, SCG; Funding Acquisition: JED, SCG; 21 Investigation: HM, DSC, THB, OST, SCG; Project Administration: SCG, JED; Supervision: SCG; 22 Visualization: HM, DSC, SCG; Writing – Original Draft Preparation: SCG; Writing – Review & 23 Editing: HM, DSC, THB, JED, SCG. 24 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 2

25 Protein phosphatase 1 (PP1) is a key regulator of cellular phosphorylation and its activity is 26 regulated via binding to cellular regulatory proteins via conserved short linear motifs (SLiMs). 27 The herpes simplex virus (HSV)-1 protein pUL21 binds PP1 via the TROPPO motif, which lacks 28 sequence similarity to canonical PP1-binding SLiMs. Here, we combine structure prediction, 29 mutagenesis, and biophysical assays to elucidate the molecular basis of this interaction. 30 AlphaFold2-Multimer structural models suggest that the TROPPO motifs of pUL21 and of 31 pORF38, the varicella-zoster virus homologue of pUL21, bind the same hydrophobic groove on 32 PP1 as RVxF and ϕϕ[xF] motifs, forming an extended β-sheet that bridges PP1 and the N-33 terminal domain of pUL21 or pORF38. Site-directed mutagenesis of both pUL21 and PP1 34 confirms key predicted interactions. Competition fluorescence polarisation confirms that 35 pUL21 competes directly with cellular PP1 RvXF motifs for PP1 binding, albeit with lower affinity. 36 Substituting key residues of the pUL21 TROPPO motif to resemble a canonical RVxF sequence 37 increases PP1 binding, suggesting that alphaherpesviruses have evolved suboptimal motifs to 38 fine-tune phosphatase recruitment that may balance viral kinase and phosphatase activities 39 during infection. Our findings reveal a novel mechanism of PP1 recruitment by viral proteins and 40 suggest that many other PP1 regulators may possess non-canonical binding motifs and thus 41 remain undiscovered. 42

43 Phosphorylation is critical to the control of most cellular processes. Viruses have learnt to 44 exploit this, encoding kinases that regulate cell cycle progression, modify gene expression, 45 overcome innate immune restriction and prevent apoptosis (1, 2). Similarly, multiple viruses 46 modulate the dephosphorylation of both cellular and viral target proteins within infected cells. 47 Herpes simplex virus (HSV)-1 protein ICP34.5 (a.k.a. γ134.5) counteracts the antiviral innate 48 immune response by stimulating dephosphorylation of eIF2α to relieve host transcriptional 49 shutdown (3, 4). The HIV Tat protein interacts with the cellular enzyme protein phosphatase 1 50 (PP1) to promote viral gene transcription (5, 6). Additionally, the measles virus V protein 51 suppresses innate immune signalling by sequestering PP1, thereby preventing 52 dephosphorylation of the cytosolic RNA sensor MDA5 that is required for its activation (7). Given 53 the multitude of cellular signalling pathways that are regulated by protein phosphorylation, 54 interfering with virus-mediated protein dephosphorylation represents a promising new avenue 55 for the development of potent antiviral therapies (6, 8). 56 PP1 is a highly abundant cellular serine/threonine phosphatase. PP1 has low intrinsic 57 specificity and it recruits its substrates via regulatory proteins (9, 10), also known as adaptor 58 proteins, PP1 Interacting Proteins (PIPs) or Regulators of Protein Phosphatase One (RIPPOs) 59 (11). These regulatory proteins utilise small linear motifs (SLiMs) to bind different surfaces of 60 PP1 (9, 10). The most abundant and well characterised of these is the RVxF motif, which binds 61 to a hydrophobic groove on the surface of PP1. Over 90% of cellular PP1 adaptors have an RVxF 62 motif (12), which has the consensus sequence [KR][KR][VI]ψ[FW], where ψ is any amino acid 63 except FIMYDP (13). Many PP1 regulatory proteins possess more than one PP1-interacting SLiM 64 (reviewed in (10)) allowing combinatorial control of PP1 binding (14). One example of an 65 ‘accessory’ SLiM is the ϕϕ[xF] motif, first identified via structural characterisation of PP1 in 66 complex with spinophilin (15), which lies approximately 6–8 residues downstream of the RVxF 67 motif and is present alongside the RVxF motif in most PP1 regulatory proteins (16). Like their 68 cellular counterparts, viral PP1 regulatory proteins commonly contain an RVxF motif that is 69 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 3 critical for PP1 binding (5, 7, 17), but the presence of additional SLIMs in these adaptors is less 70 well explored. 71 HSV-1 pUL21 was recently identified as a novel viral PP1 regulatory protein that promotes virus 72 replication and spread. HSV-1 pUL21 and its homologues across the alphaherpesviruses 73 comprise two folded domains (18, 19) that are linked by a flexible linker domain (20). PP1 binds 74 pUL21 via a novel short linear motif, termed the Twenty-one Recruitment Of Protein 75 Phosphatase One (TROPPO) motif, that is present in the linker region and is conserved across 76 alphaherpesviruses. Mutation of the TROPPO motif abolishes the ability of either HSV-1 pUL21 77 or the Varicella-Zoster virus (VZV) homologue (pORF38) to bind PP1. Loss of PP1 binding results 78 in hyperphosphorylation of multiple proteins, both viral and cellular, in HSV-1 infected cells. 79 pUL21 binding to PP1 regulates phosphorylation of components of the viral nuclear egress 80 complex (20), which remodels the nuclear membrane to support virus maturation (21, 22). The 81 pUL21:PP1 interaction also leads to dephosphorylation and thus hyperactivation of the 82 ceramide transport protein CERT, altering sphingolipid metabolism in infected cells (23). This 83 TROPPO motif does not bear obvious relation to known PP1 interacting motifs and pUL21 does 84 not have any other known PP1 binding motifs such as an RVxF motif. The molecular basis of 85 pUL21 binding to PP1 was thus unclear. 86 While inhibition of the ubiquitous and highly active enzyme PP1 is toxic, disrupting specific PP1 87 activities via blocking of specific interactions is a promising avenue for the design of new 88 therapies (24). The lack of homology with ‘RVxF’ motifs used by cellular PP1-binders suggested 89 that the viral TROPPO motif might bind a novel surface on PP1 and thus could be specifically 90 targeted by compounds that would inhibit virus replication and spread with minimal side-91 effects, as cellular functions of PP1 would remain uninhibited. We thus sought to determine 92 how the TROPPO motif of HSV-1 pUL21 and its homologues bind PP1. 93

94 Plasmids 95 Bacterial expression plasmids encoding the mouse PP1γ catalytic domain (residues 7–300)-H6 96 and pUL21-H6, and mammalian expression vectors encoding pUL21(FV242AA)-GFP , pORF38-97 GFP and pORF38(FV255AA)-GFP , was described previously (20). We note that the mouse and 98 human PP1γ catalytic domains share 100% amino acid identity. pORF38-NLT-H6 was generated 99 by sub-cloning residues 1–263 of ORF38 into the vector pOPTnH (25). pUL21-RVxF-H6 was 100 generated by inserting the mutations A237V and V239F into pUL21-H6 by QuikChange site-101 directed mutagenesis. Human PP1α was cloned from HeLa cell cDNA into a vector derived from 102 pF5K (Promega) with an N-terminal Myc epitope tag via restriction-based cloning, and point 103 mutations were introduced into Myc-PP1α by QuikChange mutagenesis. A plasmid encoding 104 pUL21-GFP was generated by sub-cloning a codon-optimised synthetic UL21 gene (GeneArt) 105 into pEGFP-N1 (Clontech), yielding an identical pUL21-GFP amino acid sequence as used in 106 (20). pUL21-ΔL, -ΔR and -ΔLR were generated from this vector via inverse PCR, and point 107 mutations were introduced via QuikChange mutagenesis. 108 Peptides 109 Peptides were commercially synthesised to >95% purity (GenScript) as follows: GADD34 110 RVxF+ϕϕ[xF] (residues 552–568) with fluorescein isothiocyanate (FITC) attached via an N-111 terminal aminohexanoic (Ahx) linker (sequence: [FITC-Ahx]-KARKVRFSEKVTVHFLA); pUL21 112 TROPPO (residues 234–250, sequence: AKRATVSEFVQVKHIDR); pOPF38 TROPPO (residues 113 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 4 246–263; sequence: KSDHITLSNFVQIRTIPR). TROPPO motif peptides were dissolved in aqueous 114 buffers and GADD34 RVxF+ϕϕ[xF] was dissolved in dimethyl sulfoxide (DMSO). 115 Recombinant protein purification 116 Proteins were expressed using Escherichia coli T7 Express lysY/Iq cells (New England Biolabs) 117 grown in 2×TY medium at 37°C to OD600 0.8–1.2 before cooling to 22°C and induction of protein 118 expression using 0.4 mM IPTG. Cells were harvested at 16–20 hours post-induction and pellets 119 stored at -70°C until required. Cells were resuspended in chilled lysis buffer (20 mM Tris pH 7.5, 120 20 mM imidazole, 500 mM NaCl, 1.4 mM β-mercaptoethanol 0.5 mM MgCl2 and 0.05% TWEEN-121 20 [plus 1 mM MnCl2 for PP1]) that was supplemented with 200 µL EDTA-free protease inhibitor 122 cocktail (Merck) and 400 U bovine DNase I (Merck). Cells were lysed using a TS series cell 123 disruptor (Constant Systems) at 24 kpsi and lysates were clarified (40,000×g, 30min, 4 °C) 124 before incubation with Ni-NTA agarose (Qiagen) for 1 h at 4°C. The resin was washed with ≥20 125 column volumes of wash buffer (20 mM Tris, 20 mM imidazole and 500 mM NaCl [plus 1 mM 126 MnCl2 for PP1]) at pH 7.5 (PP1γ and pORF38-NLT) or pH 8.5 (pUL21 and pUL21-RVxF). Protein 127 was eluted using wash buffer supplemented with additional imidazole (final concentration 250 128 mM) and proteins were subjected to size-exclusions chromatography using HiLoad 16/600 129 Superdex 75 (PP1) or Superdex 200 (other) columns equilibrated in 20 mM Tris pH 8.5, 500 mM 130 NaCl, 1 mM DTT (for pUL21 and pUL21-RVxF) or 50 mM HEPES pH 8.0, 500 mM NaCl, 0.5 mM 131 TCEP (for PP1 and pORF38-NLT). Fractions were analysed by SDS-PAGE and those containing 132 the desired protein were pooled, concentrated using centrifugal concentrators (Millipore) and 133 stored at 4°C (short-term) or snap-frozen in liquid nitrogen for long-term storage at -70°C. 134 Isothermal Titration Calorimetry (ITC) 135 ITC experiments were performed using a MicroCal PEAQ-ITC automated calorimeter (Malvern 136 Panalytical). The solvent (ITC buffer) was 50 mM HEPES pH 8.0, 500 mM NaCl, 0.5 mM TCEP. 137 Lyophilised peptides were dissolved into ITC buffer and their concentration was estimated from 138 dry mass. Proteins were diluted in ITC buffer and their concentration was estimated from the 139 theoretical extinction coefficient at 280 nm (26). Titrations were conducted at 25°C using 12 x 140 3 µL injections with syringe and cell contents as listed in Table S1. For all reagents, control 141 titrations (syringe into buffer or buffer into cell) showed very low non-specific heat evolution. 142 Data were analysed using MicroCal PEAQ-ITC analysis software (Malvern Panalytical) and fitted 143 using a one-site binding model. 144 Fluorescence polarisation anisotropy 145 Fluorescence anisotropy was measured using a SpectraMax i3 microplate reader (Molecular 146 Devices) with wavelengths 485 ± 20 nm (excitation) and 535 ± 25 nm (emission). Samples 147 (100 µL per well) were dispensed into low-bind black half-area 96-well microtitre plates 148 (Corning) and fluorescence anisotropy was recorded at 27°C. Experiments were performed in 149 50 mM HEPES pH 8.0, 500 mM NaCl, 0.05% TWEEN-20 unless otherwise noted. For all 150 titrations, fluorescent GADD34 RVxF+ϕϕ[xF] peptide was diluted from a 5 µM stock in DMSO 151 into buffer for a final concentration of 4 nM (0.08% DMSO). To measure association, 2 nM 152 peptide was incubated with a serial dilution of PP1γ catalytic domain. For competition assays a 153 serial dilution of the competitor was mixed 1:1 with a pre-formed complex of 4 nM GADD34 154 peptide and 500 nM PP1γ catalytic domain, to yield a final concentrations of 2 nM GADD34 155 peptide, 250 nM PP1γ and competitor as indicated. The PP1γ dissociation constant (KD) was 156 calculated by fitting the grating factor corrected anisotropy data to a one-site binding 157 equilibrium model in Prism version 7 (GraphPad). For competition experiments, IC50 values 158 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 5 were calculated by fitting the grating factor corrected anisotropy data to a four-parameter 159 inhibitor concentration response curve using Prism version 7 (GraphPad). 160 Mammalian cell culture 161 Mycoplasma-free, human embryonic kidney 293 T (HEK 293T) cells (American Type Culture 162 Collection #CRL-3216) were maintained in Dulbecco’s modified Eagle’s medium (DMEM) with 163 high glucose (Merck), supplemented with 10% (v/v) heat-inactivated fetal calf serum and 2 mM 164 L-glutamine (complete DMEM) in a humidified 5% CO2 atmosphere at 37 °C. 165 Immunoprecipitation of GFP-tagged bait protein from transfected cells 166 Monolayers of HEK 293T cells were transfected with 7.7 μg of DNA per 9 cm dish, using 167 18 μg/mL 25 kDa branched polyethylenimine (Merck) or using TransIt-LT1 (Mirus) in accordance 168 with the manufacturer’s instructions. Cells were harvested 24 h post-transfection and 169 incubated for 30 min in ice-cold 10 mM Tris pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 0.5% Igepal 170 CA-630 (a.k.a. NP-40), 1% (v/v) EDTA-free protease inhibitor cocktail (Merck) and incubated 171 (4°C, 30 min) before clarification at 21,000×g for 10 min at 4 °C. Lysate protein concentrations 172 were measured using the bicinchoninic acid (BCA) assay (Pierce) and normalised before affinity 173 capture using GFP-Trap resin (ChromoTek) following the manufacturer’s protocol. Samples were 174 eluted by heating the resin to 95°C for 5 min in 45 μL 2×SDS-PAGE loading buffer. Proteins were 175 separated by SDS-PAGE and transferred to 0.45 μm nitrocellulose membranes (PerkinElmer or 176 Cytiva). Membranes were stained with Ponceau S and imaged using a G:Box XX9 (Syngene) 177 before blocking in Tris-buffered saline with 0.1% TWEEN-20 (TBS-T) supplemented with 5% (w/v) 178 non-fat milk powder. Immunoblotting was performed using antibodies diluted in blocking buffer 179 as listed below and immunoblots were imaged using an Odyssey CLx (LI-COR 180 Biosciences). Densitometry was performed using Image Studio Lite version 5.2 (LI-COR 181 Biosciences) and statistical tests were performed using Prism version 7 (GraphPad). 182 Antibodies 183 The following primary antibodies and dilutions were used for immunoblotting: rabbit anti-CERT 184 1:10,000 (Abcam #Ab72536), mouse anti-PP1α 1:1000 (Santa Cruz #sc-271762), mouse anti-185 Myc 1:4000 (Millipore #05-724), rat anti-tubulin (clone YL1/2) hybridoma supernatant 1:40 (27). 186 Secondary antibodies from LI-COR Biosciences were diluted 1:10,000 as follows: IRDye 680RD 187 goat anti-rat (#926-68029) and goat anti-mouse (#926-68020), or IRDye 800CW goat anti-rabbit 188 (#926-32221). 189 Structure prediction and analysis 190 Structures were predicted using a locally installed version of ColabFold version 1.5.3. Input 191 sequences were human PP1γ catalytic domain (residues 7-300, UniProt P36873) plus either 192 HSV-1 strain KOS pUL21 (residues 1-535, UniProt ID: F8RG07) or VZV strain Dumas pORF38 193 (residues 1-541, Uniprot P09289) in complex with PP1. Mean lipophilicity maps were calculated 194 using ChimeraX (28) and molecular graphics were generated using an open-source build of 195 PyMOL (Schrödinger). 196

197 The TROPPO motif alone is insufficient to bind PP1 198 Prior studies of cellular regulatory proteins showed that peptides containing the RXvF and 199 ϕϕ[xF] motifs are sufficient to bind the catalytic domain of PP1 (29). Using fluorescence 200 anisotropy, we confirmed that a peptide containing the RVxF and ϕϕ[xF] motifs of GADD34 201 (a.k.a. PP1 regulatory subunit 15A) binds purified PP1γ catalytic domain (residues 7–300) with 202 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 6 144 ± 84 nM affinity (mean ± SEM, n = three independent experiments each performed in 203 technical triplicate) (Fig. 1A), similar to previous isothermal titration calorimetry (ITC) affinity 204 measurements for the same peptide sequence (29). Competition fluorescence anisotropy 205 experiments were performed, whereby peptides containing the TROPPO motifs of pUL21 or 206 pORF38 were titrated into a pre-formed complex of PP1 plus the fluorescent GADD34 peptide. 207 Neither TROPPO-containing peptide was able to displace the GADD34 peptide, even when 208 present at over 1000-fold molar excess (Fig. 1B). This suggested that either the TROPPO motifs 209 do not bind PP1, or they bind a different site on the PP1 surface that does not overlap with the 210 RVxF binding groove. ITC was employed to distinguish between these possibilities. Surprisingly, 211 titrating high concentrations (500–1000 µM) of pUL21 or pORF38 TROPPO-containing peptides 212 against the PP1 catalytic domain did not evolve heats above background levels (Fig. 1C,D, Table 213 S1), suggesting that neither peptide is sufficient to bind PP1. 214 The C-terminal region of the pUL21 linker is dispensable for PP1 binding 215 The N-terminal domain and linker region of pUL21 are necessary for immunoprecipitation (IP) of 216 PP1, with mutation of the TROPPO motif within the linker domain efficiently disrupting this 217 binding (20). While the TROPPO motif is absolutely conserved across alphaherpesvirus pUL21 218 homologues, the remainder of the pUL21 linker regions are only poorly conserved (Fig. 2A). 219 Strikingly, the region of the linker between the TROPPO motif and the C-terminal domain is 220 significantly longer in HSV-1 and HSV-2 pUL21 than in other herpesviruses such as VZV , the 221 pUL21 homologue of which (pORF38) is also capable of binding PP1 (20). 222 To probe the contribution of the linker region outside the TROPPO motif to PP1 binding, deletion 223 mutants of HSV-1 pUL21 were designed lacking linker residues that precede the TROPPO motif 224 (residues 216–237; pUL21ΔL), linker residues that follow the TROPPO motif (residues 257–276; 225 pUL21ΔR), or both flanking regions (pUL21ΔLR). Wild-type and mutant pUL21, and the pUL21 226 homologue ORF38 from VZV , were expressed as GFP fusion proteins and used for co-IP analysis 227 following transient transfection in HEK293T cells (Fig. 2B). As previously reported, pUL21 and 228 pORF38 efficiently precipitate endogenous PP1α and the interaction is disrupted by mutation of 229 the TROPPO motif (pUL21-FV242AA and pORF38-FV255AA). Strikingly, removal of the N-230 terminal region of the pUL21 linker (pUL21ΔL or pUL21ΔLR) completely abolishes PP1α binding, 231 whereas binding is retained when the C-terminal region of the pUL21 linker is removed 232 (pUL21ΔR). All pUL21 constructs retain CERT binding, consistent with localisation of CERT-233 binding to the C-terminal domain of pUL21 (20, 23) and confirming the correct expression and 234 folding of these mutants. 235 Structural model of the pUL21:PP1 and pORF38:PP1 complexes 236 Extensive attempts to co-crystallise WT or ΔR pUL21 with the catalytic subunit of PP1 proved 237 unsuccessful. Therefore, to investigate the relative contributions of the N-terminal domain, 238 linker and TROPPO motif to PP1 binding, the structure of HSV-1 pUL21 (Fig. 3A) and VZV pORF38 239 (Fig. 3B) in complex with the catalytic domain of human PP1γ were predicted using AlphaFold2-240 Multimer (30) via the Colabfold pipeline (31). The models for pUL21 (Fig. 2A, pLDDT/ipTM = 241 86.0/0.936) and pORF38 (Fig. 2B, pLDDT/ipTM = 85.4/0.918) predict that the TROPPO motif and 242 several segments of the linker region preceding it associate with the N-terminal domain and 243 with PP1. The per-residue confidence of the prediction (pLDDT) is high for residues of the 244 ordered pUL21 and pORF38 N- and C-terminal domains and for the PP1 catalytic domain, plus 245 for residues of the linker region that are predicted to interact with these domains (Fig. 3C). 246 Analysis of the Predicted Aligned Error (PAE) for both models confirms that the orientation of the 247 N-terminal domain and linker region up to the TROPPO motif with respect to PP1 is confidently 248 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 7 predicted, while the relative orientations of linker region following the TROPPO motif and the C-249 terminal domains are not predicted with confidence (Fig. 3D). 250 Previously, small angle X-ray scattering showed that pUL21 comprises two ordered domains 251 separated by a highly flexible linker (20). Crystallographic structures showed the N-terminal 252 domain of pUL21 (pUL21-N) to possess a distinctive α/β fold, with two anti-parallel β-sheets 253 that pack against surface-exposed α-helices (the ‘inner sheet’ and ‘lower outer sheet’) plus a 254 third β-sheet that packs against the inner sheet (the ‘upper outer sheet’) (18). Despite pUL21 255 residues 1–216 being present in the crystallisation experiment and being conserved across 256 alphaherpesviruses, only residues 1–198 were resolved in electron density. In the predicted 257 complex, N-terminal domain residues 209–212 and residues 242–248 of the TROPPO motif are 258 predicted to become ordered, forming two strands of anti-parallel beta sheet that serve as a 259 bridge between the three-stranded pUL21-N upper outer sheet and the final β-sheet of the PP1 260 catalytic domain (Fig. 3A). Residues 221-223 of the Linker (L) region are also predicted to 261 become ordered, forming a short β-strand that extends the pUL21-N inner sheet. We note that 262 these newly-formed sheets are proximal to the amino terminus of pUL21, consistent with an 263 inability of N-terminally GFP-tagged pUL21-N-plus-linker to bind PP1 (20). 264 All residues of the pUL21 TROPPO motif (239VSFVQVKHI248) are predicted to be in close 265 association with PP1 (Fig. 3E). Surprisingly, the TROPPO motif is predicted to bind at the same 266 surface hydrophobic groove as cellular RVxF-containing peptides, adopting a similar loop-plus-267 sheet conformation as GADD34 residues 553–568 (29). The first residue of the TROPPO and the 268 three preceding (236RATV239) of pUL21 align structurally with residues 555KVRF558 of the GADD34 269 RVxF motif (Fig. 3F), with pUL21 A237 and V239 predicted to bind the PP1 hydrophobic pockets 270 occupied by GADD34 V556 and F558 (Fig. 3G). Most cellular regulatory proteins that bind PP1 271 possess a second PP1-binding SLiM, ϕϕ[xF], that is C-terminal to the RVxF (16). In the predicted 272 complex, residues 244VKHI247 of the pUL21 TROPPO adopt a similar conformation to the GADD34 273 ϕϕ[xF] residues 564VHFL567 (Fig. 3G). 274 We previously showed that pUL21 mutations F242E or V243D are sufficient to abolish PP1 275 binding (23). The severity of the V243D mutation is readily explained by the complex prediction, 276 this residue is predicted to interact with hydrophobic residues on the surface of PP1 (Fig. S1) 277 and there is a valine at the equivalent position in GADD34 (Fig. 3G). The severity of the F242E 278 mutation is less obvious when considering only the TROPPO motif, as the side chain is 279 predicted to point away from PP1 (Fig. 3G). However, it becomes clearer when one considers 280 the context of pUL21-N and the residues of the linker that are predicted to become ordered. The 281 side chain of F242 is predicted to be largely buried, interacting with hydrophobic side chains of 282 pUL21 A48 and F50, plus the hydrophobic face of the R24 guanidyl group (Fig. 3H). Mutation to a 283 charged residue would be predicted to disrupt this packing. The TROPPO motif of pORF38 forms 284 very similar interactions with both the pORF38 N-terminal domain and PP1 in the model of the 285 pORF38:PP1 complex (Fig. S1). Loss of hydrophobic interactions would thus similarly explain 286 why the FV255AA mutations abolished binding to PP1 (Fig. 2B) (20). 287 Site-d irected mutagenesis confirms the predictive power of the pUL21:PP1 model 288 While the models of pUL21 and pORF38 in complex with PP1 were compellingly consistent with 289 both each other and with previous results, structurally informed mutagenesis is required to 290 rigorously test their predictive power. To this end, single amino acids mutations were introduced 291 in both PP1 and pUL21, and their ability to interact in cells was probed by c o I P. 292 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 8 For PP1, we utilised three published mutations that have all been show to severely decrease 293 PP1 regulatory protein binding: F257A (32), which is at the RVxF hydrophobic pocket into which 294 the side chain of pUL21 V239 is predicted to bind (Fig. 4A); and L289R or C291R (33), which are 295 on the final β-sheet of the PP1 catalytic domain and contribute to the RVxF hydrophobic surface 296 with which pUL21 A237 is predicted to associate (Fig. 4A). HEK293T cells were co-transfected 297 with pUL21-GFP and WT or mutant Myc-PP1, and lysates were subjected to GFP affinity capture 298 before SDS-PAGE and immunoblotting. The coIP of all three mutants was profoundly less 299 efficient than for WT Myc-PP1 (Fig. 4B). However, we consistently observed lower abundance of 300 these mutants in the cell lysates used for IP analysis, suggesting lower expression or increased 301 turnover in transfected cells. To account for this, the amount of input and bound Myc-PP1 was 302 quantified and the ratio of bound:input was calculated (Fig. 4C), confirming that all three 303 mutations significantly disrupt binding to pUL21-GFP. 304 For pUL21, a panel of mutations were designed to probe the predicted interactions of residues 305 at the interface with both PP1 and pUL21-N. Five mutations were designed to disrupt the 306 interaction: Q174R, which would replace a buried residue on pUL21-N sheet β14 that contacts 307 Q244 of the TROPPO motif (Fig. 4D); T209R, which would replace a small buried side chain that 308 contacts pUL21-N with a large charged side chain (Fig. 4D); VV211AA, where two hydrophobic 309 side chains (including one that is buried) in the sheet that bridges the TROPPO motif and pUL21-310 N β1 are removed (Fig. 4D); A218E, where a small hydrophobic residue that interacts with a 311 surface hydrophobic patch on PP1 is replaced with a charged residue (Fig. 4D); and R250E, 312 where the charge of a side chain that interacts with the PP1 D277 side chain is inverted (Fig. 4D). 313 Three mutations were designed to enhance PP1 binding: A237V , substituting alanine for the 314 canonical RVxF valine residue (Fig. 4A); V239F , substituting valine for the canonical RVxF 315 phenylalanine residue (Fig. 4A); and H247F , substituting for the equivalent residue in GADD34 316 where the larger hydrophobic side chain could make more extensive interactions with the aryl 317 portion of the PP1 K297 side chain (Fig. 4D). A control mutation A232K was also introduced, as 318 substitution of a small hydrophobic side chain for a large charged one on a mobile surface-319 exposed loop (Fig. 4A) would be predicted to leave PP1 binding unaltered. 320 HEK293T cells were transfected with WT or mutant pUL21-GFP and the coIP of endogenous 321 PP1α was monitored by immunoblotting (Fig. 4E). Pleasingly, all designed mutations had the 322 predicted effect: PP1α coIP was abolished by the Q174R, T209A, VV211AA and A218E 323 substitutions, in addition to the control FV242AA mutation, and the R250E substitutions 324 dramatically reduced binding. All proteins were present at similar abundance in the input 325 samples and all retained binding to CERT, confirming their correct folding. As predicted, the 326 surface-loop substitution A232K did not dramatically alter PP1α coIP . The A237V, V239F and 327 H247F mutants all co-precipitated more PP1α than did WT pUL21-GFP , suggesting that all three 328 have higher affinity for the PP1 catalytic subunit. Taken together, these mutagenesis results 329 strongly support the predicted structure of the pUL21:PP1 complex. 330 Alphaherpesvirus proteins compete directly with cellular regulatory proteins for PP1 binding 331 Site-directed mutagenesis confirmed the predictive power of the pUL21:PP1 complex, but the 332 affinity of the interaction between PP1 and alphaherpesvirus pUL21 homologues remained 333 unknown. Attempts to perform ITC using recombinant full-length pUL21 and PP1 catalytic 334 domain were unsuccessful as neither protein could be concentrated sufficiently to serve as 335 titrant in the syringe. Similarly, attempts to purify full-length pORF38 following recombinant 336 expression in E. coli were frustrated by low protein expression and solubility. As the R region of 337 the linker and C-terminal domain are dispensable for PP1 binding, we designed expression 338 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 9 constructs encoding the pUL21 and pORF38 N-terminal domains plus the Linker region up to 339 the end of the TROPPO motif (the NLT region) with a C-terminal hexahistidine tag (Fig. 5A). The 340 pORF38-NLT construct was readily expressed and purified to concentrations suitable for ITC 341 (Fig. 5B). ITC titration of pORF38-NLT into PP1 showed that the proteins form a 1:1 complex with 342 affinity of 497 ± 143 nM (mean ± SEM, three independent experiments) (Fig. 5C and Table S1). 343 This is substantially weaker binding than observed for the GADD34 peptide binding PP1 via 344 fluorescence anisotropy (Fig. 1A) or ITC (29). 345 To definitively test whether pORF38 binds the same hydrophobic groove as cellular RVxF motifs, 346 competition fluorescence anisotropy was performed via titration of pORF38-NLT into a pre-347 formed complex of PP1 plus the fluorescent GADD34 peptide. The liberation of GADD34 peptide 348 at increasing concentrations of pORF38-NLT, as evidenced by a decrease in fluorescence 349 anisotropy, confirmed that pORF38 competes with the RVxF peptide for PP1 binding (Fig. 5D), 350 with an IC50 of 417 ± 11 nM (mean ± SEM, three independent experiments each performed in 351 technical triplicate). 352 Similar competition fluorescence anisotropy experiments with purified full-length pUL21 (Fig. 353 5E) were performed in buffer optimised to enhance solubility of the protein (Tris pH 8.5 in place 354 of HEPES pH 8). Under these conditions PP1 binds the GADD34 peptide with slightly higher 355 affinity (55 ± 21 nM, mean ± SEM of three independent experiments performed in technical 356 triplicate; Fig. S2). Full-length pUL21 is unable to out-compete GADD34 for binding to PP1 (Fig. 357 5F), suggesting that the affinity of PP1 for full-length pUL21 is lower than for pORF38 – 358 consistent with the observation that PP1α is consistently co-immunoprecipitated more 359 efficiently by pORF38-GFP than by pUL21-GFP (Fig. 2B and Fig. 3H of (20)). To test whether 360 optimisation of the TROPPO sequence to better match a canonical RVxF motif could enhance 361 binding, we generated, expressed and purified a pUL21-H6 mutant where residues A237 and 362 V239 had been substituted for valine and phenylalanine, respectively (pUL21-RVxF , Fig. 5E). 363 Fluorescence anisotropy confirms that pUL21-RVxF competes with the GADD34 peptide for 364 binding to the PP1 hydrophobic groove with an IC50 of 337 ± 31 nM (mean ± SEM, three 365 independent experiments each performed in technical triplicate). 366

367 pUL21 is required for efficient replication and spread of HSV-1 (20, 34–36). We previously 368 identified HSV-1 pUL21 as a PP1 regulatory protein and that it contained a novel SLiM required 369 for PP1 binding, the TROPPO motif (20). The absence of identifiable sequence homology with 370 other known PP1-associated SLiMs (10) raised the tantalising possibility that the TROPPO motif 371 binds PP1 via a novel surface that could be targeted for the development of antiviral 372 compounds. Combining structure prediction, site-directed mutagenesis, immunoprecipitation 373 and biophysical characterisation, we now show that the TROPPO motif binds the same 374 hydrophobic surface of PP1 as cellular RVxF and ϕϕ[xF] SLiMs. Competition fluorescence 375 anisotropy confirms that TROPPO-containing proteins compete directly with cellular RVxF and 376 ϕϕ[xF] motifs for binding to PP1 (Fig. 5). 377 PP1 catalytic domains are unlikely to exist in their apo form within cells (10) and, as such, 378 pUL21 and pORF38 must compete with cellular regulatory proteins to access PP1 in infected 379 cells. There are over 200 mammalian PP1 regulatory proteins (9, 12) and their affinities for PP1 380 range from ~10 nM (15, 37) to ~1 µM (38). Given the competition for binding, it is puzzling that 381 pORF38 has evolved to bind PP1 with only modest affinity (~500 nM) (Fig. 5C). The PP1-binding 382 of pUL21 must be even lower, as unlike pOPF38-NLT wild-type pUL21 cannot displace a 383 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 10 GADD34 RVxF- and ϕϕ[xF]-containing peptide from PP1. One striking feature of the predicted 384 pUL21:PP1 complex is that the TROPPO motif containing the ϕϕ[xF] motif lies sandwiched 385 between PP1 and the pUL21-N, forming part of an extended β-sheet that bridges core β-sheets 386 across the two proteins (Fig. 2A). While the structure of regulatory protein spinophilin (15) 387 showed a sheet-turn-sheet binding PP1 at the ϕϕ[xF] interacting region, adding two strands to 388 the PP1 core β-sheet, we are unaware of any examples where the RVxF and ϕϕ[xF] form a 389 structured bridge between PP1 and another folded domain. The formation of this extended 390 sheet, with concomitant burial of hydrophobic residues like F242 into a newly-extended 391 hydrophobic core of pUL21-N (Fig. 3H), would stabilise the interaction between pUL21 and PP1 392 once formed. Such cooperativity would restrain the TROPPO motif in the correct conformation 393 for PP1 binding despite the modest affinity (39). 394 Published RVxF consensus sequences (13, 40, 41) invariably have a valine or isoleucine and 395 phenylalanine or tryptophan in the second and fourth position, respectively, whereas in pUL21 396 and pORF38 the equivalent residues are alanine and valine (Fig. 3). It is not clear that there are 397 structural constraints preventing the virus from evolving an optimal RVxF motif since, like most 398 regulatory protein regions that bind PP1 (12), the linker region of pUL21 is intrinsically 399 disordered (20). Mutation of A237 to valine or V239 to phenylalanine enhances the coIP of PP1α 400 by pUL21 in transfected cells (Fig. 4E) and the A237V+V239R double substitution gives pUL21-401 RVxF the ability to compete with the GADD34 peptide for PP1 binding (Fig. 5F). HSV-1 encodes a 402 second PP1 regulatory protein, ICP34.5, that has high sequence homology to GADD34 and 403 binds PP1 via a canonical RVxF motif (17, 42). There is thus nothing to prevent HSV-1 encoding a 404 protein with a bona fide RVxF motif, and we hypothesise that selective pressure related to the 405 molecular function of pUL21 has driven it to retain a suboptimal RVxF motif. Given the high 406 degree of TROPPO sequence conservation across pUL21 homologues and corresponding 407 absence of canonical RVxF motifs, we conclude that this evolutionary constraint must be 408 common to all alphaherpesviruses. 409 What might drive these viruses to maintain a low-affinity PP1 recruitment domain? One clue 410 comes from in vitro evolution experiments, where serial passage of HSV-1 encoding pUL21 411 mutated to prevent PP1 binding led to compensatory mutations in the virus-encoded kinase 412 pUS3 (20). These two proteins have overlapping substrates, with pUL21-mediated 413 dephosphorylation directly antagonising pUS3 kinase activity. A balance of activity is clearly 414 critical for the virus: abolishing expression of either pUS3 or pUL21 severely diminishes virus 415 replication in cultured keratinocytes, as does removal of pUL21 PP1 binding, but passage of the 416 PP1-binding pUL21 mutants led to rapid accumulation of mutations that lowered pUS3 kinase 417 activity and restored virus fitness. The peak of pUL21 expression is at late times (18 hours) post-418 infection in cultured keratinocytes, whereas pUS3 abundance peaks at 6 hours post-infection 419 and then declines slightly (43). We hypothesise that the PP1-binding motif of pUL21 is 420 specifically tuned to prevent effective PP1 binding at early times post-infection, when pUS3 421 activity is presumably pro-viral. Instead, the presence of a suboptimal PP1 interaction motif 422 would allow pUL21 to effectively compete for PP1 binding only at late times post infection and 423 in specific subcellular locations, for example the nuclear envelope, where the protein is highly 424 abundant (20). This fine-tuning of activity would be especially important for a multi-functional 425 protein like pUL21, with roles cell-to-cell spread (20, 34, 36, 44–46), preventing futile genome 426 release by nascent viral capsids (35, 47) and virus assembly (45, 46, 48, 49), in addition to its 427 role in regulating viral nuclear egress (44, 45, 50–52). High affinity binding to PP1 could recruit 428 phosphatase activity to the wrong cellular or viral complexes, or it could inappropriately 429 sequester PP1 and thereby prevent other PP1-mediated dephosphorylation events within the 430 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 11 cell that promote virus replication. Similar fine-tuning of PP1 recruitment has been observed for 431 HIV , where mutation of the suboptimal QVCF motif to RVCF enhanced PP1 binding but reduced 432 the ability of Tat to stimulate HIV transcription (5). 433 In conclusion, we show that the alphaherpesvirus pUL21 homologues bind the same PP1 434 hydrophobic surface groove as cellular RVxF and ϕϕ[xF] motifs. Extensive site-directed 435 mutagenesis supports a predicted structure of the pUL21:PP1 complex where intrinsically 436 disordered regions of the pUL21 linker fold into an extended β-sheet that bridges the cores of 437 PP1 and the pUL21 N-terminal domain. The conservation of suboptimal binding residues within 438 the region structurally equivalent to the cellular RVxF implies that the viruses have deliberately 439 regulated PP1 binding affinity to support optimal virus replication. This highlights the careful 440 regulation of kinase and phosphatase activity required to support efficient virus replication and 441 spread. Our work also highlights how PP1 can be recruited by proteins lacking any identifiable 442 PP1-binding SLiM, suggesting that the pool of viral and cellular PP1 regulatory proteins may be 443 larger than previously imagined. 444

445 We thank Owen Tutt (University of Cambridge) for technical assistance. For the purpose of open 446 access, the authors have applied a Creative Commons Attribution (CC BY) licence to any Author 447 Accepted Manuscript version arising from this submission. 448 Conflicts of interest 449 The author(s) declare that there are no conflicts of interest. 450 Funding information 451 This work was funded by a Sir Henry Dale Fellowship, jointly funded by the Wellcome Trust and 452 the Royal Society (098406/Z/12/B) to SCG. HM is funded by the University of Cambridge School 453 of Clinical Medicine Doctoral Training Programme in Medical Research. DSC is funded by a 454 British Skin Foundation PhD studentship (028/S/22). THB was funded by a University of 455 Cambridge Department of Pathology PhD studentship. JED is a Wellcome Trust Senior Research 456 Fellow (219447/Z/19/Z). 457 Data availability 458 The models of pUL21 and pORF38 in complex with PP1 have been deposited to the University of 459 Cambridge Data Repository (DOI). The authors confirm that all other data supporting the 460 findings of this study are available within the article and/or its supplementary materials. 461 462 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 12

463 1. Jacob, T., Van den Broeke, C., and Favoreel, H. W. (2011) Viral serine/threonine protein 464 kinases. J Virol. 85, 1158–1173 465 2. Kato, A., and Kawaguchi, Y . (2018) Us3 Protein Kinase Encoded by HSV: The Precise 466 Function and Mechanism on Viral Life Cycle. Adv Exp Med Biol. 1045, 45–62 467 3. He, B., Gross, M., and Roizman, B. (1998) The gamma134.5 protein of herpes simplex virus 468 1 has the structural and functional attributes of a protein phosphatase 1 regulatory 469 subunit and is present in a high molecular weight complex with the enzyme in infected 470 cells. J Biol Chem. 273, 20737–20743 471 4. He, B., Gross, M., and Roizman, B. (1997) The gamma(1)34.5 protein of herpes simplex 472 virus 1 complexes with protein phosphatase 1alpha to dephosphorylate the alpha subunit 473 of the eukaryotic translation initiation factor 2 and preclude the shutoff of protein 474 synthesis by double-stranded RNA-activated protein kinase. Proc Natl Acad Sci U S A. 94, 475 843–848 476 5. Ammosova, T., Jerebtsova, M., Beullens, M., Lesage, B., Jackson, A., Kashanchi, F ., 477 Southerland, W., Gordeuk, V . R., Bollen, M., and Nekhai, S. (2005) Nuclear targeting of 478 protein phosphatase-1 by HIV-1 Tat protein. J Biol Chem. 280, 36364–36371 479 6. Lin, X., Ammosova, T., Choy, M. S., Pietzsch, C. A., Ivanov, A., Ahmad, A., Saygideğer, Y ., 480 Kumari, N., Kovalskyy, D., Üren, A., Peti, W., Bukreyev, A., and Nekhai, S. (2019) Targeting 481 the Non-catalytic RVxF Site of Protein Phosphatase-1 With Small Molecules for Ebola Virus 482 Inhibition. Front Microbiol. 10, 2145 483 7. Davis, M. E., Wang, M. K., Rennick, L. J., Full, F ., Gableske, S., Mesman, A. W., Gringhuis, S. 484 I., Geijtenbeek, T. B. H., Duprex, W. P ., and Gack, M. U. (2014) Antagonism of the 485 phosphatase PP1 by the measles virus V protein is required for innate immune escape of 486 MDA5. Cell Host Microbe. 16, 19–30 487 8. Lin, X., Sajith, A. M., Wang, S., Kumari, N., Choy, M. S., Ahmad, A., Cadet, D. R., Gu, X., 488 Ivanov, A. I., Peti, W., Kulkarni, A., and Nekhai, S. (2020) Structural Optimization of 2,3-489 Dihydro-1H-cyclopenta[b]quinolines Targeting the Noncatalytic RVxF Site of Protein 490 Phosphatase 1 for HIV-1 Inhibition. ACS Infect Dis. 6, 3190–3211 491 9. Peti, W., Nairn, A. C., and Page, R. (2013) Structural basis for protein phosphatase 1 492 regulation and specificity. FEBS J. 280, 596–611 493 10. Casamayor, A., and Ariño, J. (2020) Controlling Ser/Thr protein phosphatase PP1 activity 494 and function through interaction with regulatory subunits. Adv Protein Chem Struct Biol. 495 122, 231–288 496 11. Wu, D., De Wever, V ., Derua, R., Winkler, C., Beullens, M., Van Eynde, A., and Bollen, M. 497 (2018) A substrate-trapping strategy for protein phosphatase PP1 holoenzymes using 498 hypoactive subunit fusions. J Biol Chem. 293, 15152–15162 499 12. Bollen, M., Peti, W., Ragusa, M. J., and Beullens, M. (2010) The extended PP1 toolkit: 500 designed to create specificity. Trends Biochem Sci. 35, 450–458 501 13. Hendrickx, A., Beullens, M., Ceulemans, H., Den Abt, T., Van Eynde, A., Nicolaescu, E., 502 Lesage, B., and Bollen, M. (2009) Docking motif-guided mapping of the interactome of 503 protein phosphatase-1. Chem Biol. 16, 365–71 504 14. Bollen, M. (2001) Combinatorial control of protein phosphatase-1. Trends Biochem Sci. 26, 505 426–431 506 15. Ragusa, M. J., Dancheck, B., Critton, D. A., Nairn, A. C., Page, R., and Peti, W. (2010) 507 Spinophilin directs protein phosphatase 1 specificity by blocking substrate binding sites. 508 Nat Struct Mol Biol. 17, 459–464 509 16. Srivastava, G., Bajaj, R., Kumar, G. S., Gaudreau-Lapierre, A., Nicolas, H., Chamousset, D., 510 Kreitler, D., Peti, W., Trinkle-Mulcahy, L., and Page, R. (2022) The ribosomal RNA processing 511 1B:protein phosphatase 1 holoenzyme reveals non-canonical PP1 interaction motifs. Cell 512 Rep. 41, 111726 513 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 13 17. Zhang, C., Tang, J., Xie, J., Zhang, H., Li, Y ., Zhang, J., Verpooten, D., He, B., and Cao, Y . 514 (2008) A conserved domain of herpes simplex virus ICP34.5 regulates protein phosphatase 515 complex in mammalian cells. FEBS Lett. 582, 171–176 516 18. Metrick, C. M., Chadha, P ., and Heldwein, E. E. (2015) The unusual fold of herpes simplex 517 virus 1 UL21, a multifunctional tegument protein. J Virol. 89, 2979–84 518 19. Metrick, C. M., and Heldwein, E. E. (2016) Novel Structure and Unexpected RNA-Binding 519 Ability of the C-Terminal Domain of Herpes Simplex Virus 1 Tegument Protein UL21. J Virol. 520 90, 5759–69 521 20. Benedyk, T. H., Muenzner, J., Connor, V ., Han, Y ., Brown, K., Wijesinghe, K. J., Zhuang, Y ., 522 Colaco, S., Stoll, G. A., Tutt, O. S., Svobodova, S., Svergun, D. I., Bryant, N. A., Deane, J. E., 523 Firth, A. E., Jeffries, C. M., Crump, C. M., and Graham, S. C. (2021) pUL21 is a viral 524 phosphatase adaptor that promotes herpes simplex virus replication and spread. PLoS 525 Pathog. 17, e1009824 526 21. Draganova, E. B., Thorsen, M. K., and Heldwein, E. E. (2021) Nuclear Egress. Curr Issues 527 Mol Biol. 41, 125–170 528 22. Klupp, B. G., and Mettenleiter, T. C. (2023) The Knowns and Unknowns of Herpesvirus 529 Nuclear Egress. Annu Rev Virol. 10, 305–323 530 23. Benedyk, T. H., Connor, V ., Caroe, E. R., Shamin, M., Svergun, D. I., Deane, J. E., Jeffries, C. 531 M., Crump, C. M., and Graham, S. C. (2022) Herpes simplex virus 1 protein pUL21 alters 532 ceramide metabolism by activating the interorganelle transport protein CERT. J Biol Chem. 533 298, 102589 534 24. Vagnarelli, P ., and Alessi, D. R. (2018) PP1 Phosphatase Complexes: Undruggable No 535 Longer. Cell. 174, 1049–1051 536 25. Neidel, S., Maluquer de Motes, C., Mansur, D. S., Strnadova, P ., Smith, G. L., and Graham, 537 S. C. (2015) Vaccinia virus protein A49 is an unexpected member of the B-cell Lymphoma 538 (Bcl)-2 protein family. J Biol Chem. 290, 5991–6002 539 26. Wilkins, M. R., Gasteiger, E., Bairoch, A., Sanchez, J. C., Williams, K. L., Appel, R. D., and 540 Hochstrasser, D. F . (1999) Protein identification and analysis tools in the ExPASy server. 541

Mol Biol. 112, 531–552 542 27. Kilmartin, J. V ., Wright, B., and Milstein, C. (1982) Rat monoclonal antitubulin antibodies 543 derived by using a new nonsecreting rat cell line. J Cell Biol. 93, 576–582 544 28. Meng, E. C., Goddard, T. D., Pettersen, E. F ., Couch, G. S., Pearson, Z. J., Morris, J. H., and 545 Ferrin, T. E. (2023) UCSF ChimeraX: Tools for structure building and analysis. Protein Sci. 546 32, e4792 547 29. Choy, M. S., Yusoff, P ., Lee, I. C., Newton, J. C., Goh, C. W., Page, R., Shenolikar, S., and 548 Peti, W. (2015) Structural and Functional Analysis of the GADD34:PP1 eIF2α Phosphatase. 549 Cell Rep. 11, 1885–1891 550 30. Evans, R., O’Neill, M., Pritzel, A., Antropova, N., Senior, A., Green, T., Žídek, A., Bates, R., 551 Blackwell, S., Yim, J., Ronneberger, O., Bodenstein, S., Zielinski, M., Bridgland, A., 552 Potapenko, A., Cowie, A., Tunyasuvunakool, K., Jain, R., Clancy, E., Kohli, P ., Jumper, J., 553 and Hassabis, D. (2021) Protein complex prediction with AlphaFold-Multimer, BioRxiv, 554 10.1101/2021.10.04.463034 555 31. Mirdita, M., Schütze, K., Moriwaki, Y ., Heo, L., Ovchinnikov, S., and Steinegger, M. (2022) 556 ColabFold: making protein folding accessible to all. Nat Methods. 19, 679–682 557 32. Lesage, B., Beullens, M., Nuytten, M., Van Eynde, A., Keppens, S., Himpens, B., and Bollen, 558 M. (2004) Interactor-mediated nuclear translocation and retention of protein phosphatase-559 1. J Biol Chem. 279, 55978–55984 560 33. Gibbons, J. A., Weiser, D. C., and Shenolikar, S. (2005) Importance of a surface 561 hydrophobic pocket on protein phosphatase-1 catalytic subunit in recognizing cellular 562 regulators. J Biol Chem. 280, 15903–15911 563 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 14 34. Finnen, R. L., and Banfield, B. W. (2018) CRISPR/Cas9 Mutagenesis of UL21 in Multiple 564 Strains of Herpes Simplex Virus Reveals Differential Requirements for pUL21 in Viral 565 Replication. Viruses. 10, 258 566 35. Thomas, E. C. M., Finnen, R. L., Mewburn, J. D., Archer, S. L., and Banfield, B. W. (2023) The 567 Herpes Simplex Virus pUL16 and pUL21 Proteins Prevent Capsids from Docking at Nuclear 568 Pore Complexes. PLoS Pathog. 19, e1011832 569 36. Sarfo, A., Starkey, J., Mellinger, E., Zhang, D., Chadha, P ., Carmichael, J., and Wills, J. W. 570 (2017) The UL21 Tegument Protein of Herpes Simplex Virus 1 Is Differentially Required for 571 the Syncytial Phenotype. J Virol. 91, 01161–17 572 37. Choy, M. S., Hieke, M., Kumar, G. S., Lewis, G. R., Gonzalez-DeWhitt, K. R., Kessler, R. P ., 573 Stein, B. J., Hessenberger, M., Nairn, A. C., Peti, W., and Page, R. (2014) Understanding the 574 antagonism of retinoblastoma protein dephosphorylation by PNUTS provides insights into 575 the PP1 regulatory code. Proc Natl Acad Sci U S A. 111, 4097–4102 576 38. Choy, M. S., Nguyen, H. T., Kumar, G. S., Peti, W., Kettenbach, A. N., and Page, R. (2024) A 577 protein phosphatase 1 specific phosphatase targeting peptide (PhosTAP) to identify the 578 PP1 phosphatome. Proc Natl Acad Sci U S A. 121, e2415383121 579 39. Ivarsson, Y ., and Jemth, P . (2019) Affinity and specificity of motif-based protein-protein 580 interactions. Curr Opin Struct Biol. 54, 26–33 581 40. Meiselbach, H., Sticht, H., and Enz, R. (2006) Structural analysis of the protein 582 phosphatase 1 docking motif: molecular description of binding specificities identifies 583 interacting proteins. Chem Biol. 13, 49–59 584 41. Wakula, P ., Beullens, M., Ceulemans, H., Stalmans, W., and Bollen, M. (2003) Degeneracy 585 and function of the ubiquitous RVXF motif that mediates binding to protein phosphatase-1. 586 J Biol Chem. 278, 18817–18823 587 42. Chou, J., and Roizman, B. (1994) Herpes simplex virus 1 gamma(1)34.5 gene function, 588 which blocks the host response to infection, maps in the homologous domain of the genes 589 expressed during growth arrest and DNA damage. Proc Natl Acad Sci U S A. 91, 5247–5251 590 43. Soh, T. K., Davies, C. T. R., Muenzner, J., Hunter, L. M., Barrow, H. G., Connor, V ., Bouton, C. 591 R., Smith, C., Emmott, E., Antrobus, R., Graham, S. C., Weekes, M. P ., and Crump, C. M. 592 (2020) Temporal Proteomic Analysis of Herpes Simplex Virus 1 Infection Reveals Cell-593 Surface Remodeling via pUL56-Mediated GOPC Degradation. Cell Rep. 33, 108235 594 44. Le Sage, V ., Jung, M., Alter, J. D., Wills, E. G., Johnston, S. M., Kawaguchi, Y ., Baines, J. D., 595 and Banfield, B. W. (2013) The herpes simplex virus 2 UL21 protein is essential for virus 596 propagation. J Virol. 87, 5904–5915 597 45. Roddy, K., Grzesik, P ., Smith, B. J., Ko, N., Vashee, S., and Desai, P . J. (2025) The loss of both 598 pUL16 and pUL21 in HSV-1-infected cells alters capsid-tegument composition, nuclear 599 membrane architecture, cytoplasmic maturation and cell-to-cell spread. J Gen Virol. 106, 600 002083 601 46. Finnen, R. L., Muradov, J. H., Le Sage, V ., and Banfield, B. W. (2024) Disruption of herpes 602 simplex virus type 2 pUL21 phosphorylation impairs secondary envelopment of 603 cytoplasmic nucleocapsids. J Virol. 98, e0065624 604 47. Thomas, E. C. M., Bossert, M., and Banfield, B. W. (2022) The herpes simplex virus 605 tegument protein pUL21 is required for viral genome retention within capsids. PLoS 606 Pathog. 18, e1010969 607 48. Han, J., Chadha, P ., Starkey, J. L., and Wills, J. W. (2012) Function of glycoprotein E of 608 herpes simplex virus requires coordinated assembly of three tegument proteins on its 609 cytoplasmic tail. Proc Natl Acad Sci U S A. 109, 19798–803 610 49. Ahmad, I., and Wilson, D. W. (2020) HSV-1 Cytoplasmic Envelopment and Egress. Int J Mol 611 Sci. 21, E5969 612 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 15 50. Gao, J., Finnen, R. L., Sherry, M. R., Le Sage, V ., and Banfield, B. W. (2020) Differentiating 613 the Roles of UL16, UL21, and Us3 in the Nuclear Egress of Herpes Simplex Virus Capsids. J 614 Virol. 94, 00738–20 615 51. Muradov, J. H., Finnen, R. L., Gulak, M. A., Hay, T. J. M., and Banfield, B. W. (2021) pUL21 616 regulation of pUs3 kinase activity influences the nature of nuclear envelope deformation 617 by the HSV-2 nuclear egress complex. PLoS Pathog. 17, e1009679 618 52. Nahas, K. L., Connor, V ., Wijesinghe, K. J., Barrow, H. G., Dobbie, I. M., Harkiolaki, M., 619 Graham, S. C., and Crump, C. M. (2025) Applying 3D correlative structured illumination 620 microscopy and X-ray tomography to characterise herpes simplex virus-1 morphogenesis. 621 10.7554/eLife.105209.1 622 623 624 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 16 Figure 1. Alphaherpesvirus TROPPO peptides do not bind PP1. 625 A Fluorescence anisotropy of PP1γ catalytic domain (residues 7–300) binding 2 nM fluorescently 626 labelled peptide containing the GADD34 RVxF and ϕϕ[xF] motifs (residues 552–568). Mean ± SD 627 is shown for technical triplicate measurements. Affinity (KD) is average of three independent 628 experiments. B Competition fluorescence anisotropy of peptides containing the 629 alphaperhesvirus TROPPO motif . Pre-formed complexes of 2 nM fluorescent GADD34 peptide 630 with 250 nM PP1 were incubated with increasing concentrations of TROPPO-containing 631 peptides from pUL21 (residues 234–250; left) and pORF38 (residues 246–263; right). Neither 632 peptide successfully displaced GADD34 RVxF . Mean ± SD is shown for technical triplicate 633 measurements. C, D Isothermal titration calorimetry of PP1γ catalytic domain with TROPPO 634 containing peptides from C pUL21 and D pORF38. For each, baseline-corrected differential 635 power (DP) versus time (left) and integrated change in enthalpy (ΔH) versus molar ratio (right) 636 are shown. Figures are representative of three (C) or two (D) independent experiments. 637 638 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 17 Figure 2. Contribution of the pUL21 linker region to PP1 binding. 639 A Alignment of alphaherpesvirus pUL21 homologue sequences spanning the linker between 640 pUL21 N- and C-terminal domains, plus the N-terminal domain residues that were disordered in 641 the pUL21-N crystal structure (18). The following sequences were aligned using ClustalW 642 (Abbreviation and Uniprot ID are shown in parentheses): HSV-1 (HSV1, P10205), HSV-2 (HSV2, 643 G9I242), cercopithecine herpesvirus 2 (CHV-2, Q5Y0T2), saimiriine herpesvirus 1 (SHV-1, 644 E2IUE9), bovine alphaherpesvirus 1 (BHV-1, Q65563), leporid alphaherpesvirus 4 (LHV-4, 645 J9QYM9), feline herpesvirus 1 (FHV-1, D1FXW1), equine herpesvirus 1 (EHV-1, P28972), 646 pseudorabies virus (PRV , Q04532), anatid herpesvirus 1 (AHV-1, A4GRJ2), varicella-zoster virus 647 (VZV , Q6QCT9), turkey herpesvirus (MHV-1, Q9DPR5). The regions of the linker preceding (L 648 region, residues 216–237, grey) or following (R region, residues 257–276, grey) the TROPPO motif 649 (pink) are highlighted. B HEK293T cells were transfected with plasmids expressing wild-type 650 (WT) or mutated HSV-1 pUL21-GFP or VZV pORF38-GFP . At 24 hours post-transfection the cells 651 were lysed, subjected to immunoprecipitation using a GFP affinity resin, and captured proteins 652 were subjected to SDS-PAGE and immunoblotting using the listed antibodies. Ponceau S (PonS) 653 staining of the nitrocellulose membrane before blocking is shown, confirming efficient capture 654 of GFP-tagged proteins. Figure is representative of two independent experiments. 655 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 18 Figure 3. HSV-1 pUL21 and VZV pORF38 are predicted to bind the ‘RVxF’ hydrophobic groove 656 of PP1. 657 A,B AlphaFold2-Multimer models of human PP1γ catalytic domain (residues 7–300; yellow 658 ribbons) in complex with A HSV-1 pUL21 (purple ribbons) or B VZV pORF38 (rose blush ribbons), 659 with N- and C-terminal domains in lighter and darker shades, respectively. For each model, 660 regions of the linker between the N- and C-terminal domains are coloured grey and the TROPPO 661 motif is in bright pink. In A the region of the conserved N-terminal domain missing from the 662 crystal structure (18) is light pink and inset shows the packing of predicted β-sheets at the 663 pUL21-N:PP1 interface. C Per-residue predicted local distance difference test (pLDDT) of the 664 pUL21:PP1 (top) and pORF38:PP1 (bottom) models, where values above 70 represent residues 665 predicted with high local confidence. Domains are coloured as in (A, B). D Predicted aligned 666 error of pUL21:PP1 (left) and pORF38:PP1 (right) models. Regions of the matrix where the 667 relative orientations of residues are predicted with high confidence are shown in green. E 668 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 19 Superposition of the pUL21:PP1 predicted structure on the crystal structure of PP1α in complex 669 with the GADD34 RVxF+ϕϕ[xF] motif (PDB 4XPN) (29). PP1 is shown as a molecular surface 670 coloured by mean lipophilicity, from orange (hydrophobic) to cyan (polar), and PP1-binding 671 motifs are shown as ribbons. F Structure-based alignment of the pUL21 and GADD34 PP1-672 binding sequences, with SLiMs highlighted. G Predicted molecular interactions at the RVxF 673 binding groove of PP1. Selected sticks are shown with carbon atoms green (GADD34) or 674 coloured as in (A) (PP1 and pUL21). H Predicted interactions between pUL21-N and residues of 675 the TROPPO motif. Selected sticks are shown with carbon atoms coloured as in (A). 676 677 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 20 678 Figure 4. Site-directed mutagenesis supports pUL21 binding the hydrophobic RVxF groove 679 of PP1. 680 A Model of the pUL21:PP1 complex, highlighting predicted interactions with PP1 residues 681 important for binding regulatory subunits containing RVxF motifs. Structures are shown as 682 ribbons for PP1 (yellow), the pUL21 linker (grey), TROPPO (bright pink), and residues of pUL21-N 683 missing from the crystal structure (light pink) (18). Selected residues shown as sticks. B 684 HEK293T cells were co-transfected with plasmids expressing pUL21-GFP and wild-type (WT) or 685 mutated Myc-tagged human PP1α. At 24 hours post-transfection the cells were lysed, subjected 686 to immunoprecipitation using a GFP affinity resin, and captured proteins were subjected to 687 SDS-PAGE and immunoblotting using the listed antibodies. Ponceau S (PonS) staining of the 688 nitrocellulose membrane before blocking is shown, confirming efficient capture of GFP-tagged 689 proteins. Figure is representative of at least three independent experiments (see below). C 690 Quantitation of Myc-PP1 co-immunoprecipitation by pUL21-G F P. Ratio of bound to input signals 691 as determined by densitometry is shown as individual points and mean ± SD from seven (WT), 692 five (F257A and L289R) or three (C291R) independent experiments. One-way ANOVA with Holm-693 Sidak’s multiple comparisons test was used for the statistical analysis (***p < 0.001). D Model 694 of the pUL21:PP1 complex, highlighting pUL21 residues at the interfaces with PP1 and pUL21-N. 695 Ribbons are shown with selected residues represented as sticks, coloured as in (A). E HEK293T 696 cells were transfected with plasmids expressing wild-type (WT) or mutated HSV-1 pUL21-GFP . 697 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 21 At 24 hours post-transfection the cells were lysed, subjected to immunoprecipitation using a 698 GFP affinity resin, and captured proteins were subjected to SDS-PAGE and immunoblotting 699 using the listed antibodies. Ponceau S (PonS) staining of the nitrocellulose membrane before 700 blocking is shown, confirming efficient capture of GFP-tagged proteins. Figure is representative 701 of two independent experiments. 702 703 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 22 704 Figure 5. Alphaherpesvirus pUL21 homologues compete with cellular RVxF motifs for 705 binding PP1. 706 A Schematic of pORF38-NLT, which encodes the N-terminal domain plus Linker region to the 707 end of the TROPPO motif. B Coomassie-stained SDS-PAGE of purified C-terminally 708 hexahistidine-tagged pORF38-NLT. C ITC of PP1γ catalytic domain with pORF38-NLT. Baseline-709 corrected differential power (DP) versus time (left) and integrated change in enthalpy (ΔH) 710 versus molar ratio (right) is shown. Figure is representative of three independent experiments 711 (Table S1), and average values for affinity (KD), stoichiometry (N) and ΔH are shown. D 712 Competition fluorescence anisotropy confirms that pORF38-NLT competes with RVxF+ϕϕ[xF] 713 peptides for PP1 binding. A pre-formed complex of 2 nM fluorescent GADD34 RVxF+ϕϕ[xF] 714 peptide with 250 nM PP1 was incubated with increasing concentrations of pORF38-NLT, 715 resulting in displacement of the GADD34 peptide. Mean ± SD is shown for technical triplicate 716 measurements. Average IC50 from three independent experiments is shown. E Coomassie-717 stained SDS-PAGE of purified C-terminally hexahistidine-tagged pUL21 WT and A237V+V239F 718 (RVxF) mutant. F Competition fluorescence anisotropy of WT (purple) and RVxF (blue) pUL21. A 719 pre-formed complex of fluorescent 2 nM GADD34 peptide with 250 nM PP1 was incubated with 720 increasing concentrations of pUL21. Mean ± SD is shown for technical triplicate measurements. 721 Average IC50 from three independent experiments is shown. 722 723 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 23 724 Figure S1. Comparison of (predicted) peptide conformations at the PP1 hydrophobic 725 groove. 726 A–C PP1 is shown as a molecular surface coloured by mean lipophilicity, from orange 727 (hydrophobic) to cyan (polar). A Structure of the crystal structure of PP1α catalytic domain in 728 complex with the GADD34 RVxF+ϕϕ[xF] motif (PDB 4XPN) (29). The GADD34 peptide is shown 729 as sticks with green carbon atoms. B AlphaFold2-Multimer model of PP1γ catalytic domain in 730 complex with HSV-1 pUL21. The TROPPO motif (pink carbon atoms) and flanking residues (grey 731 alpha carbons) are shown as sticks. C AlphaFold2-Multimer model of PP1γ catalytic domain in 732 complex with VZV pORF38. The TROPPO motif (pink carbon atoms) and flanking residues (grey 733 alpha carbons) are shown as sticks. 734 735 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint 24 736 Figure S2. Association of GADD34 peptide and PP1 in Tris pH 8.5 buffer. 737 Fluorescence anisotropy of PP1γ catalytic domain (residues 7–300) binding 2 nM fluorescently 738 labelled peptide containing the GADD34 RVxF and ϕϕ[xF] motifs (residues 552–568). Mean ± SD 739 is shown for technical triplicate measurements. Affinity (KD) is average of three independent 740 experiments. 741 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 25, 2025. ; https://doi.org/10.1101/2025.07.22.666160doi: bioRxiv preprint