Human cytomegalovirus protein pUL36: A dual cell death pathway inhibitor

Significance Cell death is a key defense against viral infection, preventing spread from infected to uninfected cells. Correspondingly, certain viruses encode inhibitors of apoptotic and necroptotic cell death pathways in order to facilitate their persistence. Human cytomegalovirus (HCMV) is an important human pathogen that can block apoptosis, but hitherto it has been unclear whether or how the virus blocks necroptosis. Here, we used a proteomic screen to identify human proteins targeted for destruction by HCMV, finding that the key necroptosis mediator MLKL is degraded throughout infection. MLKL is targeted for degradation by HCMV protein pUL36, which is also instrumental in inhibiting apoptosis. Thus, pUL36 is a dual cell death pathway inhibitor, and may represent an important therapeutic target.


Cells and cell culture
All fibroblast cell lines generated in this paper were derived from human foetal foreskin fibroblasts immortalised with human telomerase (HFFF-TERTs) (1). HFFF

Viruses
The genome sequence of HCMV strain Merlin (GenBank accession AY446894) is designated the reference HCMV sequence (RefSeq accession NC_006273.2) by the National Center for Biotechnology Information (3,4). A recombinant version (RCMV1111) of this strain was derived by transfection of a sequenced BAC clone (4). RCMV1111 contains point mutations in two genes (RL13 and UL128) that enhance replication in fibroblasts (4). The block HCMV deletion mutants are described in the Nightingale et al. study (5). The AD169-GFP virus (RCMV288) is described in the McSharry et al. study (1). RCMV1502 is an RCMV1111 recombinant that has tet-operators 5' to the RL13 and UL128 coding sequences (Merlin-RL13tetO-UL128tetO,) and is described in the Stanton et al. study (4). UL36 deletion mutants were generated by recombineering RCMV1502 (4).
Two deletion mutants were generated, one lacking the whole UL36 coding sequence (ΔUL36; RCMV2288), and the other lacking the second UL36 exon (ΔUL36ex2; RCMV2289). Virus stocks were prepared in HFFF-TERTs or HFFF-Tets (for tet-regulated viruses) as described in the Stanton et al. study (6). Whole-genome consensus sequences of passage 1 of each RCMV were determined using the Illumina platform as described previously (7).

Viral infections
The required volume of viral stock to achieve the multiplicity of infection (MOI) described in the results section was diluted in serum-free DMEM, mixed gently and applied to HFFF-TERTs. Mock infections were performed identically but with DMEM instead of viral stock. Time zero was considered the time at which cells first came into contact with virus. Cells were incubated with virus for 2 h at 37 °C on a rocking platform, and then the medium was replaced with DMEM/FBS.

Proteomic Screen
The 48 h degradation screen in this paper was performed in biological duplicate. Cellular lysates from the second 48 h biological replicate were analysed simultaneously with residual lysates from the 12 h degradation screen in the Nightingale et al. study (5), thus facilitating a direct comparison.

Infection
In Experiment 1, 3×10 6 HFFF-TERTs were plated in DMEM/FBS/PS in 75 cm 2 flasks. After 24 h, the medium was replaced by serum-free DMEM containing 4 μg/ml dexamethasone, which has been shown to improve infection efficiency (8,9). After a further 24 h, the medium was removed and the cells were infected as described above at an MOI of 10. At 36 hours post-infection (hpi), 10 µM MG132 (Merck) or the equivalent volume of DMSO (Sigma-Aldrich) was added to the cells. In Experiment 2, 8×10 5 HFFF-TERTs were seeded into 25 cm 2 flasks, and then treated as in Experiment 1.

Whole Cell Lysate Protein Digestion
Methods for whole cell lysate protein preparation and digestion, peptide labelling with tandem mass tags, HpRP fractionation, liquid chromatography-mass spectrometry and data analysis are discussed in detail in the Nightingale et al. study (5), and are recapitulated below including modifications for the present study.
At 48 hpi, cells were washed twice with PBS, and 500 µl lysis buffer was added (6 M guanidine/50 mM HEPES pH 8.5). Cell lifters (Corning) were used to scrape the cells into lysis buffer, which was then removed to an eppendorf tube, vortexed extensively and sonicated. Cell debris was removed by centrifuging twice at 21,000 g for 10 min.
From this point onward, lysates from the second biological replicate (Experiment 2) were treated identically to residual lysates from a 12 h degradation screen that we performed previously (5).
Dithiothreitol (DTT) was added to a final concentration of 5 mM and incubated at room temperature for 20 mins. Cysteine residues were alkylated with 15 mM iodoacetamide and incubated for 20 min at room temperature in the dark. Excess iodoacetamide was quenched with DTT for 15 mins.
Samples were diluted with 200 mM HEPES pH 8.5 to 1.5 M guanidine followed by digestion at room temperature for 3 h with LysC protease at a 1:100 protease-to-protein ratio. Samples were further diluted with 200 mM HEPES pH 8.5 to 0.5 M guanidine. Trypsin was then added at a 1:100 protease-to-protein ratio followed by overnight incubation at 37°C. The reaction was quenched with 5% formic acid and centrifuged at 21,000 g for 10 min to remove undigested protein. Peptides were subjected to C18 solid-phase extraction (SPE, Sep-Pak, Waters) and vacuum-centrifuged to near-dryness.

Peptide Labelling with Tandem Mass Tags
In preparation for TMT labelling, desalted peptides were dissolved in 200 mM HEPES pH 8.5.
Peptide concentration was measured by microBCA (Pierce), and 25 µg of peptide was labelled with TMT reagent. TMT reagents (0.8 mg, Thermo Scientific) were dissolved in 43 µl anhydrous acetonitrile and 3 µl were added to each peptide sample at a final acetonitrile concentration of 30% (v/v). Samples were labelled as follows:  (TMTpro 133N). Following incubation at room temperature for 1 h, the reaction was quenched with hydroxylamine to a final concentration of 0.5% (v/v). TMT-labelled samples were combined at a 1:1:1 ratio (Experiment 1) or 1:1:1:1:1:1 ratio (Experiment 2). The sample was vacuum-centrifuged to near dryness and subjected to C18 SPE (Sep-Pak, Waters). An unfractionated single-shot was analysed initially to ensure similar peptide loading across each TMT channel, thus avoiding the need for excessive electronic normalisation. In Experiment 1, data from the single-shot experiment was analysed with data from the corresponding fractions to increase the overall number of peptides quantified. In Experiment 2, data from 13 fractions were used in the analysis. Normalisation is discussed in 'Data Analysis' and high pH reversed-phase (HpRP) fractionation is discussed below.

Offline HpRP Fractionation
TMT-labelled tryptic peptides were subjected to HpRP fractionation using an Ultimate 3000 RSLC UHPLC system (Thermo Fisher Scientific) equipped with a 2.1 mm internal diameter (ID) x 25 cm long, 1.

Generation of lentiviral expression vectors
cDNA was generated from HFFF-TERTs by reverse transcription of RNA using an RNeasy Mini Kit (Qiagen) followed by GoScript Reverse Transcriptase (Promega) according to the manufacturer's protocol. To generate an expression construct for MLKL-HA, primers were designed to recognise the 3' and 5' ends of the MLKL gene and contained flanking Gateway attB sequences to facilitate cloning into pDONR223 using the Gateway system (Thermo Scientific). The reverse primer additionally contained a 6 bp linker region, followed by the coding sequence for an HA tag and a stop codon (Dataset S3A).
For expression of the V5-tagged viral genes, recombinant adenovirus vectors (RAds) were used as a template as described previously (9). Each template expressed a C-terminally V5-tagged gene under the control of the HCMV major immediate early promoter, with a 6 bp linker region between the end of the gene and the tag. To amplify genes from the RAds, primers were designed to recognise the 3' end of the HCMV promoter (forward 'GAW-CMVp-F') and the 3' end of the V5 tag (reverse 'attB2-V5-R') (Dataset S3A). Both primers had flanking Gateway attB sequences.
Spliced UL150A was synthesized as double-stranded DNA fragment (gBlocks®, Integrated DNA Technologies) comprising the viral gene succeeded by a 6 bp linker region, the coding sequence for the V5 tag and then the stop codon, and Gateway attB sequences.
PCR amplification of MLKL-HA and the viral genes was achieved using PfuUltra II fusion HS DNA polymerase (Agilent). PCR products were subsequently purified using a QIAquick Gel Extraction Kit (Qiagen), cloned into the pDONR223 entry vector, and then into the lentiviral destination vector pHAGE-pSFFV (described in the Nightingale et al. study (5)) (12). The sequence of MLKL was confirmed as the long, necroptotic isoform and shown to contain two non-diseasecausing natural variants (S52T and M169L); this and the confirmed sequences of HCMV genes used in this study are shown in Dataset S3B. The UL36 gene sequence contains a single intron.
Confirmation of splicing was achieved by reverse transcription of RNA from UL36-expressing HFFF-TERTs and sequencing of the cDNA (Dataset S3B). The pHAGE-SFFV control vector contains a short randomized DNA sequence (5).

PCR site-directed mutagenesis to generate UL36 point mutants
In the first round of PCR, two pairs of primers were used in two separate PCRs using strain Merlin UL36-V5 in the pDONR223 vector as a template. In the first reaction, reverse PCR primers encompassing each of the point mutants were combined with the UL36 attB1 fwd primer (Dataset S3C). In the second reaction, forward PCR primers encompassing each of the point mutants were combined with the UL36 V5 attB2 rev primer (Dataset S3C). A second round of PCR joined the two halves of UL36 by using the UL36 attB1 fwd and UL36 V5 attB2 rev primers. Five resulting PCR products encoding each of the different point mutants flanked by attB sites were cloned into pDONR223 and then pHAGE-pSFFV by Gateway cloning, and the sequences were confirmed.

LC-MS/MS for IP experiments
Mass spectrometry data were acquired using an Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific, San Jose, CA). Peptides were separated using an Ultimate 3000 RSLC nano UHPLC equipped with a 300 µm ID x 5 mm Acclaim PepMap µ-Precolumn (Thermo Fisher Scientific) and a 75 µm ID x 50 cm 2.1 µm particle Acclaim PepMap RSLC analytical column. Peptides were separated using a 3 hr gradient of 3-37% acetonitrile in 0.125% formic acid at a flow rate of 250 nL/min. The following settings were used: MS1: 350-1500 Th, Quadrupole isolation, 120,000 Resolution, 2×10 5 AGC target, 50 ms maximum injection time, ions injected for all parallisable time. MS2: Quadrupole isolation at an isolation width of m/z 0.7, HCD fragmentation (NCE 34) with ion trap scanning out in rapid mode from m/z 120, 10000 AGC target, 250 ms maximum injection time, in centroid mode. Dynamic exclusion was set to +/-10 ppm for 25 s. MS2 fragmentation was trigged on precursors 5×10 3 counts and above.

Sample preparation for immunoblotting
Proteins bound to the resin were eluted once with 40 µl 2.5 mg/ml V5 or HA peptide at 37 °C for 1 hour with agitation. Eluted proteins were reduced with 6× Protein Loading Dye (Tris 375 mM pH 6.8, 12% SDS, 30% glycerol, 0.6M DTT, 0.06% bromophenol blue) at 95 °C for 5 minutes prior to separation by SDS-PAGE as described above. For the input blot, 2% of the original lysates were removed prior to IP and heated in 6× Loading Dye at 95 °C for 5 minutes prior to SDS-PAGE.

Cell death assays
All experiments were performed in biological triplicate conducted in parallel and repeated in 2 or

Data Analysis
Mass spectra were processed using a Sequest-based software pipeline for quantitative proteomics,  (14). Peptide spectral matches (PSMs) were filtered to an initial peptide-level false discovery rate (FDR) of 1% with subsequent filtering to attain a final protein-level FDR of 1%. PSM filtering was performed using a linear discriminant analysis, as described previously (14). This distinguishes correct from incorrect peptide IDs in a manner analogous to the widely used Percolator algorithm (https://noble.gs.washington.edu/proj/percolator/), though employing a distinct machine-learning algorithm. The following parameters were considered: XCorr, DCn, missed cleavages, peptide length, charge state, and precursor mass accuracy. Protein assembly was guided by principles of parsimony to produce the smallest set of proteins necessary to account for all observed peptides (algorithm described in (14)). Where all PSMs from a given HCMV protein could be explained either by a canonical gene or non-canonical ORF, the canonical gene was picked in preference. In a small number of cases, PSMs assigned to a non-canonical or 6FT-ORF were a mixture of peptides from the canonical protein and the ORF. This most commonly occurred where the ORF was a 5'terminal extension of the canonical protein (thus meaning that the smallest set of proteins necessary to account for all observed peptides included the ORFs alone). In these cases, the peptides corresponding to the canonical protein were separated from those unique to the ORF, generating two separate entries. In a single case, PSM were assigned to the 6FT-ORF 6FT_6_ORF1202_676aa, which is a 5'-terminal extension of the non-canonical ORF ORFL147C.
The principles described above were used to separate these two ORFs. Proteins were quantified by summing TMT reporter ion counts across all matching peptide-spectral matches using ''MassPike'', as described previously (10). Briefly, a 0.003 Th window around the theoretical m/z of each reporter ion was scanned for ions and the maximum intensity nearest to the theoretical m/z was used. The primary determinant of quantitation quality is the number of TMT reporter ions detected in each MS3 spectrum, which is directly proportional to the signal-to-noise (S:N) ratio observed for each ion. Conservatively, every individual peptide used for quantitation was required to contribute sufficient TMT reporter ions so that each on its own could be expected to provide a representative picture of relative protein abundance (10). An isolation specificity filter with a cutoff of 50% was additionally employed to minimise peptide co-isolation (10). Peptide-spectral matches with poor quality MS3 spectra (a combined S:N ratio of less than 120 (Experiment 1) or 240 (Experiment 2) across all TMT reporter ions) or no MS3 spectra at all were excluded from quantitation. Peptides meeting the stated criteria for reliable quantitation were then summed by Although peptides were assigned appropriately to HLA-A alleles, it was not possible to assign peptides confidently to only two HLA-B or HLA-C alleles, and signal:noise values were further summed for each of these alleles to give a single combined result for HLA-B or HLA-C.

Statistical analysis
Figures 2, S1A and S4A. The method of Significance B was used to estimate the p-value that the fold downregulation or rescue was significantly different to 1 (16). Values were calculated and corrected for multiple hypothesis testing using the method of Benjamini-Hochberg in Perseus version 1.5.2.20 (16).   were derived from ratios generated from averaged data from experiments 1 and 2 using Significance B values ( Figure 2B) *p<0.05, **p<0.001, ***p<1×10 -7 . A full list of proteins degraded late in infection can be found in Dataset S1, including details of which are also degraded early. (B) K-means-based hierarchical cluster analysis of all human proteins quantified ( Figure   2B). K-means clustering with 1-20 classes was used to assess the summed distance of each protein from its cluster centroid. While this summed distance necessarily became smaller as more clusters were added, the rate of decline decreased with each added group, eventually settling at a fairly constant rate of decline that reflected over-fitting; clusters added prior to this point reflected underlying structure in the protein data, while clusters subsequently added through over-fitting were not informative. The point of inflexion fell at or after nine classes, indicating that there were at least nine classes of proteins displaying different expression profiles across the protein samples.
The panels on the right show examples of proteins belonging to classes 1, 4, 5 and 9, which contain the 7 proteins degraded early and late during infection ( Figure 2C). A full list of proteins in each cluster can be found in Dataset S1C.  (19). Ser125 is phosphorylated in a cell-cycle dependent manner but its effect on MLKL function is unknown (20,21). No phospho-MLKL peptides were identified.

Supplementary Information Legends for Datasets
Dataset S1 (separate .xls file). Interactive spreadsheet of all TMT-based proteomic data in this paper. The 'Plotter' worksheet generates graphs of protein abundance for any human or viral protein quantified across each experiment. The 'Data' worksheet shows minimally annotated protein data, for which the only modifications are formatting, deletion of contaminants, normalisation and reassignment of non-canonical HCMV ORFs. The 'K-means clusters' worksheet shows the human proteins belonging to each cluster from Figure