Cell culture, treatment, harvest, and lysis

HEK293 (human, DMSZ, ACC 635) and U2OS (human, American Type Culture Collection [ATCC], HTB-96) cells were cultivated in DMEM (Gibco, Invitrogen) supplemented with 10% fetal bovine serum (Gibco, Invitrogen), 100 U/ml penicillin (Gibco, Invitrogen), and 100 µg/ml streptomycin (Gibco, Invitrogen) at 37 °C in a humidified incubator with a 5% CO₂ atmosphere. For cell harvest, cells were washed twice with ice-cold PBS (Gibco, Invitrogen), centrifuged, snap-frozen in liquid nitrogen, and stored at −80 °C until lysis. Frozen cell pellets were lysed by adding lysis buffer (1% SDC in 100 mM Tris/HCl, pH 8.5) directly onto frozen cell pellets, followed by repeated aspiration and boiling for 5 min at 95 °C.

For proteasome inhibition, HEK293 or U2OS cells were treated with 10 µM MG132 (InvivoGen) at ~80% confluence for 4 h and successively harvested. For circadian cycle experiments, cells were synchronized, when they reached at least 90% confluence, with dexamethasone (1 μM) for 1 h. Following this, U2OS were washed once with PBS and the medium was replaced. The first time point was collected after 24 h of synchronization continuing the collection every 4 h across 32 h for each of the 4 biological replicates. Collected cells were stored and lysed as described. For TNF stimulation of U2OS cells, confluent cultures were either stimulated with 100 ng/ml TNF for 10 min or left unstimulated. Cells were washed 3× with ice-cold PBS, directly lysed with lysis buffer, and boiled for 5 min at 95 °C.

Western blot analysis

To validate TNF-signaling induction U2OS cells were plated in 6-well plates and when confluent stimulated for 5, 10, 15, 30, and 60 min with 100 ng/ml TNF or left untreated. After stimulation, cells were washed in PBS and lysed in 4% SDS in 100 mM Tris/HCl, pH 8. Lysates were boiled, sonicated, and protein concentrations were estimated using BCA. SDS sample loading buffer (450 mM Tris-HCl, pH 8, 60% (v/v) glycerol, 12% (w/v) SDS, 0.02% (w/v) bromophenol blue, 600 mM DTT) was added to lysates before separation on 12% Novex Tris-glycine gels (Thermo Fisher Scientific, XP00120BOX). Separated proteins were transferred onto PVDF membranes (Merck Millipore, IPVH00010). Membranes were blocked in 5% BSA in PBST and antibodies diluted in 2% BSA in PBST.

For validating increased RIPK1 and TRAF2 ubiquitination upon TNF treatment U2OS cells were either left untreated or stimulated with TNF (100 ng/ml) for 10 min, washed in PBS, and lysed in 1 ml DISC buffer (150 mM NaCl, 30 mM Tris pH 7.5, 10% glycerol, 1% Triton X-100) with protease inhibitors (Roche) and 10 mM n-ethyl-maleimide (NEM). Samples were clarified by centrifugation at 16,000 × g for 15 min, equalized to 1 mg of protein, and added directly to 20 µl packed glutathione sepharose beads pre-bound with 100 µg GST-UBA (Ubiquillin-UBA x1)¹³. Beads were incubated on a rotating wheel at 4 °C overnight, washed five times with DISC buffer, and eluted with 2× SDS sample buffer. Proteins were separated on a 10% Novex Tris-glycine gel (Thermo Fisher Scientific, XP00105BOX) and transferred onto a nitrocellulose membrane (GE Healthcare Lifescience, 10600002).

Antibodies (diluted 1:1000) used for immunoblotting were as follows: anti-phospho p65 (CST, 3033 P), anti-p65 (CST, 4764 P), anti-IκBα (CST, 9242), anti-phospho p38 (CST, 9215), anti-p38 (CST, 9212), anti-β-actin (CST, 4970) for TNF-signaling validation and anti-RIP (BD Bioscience, 610458), anti-TRAF2 (CST, 4712), and anti-β-actin (Santa Cruz, sc-47778) for validation of increased RIPK1 and TRAF2 ubiquitination.

RNA isolation and QPCR

RNA was isolated from three biological replicates of each U2OS time point according to manufacture instruction using the RNeasy Plus Mini Kit (QIAGEN, #74134). Isolated RNA was reversely transcribed by using first-strand cDNA synthesis kit (Thermo Fisher Scientific, #K1612). QPCR was performed at the C1000 Thermal Cycler (Bio-Rad) with iQ^TM SYBR Green Supermix (Bio-Rad, #170-8862) with primers for Bmal1 (froward: caggaaaaataggccgaatg; reverse: gcgatgaccctcttatcctg), Per1 (forward: ggacactcctgcgaccag; reverse: gggagtgaggtggaagatctaa), and Gapdh (forward: agccacatcgctcagacac; reverse: gcccaatacgaccaaatcc). The in-build analysis tool of the CFX Manager Software (Version 3.1, Bio-Rad) was used to determine the normalized expression with the ∆∆Cq method of Bmal1 and Per1 compared to Gapdh in technical triplicates for all three biological replicates of each time point. The technical triplicates were further averaged and adjusted so that the highest value was set to 1. Following this, the average of all biological replicates and the SEM (standard error of the mean) was calculated for all the time points.

Protein digestion and peptide cleanup

Lysates were sonicated for 1 min (Branson Sonifier) and protein concentrations were estimated by tryptophan assay. After addition of CAA and TCEP to a final concentration of 10 and 40 mM, respectively, samples were incubated for 5 min at 45 °C for protein reduction and alkylation. Thereafter, Samples were digested overnight at 37 °C using trypsin (1:100 w/w, Sigma-Aldrich) and LysC (1/100 w/w, Wako).

For proteome analysis, sample aliquots (~15 µg) were desalted in SDB-RPS StageTips (Empore). Briefly, samples were first diluted with 1% TFA in isopropanol to a final volume of 200 µl. Thereafter, samples were loaded onto StageTips and sequentially washed with 200 µl of 1% TFA in isopropanol and 200 µl 0.2% TFA/2% ACN. Peptides were eluted with 60 µl of 1.25% ammonium hydroxide (NH₄OH)/80% ACN and dried using a SpeedVac centrifuge (Eppendorf, Concentrator plus). Dried peptides were resuspended in buffer A* (2% ACN/0.1% TFA) supplemented with iRT peptides (1/30 v/v) (iRT Standard, Biognosys).

For diGly peptide enrichment, samples were four-fold diluted with 1% TFA in isopropanol and loaded onto SDB-RPS cartridges (Strata™-X-C, 30 mg/3 ml or Strata™-X-C, 200 mg/6 ml, Phenomenex Inc.). Before peptide loading, cartridges were equilibrated with 8 bed volumes (BV) of 30% MeOH/1% TFA and washed with 8 BV of 0.2% TFA. Samples were loaded by gravity flow and sequentially washed twice with 8 BV 1% TFA in isopropanol and once with 8 BV 0.2% TFA/2% ACN. Peptides were eluted twice with 4 BV 1.25% NH₄OH/80% ACN and diluted with ddH₂O to a final ACN concentration of 35% ACN. Thereafter, samples were snap-frozen in liquid nitrogen, lyophilized, and stored at 4 °C until diGly peptide enrichment.

DiGly peptide enrichment

Lyophilized peptides were resuspended in immunoaffinity purification buffer (IAP) (50 mM MOPS, pH 7.2, 10 mM Na₂HPO₄, 50 mM NaCl) and sonicated for 2.5 min (Bioruptor plus, Diagenode). Peptide concentration was estimated by tryptophan assay. DiGly remnant containing peptides were enriched using the PTMScan® Ubiquitin Remnant Motif (K-ɛ-GG) Kit (Cell Signaling Technology (CST)), which was kindly provided by CST. First, antibodies were cross-linking to beads. Following Udeshi et al.²² 1 vial of antibody coupled beads were first washed 3 times with 1 ml cold cross-linking wash buffer (100 mM sodium tetraborate decahydrate, pH 9.0), followed by 30 min incubation in 1 ml cross-linking buffer (20 mM dimethylpimipimidate cross-linking wash buffer) for 30 min at room temperature and gentle agitation. The cross-linking reaction was quenched by two consecutive washes with 1 ml cold quenching buffer (200 mM ethanolamine, pH 8.0) and 2 h incubation in 1 ml quenching buffer at room temperature under gentle agitation. After quenching cross-linked beads were washed three times with 1 ml of cold IAP and used directly for diGly peptide enrichment or stored in 1 ml 0.02% sodiumazide in phosphate-buffered saline, pH 7.4. Unless otherwise stated, 1/8 of a vial of cross-linked antibody beads and 1 mg of peptide material were used for diGly peptide enrichments. For this, peptides were added to cross-linked antibody beads and the volume was adjusted to 1 ml with IAP buffer. After 1 h of incubation at 4 °C and gentle agitation, beads were washed twice with cold IAP and five times with cold ddH₂O. For this, beads were transferred into GF-filter StageTips and for each wash step, the according wash solution was added and passed through by centrifugal force. Thereafter, GF-StageTips were stacked onto SDB-RPS StageTips and peptides were directly eluted into SDB-RPS StageTips. For this, 50 µl 0.15% TFA were added twice onto the beads and passed through by centrifugation for 5 min at 100 × g. Thereafter, 100 µl of 0.2% TFA was added on top of peptide eluates, followed by sample loading onto the stationary material of SDB-RPS StageTips. Peptides were washed, eluted, and dried as described for proteomes samples, with the difference, that 0.2% TFA was used for the first wash step. Dried peptides were resuspended in 9 µl buffer A*, supplemented with iRT peptides (1/30 v/v) for LC/MS-MS analysis.

Basic reversed-phase fractionation

Basic reversed-phase (bRP) fractionation for diGly peptide and proteome spectral libraries were performed on an UFLC System (Shimadzu) and EASY-nLC 1000 (Thermo Fisher Scientific, Germany), respectively.

For diGly peptide separation, lyophilized samples were resuspended in Buffer A (5 mM NH₄HCO₂/2% ACN) and 5 mg peptide material (5 mg/ml) was loaded onto a reversed-phase column (ZORBAX 300Extend-C18, Agilent). Peptides were separated at a flow rate of 2 ml/min and a constant column temperature of 40 °C using a binary buffer system, consisting of buffer A and buffer B (5 mM NH₄HCO₂/90% ACN). An elution gradient at 0% B stepwise increased to 28 in 53 min and to 78 in 6 min was deployed. Eluting peptides were automatically collected into a 96-deepwell plate while well positions were switched in 40 s intervals.

For peptide fractionation on the EASY-nLC 1000 system, ~55 µg peptide material were loaded onto a 30 cm in-house packed, reversed-phase columns (250-μm inner diameter, ReproSil-Pur C18-AQ 1.9 μm resin [Dr. Maisch GmbH]). Peptides were separated at a flow rate of 2 µl/min using a binary buffer system of buffer A (PreOmics) and buffer B (PreOmics). An elution gradient at 3% B stepwise increased to 30% in 45 min, 60% in 17 min, and 95% in 5 min was used. Eluting peptides were concatenated into 24 fractions by switching the rotor valve of an automated concatenation system (Spider fractionator, PreOmics)⁸⁰ in 90 s intervals.

Library sample preparation

For individual deep diGly libraries, 2 × 5 mg of peptide was fractionated by bRP fractionation. For K48-peptide containing fraction identification, 100 µl aliquots of fractions 46 to 54 were dried in a SpeedVac, resuspended in A*, and measured on an LTQ Orbitrap XL mass spectrometer. K48-peptide containing fractions of both plates were pooled in sample pool “K48” (Supplementary Fig. 1a). Remaining fractions of both pates were concatenated into P1–P8 (Supplementary Fig. 1a), snap-frozen, and lyophilized. Lyophilized peptides were resuspended in 1 ml IAP buffer and diGly peptides were enriched as described above. In case of HEK293 library generation, an optional second supernatant IP was conducted. For this, 500 µl of previous diGly peptide enrichment supernatants were pooled as indicated (Supplementary Fig. 1a) and used for sequential diGly peptide enrichment.

For the proteome library, aliquots of U2OS samples for proteome cycling analysis were used. Approximately 3 µg peptide material of individual time points of two biological replicates, after SDB-RPS cleanup, were pooled and fractionate via bRP fractionation as described above. Fractionated samples were dried using a SpeedVac and resuspended in A* supplemented with iRT peptides (1/30 v/v) for LC-MS/MS measurement and spectral library generation.

Nanoflow LC-MS/MS proteome measurements

Peptides were loaded onto a 50 cm, in-house packed, reversed-phase columns (75 μm inner diameter, ReproSil-Pur C18-AQ 1.9 μm resin [Dr. Maisch GmbH]). The column temperature was controlled at 60 °C using a homemade column oven and binary buffer system, consisting of buffer A (0.1% formic acid (FA)) and buffer B (0.1% FA in 80% ACN), was utilized for low pH peptide separation. An EASY-nLC 1200 system (Thermo Fisher Scientific), directly coupled online with the mass spectrometer (Q Exactive HF-X, Thermo Fisher Scientific) via a nano-electrospray source, was employed for nanoflow liquid chromatography, at a flow rate of 300 nl/min. For individual measurements, 500 ng of peptide material was loaded and eluted with a gradient starting at 5% buffer B and stepwise increased to 30% in 95 min, 60% in 5 min, and 95% in 5 min.

The same general setup was used, for K48-peptide containing fraction identification, while the column and mass spectrometer were changed to a 20 cm column and an LTQ Orbitrap XL, respectively.

For DDA experiments, the Thermo Xcalibur (4.0.27.19) and LTQ Tune plus (2.5.5 SP2) software were used for Q Exactive HF-X and LTQ Orbitrap XL instruments, respectively. The Q Exactive HF-X was operated in Top12 mode with a full scan range of 300–1650 m/z at a resolution of 60,000. The automatic gain control (AGC) was set to 3e6 at a maximum injection time of 20 s. Precursor ion selection width was kept at 1.4 m/z and fragmentation was achieved by higher-energy collisional dissociation (HCD) (NCE 27%). Fragment ion scans were recorded at a resolution of 15,000, an AGC of 1e5 and a maximum fill time of 60 ms. Dynamic exclusion was enabled and set to 20 s. The LTQ Orbitrap XL was operated in Top10 mode with a full scan range of 300–1700 m/z at a resolution of 60,000. Precursor ion selection width was kept at 2.0 m/z and fragmentation was achieved by collision-induced dissociation (CID) (NCE 35%).

For DIA analysis, the MaxQuant Live software suite was utilized for data acquisition⁸¹. The full scan range was set to 300–1650 m/z at a resolution of 120,000. The AGC was set to 3e6 at a maximum injection time of 60 ms. HCD (NCD 27%) was used for precursor fragmentation and fragment ions were analyzed in 33 DIA windows at a resolution of 30,000, while the AGC was kept at 3e6.

Nanoflow LC-MS/MS diGly measurements

DiGly peptide enriched samples were measured on a Q Exactive HF-X using the same instrumental setup as for proteome analysis. For diGly single-run measurements one quarter (2 µl) and for diGly library preparation one-half (4 µl) of enriched samples were loaded for LC-MS/MS analysis, unless stated otherwise. Loaded peptides were eluted using a gradient starting at 3% buffer B and stepwise increased to 7% in 6 min, 20% in 49 min, 36% in 39 min, 45% in 10 min, and 95% in 4 min.

For DDA analysis, the MS was operated in Top12 mode with a full scan range of 300–1350 m/z at a resolution of 60,000. AGC was set to 3e6 at a maximum injection time of 20 s. Precursor ion selection width was kept at 1.4 m/z and fragmentation was achieved by HCD (NCE 28%). Fragment ion scans were recorded at a resolution of 30,000, an AGC of 1e5 and a maximum fill time of 110 ms. Dynamic exclusion was enabled and set to 30 s.

For DIA analysis, the MaxQuant Live software suite was employed for data acquisition⁸¹. The full scan range was set to 300–1350 m/z at a resolution of 120,000. The AGC was set to 3e6 at a maximum injection time of 60 ms. HCD (NCD 28%) was used for precursor fragmentation and resulting fragment ions were analyzed in 46 DIA windows at a resolution of 30,000 (unless otherwise stated) and an AGC of 3e6. DIA window distribution parameters PdfMu and PdfSigma were set to 6.161865 and 0.348444, respectively, unless stated otherwise (Supplementary Data 6).

Raw data analysis

DDA raw data used for K48-peptide fraction identification and DIA and DDA comparisons were analyzed with MaxQuant (1.6.2.10) using default settings and enabled match between runs (MBR) functionality. Carbamidomethyl (C) was defined as fixed modification and Oxidation (M), Acetyl (Protein N-term), and DiGly (K) were set as variable modifications.

DDA raw data, used for spectral library construction, were processed with Spectronauts build in search engine pulsar (13.12.200217.43655)²⁸. Default settings were used for proteome spectral libraries. For diGly spectral libraries, the “Best N Fragments per peptides” maximum value was adjusted to 25. For hybrid library construction DIA raw files were processed together with DDA library raw files using the same search settings.

DIA raw files were processed using Spectronaut (13.12.200217.43655)²⁸. Proteome analysis was performed with default settings. For diGly analysis, diGly (K) was defined as an additional variable modification and PTM localization was enabled and set to 0. For dilution experiments, “XIC RT extraction window” was set to “static” with a window width of 10 min. Direct DIA searches used the same settings as described above.

Bioinformatics analysis

Data analysis was primarily performed in the Perseus software suite (1.6.7.0). For diGly site analysis, Spectronaut normal report output tables were aggregated to diGly sites using the peptide collapse plug-in tool for Perseus³⁰. DiGly sites were aggregated using the linear model-based approach and filtered for a localization probability >0.5. Data sets of both acquisition strategies, DIA and DDA, were filtered to contain >50% valid values in at least one experimental condition. Missing values were imputed based on a Gaussian normal distribution with a width of 0.3 and a downshift of 1.8. Student t-test statistics (FDR cutoff 1% or 5%; s0 = 0.1) for TNF-stimulation experiments were performed in Perseus. Fisher’s Exact GOBP Term enrichment of upregulated diGly sites and cycling diGly sites was performed on the pantherdb website (http://pantherdb.org/) and in perseus, respectively, with Benjamini Hochberg FDR correction enabled and set to a 5% cutoff. Network representation of upregulated diGly sites was performed with the STRING app (1.5.1) in Cytoscape (3.7.2). The power analysis was performed in R (3.6.2), using the ‘pwr’⁸² and ‘effectsize’⁸³ packages. The Cohen’s distance was calculated based on a fixed power of 80%, a sample size of 6 per condition and a desired significance threshold of 1%. The test was set to a “two.sample” and “two.sided” t-test. A fold-change threshold was subsequently estimated by multiplying Cohens’s distance with the pooled standard deviation separately for each peptide. We only considered sites without missing values for this analysis.

For the cycling analysis of diGly sites, data were first filtered for diGly sites identified in at least 50% across all measurements. Proteins and diGly sites raw intensities were log₂ transformed and normalized by median subtraction. For diGly site protein normalization the median values of biological quadruplicates were subtracted from normalized diGly sites. Missing values of protein data for subtraction were imputed based on a Gaussian normal distribution with a width of 0.3 and a downshift of 1.8. Cycling analysis of normalized protein and diGly site data was performed as previously described, but in this case with a period time of 24.8 h^52,53. A q-value cutoff of <0.1 and <0.33 was used to define cycling DiGly sites and proteins, respectively.

Website tool

For profile plots individual z-scores for each protein abundance normalized diGly site and the median z-score and standard error of means (SEM) were subsequently determined for each time point. The resulting median z-scores and SEM values were multiplied with the cycling amplitude of each diGly site (Perseus periodicity analysis output). For sequence visualization and protein domain annotation each diGly site location was mapped to the first UniProt ID of its assigned protein group and was visualized based on its respective protein sequence stored in the fasta file that was used for MS/MS data analysis (human fasta, downloaded 2015). The protein sequences for visualization were obtained using the ‘fasta’ functions from pyteomics^84,85. Information about protein domains was obtained from UniProt (https://www.uniprot.org/, accessed 25.05.2020), including the following categories: ‘Topological domain’, ‘Motif’, ‘Region’, ‘Repeat’, ‘Zink finger’, and ‘Domain [FT]’.

To evaluate whether multiple observed cycling diGly sites are located in a specific region on the protein, we performed a proximity analysis. Three different metrics were evaluated: (1) the average distance (In amino acids) between all observed cycling diGly sites, (2) the minimum distance between any two observed cycling diGly sites, and (3) the maximum distance between any two observed cycling diGly sites. The observed distance metrics were compared to the distances expected from a random distribution of the diGly sites of a protein across all of its lysines. 10,000 random distributions were considered, and an empirical p-value was estimated based on the fraction of random samples with a smaller or equally small distance metric as the observed cycling diGly sites. For the main analysis, diGly sites with a q-value ≤ 0.1 were considered as cycling diGly sites.

Data preprocessing and visualization for the dashboard was performed using the python programming language. Following libraries were utilized for data processing: numpy (1.18.1), pandas (0.24.2), re, random, and pyteomics^84,85 (4.2). Several libraries from the HoloViz (0.11.3) family of tools were used for data visualization and creation of the dashboard, including panel and holoviews (1.13.2), but also bokeh (2.0.1), plotly (4.6.0), and matplotlib (3.0.3).

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.