Omics Assisted N-terminal Proteoform and Protein Expression Profiling On Methionine Aminopeptidase 1 (MetAP1) Deletion*

Excision of the N-terminal initiator methionine (iMet) residue from nascent peptide chains is an essential and omnipresent protein modification carried out by methionine aminopeptidases (MetAPs) that accounts for a major source of N-terminal proteoform diversity. Although MetAP2 is known to be implicated in processes such as angiogenesis and proliferation in mammals, the physiological role of MetAP1 is much less clear. In this report we studied the omics-wide effects of human MetAP1 deletion and general MetAP inhibition. The levels of iMet retention are inversely correlated with cellular proliferation rates. Further, despite the increased MetAP2 expression on MetAP1 deletion, MetAP2 was unable to restore processing of Met-Ser-, Met-Pro-, and Met-Ala- starting N termini as inferred from the iMet retention profiles observed, indicating a higher activity of MetAP1 over these N termini. Proteome and transcriptome expression profiling point to differential expression of proteins implicated in lipid metabolism, cytoskeleton organization, cell proliferation and protein synthesis upon perturbation of MetAP activity.

Protein biogenesis is one of the most fundamental and complex biological processes, with important implications for human health and disease. The concerted action of the ribosome and numerous ribosome-associated factors is essential to ensure proper initiation of translation, protein folding, processing and targeting. By the successive steps of ternary complex recruitment, ribosome scanning, start codon selection and ribosomal subunit joining, translation initiation acts as the gate-keeping step of translation. Here, protein biogenesis initiates with either an initiator N-formylmethionine (fMet) 1 in case of bacterial and eukaryotic organellar protein synthesis, or an initiator methionine (iMet) in case of archaeal and eukaryotic cytosolic protein synthesis.
Typically, for more than half of all nascent protein chains, the iMet is cotranslationally removed by the action of methionine aminopeptidases (MetAPs), a process referred to as N-terminal (Nt-) methionine excision (NME) (1). MetAPs represent a family of conserved and essential enzymes, as inferred from the lethality of the map null mutant in case of Salmonella typhimurium and Escherichia coli (2). Although typically only one MetAP gene is found in the genome of prokaryotes, eukaryotes express at least two types of cytosolic MetAPs belonging to a family of evolutionary conserved metalloproteases, being a type 1 and type 2 MetAP. Although MetAP2 by itself was shown to be essential for the development of specific tissues in multicellular organisms (3), and MetAP1 appears to be vital for cell proliferation, a clear redundancy between MetAP1 and MetAP2 has been observed in Arabidopsis thaliana and Saccharomyces cerevisiae. Not surprisingly, double deletion mutants are nonviable, again pointing to their essential nature (4,5). Further, and irrespective of their limited sequence identity, the catalytic domains of both MetAPs adopt a similar protein fold encompassing a highly conserved pocket for iMet binding (6,7).
In case of eubacterial, mitochondrial and plastid NME, MetAPs remove the iMet of nascent polypeptide chains only after deformylation by the action of peptide deformylases (PDF) (8,9). Further, the manifestation of (the extent of) iMet processing by MetAPs typically relates to the identity of the extreme N-terminal residues of the nascent polypeptide chain. More specifically, MetAPs only cleave iMet from nascent polypeptide chains in case of penultimate N-terminal residues (P1Ј) with (relatively) small and uncharged side chains (i.e. Gly, Ser, Ala, Pro, Cys, Thr, or Val) (10). The latter referred to as NME compatible residues. Furthermore, amino acid residues beyond P1Ј may also contribute to the efficiency of NME (11)(12)(13)(14)(15). Interestingly and although still unclear, the need for having a MetAP of eubacterial and archaeal origin, being MetAP1 and MetAP2 respectively, might be attributed to (slight) differences in their substrate specificity profiles.
Further, the N-terminal amino acid sequence is also the major feature determining whether or not a given protein is Nt-acetylated and by which Nt-acetyltransferase or NAT (16). As such, the interplay among other co-translational acting modifiers, such as the NATs responsible for the cotranslational Nt-acetylation of a large cohort of proteins in case of eukaryotes (17)(18)(19), has also been implicated in determining MetAP susceptibility and thus (the extend of) iMet retention. More specifically we have previously shown Nt-acetylated iMet to be refractory toward the action of MetAPs (20).
By having different processing enzymes operating at the ribosomal exit tunnel, each nascent protein may undergo several cotranslational modifications before being fully matured.
In this study, we studied the omics-wide effects of human MetAP1 deletion using a human haploid HAP1 MetAP1 knockout cell line model and compared it to general MetAP inhibition using Bengamide B, a marine-derived natural product with broad spectrum antitumor activity which functions as a nanomolar inhibitor of MetAPs 1 and 2 (24). The use of these different setups additionally enable discriminating more direct from indirect MetAP actions. By enriching for protein N termini using positional proteomics, we revealed the in vivo substrate profile of MetAP1 at the proteome-wide level, thereby aiding in elucidating the functional role of MetAPs. Furthermore, our proteome analysis showed that the increased expression of MetAP2 observed on MetAP1 deletion was unable to restore processing of MS-, MP-and MA-N termini, overall pointing to MetAP1 as the MetAP being exclusively or more active over these N termini and providing increasing knowledge on the in vivo substrate repertoire of MetAPs.
Only in specific cases of which the iMet-processed and iMet unprocessed Nt-proteoforms could be observed, iMet processing affected the steady-state Nt-acetylation levels of Nt-proteoforms. Interestingly, changes in steady-state protein expression levels were typically explained at the transcript level and not related to protein N-terminal amino acid identities. More specifically, upon perturbation of MetAP activity, omics expression profiling revealed differential expression of proteins implicated in lipid metabolism, cytoskeleton organization, cell proliferation and protein synthesis.
Proteome Sample Preparation for Shotgun Analyses-Cell pellets of WT (HAP1 WT) and METAP1 KO (HAP1 MetAP1 KO) cells obtained from 3 independent cell cultures and containing approx. 10 ϫ 10 6 cells per pellet, were collected and stored at Ϫ80°C until further processing. HAP1 cell pellets resuspended in 400 l Gu.HCl lysis buffer (4 M GdmCl, 50 mm NH4HCO3 (pH 7.9)) were lysed by three rounds of freeze-thaw lysis in liquid in N 2 . The lysates were sonicated (Branson probe sonifier output 4, 50% duty cycle, 3 ϫ 30 s, 1 s pulses) followed by centrifugation for 10 min at 16,100 ϫ g (4°C), the supernatant removed and protein concentration determined by Bradford measurement according to the manufacturer's instructions. An aliquot equivalent of 200 g (ϳ5 ϫ 10 6 cells) was transferred to a clean tube, diluted to 1 mg/ml with lysis buffer, 2ϫ diluted with HPLC grade water, and precipitated with 4ϫ volumes of Ϫ20°C acetone overnight. The precipitated protein material was recovered by centrifugation for 15 min at 3500 ϫ g (4°C), pellets washed twice with Ϫ20°C 80% acetone, and air dried upside down for ϳ10 min at RT or until no residual acetone odor remained. Pellets were resuspended in 200 l TFE (2,2,2-trifluoroethanol) digestion buffer (10% TFE, 100 mM ammonium bicarbonate) with sonication until a homogenous suspension was reached. All samples were digested overnight at 37°C using mass spec grade trypsin/Lys-C Mix (Promega, Madison, WI) (enzyme/ substrate of 1:50 w/w) while mixing (550 rpm). Samples were acidified with TFA to a final concentration of 0.5%. Samples were cleared from insoluble particulates by centrifugation for 10 min at 16,100 ϫ g (4°C) and the supernatant transferred to clean tubes. Methionine oxidation was performed by the addition of H 2 O 2 to reach a f.c. of 0.5% for 30Ј at 30°C. Solid phase extraction of peptides was performed using C18 reversed phase sorbent containing 100 l pipette tips (Bond Elut OMIX 100 l C18 tips (Agilent, Santa Clara, CA) according to the manufacturer's instructions. The pipette tip was conditioned by aspirating the maximum pipette tip volume of water:acetonitrile, 50:50 (v/v) and the solvent discarded. After equilibration of the tip by washing 3 times with the maximum pipette tip volume in 0.1% TFA in water, 100 l of the acidified samples (ϳ200 g) were dispensed and aspirated for 10 cycles for maximum binding efficiency. The tip was washed 3 times with the maximum pipette tip volume of 0.1% TFA in water:acetonitrile, 98:2 (v/v) and the bound peptides eluted in LC-MS/MS vials with the maximum pipette tip volume of 0.1% TFA in water:acetonitrile, 30:70 (v/v). The samples were vacuum-dried in a SpeedVac concentrator and re-dissolved in 20 l of 2 mM tris(2carboxyethyl)phosphine in 2% acetonitrile.
Proteome Sample Preparation for N-terminal Peptide Enrichment Analyses-Pellets of WT (HAP1 WT) and METAP1 KO (HAP1 MetAP1 KO) cells obtained from 3 independent cell cultures (i.e. corresponding cultures of the shotgun proteome analysis) and containing approx. 50 ϫ 10 6 cells per pellet, were collected and stored at Ϫ80°C until further processing. HAP1 cell pellets resuspended in 50 mm NH4HCO3 (pH 7.9) with protease inhibitors added (Roche) were lysed by three rounds of freeze-thaw lysis in liquid N 2 and the lysates cleared by centrifugation for 10 min at 16,000 ϫ g. Solid guanidinum hydrochloride was added to the supernatant to a final concentration of 4 M and the protein samples (equivalent of 2 mg) were reduced and S-alkylated. All primary protein amines were blocked using an Nhydroxysuccinimide ester of (stable isotopic encoded) acetate at the protein level (i.e. an NHS ester of 13 C 2 D 3 acetate) to enable the assignment and the calculation of the degree of in vivo Nt-acetylation events (25). All modified proteome samples were digested overnight at 37°C using mass spec grade trypsin/Lys-C Mix (enzyme/substrate of 1:50 w/w) while mixing (550 rpm) and the resulting peptide mixtures vacuum dried. N-terminal peptide enrichment by SCX was performed as described previously (26). The eluted fraction enriched for Ntpeptides were vacuum dried and re-dissolved in 1 ml of 0.5% TFA in 2% acetonitrile. Samples were cleared from insoluble particulates by centrifugation for 10Ј at 16,100 g and 4°C, and the supernatant transferred to clean tubes. Methionine oxidation was performed on 100 l of acidified sample (ϳ200 g) by the addition of H 2 O 2 to reach a f.c. of 0.5% for 30Ј at 30°C. Solid phase extraction of peptides was performed using C18 reversed phase sorbent containing 100 l pipette tips as described above.
RNA Isolation, RNA Sequencing Library Preparation, Sequencing and mRNA-seq Data Processing-RNA was isolated from HAP1 cells using the TRIzol reagent (Invitrogen) according to manufacturer's instructions. RNA yields were determined using a NanoDrop spectrophotometer (Wilmington, Delaware) and RNA quality was assessed by Agilent Bioanalyzer RNA 6000 Nano Kit. Only samples with RIN values above 9 were accepted. Library generation including random fragmentation, cDNA synthesis and sequencing was performed at the VIB Nucleomics Core (www.nucleomics.be) using the TruSeq stranded total RNA sample preparation kit (Illumina, San Diego, CA) and including a Ribo-Zero (Illumina) depletion step as to remove ribosomal RNA from total RNA by the use of biotinylated Ribo-Zero oligos. Libraries were subjected to sequencing on a NextSeq 500 instrument (Illumina) to yield 75 bp single-end reads, resulting in ϳ23-28 million 75 bp single end reads per sample. The 3Ј adapter sequence (AGATCGGAAGAGCACAC) was trimmed using the fastx_clipper. Reads were pre-mapped onto small nuclear RNA, tRNA and rRNA. The remaining unmapped reads were then mapped onto the human GRCh37 reference genome (Ensembl annotation bundle 75) using STAR 2.4.0i allowing only unique mapping with a maximum of two mismatches. Separate BedGraph files were generated for the sense and antisense strand. Reads were counted across annotated genes using HTseq set to mode "union," feature type "exon" and minimum alignment quality of 10. For the MetAP1 KO cell line, two biological replicate samples were sequenced and for the WT cell line three. The unfiltered reads per kilobase of exon per million fragments (RPKM) values of protein-coding genes displayed a bimodal distribution, where approximately one third of genes had poor mRNA expression (i.e. genes with at least one read mapped but with RPKM values below 1), in line with previous reports, (27). Moreover, typically less than 2% of these poorly expressed genes have a corresponding protein expression detected using label-free shotgun proteomics. Therefore, and besides considering only unique reads from the mapping, only protein-coding genes with Ն 1 RPKM in all biological replicates of both conditions were retained for differential expression analysis. Differential expression analysis was performed using the Bioconductor package DESeq2 (Stanford, CA) in the R statistical programming environment. To determine significantly regulated genes, an adjusted p value threshold of 0.01 was applied. The fold change calculations for differentially expressed genes was determined by DESeq2 and represent the log2 change in expression level (i.e. counts) for the MetAP1 KO versus the HAP1 control setup. RNA-seq sequencing data is accessible through GEO Series accession number GSE103405 in NCBI's Gene Expression Omnibus (28) with reviewer password: wpuxikegfnkjjkx.
LC-MS/MS Analysis and Peptide Identification-Peptides were separated by nano-LC and directly analyzed with a Q Exactive instrument (Thermo Scientific, Bremen, Germany) operating in MS/MS mode as described before (29). More specifically, all samples were analyzed via LC-MS/MS on an Ultimate 3000 RSLC nano LC (Thermo Fisher Scientific) in-line connected to a Q Exactive mass spectrometer (Thermo Fisher Scientific). The sample mixture was first loaded on a trapping column (made in-house, 100 m i.d. ϫ 20 mm, 5 m beads C18 Reprosil-HD, Dr. Maisch, Ammerbuch-Entringen, Germany). After flushing from the trapping column, the sample was loaded on an analytical column (made in-house, 75 m i.d. ϫ 150 mm, 3 m beads C18 Reprosil-HD, Dr. Maisch). Peptides were loaded with loading solvent (0.1% TFA in water) and separated with a linear gradient from 98% solvent AЈ (0.1% formic acid in water) to 55% solvent BЈ (0.1% formic acid in water/acetonitrile, 20:80 (v/v)) in 30 min at a flow rate of 300 nL/min. This was followed by a 5 min wash reaching 99% solvent BЈ. The mass spectrometer was operated in data-dependent, positive ionization mode, automatically switching between MS and MS/MS acquisition for the 10 most abundant peaks in each MS spectrum. The source voltage was 3.4 kV, and the capillary temperature was 275°C. One MS1 scan (m/z 400 -2000, AGC target 3 ϫ 10 6 ions, maximum ion injection time 80 ms) acquired at a resolution of 70 000 (at 200 m/z) was followed by up to 10 tandem MS scans (resolution 17 500 at 200 m/z) of the most intense ions fulfilling predefined selection criteria (AGC target, 5 ϫ 10 4 ions; maximum ion injection time, 60 ms; isolation window, 2 Da; fixed first mass, 140 m/z; spectrum data type, centroid; underfill ratio, 2%; intensity threshold, 1.7 ϫ 10 4 ; exclusion of unassigned, 1, 5-8, Ͼ8 charged precursors; peptide match preferred; exclude isotopes, on; dynamic exclusion time, 20 s). The HCD collision energy was set to 25% normalized collision energy, and the polydimethylcyclosiloxane background ion at 445.120025 Da was used for internal calibration (lock mass).
Raw data files were searched with MaxQuant (30) using the Andromeda search engine (31) (version 1.5.4.1) and MS/MS spectra searched against the Swiss-Prot database (taxonomy Homo sapiens; 20,195 entries; May 2016 version) for N-terminal proteome analyses, or a two amino acids N-terminally truncated version of the Swiss-Prot database was used to avoid contribution of N-terminal peptide abundances to the overall LFQ protein calculations in case of shotgun proteome analyses. In both cases potential contaminants present in the contaminants.fasta file that comes with MaxQuant were automatically added.
A precursor mass tolerance was set to 20 ppm for the first search (used for nonlinear mass recalibration) and set to 4.5 ppm for the main search. For shotgun proteome analyses, trypsin/P (i.e. cleavages among lysine/arginine-proline residues were allowed) was selected as enzyme setting and up to two missed cleavage were allowed. Methionine oxidation was searched as fixed modifications, whereas N-terminal protein acetylation was set as variable modification. In case of N-terminal proteome enriched samples, ArgC was selected as enzyme setting as cleavage after lysine residues is obstructed by acylation of their side chains. Cleavages among arginine-proline residues were considered and semi-specific free N terminus was chosen as digestion mode. Methionine oxidation, carbamidomethylation of cysteines and heavy acetylation at lysine side-chains (Acetyl:2H(3)C13(2) (K)) were searched as fixed modifications, whereas acetylation and heavy (Acetyl:2H(3)C13(2) (N-term)) acetylation on protein N termini were set as variable modifications. The false discovery rate for peptide and protein identification was set to 1%, and the minimum peptide length was set to 7. The minimum score threshold for both modified and unmodified peptides was set to 40. The match between runs function was enabled and proteins were quantified by the Max-LFQ algorithm integrated in the MaxQuant software (32) in case of shotgun proteome analysis. Here, a minimum of two ratio counts and only unique peptides were considered for protein quantification.
The mass spectrometry proteomics data (i.e. the total list of identified proteins and modified peptides (next to the evidence file) in case of shotgun and N-terminal proteomics respectively)) have been deposited to the ProteomeXchange Consortium via the PRIDE (33) partner repository with the data set identifier PXD006638. Annotated, mass-labeled MS/MS spectra can be viewed on MS-Viewer (http:// msviewer.ucsf.edu/prospector/cgi-bin/msform.cgi?form ϭ msviewer) (34) using the search keys umhkpodlaj (supplemental Table S2A), 4adxtut9up (supplemental Table S2B) and 7balnvbgbr (for the 33 peptides corresponding to single-peptide based protein identifications in supplemental Table S3). In case of the supplemental Table  S2B data set, only assigned ion peaks are visualized on MS-Viewer. Complete annotated spectra can be viewed using the deposited data and MaxQuant Viewer.
Quantification of the Degree of Protein Nt-acetylation-To quantify the degree of protein in vivo Nt-acetylation, raw data files were searched with Mascot using the Mascot Daemon interface (version 2.5.1, Matrix Science) and MS/MS spectra searched against the Swiss-Prot database (taxonomy Homo sapiens; 20,172 entries; January 2017 version). The Mascot search parameters were set as follows; methionine oxidation to methionine-sulfoxide, heavy acetylation at lysine side-chains (Acetyl:2H(3)C13(2) (K)) and carbamidomethylation of cysteine were set as fixed modifications. To enable identification quantification of the degree of Nt-acetylation, acetylation versus heavy acetylation of N termini (Acetyl:2H(3)C13(2) N-term) was selected as quantification option, and carried out using the Mascot Distiller Quantitation Tool (version 2.5.1.0). Mass tolerance was set to 10 ppm on the precursor ion (with Mascot's C13 option set to 1) and to 20 mmu on fragment ions. Peptide charge was set to 1ϩ, 2ϩ, 3ϩ and instrument setting was put to ESI-QUAD. Endoproteinase semi-Arg-C/P (Arg-C specificity with arginine-proline cleavage allowed) was selected as enzyme allowing for 1 missed cleavage.
In case of N-terminal peptide analyses, only peptides that were ranked one, have a minimum amino acid length of seven, scored above the threshold score set at 99% confidence, displayed a start position of 1 or 2 and belonged to the category of N termini compliant with the rules of initiator methionine (iMet) processing (10) or represented actin N termini, were withheld. More specifically, iMet processing was considered in the case of iMet-starting N termini followed by any of the following amino acids; Ala, Cys, Gly, Pro, Ser, Thr, Met, or Val. Data management was done in ms_lims (35). Whenever the degree of protein in vivo Nt-acetylation could be calculated for Mascot identifications with Mascot Distiller based quantifications, data was considered for the analysis reported under the section 'Proteome-wide characterization of Nt-acetylation profiles' in case of matching MaxQuant N-terminal peptide identifications.
Software Packages for Statistics and Data Visualization-For basic data handling, normalization, statistics (if not stated otherwise), and annotation enrichment analysis, we used the freely available opensource bioinformatics platform Perseus (version 1.5.3.2) (36). Perseus was used for nonsupervised hierarchical clustering and the generation of heat maps, profile and scatter plots. Furthermore, 1D and 2D annotation enrichment analysis was performed using the Perseus software (37) with a two-sided test with Benjamini-Hochberg FDR correction. Data analysis after uploading the protein groups file obtained from MaxQuant database searching was performed as described previously (38). All replicate samples were grouped and LFQintensities log(2) transformed. Proteins with less than three valid values in at least one group were removed and missing values were imputed from a normal distribution around the detection limit (with 0.3 spread and 1.8 downshift).
Experimental Design and Statistical Rationale of the Proteome and Transcriptome Analyses-We analyzed the proteomes and matching N-terminal proteomes of a subset of 12 samples corresponding to 3 biological triplicates per setup i.e. WT, MetAP1 KO, and Bengamide B treated WT and MetAP1 KO HAP1 cells) using a label-free intensity based proteomics approach. In addition, we profiled matching transcriptomes of non-inhibitor treated samples corresponding to at least two biological replicates per condition. More specifically, for the MetAP1 KO cell line, two biological replicate samples were sequenced and for the WT cell line three biological replicate samples.
Lens-free Cell Imaging and Calculation of Cell Doubling-Holographic lensless imaging of cell cultures grown in parallel was performed using Cytonote devices from Iprasense with a resolution of 1-2 m. By placing the Cytonote devices consisting of a LED for illumination and a CMOS image sensor in contact with the cell culture vessels (i.e. 25 cm 2 falcons) in an incubator, the interference pattern with the sample (i.e. the interference between the partially coherent incident light and the light scattered by the cells in culture) was captured. From the holographic pattern, the sample image is reconstructed by applying the Horus software, including cell segmentation and basic analysis functions. The extremely wide field of view (30 mm 2 ) results in relevant statistics from single images.
Confluency of cell culture monitored using lens-free imaging over the course of 24 h before the addition of Bengamide B and the following 48 h, was fitted with an exponential model using the Doubling Time Online Calculator (http://www.doubling-time.com/ compute.php). Cell doubling time was calculated from the assigned growth rate as follows: ln(2)/growth rate and represented in hours.

RESULTS
To elucidate the omics-wide effects of MetAP perturbation and to study the redundant action and proteome-wide substrate specificity profiles of human MetAP1 and MetAP2 in general, an attempt was made to create human MetAP1 and MetAP2 knockout cells in the HAP1 haploid cell line. Using CRISPR/Cas9 editing, the Cas9 endonuclease is targeted by a pair of guide RNAs to generate double stranded DNA breaks, a process followed by non-homologous end-joining (NHEJ) repair that is accompanied by a precise DNA deletion of the locus of interest (39). By producing frameshift mutants of the gene of interest, CRISPR/Cas9 editing can generate custom human knockout cells. Consequently, genome engineering using CRISPR-mediated gene knockout is especially straightforward in haploid human cell lines and this way, frameshift mutants were obtained for MetAP1 and MetAP2 in HAP1 cells. However, as confirmed by Western blotting analysis and the mass spectrometric detection of MetAP1 (identified based on the exclusive detection of 6 unique peptides in merely all (inhibitor-treated) WT setups) (Fig. 1A), a functional mutant clone (i.e. a 179 bp insertion in exon 4 (see Experimental Procedures)) was obtained only in case of MetAP1. In case of MetAP2, the frameshift resulted in the production of an N-terminally truncated MetAP2 proteoform which still retained (part of) its activity as deduced from the unaltered iMet retention profiles observed (data not shown). Interestingly, despite the unaltered MetAP2 expression observed at the transcript level (supplemental Table S1), a 2-and 4-fold increase of MetAP2 was observed at the proteome level in case of Bengamide B treated wild type cells or in MetAP1 KO cells and Bengamide B treated MetAP1 KO cells, respectively (Fig. 1A).

MetAP1 Knockout Cells Display Altered Proliferation Rates and Cellular Morphology Next to Increased Sensitivity Toward
MetAP Inhibition-Because MetAP1 and/or MetAP2 inhibition was previously linked to altered cell proliferation, we first determined the effects of MetAP1 knockout and chemical MetAP inhibition-using Bengamide B as a nanomolar inhibitor of MetAPs 1 and 2 (24)on cell proliferation. Using time lapse data obtained by lensless cell culture imaging of cells in culture; cell densities, cell sizes and growth rates of WT and MetAP1 mutant cells were quantitatively evaluated and compared. As deduced from the monitored confluency measurements of cell cultures before and after the addition of Beng-amide B (Fig. 1B), cell doubling time calculations revealed a near 2-fold increase in the doubling time of MetAP1 knockout versus WT cells (i.e. from 17 h to 31 h) when monitoring confluency during the first 24h after platting. Interestingly, although the proliferation rates of WT HAP1 remained essentially unaltered during the next 24 h of cultivation, the doubling rate of MetAP1 knockout cells more closely approached that of WT cells (i.e. 20 h), indicating a delay in cell proliferation, especially during the first phase after plating. The difference in cell doubling becomes even more pronounced on addition of the MetAP inhibitor Bengamide B. Here a doubling time of 28 h was observed in the case of Bengamide B treated WT cells, whereas 75 h was required to achieve cell doubling of Bengamide B treated MetAP1 knockout HAP1 cells (i.e. an over 4-fold increase in doubling time as compared with its isogenic WT counterpart). As such, a correlation between the loss of MetAP1 expression and its inhibitory effect on cell proliferation can clearly be observed (Fig. 1B), indicating that the lack of MetAP1 expression in HAP1 cells can not entirely be compensated for by the increased MetAP2 expression levels observed (Fig. 1A). Interestingly, although a fibroblastlike morphology can be observed in case of both HAP1 cell lines, MetAP1 mutant cells show on average a 30% increase in attached cell size (i.e. from 17 to 22 m in diameter) (Fig. 2). Because MetAP1 has previously been implicated in cell cycle progression through G 2 /M phase (40), a delayed G 2 /M phase progression can likely account for this observation.
As a reference sample, Bengamide B treated cells were sampled in parallel. Here, the inhibitor was used at a final concentration of 0.1 g per ml for 24h of incubation. Using a monoclonal antibody raised against an N-terminal peptide corresponding to the iMet unprocessed N terminus of 14-3-3␥ (i.e. iMet-14-3-3␥), changes in MetAP dependent iMet processing efficiencies of iMet-14-3-3 ␥ could be monitored.
When compared with total 14-3-3 ␥ protein levels, the basal levels of iMet-14-3-3␥ increased about 50% in untreated MetAP1 KO cells (Fig. 3) whereas on Bengamide B treatment, a 3.4 fold and 4 fold increase of iMet-14-3-3␥ was observed in case of WT cells and MetAP1 KO cells, respectively. As such, MetAP1 deletion clearly reduces the efficiency of iMet-14-3-3␥ processing and the use of higher concentrations or prolonged incubation with Bengamide B didn't lead to a further increase of iMet-14-3-3␥ (data not shown). Noteworthy, a cell dependence in the increase in the ratio of unprocessed versus total 14-3-3␥ on Bengamide B treatment has been reported previously, further hinting to cell type dependent MetAP efficiencies.
Proteome-wide Characterization of iMet Retention Profiles-N-terminal proteome enriched replicate samples were analyzed by 1D nano LC-MS/MS using 3h gradients on a quadrupole Q Exactive Orbitrap instrument (42) and subjected to label-free, ion intensity-based peptide quantitation and quantification of the degree of N-terminal acetylation (see below) using MaxQuant (30) and Mascot respectively. N-terminal peptide enrichment was used to investigate proteomewide alterations in iMet retention on genetic MetAP1 deletion and/or chemical MetAP inhibition among others. More specifically, 3 biological replicate proteome samples were obtained in case of WT, MetAP1 KO and Bengamide B treated WT and MetAP1 KO HAP1 cells.
In case of N-terminal proteome enriched samples, all samples were reduced and S-alkylated and the primary protein amines blocked using an N-hydroxysuccinimide ester of (stable isotopic encoded) acetate at the protein level (i.e. an NHS ester of 13 C 2 D 3 -acetate) to enable the assignment and the calculation of the degree of in vivo Nt-acetylation (see below) (43). N-terminal peptide enrichment by SCX was performed as  Tables S2A and S2B). Of these N termini, 1506 started at position 1 or 2, whereas 261 had a start position beyond protein position 2. As reported previously, the latter category of N termini are indicative of alternative translation initiation events (44,45). Of the 1631 N termini with intensitybased peptide MaxQuant quantification data (supplemental Table S2A), 407 N termini were identified in all setups analyzed (25% of all identified protein N termini).
When performing a 1D annotation enrichment (37) on the modified N-terminal peptide intensities obtained using Max-Quant (supplemental Table S2B), and only considering iMet compliant categories of N termini (i.e. A, MA, C, MC, MD, ME, MF, G, MG etc.), significant changes (p Ͻ 0.01) in the abundance of certain N-terminal categories in the MetAP1 KO could be observed when compared with the WT setup (Fig. 4).
More specifically, the steady state levels of MS-, MP-, and MA-N termini are significantly increased in the knockout as compared with their WT levels, whereas their processed counterparts (i.e. S-and A-N termini) are significantly reduced. MV-, MT-, and MG-termini are only significantly affected on MetAP inhibition in general or by the additional inhibition of MetAP2 in MetAP1 knockout cells (Fig. 4A). Peptide intensity profile plots also reflect these significant changes observed (Fig. 4B).
Proteome-wide Characterization of Nt-acetylation Profiles-Because our previous work revealed a kinetic competition between NATs and MetAPs (20), we also assessed (potential differences in) protein Nt-acetylation levels.
Overall, as deduced from the very high correlation observed, the degree of Nt-acetylation remained essentially unaltered when comparing the Nt-acetylomes of the different setups analyzed (supplemental Fig. S1 and supplemental Table S2A). This observation hints to the fact that the degree of iMet processing has only minor effects on the steady-state Nt-acetylation levels of Nt-proteoforms whenever detectable (i.e. iMet-processed and iMet unprocessed Nt-proteoforms).

Shotgun Proteome Analysis Reveal Changes in Protein Expression Profiles on Perturbation of MetAP Activity of Proteins
Implicated in Lipid Metabolism, Cytoskeleton Organization, and Protein Synthesis-Shotgun proteome analysis was performed to monitor quantitative differences in steady-state protein expression levels by high-resolution mass spectrometry (MS) based proteomics using single LC runs and labelfree protein quantitation (32). Of the 3 biological replicate samples analyzed per setup (i.e. WT, MetAP1 KO and Bengamide B treated WT and MetAP1 KO HAP1 cells) and thus 12 proteome samples analyzed in total, 4971 unique proteins could be identified at a peptide and protein false discovery rate (FDR) of 1%. Label-free quantification was performed to enable relative proteome comparisons (see also methods) by use of the intensity values (LFQ) obtained using the MaxLFQ algorithm (30,32). Of note, N-terminal peptide intensities were not considered for the LFQ calculations to avoid contribution of N-terminal peptide abundances to the overall LFQ protein calculations (see methods). After log2 transformation of the obtained LFQ values, a multiscatter plot was calculated together with the Pearson correlation. The very high Pearson correlations observed (i.e. an average Pearson correlation of 0.98 and all above 0.95) between the different replicates and samples analyzed overall point to the high reproducibility of the chromatographic separations and MS detectability among the different runs (supplemental Fig. S2). Protein expression profiling using a two-way ANOVA test was performed to reveal proteins of which the expression level significantly differs between the genotype (p Ͻ 0.001) and treatment samples (p Ͻ 0.01). Log2 transformed LFQ values were grouped for all replicate samples and only proteins with minimally 3 valid values in at least one setup were considered, overall permit- ). Term enrichment was determined using the 1D annotation enrichment algorithm embedded in the Perseus software suite and p values were corrected for multiple hypotheses testing using the Benjamini and Hochberg false discovery rate. Only corrected p values Ͻ0.01 were considered. B, Individual profile plots of normalized N-terminal peptide intensities across all four conditions assayed are shown. N-terminal categories for which N termini with significantly altered abundancies could be observed (except for MK-N termini) were highlighted in red and blue in case of iMet-processed and iMet-retaining N termini, respectively. ting differential protein expression analysis of 3,386 unique proteins (supplemental Table S3). Missing data values were replaced with imputed values from the corresponding normal distribution around the detection limit. Overall, 278 protein groups were found significantly regulated dependent on the genotype (p Ͻ 0.001) and 183 protein groups dependent on Bengamide B treatment (p Ͻ 0.01). The intensities of significantly regulated proteins are shown as heat maps after zscore transformation and non-supervised hierarchical row clustering (Fig. 5).
Relative enrichment of the indicated gene categories (Uni-Prot keywords) calculated using the 1D annotation enrichment algorithm (37) point to significant expression changes (p Ͻ 0.02) in four main protein categories linked to the terms; "protein synthesis," "organelle," "lipids/membrane," and "cell connection/cytoskeleton" (Fig. 6). Although all four categories were affected on MetAP1 deletion (e.g. downregulation of protein synthesis and lipid metabolism, and upregulation of cytoskeletal proteins), only proteins implicated in protein synthesis and more specifically ribosome biogenesis were found to be significantly downregulated on Bengamide B treatment of control cells.
Western blot analyses were performed to substantiate some of the genotype and/or inhibitor dependent quantitative changes observed in the protein expression levels. When probing for the differentially expressed proteins COTL1, FIG. 5. Z score heat map representations of cluster analysis reveal proteins with differential abundance dependent on genotype or treatment. Non-supervised hierarchical clustering of z-scored MS intensities of significant genotype (WT or MetAP1 KO) (A) and treatment (untreated or Bengamide B treatment) (B) regulated proteins after two-way ANOVA are represented. 278 protein groups were found significantly regulated dependent on the genotype (p Ͻ 0.001) and 183 protein groups dependent on Bengamide B treatment (p Ͻ 0.01) (supplemental Table S3). Green indicates low intensities whereas red indicates high intensities.

Protein Expression Changes in the Context of Protein N-terminal Amino Acid Identity and Transcript Levels Using Deep
Transcriptomics-In line with the unaffected steady-state protein levels of the MetAP1/2 substrate 14-3-3␥ ( Fig. 3 and supplemental Table S3), when performing annotation enrichment of protein N-terminal identities and using the 1D annotation enrichment algorithm (37) on the normalized protein intensities and independent of iMet processing (i.e. all 20 possible categories were considered MA, MC, MD, ME, …), no significant difference in the occurrence of N-terminal protein identities could be observed when comparing regulated versus nonregulated proteins.
Further, we systematically compared the transcriptome and proteome changes on MetAP1 knockout. When only considering the shotgun proteomes of untreated control and MetAP1 KO cells, for 97% of the proteins quantified by MS, a transcript abundance estimate was obtained by means of RNA sequencing (supplemental Table S1). The Pearson cor- ). Term enrichment was determined using the 1D annotation enrichment algorithm embedded in Perseus (36,37) and p values were corrected for multiple hypotheses testing using the Benjamini and Hochberg false discovery rate. Only corrected p values Ͻ0.02 were considered. relation of transcript and protein abundance was found to be ϳ0.6 (Figs. 8A and 8B). The correlation of mRNA abundance ratios with protein abundance ratios was moderate (Pearson r ϭ 0.28) (Fig. 8C).
By performing a gene ontology term enrichment analysis to charachterize global effects on the transcript as well as proteome level, a 2D annotation enrichment analysis (37) was performed based on a nonparametric Mann-Whitney test (FDR Ͻ 0.02) using the Perseus software (36) (Fig. 9). Although the effect on processes linked to cell proliferation was mainly reflected in the reduced transcript levels, processes such as DNA replication and sterol biosynthesis were downregulated both at the transcriptome and proteome level in MetAP1 knockout cells. Consistent with the antiproliferative effect of MetAP1 inhibition and reflected in the lowered abundancy of a multitude of ribosomal proteins (supplemental Table S3), downregulation of (mitochondrial) translation processes was also found to occur at both omics levels.
Processes upregulated at the transcript and steady-state proteome level of MetAP1 knockout cells include the positive regulation of cell adhesion and extracellular structure organization next to secretory processes such as vesicle organization, exocytosis and Golgi vesicle transport. Furthermore, although a(n) (slight) upregulation of proteins implicated in protein folding (e.g. chaperonnes) and the biosynthesis of nucleobase-containing compounds could be observed, the corresponding mRNA levels were downregulated. DISCUSSION Although chemical inhibition and knockdown studies on the topic of MetAPs have been reported previously, the high efficiency of iMet excision and incompleteness of inhibition or knockdown next to the redundant actions of MetAPs observed precludes general conclusions of independent MetAP activities. Although both MetAPs have been listed as "core essentialome" components in the human haploid HAP1 cell line and thus likely represent highly essential proteins (46), a functional MetAP1 mutant HAP1 clone edited by CRISPR/Cas to contain a frameshift mutation in the fourth coding exon was nonetheless successfully obtained. The use of isogenic human haploid cells in a variety of genetic studies, confirms the validity of the use of the HAP1 haploid cell line (or its parental line KBM7) as a model system compared with a "standard" diploid cell line.
With this cell model at hand, we pursued to study the omics-wide effect of MetAP1 deletion and general MetAP inhibition. In line with previous reports, a correlation between MetAP expression levels and cellular MetAP inhibitory effect could be observed. Further, iMet retention levels were inversely correlated with cellular proliferation rates. Because ectopic MetAP expression and protein expression levels of MetAPs was reported to be positively correlated with the increased resistance of cell lines against MetAP inhibition (47,48), the specific increase in MetAP2 expression levels observed on MetAP1 deletion and/or MetAP inhibition, points to reprogramming of the cell enabling the cell to cope with the loss of MetAP.
Further, iMet retention profiles determined by positional proteomics revealed statistically significant changes in the steady-state abundances of (M)S-, (M)P-and MA-N termini. More specifically, iMet-retaining and iMet-processed proteoforms of these N termini were found to be respectively up- More likely however, in line with the previously determined in vitro substrate specificity profiles of MetAP1 and MetAP2, MetAP2 may display a much higher or even exclusive action over these type of N termini (11). More specifically, Xiao et al. showed that the catalytic activity of human MetAP1 toward MV-and MT-peptides was up to two orders of magnitude lower than that of MetAP2 pointing to the fact that MetAP2 is mainly responsible for processing of MV-and MT-N termini (11).
Nonetheless, MetAP2 is unable to restore levels of iMet processing observed in case of MS-, MP-, and MA-N termini, indicating the manifold higher or exclusive activity of MetAP1 over these type of N termini.
Viewing the previously revealed kinetic competition between NATs and MetAPs (20), to our surprise, only few protein N termini (e.g. MAP-starting N termini) were identified as differentially Nt-acetylated when comparing the Nt-acetylomes of WT and MetAP1 KO cells. The effect observed is likely because of Nt-acetylation being unable to keep up with the increased levels of specific iMet-retaining N termini of nascent polypeptides, making them more easily observable in case of sub-optimal NAT substrates, including MAP-N termini displaying a proline at position 3, a residue known to diminish Nt-acetylation and lower iMet cleavage susceptibility (11,49). However, because in most cases, only the steadystate levels of the main proteoforms can be detected, no general conclusions regarding the influence of iMet excision on Nt-acetylation can be drawn easily.
Interestingly, differential iMet excision did not appear to account for the differential protein expression patterns observed. On the other hand, and despite the rather moderate correlation of mRNA abundance ratios with protein abundance ratios, proteome-changes were typically reflected by altered transcript abundances.
Finally, on perturbation of MetAP activity, and besides the differential expression of proteins implicated in cell proliferation, omics expression profiling revealed differential expression of cytoskeletal proteins, proteins associated with protein and lipid and cholesterol biosynthesis. In this context, it is noteworthy that MetAP2 inhibitors, originally developed as anticancer therapies, have proven anti-obesity effects by their inhibitory action over the sterol regulatory element binding protein (SREBP), thereby resulting in reduced lipid and cholesterol biosynthesis (50). Based on our results, MetAP1 can also be put forward as a potential target for the discovery and development of new anti-obesity besides anticancer therapeutic strategies.
Acknowledgments-We thank Prof. Kris Gevaert for conceptual assistance and financial support of this research and Nucleomics Core of the VIB (Leuven) for performing the deep-sequencing experiment.

DATA AVAILABILITY
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium with the data set identifier PXD006638. Annotated, mass-labeled MS/MS spectra can be viewed on MS-Viewer (http:// msviewer.ucsf.edu/ prospector/cgi-bin/msform.cgi?form_msviewer) (34) using the search keys umhkpodlaj (supplemental Table S2A), 4adxtut9up (supplemental Table S2B) and 7balnvbgbr (for the FIG. 9. 2D annotation enrichment of proteome and transcriptome expression changes. The 2D-annotation enrichment scores for GO biological processes (GOBP names) were plotted for the proteome and transcriptome expression changes observed when comparing the MetAP1 KO versus WT setups. The fold change of log2 transformed means of corresponding normalized iBAQ and RPKM values were considered only in case of two valid replicate values for the proteome and transcriptome setups analyzed. The biological processes represented by green dots show common upregulation at both mRNA and protein levels, whereas orange dots indicate simultaneous downregulation or, in case of the points with low absolute Y-values, represent biological processes downregulated at the mRNA level only. The process represented by black dots exhibit downregulations at the mRNA level, while being upregulated (GOBP protein folding) or nonregulated at the level of the proteome. Only corrected p values Ͻ0.02 were considered.
33 peptides corresponding to single-peptide based protein identifications in supplemental Table S3). RNA-seq sequencing data is accessible through GEO Series accession number GSE103405 in NCBI's Gene Expression Omnibus with reviewer password: wpuxikegfnkjjkx (see also detailed description in Experimental Procedures). * We declare no conflict of interest.
□ S This article contains supplemental material.
¶ These authors share first authorship.