The Cytotoxic Natural Product Vioprolide A Targets Nucleolar Protein 14, Which Is Essential for Ribosome Biogenesis

Abstract Novel targets are needed for treatment of devastating diseases such as cancer. For decades, natural products have guided innovative therapies by addressing diverse pathways. Inspired by the potent cytotoxic bioactivity of myxobacterial vioprolides A–D, we performed in‐depth studies on their mode of action. Based on its prominent potency against human acute lymphoblastic leukemia (ALL) cells, we conducted thermal proteome profiling (TPP) and deciphered the target proteins of the most active derivative vioprolide A (VioA) in Jurkat cells. Nucleolar protein 14 (NOP14), which is essential in ribosome biogenesis, was confirmed as a specific target of VioA by a suite of proteomic and biological follow‐up experiments. Given its activity against ALL cells compared to healthy lymphocytes, VioA exhibits unique therapeutic potential for anticancer therapy through a novel mode of action.

Natural products display ar ich source of bioactive molecules selected by evolution to target distinct molecular pathways in eukaryotic and prokaryotic cells. [1] One of the preeminent producers of awide range of bioactive secondary metabolites are myxobacteria. [2] In 1996, the isolation of peptolides vioprolides A-D ( Figure 1A)w as reported from Cystobacter violaceus and detailed studies on their biosynthesis have recently been performed. [3] Despite their antifungal and anticancer activities in the low nanomolar range, as well as immunomodulatory activity in micromolar concentrations, [4] no molecular target elucidation has been reported up to date. [5] Despite the wealth of target identification methods available,i ncluding activity-based protein profiling (ABPP) [6,7] and thermal proteome profiling (TPP), [8,9] numer- ous natural products still lack functional characterization. This information is needed in order to decipher novel and unprecedented modes of action in cancer cells,w here drug resistance,t herapeutic failure,orrelapse require new targets and cellular pathways with high specificity towards tumor cells. [10] By applying diverse proteomic and biochemical methods we investigated the cellular mechanism of vioprolide A (VioA) in Jurkat cancer cells and recovered nucleolar protein 14 (NOP14), which is essential for ribosome biogenesis,tobe aspecific target.
Prior to target-identification studies,w er anked vioprolide A-D potencies in the acute lymphoblastic leukemia (ALL) cell line Jurkat to select the most potent growth inhibitory compound for subsequent mode of action analysis. VioA exhibited the highest antiproliferative activity,w ith ah alf maximal inhibitory concentration (IC 50 )o f4 n m ( Figure 1B). In line with ap revious report, [5] the vioprolides showed significant differences in bioactivity,thus demonstrating that the structural composition of the N-heterocycles is crucial for potency( Figure 1A).
Moreover,V ioA displayed low nanomolar IC 50 values in different cell lines,t hus illustrating its broad potency (Figure 2A, Figure S1 in the Supporting Information). To analyze whether enhanced cell death rates contribute to diminished proliferative capacity of VioA-treated cancer cells,apoptosis assays were performed. VioA exhibited an EC 50 of 15.8 and 28.9 nm in CEM and Jurkat cells,r espectively,s ignifying ahigh apoptosis rate in acute lymphoblastic leukemia (ALL) cells,c ompared to the acute myeloid leukemia (AML) cell line HL-60 (EC 50 :8 8.1 nm), as well as HeLa (EC 50 :1 34 nm) and T24 cells (EC 50 :n.a.;F igure 2B). This was confirmed by the extent of caspase-3 activation and PA RP cleavage,a s analyzed by western blotting ( Figure S2). Mechanistic studies on apoptosis induction revealed intrinsic apoptotic cell death triggered by VioA ( Figure S3, see discussion in the Supporting Information). Moreover,c ell-cycle progression in Jurkat cells treated with VioA showed asignificant decrease in cells in S-phase,a sm easured by flow cytometry analysis (Figure 2C).
To confirm the potencyofVioA in ALL and to analyze its putative therapeutic relevance,t he compound was tested in ALL patient-derived xenograft (PDX) cells.V ioA-treated ALL samples of diverse background (Table S1) exhibited an increased apoptosis rate after 24 ht reatment, as shown by flow cytometry ( Figure 2D). Notably,peripheral blood mononuclear cells (PBMCs) of healthy donors barely responded to VioA compared to PDX samples ( Figure 2D), thus clearly demonstrating that VioA exhibits auspicious antileukemic properties. To rationalize the molecular mechanism underlying the potencyofV ioA in ALL, we performed target identification through affinity based protein profiling (AfBPP). [7,11] The VioA scaffold was functionalized with aphotocrosslinker and abiorthogonal linker to yield the probe VioA-P (Scheme S1). Although the antiproliferative potencyo ft he probe against Jurkat cells only dropped by 4-fold ( Figure S4), in situ labeling with concentrations ranging from 0.5 mMt o2mM revealed known background binders [12] among the enriched proteins,b ut no clear target ( Figure S5). Since the modification with the minimal linker could have perturbed the binding properties to the dedicated target(s) and the probe ester bond suffered from limited stability ( Figure S6), TPP was selected as an alternative modification-free strategy.
Jurkat cells were incubated in situ with VioA (1 mm)o r DMSO and subsequently exposed to arange of temperatures (37 8 8C-67 8 8C). Cells were lysed and soluble proteins were isolated by ultracentrifugation, tryptically digested, and labeled using isobaric tandem mass tag labels (TMT). [12] Respective labeled samples were combined, subjected to hydrophilic-interaction chromatography (HILIC) fractionation and finally analyzed by LC-MS/MS ( Figure 3A). T m shifts were calculated from two replicates of VioA versus DMSO treatment and visualized after filtering ( Figure 3B; see the Supporting Information for filtering criteria). 67 proteins revealed aminimum temperature shift of 1 8 8Cinboth VioA-versus DMSO-treated replicates and were manually inspected for proper melting curves.S ix proteins (blue dots Figure 3B;for melting curves,see Figure S7, Table S2) passed additional significance thresholds (see the Supporting Information). Among these,U 2snRNP-associated SURP motifcontaining protein (U2SURP) and nucleolar protein 14 (NOP14) were highly stabilized upon VioA treatment (Fig-ure 3B,C). We selected these two major hits for in-depth validation. U2SURP is as pliceosomal factor and NOP14 (which showed the highest shift) is crucial for 40S ribosome subunit formation as well as maturation of 18S rRNA. [13] U2SURP was excluded as ap otential target of VioA because splicing was not affected in VioA-treated cells,a s shown by dual luciferase splicing reporter assay ( Figure S8). NOP14 plays ac rucial role in the initial steps of ribosome biogenesis,n amely,t he formation of the small subunit processome (SSU processome), an early stage of the formation of the small ribosomal subunit (40S). This processing is enhanced in cancer cells due to the high demand for continuous ribosome production for their extensive growth. [15,16] In eukaryotes,r ibosome biogenesis takes place in the nucleoli that contain genes encoding for ribosomal RNAs (rRNAs), which build the nucleic acid backbone of the ribosome. [15] Interestingly,N OP14 siRNAk nockdown cells (knockdown confirmation shown in Figure S9) as well as VioA-treated cells display reduced rRNAl evels in nucleoli, as demonstrated by 5-FU immunostaining ( Figure 4A,B). Actinomycin D( ActD), which inhibits PolI transcription, served as positive control. rDNAt ranscription is necessary for the formation of nucleoli, whereas perturbations of the transcriptional machinery leads to their disassembly. [15] Consistently,A ctD treatment led to displacement of NOP14 in the whole nucleus due to nucleolar disruption, whereas VioA treatment did not alter the localization of NOP14. Based on these results,V ioA obviously does not interfere with ribosome biogenesis at the level of transcription ( Figure 4C). Importantly,k nockdown of NOP14 impairs cancer-cell proliferation ( Figure 4D)a sshown before for VioA ( Figure 1B, Figure 2A).  [14] and thermal-response curve fitting as well as melting-pointcalculations were carried out using the TPP Rp ackage. [9] C) Thermal-response curves and calculated melting points (asterisks) for NOP14 and U2SURP proteins of VioA-treated (orange) and DMSO-treated (grey) cells. Twobiologicalr eplicates are shown.
Va rious protein-protein interactions with NOP14 are essential for af unctional ribosome biogenesis machinery in yeast. [13,17] Thus,wefocused on unravelling the interactome of NOP14 through MS-based co-immunoprecipitation (co-IP) experiments.C o-IP (with an immobilized anti-NOP14 antibody) was carried out in presence of aD SSO crosslinker to enable the trapping of transient binding partners as applied previously ( Figure 5A). [18] Analysis of pulled-down proteins by LC-MS/MS identified several significantly enriched proteins (Table S3) compared to the isotype control co-IP ( Figure 5B)i nJ urkat cells,a mongst others,n ucleolar complex protein 4homolog (NOC4L) and ribosomal RNAsmall subunit methyltransferase NEP1 (EMG1). In order to identify VioA-sensitive interactions of NOP14, the co-IP experiment was repeated with compound treatment for 24 hours prior to in situ DSSO crosslinking ( Figure S10, Table S4). Tr eatment with VioA led to the depletion of various proteins enriched in non-treated co-IP experiments (Table S4). Since it is known that interaction of NOP14 with both NOC4L and EMG1 is essential for proper maturation of 18S rRNAand 40S assembly and export, [13,19] we investigated the influence of VioA treatment on the interplay of those proteins.
Remarkably,M S-based and western blot based co-IP experiments revealed that VioA treatment led to as elective loss of interaction between NOP14 and EMG1, whereas the NOP14-NOC4L association remained unperturbed (Figure 5C,D). Of note,expression levels of neither NOC4L nor EMG1 were affected by VioA ( Figure 5E).
In yeast, EMG1 is involved in the modification of uridine 1191 of the 18S rRNA. Its catalytic activity is not essential for ribosome biogenesis,h owever the presence of the protein within the 90S pre-ribosome is inevitably required. [20] Nucleolar localization of NOP14 depends on binding to NOC4L. [22] This interaction is not disturbed by VioA, hence,N OP14 remains in the nucleolus upon VioA treatment. However, since the physical interaction of NOP14 with EMG1 is required for nuclear localization of EMG1, [13] impaired binding of those two poteins upon VioA treatment led to ad istribution of EMG1 in the entire cell in comparison to control cells,w here EMG1 is mainly located in the nucleus ( Figure 5F).
In conclusion, vioprolides represent au nique class of natural products with potent bioactivity towards cancer cells in the nanomolar range.U sing TPP,w ei dentified NOP14, which is involved in ribosome biogenesis,a sa nu nprecedented interaction partner of VioA. NOP14 plays ak ey role in the assembly of the small ribosomal subunit and in fact, VioA is the first natural product reported that addresses this crucial pathway.P BMCs showed low response to VioA in contrast to ALL cancer cells,t hus suggesting VioA as ap romising lead structure for cancer-cell targeting.C loser mechanistic studies showed that VioA alters the interaction of NOP14 with EMG1, leading to delocalization of EMG1. Our results indicate that targeting the NOP14-EMG1 interaction, and thereby ribosome biogenesis,i sapromising anticancer strategy and underline the great potential of VioA as ac hemical tool to study ribosomal processes as well as starting point for future drug development. experiments performed in duplicate, one-way ANOVA, Dunnett's test, ** P < 0.002. C) HeLa cells were treated with 10 nm VioA or 6 mm ActD for 2hand immunostained for NOP14. Nuclei were stained with Hoechst. Representativeimages out of three independentexperimentsa re shown. D) xCELLigence real-time proliferation measurement of HeLa cells either transfectedw ith nt or NOP14 siRNA for 24 hb efore seeding into E-plates. Proliferation was monitored for the subsequent 72 h. Representative curve out of three independent experiments is shown. Respectivedoubling time values were calculated using xCELLigence RTCA software. Bars represent the mean AE SEM of three independentexperiments performed in triplicate, two-tailed unpaired Student's t-test, ** P < 0.002. Figure 5. Co-immunoprecipitationw ith anti-NOP14 antibody.A)Schematic workflow of MS-based co-immunoprecipitation. Living cells were treated with 10 nm VioA or DMSO, respectively,for 24 hours. Subsequently,DSSO linker was added to covalently link protein complexes within cells. NOP14 and linked proteins were enriched using an anti-NOP14 antibody immobilizedo nprotein A/G beads. Tryptic peptides were measured via LC-MS/MS, and analyzed using MaxQuant [14] and Perseus. [21] B) Volcano plot represents two-sample t-test results of anti-NOP14 co-IP compared to isotype control co-IP (n = 4). Cutoff criteria were defined as log 2 = 2(4-fold enrichment) enrichment factor and -log 10 (t-test pvalue) = 1.3 (dotted lines). NOP14 is colored in blue, NOC4L in dark grey and EMG1 in orange. C) Comparison of normalizedL FQ intensities of NOP14, NOC4L and EMG1 in isotype control and anti-NOP14 co-IP samples treated with 10 nm VioA prior to co-IP after missing value imputation. D) Co-IP of NOP14 in Jurkat cells and detection of NOP14 and interactors NOC4L/ EMG1 by western blot analysis. Arepresentative experimentout of three independentexperiments is shown. Bars represent mean AE SEM of three independente xperiments, two-tailed unpaired Student's ttest, *** P < 0.001 E) Western blot analysis of NOP14, NOC4L and EMG1 expression levels of Jurkat cells either treated with DMSO or 10 nm VioA for 24 h. Bars represent the mean AE SEM of three independentexperiments, two-tailed unpaired Student's ttest, ns P > 0.05. F) HeLa cells treated with VioA were co-stained for NOP14 (red) and EMG1 (green). Nuclei were stained with Hoechst 33342. Representative images out of three independentexperiments performed in duplicate are shown.
TheM Sd ata have been deposited to the ProteomeXchange Consortiom via the PRIDE [23] partner repository with the dataset identifier PXD015196.