Peptide-RNA photo-crosslinks with tunable RNA chain map protein-RNA interfaces

Peptide-RNA photo-crosslinks with tunable RNA chain map protein-RNA interfaces | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Peptide-RNA photo-crosslinks with tunable RNA chain map protein-RNA interfaces View ORCID Profile Shuyao Sha , View ORCID Profile Bernhard Kuster , View ORCID Profile Jakob Trendel doi: https://doi.org/10.1101/2025.09.05.674461 Shuyao Sha 1 Chair of Proteomics and Bioanalytics, TUM School of Life Sciences, Technical University of Munich (TUM) , Freising, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Shuyao Sha Bernhard Kuster 1 Chair of Proteomics and Bioanalytics, TUM School of Life Sciences, Technical University of Munich (TUM) , Freising, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Bernhard Kuster Jakob Trendel 1 Chair of Proteomics and Bioanalytics, TUM School of Life Sciences, Technical University of Munich (TUM) , Freising, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Jakob Trendel For correspondence: jakob.trendel{at}tum.de Abstract Full Text Info/History Metrics Supplementary material Preview PDF Abstract Photo-crosslinking mass spectrometry enables the identification of protein-RNA interactions in living cells, pinpointing interaction interfaces at single-amino acid resolution. However, current isolation procedures for peptide-RNA crosslinks eliminate the RNA moiety, prohibiting sequencing of the RNA alongside the crosslinked peptide. Here, we introduce peptide-RNA crosslink isolation for sequencing by mass spectrometry or pepR-MS, a method that enriches peptide-RNA crosslinks with RNA chains of tunable length. Applied to breast cancer cells, pepR-MS identifies over 21,000 unique crosslinks at 4,757 crosslinking sites in 744 proteins. Employing different nucleases, we capture crosslinks with RNA moieties up to six nucleotides, revealing RNA crosslinking preferences at domain and subdomain resolution. Finally, we demonstrate mass spectrometry-based sequential sequencing of both peptide and RNA from the same crosslink, providing a starting point for the analysis of long-chain peptide-RNA crosslinks that map interaction interfaces across the proteome and transcriptome. Introduction In the human proteome, over 3,000 proteins have been reported to be RNA-binding 1 . About 60% of these contain one or several of the canonical RNA-binding domains, while the remaining 40% lacking them often interact with RNA via intrinsically disordered regions (IDRs). To understand the biological processes RNA-binding proteins participate in it is crucial to elucidate these interactions at molecular resolution. Photo-crosslinking can be used to fix protein and RNA interactions in living cells, where UV light induces the formation of a covalent bond between the two molecules. The peptide and RNA sequences on either side of the covalent bond can be analysed to pinpoint the interaction interface to the exact location within a protein or an RNA, respectively. RNA sequencing-based methods such as CLIP-seq 2 – 5 map these interaction sites across the transcriptome. In turn, liquid chromatography-mass spectrometry (LC-MS) of peptide-RNA crosslinks has been used to identify protein-RNA interaction sites across the proteome 6 – 10 . The purification of peptide-RNA crosslinks, however, has proven difficult due to their low abundance and bipartite composition. In addition, photo-crosslinking can occur at any amino acid, challenging conventional peptide identification software because of the enormous search space created by the great number of possible modification sites and nucleotide permutations. Early studies avoided direct analysis of the crosslinked peptide and instead analysed neighbouring non-crosslinked peptides released by sequential digestion with different peptidases 11 , 12 . Other approaches used differential photo-crosslinking, inferring RNA interaction from the absence of particular peptides after irradiation 13 , 14 . More recent strategies narrowed the search space by nuclease digestion of peptide-RNA crosslinks, leaving only one or a few nucleotides attached to the peptides 6 – 10 . These efforts each identified not more than 400 unique peptides from roughly 100 human proteins that were crosslinked to at most three nucleotides. These modest yields led to the assumption that the large number of possible peptide-RNA combinations, even with short RNA chains, exceeded the capacity of current MS search software 15 . A pragmatic solution to this was the removal of the RNA chain from peptide-RNA crosslinks by chemical digestion with hydrofluoric acid (HF), which reliably cleaves RNA to a single nucleotide or only the crosslinked base. This approach mapped approximately 2,000 RNA binding sites in 600 human proteins using 254 nm UV irradiation 15 , and 1,500 RNA binding sites in around 300 proteins when employing metabolic labeling with the photo-activatable nucleotide 4-thiouridine (4SU) and 365 nm UV irradiation 16 . However, complete chemical RNA digestion of course prevented identification of the crosslinked RNA sequences. Here, we show that isolation procedures for peptide-RNA crosslinks using nuclease digestion often contain large quantities of non-crosslinked RNA that impair LC-MS detection. We present a specialized workflow termed pepR-MS (peptide-RNA crosslink isolation for sequencing by MS) that removes non-crosslinked RNA to a similar degree as chemical digestion with HF, yet leaves the RNA moiety of the crosslink intact and tunable in length. PepR-MS strongly improves MS detection, enabling the identification of overall more than 21,000 unique crosslinks from MCF7 breast cancer cells, as well as the quantitative analysis of differential RNA interaction upon epigenetic drug treatment. Employing different nucleases, we capture crosslinks with RNA moieties up to six nucleotides, and use the immense resulting crosslink diversity to characterize the reactivity of 4SU. Finally, we present a simple sequential MS sequencing strategy to derive peptide and RNA sequences from the same crosslinked molecules, some of which could be mapped to the structure of the human 80S ribosome. Results Non-crosslinked RNA suppresses detection of peptide-RNA crosslinks Current workflows for isolating peptide-RNA crosslinks for LC-MS analysis generally involve the following main steps: in-cell photo-crosslinking, protein digestion, RNA enrichment and digestion, and final removal of non-crosslinked RNA ( Figure 1A ). A recent variation that used HF instead of nucleases for RNA digestion remarkably improved the identification of peptide-RNA crosslinks 15 , 16 . We compared MS data from these studies to previous isolation protocols. In representative MS runs from each study, averaging the MS2 spectra showed a substantial contribution of ions corresponding to RNA fragments when nucleases were used, which was not observed with HF digestion ( Figure 1B ). Further inspection of individual MS2 spectra revealed abundant and intense signals for adenine and guanine bases when nucleases were used, even in spectra matched to peptides crosslinked to only uridine ( Figure S1A ) or non-crosslinked peptides ( Figure S1B and S1C ). We speculated that these signals might originate from contaminating non-crosslinked RNA fragments rather than from the peptide-RNA crosslinks. This was also supported by observations on the MS1 level, where ions corresponding to masses of non-crosslinked RNA fragments captured a lot of MS intensity ( Figure S1D ). We suspected that these highly abundant non-crosslinked RNA fragments interfered with peptide identification, but were removed by chemical digestion with HF. Download figure Open in new tab Figure S1: Non-crosslinked RNA contamination in isolates of peptide-RNA crosslinks. A-D) Published LC-MS data of nucleotide-crosslinked peptides using a nuclease-based isolation procedure (Trendel et al., 2019). A) Exemplary MS2 spectrum of a peptide-RNA crosslink showing contamination by foreign RNA fragments. B) Same as in panel A but for peptides identified as unmodified (non-crosslinked). C) Averaged MS2 spectra of all peptides identified as unmodified (non-crosslinked). Peaks for adenine, guanine and cytosine account for 5.0%, 3.6% and 1.7% of the total ion current. D) Averaged MS1 spectra of the entire MS run. Inserts show corresponding MS2 spectra of two highlighted RNA fragments. E-I) LC-MS data from this study using a HeLa standard with or without nuclease-digested RNA spike-ins. E) Bar graph showing number of identified peptide sequences (top) and the proportion of repeatedly sequenced MS1 peaks (bottom), related to Figure 1C–E . Statistical analysis was performed using a two-tailed unpaired Student’s t-test. **, p < 0.01; ***, p < 0.001; ****, p < 0.0001. Error bars indicate standard deviation across triplicates. F) Average of identified MS2 spectra for each LC-MS run. G) Same as in panel F but for unidentified MS2 spectra. H) Bar graph comparing the occurrence of the most intense MS2 peaks in one MS run. Percentages indicate the fraction of MS2 spectra, among all acquired MS2 spectra, in which a peak at the specified mass was the most intense peak (base peak). Error bars indicate standard deviation across triplicates. I) Exemplary MS2 spectra of precursors with the same m/z and retention time in LC-MS runs with and without RNA spike-in. Download figure Open in new tab Figure 1: Non-crosslinked RNA contamination in peptide-RNA photo-crosslinks impairs MS analysis. A) Generalized experimental workflow summarizing published methods for isolating peptide-RNA crosslinks for LC-MS analysis. B) Averaged MS2 spectra of four representative MS runs from previous studies. Peaks at m/z value of RNA fragments (orange) and amino acid immonium or y1 ions (blue) are highlighted. C) Bar graph of peptide identification rate upon spike-in of nuclease-digested total RNA into a HeLa peptide standard. Statistical analysis was performed using a two-tailed unpaired Student’s t-test with *** for p < 0.001 and **** for p < 0.0001. Error bars indicate standard deviation across triplicates. D) MS1 chromatogram comparing total ion current (grey) and integrate intensity of identified peptides (blue) of a HeLa peptide standard with or without RNA spike-in. E) Averaged MS1 spectra of a HeLa standard with or without RNA spike-in. Peaks of highly abundant RNA fragments are highlighted in orange. To test this, we spiked RNase-digested total RNA into a HeLa peptide standard. Indeed, addition of increasing amounts of RNA digest progressively reduced the MS2 identification rate from 34% to 15% ( Figure 1C , p = 5.9E-5, two-sided Student’s t-test). Peptide identification was impaired throughout the gradient ( Figure S1E ), with the strongest effect at early retention times ( Figure 1D ). At the MS1 level, we detected ions corresponding to RNA oligonucleotides, similar to those seen in published datasets of nuclease-digested crosslinks, which accounted for a large portion of the total ion current, masking lower intense peptide ions ( Figure 1E and Figure S1D ). On the MS2 level, we observed that unidentified spectra were strongly dominated by RNA fragment ions ( Figure S1F ). Moreover, even identified spectra of HeLa peptides were heavily contaminated by these ions, albeit, with lower intensities compared to unidentified MS2 spectra ( Figure S1G ). The 136 Th adenine ion was by far the most frequently observed and most intense peak of all MS2s in the spike-in samples ( Figure S1H ). Overall, this implied that RNA fragments produced by end-point nuclease digestion were very commonly co-isolated for MS2 scans. We could reconstruct this co-isolation from scans of pure HeLa peptides, pure RNA fragments, and their co-isolated mixture, demonstrating that the addition of highly intense RNA ions in the MS2 strongly suppressed peptide fragment peaks ( Figure S1I ). Co-isolation resulted from m/z overlap between tryptic peptides and the highly diverse RNA fragments generated during positive-mode electrospray ionization (ESI), which include species formed by the loss of individual bases, terminal phosphates, or sugars, as well as the various nucleotide permutations 17 . In addition, we noticed a strong increase in repeated sequencing of the same precursor ions upon RNA spike-in ( Figure S1E ), implying that abundant non-crosslinked RNA fragments also consumed a notable amount of instrument time. These reduced the opportunity to analyse peptide-RNA crosslinks, which are typically low abundant and missed during standard data-dependent acquisition (DDA) when higher-intensity ions are preferentially selected for fragmentation and sequencing. In summary, these observations confirmed that non-crosslinked RNA fragments impair peptide identification by capturing MS1 and MS2 intensity as well as analysis time, thereby especially disadvantaging detection of low abundant peptide-RNA crosslinks. Removal of non-crosslinked RNA enhances detection of peptide-RNA crosslinks We systematically adapted our previously published nuclease-based workflow 9 for the isolation of peptide-RNA crosslinks, implementing several key modifications targeted at free RNA removal ( Figure 2A , S2A). Specifically, we incorporated strong cation-exchange (SCX) chromatography after nuclease digestion to selectively retain peptide-RNA crosslinks while removing non-crosslinked RNA fragments, which are otherwise retained by conventional C18-based cleanups universally used in previous methods 6 , 7 , 9 , 10 . We also optimized the preceding RNA enrichment step with multiple rounds of isopropanol precipitation, which more effectively removed non-crosslinked peptides. UV spectroscopy revealed a striking reduction of RNA in the injection-ready LC-MS sample from previously 6 μg using a C18-based cleanup to less than 0.1 μg using the SCX-based one ( Figure 2B ). We conducted LC-MS/MS analysis on samples prepared with and without removal of RNA fragments starting from an equal amount of MCF7 cells, which had been metabolically labeled with the photo-activatable nucleotide 4SU and irradiated with 365 nm UV bulbs ( Figure 2C-E ). The optimized workflow effectively removed large smearing peaks of non-crosslinked RNA in the MS1 chromatogram, revealing sharp underlying peaks, many of which could be identified as peptide-RNA crosslinks ( Figure 2C ). It also substantially reduced RNA fragment ion intensity across the run, with summed intensities of representative RNA fragments dropping by 54% ( Figure 2D ). The number of successfully identified peptide-RNA crosslinks (defined throughout the text as unique modified peptide sequence linked to a unique RNA mass offset, see Methods) more than doubled from approximately 1,700 to more than 4,600 once non-crosslinked RNA was removed ( Figure 2E , Table S1). To improve the crosslinking efficiency, we integrated our recently reported UV irradiation device for enhanced photo-activation (UVEN 18 , 19 ) which dramatically reduced the irradiation time from minutes to seconds and enabled irradiation in the original cell culture media. Strikingly, 1 s of UVEN irradiation achieved comparable crosslinking efficiency as 100 seconds of conventional UV bulb irradiation ( Figure S2B ). Applying rapid photo-activation combined with improved peptide and RNA removal, we identified 8,725 RNA crosslinks at 2,533 crosslinking sites in 250 proteins in a 90-minute MS run, outperforming the leading published method employing 4SU crosslinking and chemical RNA digestion in all metrics ( Figure 2F - 2H , Figure S2C , Table S2). Importantly, because of its ability to preserve crosslinked RNA chains, our optimized approach was able to produce a much more diverse set of peptide-RNA combinations ( Figure 2I ). We named this method ‘peptide-RNA crosslink isolation for sequencing by MS’ or pepR-MS. Download figure Open in new tab Figure S2: PepR-MS identifies peptide-RNA crosslinks with enhanced robustness and sensitivity. A) Detailed experimental scheme for the purification of peptide-nucleotide crosslinks with (lower panel) and without (upper panel) dedicated removal of non-crosslinked RNA. B) Bar graph showing identified crosslinks using the high-power UV irradiation device (UVEN) and conventional UV bulb (Vilber crosslinker). C) Identified and quantified crosslinked proteins in previous and this study, using 365 nm crosslinking (main crosslinking base 4SU) or 254 nm crosslinking (main crosslinking base U). D) Cumulative distribution of coefficient of variation (CV) across four replicates of DMSO control (grey) and Romidepsin treatment (blue). E) Schematic representation of proteins with the locations of regulated crosslinks and PTM sites previously reported by Chang et al., 2024. Download figure Open in new tab Figure 2: PepR-MS identifies peptide-RNA crosslinks with enhanced robustness and sensitivity. A) Experimental scheme of the pepR-MS workflow: Cells are labeled with 4-thiouridine and irradiated using high-power 365 nm irradiation. After TRIzol lysis, the interphase containing protein-RNA crosslinks is isolated and digested with DNase and trypsin. Non-crosslinked peptide and DNA fragments are removed by silica column enrichment and isopropanol precipitation. Nuclease digestion is applied to produce nucleotide-peptide crosslinks with the desired RNA chain length. Non-crosslinked RNA is removed by SCX desalting. B-E) Comparisons of the LC-MS/MS results with and without removal of non-crosslinked RNA. B) Line graphs comparing RNA content of the injection-ready MS sample measured by spectroscopy. C) MS1 base peak chromatograms. D) Averaged MS2 spectra comparing fragment ion intensities across all MS2s. Peaks of known RNA fragments are highlighted in orange. E) Bar graphs comparing identifications by MSFragger using RNA mass offsets up to 3 nucleotides. F-H) Bar graphs comparing previously published and this study. Note that each study either used 254 nm (main crosslinking base U) or 365 nm crosslinking (main crosslinking base 4SU). All data was re-searched with MSFragger (see Methods). F) Bar graph of all PSMs (crosslinked and non-crosslinked). G) Bar graph of identified and quantified crosslinks. H) Same as in G for crosslinking sites. I) Distribution of RNA moiety lengths (PSMs) using 365 nm crosslinking and HF digestion (Bae et al., 2021) to 365 nm crosslinking and pepR-MS with nuclease digestion. J) Volcano plot showing differential peptide-RNA crosslink abundance from MCF7 cells upon 6 hours of 10 µM romidepsin treatment. Model-based p-values were calculated by limma with Benjamini-Hochberg correction (see Methods). K) Schematic representation of proteins highlighted in panel J indicating location of regulated crosslinks and PTM sites reported by Chang et al., 2024. To test the quantitative abilities of our method, we treated MCF7 cells with the histone deacetylase inhibitor Romidepsin, which we recently showed promotes lysine acetylation across a subproteome strongly enriched in RNA-binding proteins 20 . As longer Romidepsin exposure has been shown to induce global mRNA decay starting from 18 hours of treatment 21 , 22 , we applied a relatively short treatment of six hours to 4SU-labeled MCF7 cells and compared protein-RNA interaction sites to mock-treated control cells using pepR-MS. Despite the extensive sample preparation, we found good quantitative precision of crosslink intensities with a median coefficient of variation of 28% ( Figure S2D ). Among approximately 7,000 robustly detected crosslinks, 39 were significantly modulated by Romidepsin treatment, with 34 decreasing and 5 increasing (adj. p<0.05, limma testing with Bonferroni-Holm correction, Figure 2J , Table S3). To investigate potential relationships between Romidepsin-regulated protein-RNA interactions and post-translational modifications (PTMs), we cross-referenced the RNA crosslinking sites with acetylation and phosphorylation sites that we had previously found regulated upon 6-hour Romidepsin treatment 20 . In ten out of 13 regulated sites, we observed that RNA crosslinking decreased while nearby acetylation or phosphorylation was increased ( Figure 2J and 2K , S2E). For example, in the RNA helicase DHX9, decreased RNA crosslinking was observed in the direct vicinity of an acetylation site upregulated by Romidepsin (K1163). Both sites were five amino acids in distance and part of a nuclear localization signal (NLS) 23 . Mutations at the acetylation site K1163 have been reported to impair nuclear import of DHX9 and are associated with severe neurodevelopmental disorders 24 . In SND1, two sites with decreased RNA crosslinking were detected within its core SN-like domain, adjacent to a phosphorylation site at S426, which has been reported to play a key role in cell cycle regulation 25 . The SN-like domains of SND1 are responsible for double-stranded RNA binding 26 and interaction with numerous proteins involved in transcription, splicing and gene silencing 27 , including DHX9 28 . In HNRNPA1, we found decreased RNA crosslinking at three C-terminal sites, close to two serine residues with opposing phosphorylation changes (S364 decreased, S368 increased). Phosphorylation at multiple C-terminal serines in HNRNPA1 has been reported to repress RNA interaction and its ability to form nuclear condensates 29 . Collectively, these observations suggested that PTMs may remodel the RNA binding interface in different ways, including masking charges, causing steric hinderance, or creating a binding surface that favours protein partners over RNA. Tuning RNA chain length in peptide-RNA crosslinks Conceptually peptide-RNA crosslinks with sufficient nucleotide length have the potential to map a protein-RNA interaction to an individual amino acid position in the proteome via the peptide sequence, and at the same time an individual nucleotide position in the transcriptome via the RNA sequence. While unambiguous mapping to the human transcriptome would require a median RNA sequences length of 14 nucleotides 30 , shorter sequences could already provide insight on the binding preferences of individual RNA-binding domains, which typically interact with degenerate motifs between 3-8 nucleotides 31 , 32 . We tested our isolation for crosslinks with longer RNA chains using a series of different nuclease treatments, including RNase A, T1, benzonase and a combination of all three. Figure 3A illustrates how mass offsets detected by MSFragger on peptide-RNA crosslinks aligned with the known cleavage specificity of the individual nuclease: RNase A digestion yielded mostly terminal C and U, RNase T1 produced terminal G, and benzonase generated a variety of unspecific terminals. To search for longer RNA moieties of up to 5 nucleotides, we employed NuXL, a search engine specialized for nucleic acid-peptide crosslinks 33 , 34 . Notably, the RNA chain length increased from digestion with all nucleases or RNase A (2 nt dominant), to RNase T1 (2-3 nt dominant), to benzonase (3-4 nt dominant). Indeed, this indicated the presence of even longer RNA chains in the same samples and mirrored the trend in nucleotide lengths we had observed in the MSFragger offset search ( Figure 3B ). Download figure Open in new tab Figure 3: Tuning of RNA chain length in peptide-RNA hybrids by nuclease digestion. A) Histogram of PSMs with RNA masses identified by MSFragger offset search on samples using pepR-MS and various nuclease combinations indicated on the left. B) Histogram of PSMs with nucleotide moieties identified by NuXL in the same samples as in panel A. C-E) Exemplary RNA motifs at crosslinking sites derived from RNA chains identified by NuXL. Each motif shows the probability of the crosslinked 4SU (X) and flanking nucleotides based on all crosslinks of the same peptide (see Figure S3A and Methods). See also Figures S3B-D . Download figure Open in new tab Figure S3: RNA chains with tunable length reveal domain-resolved crosslinking preference. A) Schematic of region-specific RNA motif generation. For each PSM, all possible permutations of the RNA moiety are considered to calculate position-specific nucleotide probabilities. PSMs mapping to the same protein region are then combined to derive representative RNA motifs for that region (see Methods). B) Schematic representation of SRSF1 and SRSF5 locating the motifs identified in RRM1. Compared are motifs derived by MSFragger or NuXL, respectively, with numbers indicating PSM counts supporting each motif. See also Figure 2C . C) Same as in B for HNRNPC. See also Figure 2D . D) Same as in B for NONO. See also Figure 2E . For many highly abundant proteins, crosslinks with diverse RNA sequences were detected, which we used to derive RNA sequence motifs for individual RNA-binding domains. Notably, this was limited by the fact that mass offsets detected by MSFragger or NuXL can only be used to deduce nucleotide composition of a crosslink and much less information on its sequence. Nevertheless, we calculated the probability of nucleotides at specific positions relative to the crosslinking site considering all possible permutations of the detected RNA moiety from a single PSM. By integrating all PSMs from one particular protein region, we derived domain-resolved nucleotide preferences ( Figure S3A , Table S4 and S5, see Methods for details). For example, the RRM1 domain of the splicing factors SRSF1 and SRSF5 have previously been shown to preferentially bind C-rich sequences, particularly CN motifs (N for any nucleotide) in vitro 35 . Consistent with this, we observed C-rich motifs at crosslinking sites within the RRM1 domains of both proteins ( Figure 3C and S3B ). Interestingly, the motif associated with the RRM1 of SRSF1 showed slightly higher preference to G than that of SRSF5, which was mainly restricted to C and A, aligning with the functional divergence of the two proteins 36 . We also observed distinct RNA sequence motifs within different domains of the same protein. In HNRNPC, crosslinks in the C-terminal RRM showed strong enrichment for polyuridine sequences ( Figure 3D and S3C ), in agreement with previous findings 37 – 40 . In contrast, crosslinks within its basic leucine zipper-like motif (bZLM) displayed a preference for alternating U/C or UUC motifs ( Figure 3D and S3C ). Another example for this was NONO, which crosslinked to U-rich RNA sequences within its RRM1 domain ( Figure 3E and S3D ), consistent with previous in vitro data showing NONO homodimers binding G or U-rich RNA oligos in an RRM1-dependent manner 41 . Its RRM2, in contrast, lacks the conserved aromatic residues of canonical RRMs and has therefore been questioned to bind RNA at all, instead mediating interaction with protein or highly structured RNA 42 , 43 . We observed crosslinks in RRM2 with a preference for A, confirming its direct RNA interaction in living cells but implying a distinct RNA binding specificity compared to RRM1 ( Figure 3E and S3D ). Peptide-RNA crosslink diversity reveals 4SU reactivity Because of the exponentially rising number of unique peptide-RNA sequence combinations with increasing RNA chain length, our nuclease-based workflow produced many more crosslinks than previously observed by chemical RNA digestion ( Figure 2I ). Nucleases with distinct RNA cleavage specificities generated 11,609 unique crosslinks confidently identified with three or fewer RNA nucleotides by MSFragger, providing a rich resource to characterize 4SU crosslinking preferences at the protein-RNA interface. We detected all 20 amino acids at crosslinking sites, with a clear preference for glycine, proline, histidine, tyrosine and phenylalanine compared to the overall amino acid composition of the crosslinked proteins ( Figure 4A and 4B , Table S5). The amino acid frequency at the crosslinking site varied notably between disordered regions and different types of RNA-binding domains ( Figure 4C ). The increased frequency of certain amino acids reflected a 4SU crosslinking bias towards π-stacking aromatic residues (histidine, tyrosine and phenylalanine) 44 as well as disorder promoting amino acids (especially glycine and proline) 45 , 46 . Most crosslinking sites were located within canonical RNA-binding domains, primarily the RNA recognition motif (RRM), the K homology domain (KH) and the cold shock domain (CSD) ( Figure 4D , Table S5). The RRM was the most frequently crosslinked domain, with over 4,800 unique crosslinks mapping to 99 different RRMs from 64 proteins. Aligning these crosslinks to the hidden Markov model-based domain definition of the RRM (Pfam HMM profile) revealed 4SU reactivity hotspots across positions corresponding to the two central beta sheets ( Figure 4E ). To obtain a three-dimensional view of these interaction interfaces, we used AlphaFold 3 47 and predicted the structure of a generic RRM based on the consensus sequence of the Pfam HMM profile in complex with the oligonucleotide CAUC. The resulting model showed high prediction confidence (average pLDDT = 91) and resembled experimentally determined RRM-RNA complexes such as the NMR solution structure of the SRSF3 RRM in complex with CAUC 48 ( Figure 4F and 4G ). Projecting the crosslinking frequency at each position of the RRM onto the 3D structure indicated strong 4SU reactivity within the central beta sheet formed by beta strands two and three, and the flexible loop between them. This aligned with previous reports on experimentally determined protein-RNA interfaces of individual RRMs 40 , 48 – 51 . Apart from canonical RNA-binding domains, nearly half of the crosslinking sites localized to IDRs ( Figure 4D ). Among proteins crosslinking to RNA exclusively with their IDR, we found an enrichment of ribosomal proteins ( Figure S4A -C), which are known to carry highly disordered, non-canonical RNA-binding domains 52 . We identified some crosslinking sites mapping to short linear motifs (SLiMs) 53 in IDRs, which are evolutionary conserved peptide sequences with a defined function embedded in the otherwise poorly conserved region. For example, we observed crosslinking in a CDK phosphorylation motif within one IDR of NPM1, in line with previous findings that phosphorylation of this IDR reduced the RNA interaction of NPM1 54 , influencing its localization and function in ribosome biogenesis 55 , 56 . In summary, the large number of unique crosslinks isolated by pepR-MS revealed detailed 4SU crosslinking preferences towards canonical RNA-binding domains and IDRs. Download figure Open in new tab Figure S4: Molecular functions associated with protein-RNA crosslinks in canonical domains and IDRs. A) Bar graph comparing top five enriched Gene Ontology (GO) terms of proteins crosslinked within canonical domains (RRM, KH, CSD, KOW, DZF, Helicase C-like, B30.2/SPRY, and dsRBD) compared to all MCF7 proteins. B) Same as in panel A but for proteins crosslinked exclusively within disordered region. C) Bar graph comparing relative occurrence of proteins with GO terms in proteins crosslinked exclusively through disordered regions relative to those crosslinked within structured domains. Download figure Open in new tab Figure 4: Localization of peptide-RNA crosslinking sites and 4SU crosslinking reactivity. A) Scatter plot comparing amino acid occurrences in crosslinked proteins and at the RNA crosslinking site. B) Bar plot illustrating frequencies of crosslinked amino acids relative to their overall frequencies in all crosslinked proteins. Colouring indicates p-value of Fisher’s exact tests after Benjamini-Hochberg correction (based on crosslinking sites). C) Pie diagrams comparing amino acid composition of all crosslinked proteins to their composition at crosslinking sites in IDRs or exemplary protein domains. D) Bar plot comparing number of crosslinks observed in various structured domains, ribosomal proteins and IDRs. E) Bar plots quantifying crosslinking sites within the RNA recognition motif (RRM) domain defined by Pfam domain definition (HMM profile on top). Bars show the number of crosslinked RRMs (top) and the number of unique crosslinks (bottom) at each position. F) Alphafold 3-predicted structure of the Pfam RRM consensus sequence in complex with CAUC (left), compared to the NMR solution structure of the SRSF3 RRM bound to CAUC (PDB 2I2Y, right). Red colouring refers to number of crosslinked RRMs displayed in panel E. G) Secondary structure prediction of sequences in panel F. All calculations based on crosslink count if not explicitly specified otherwise, see Methods. Sequential sequencing of peptide and RNA in crosslinks Using different nucleases, we demonstrated that pepR-MS could retain longer peptide-crosslinked RNA fragments while maintaining LC-MS compatibility of the sample. Mass offsets found by MSFragger or NuXL identified RNA chain lengths as long as five nucleotides ( Figure 3A and 3B ), however, only revealing the composition of the RNA but not its sequence. To determine the RNA sequence alongside the peptide sequence, we first characterized the MS fragmentation behaviour of non-crosslinked synthetic RNA under the same conditions used for crosslink detection. Previous reports had demonstrated that oligonucleotides fragment more readily than peptides in artificial peptide-oligonucleotide conjugates under positive ESI-MS conditions 57 . Indeed, we found that the synthetic RNA oligo UAGUAG fragmented at much lower collision energies compared to a HeLa peptide standard or synthetic PROCAL peptides 58 ( Figure 5A ). This suggested that a low collision energy that produced a viable RNA fragment ladder would leave a peptide intact, allowing us to decouple sequencing of the crosslinked RNA and peptide moieties entirely. We implemented a simple data-dependent acquisition method that recorded two consecutive MS2 scans for the same precursor with two distinct collision energies ( Figure 5B ). We tested the method on benzonase-digested peptide-RNA crosslinks, for which we had previously found mass-offsets up to five nucleotides ( Figure 3B ). For the data recorded with our sequential sequencing method, we used MSFragger and NuXL to find MS2 spectra for peptides with mass-offsets corresponding to RNA chains up to six nucleotides ( Figure S5A and S5B ). This resulted in the identification of 4,701 MS2 spectra for peptide-RNA crosslinks using the high collision energy, and, as anticipated, a lower number of 2,356 MS2 spectra using the low collision energy ( Figure S5C ). As proof of concept, we used the peptide identifications in the high-collision-energy MS2 as entry points to find MS2 pairs, where peptide and RNA could be sequenced. Indeed, manual interpretation of low-collision-energy MS2 spectra allowed in many cases for sequencing of the RNA moiety. Figure 5C presents one particularly notable example of an MS2 pair, which we could validate in the cryo-electron microscopy structure of the human ribosome 59 . After NuXL had identified an RPS9 peptide with a four-nucleotide mass offset in a high-collision-energy MS2 spectrum, manual interpretation of the corresponding low-collision-energy MS2 spectrum readily revealed the underlying RNA sequence GAAX (X for crosslinked 4SU). As RNA preferentially cleaves at the 5’ phosphate bond during collision-induced dissociation, producing c and y-type ions 17 , the direction of the RNA sequence was unambiguous ( Figure S5D ). Combined with a separate MS2 pair of the same peptide ( Figure S5E ), the extended sequence GAAXGA could be located to the 18S rRNA in the structure of the human 80S ribosome 59 in immediate proximity to the identified RPS9 peptide ( Figure 5D ). This demonstrated that the distinct fragmentation behaviour of peptides and RNA can be exploited for the sequential sequencing of peptide-RNA photo-crosslinks, and that this paired sequence information can be used to pinpoint peptide-RNA interfaces present in the cell. Download figure Open in new tab Figure S5: Sequential sequencing of peptide and RNA in peptide-RNA crosslinks in the same MS run. A) Histogram of RNA masses identified by MSFragger open search on pepR-MS isolated peptide-RNA crosslinks using benzonase-digestion. The MS method shown in Figure 5B was used. B) Histogram of nucleotide moieties identified by NuXL on the same data as in panel A. C) Venn diagram comparing peptide identifications acquired with high and low collision energy (CE) between MSFragger and NuXL. Overlap indicates successfully identified peptides in both MS2 spectra of the sequential pair from same precursor. D) Schematic representations of RNA fragmentations derived from the MS2 spectrum shown in Figure 5C . HCD fragmentation primarily generates c and y-type ions, implying only a single possible RNA strand orientation (right). E) Pair of MS2 spectra demonstrating sequential sequencing of the crosslink SPYGGGRPGRVK-XGA mapping to RPS9, which extends the RNA sequence identified in the MS2 pair shown in Figure 5C . Download figure Open in new tab Figure 5: Sequencing peptide and RNA chains of peptide-RNA crosslinks in the same MS run. A) Scatter plot illustrating the fragmentation behaviour of the synthetic RNA oligo UAGUAG, a HeLa peptide standard, and the synthetic peptide standard PROCAL under increasing HCD collision energy. Average ratio of the precursor MS2 intensity and total MS2 ion current is shown. B) Schematic of the MS method for sequencing the RNA moiety of peptide-RNA crosslinks at low collision energy (CE) and the peptide moiety at high CE. C) Pair of MS2 spectra for sequential sequencing of the crosslink SPYGGGRPGRVK-GAAX mapping to RPS9. D) Validation of the sequenced crosslink in the cryo-electron microscopy structure of human 80S ribosome (PDB 4UG0. light blue: identified RPS9 peptide; dark blue: crosslinking site; orange: crosslinked section of 18S rRNA). Discussion MS analysis of peptide-RNA crosslinks provides a direct approach to record snapshots of protein-RNA interactions in living cells. However, its broader application has been limited by several technical challenges. First, low abundance of peptide-RNA crosslinks due to both the low efficiency of UV crosslinking and the predominance of non-crosslinked peptides requires highly efficient enrichment methods to recover pure crosslinked species 6 , 60 , 61 . Second, it has been argued that LC-MS/MS analysis of peptide-RNA crosslinks is inherently constrained by the heterogeneity of RNA chains, which would lead to highly complex spectra and dramatically expand the search space, inevitably impairing crosslink identification 11 – 13 , 15 , 16 . Our findings reveal that much of the difficulty in the LC-MS analysis of photo-crosslinked peptide-RNA hybrids were sample-related. We demonstrate that contamination by non-crosslinked RNA, rather than MS search space, is the main obstacle to detecting peptide-RNA conjugates by MS. We introduced an optimized workflow that enhances sample purity to the benefit of MS robustness and sensitivity, while preserving parts of the crosslinked RNA chain. Removal of non-crosslinked RNA by pepR-MS led to four decisive improvements, two in respect to LC separation and two in respect to MS detection. First, the much lower RNA content of the injection-ready MS sample enabled the analysis of much larger amounts of peptide-RNA crosslinks derived from more input material of cultured cells without saturating the LC system. Second, sample purity also heavily decreased carry-over and performance decreases of the column, allowing for many consecutive injections without compromising LC performance. Third, by removing highly abundant RNA fragments, we strongly decreased the dynamic range of the sample, allowing more sensitive detection of lower-intensity peptide-RNA crosslinks. Last, pervasive MS2 co-isolation of RNA fragments was effectively removed, resulting in much higher peptide fragment intensities and better MS2 quality. These improvements multiplied the identifications of peptide-RNA crosslinks leading to as many as 8,725 identified crosslinks in 250 proteins from a single biological replicate. Analysis with the latest instrument generation with higher MS sensitivity will further improve these numbers. We anticipate that pepR-MS should also be applicable to other crosslinking types including chemical crosslinking 34 . Using nucleases with distinct cleavage specificities, we demonstrate that the RNA moiety in peptide-RNA crosslinks can be adjusted, which lays the foundation for the sequencing of longer RNA chains. While the successful peptide and RNA sequencing of crosslinks mapping to the structure of the human ribosome suggests that proteome-transcriptome-wide mapping of protein-RNA interactions might be possible, it also reveals the great challenges ahead, including sample preparation, MS acquisition and data analysis. Unambiguous assignment of RNA moieties to their transcriptomic origin requires crosslinks with longer RNA sequences. Thus, shorter nuclease digestion times or physical shearing methods such as sonication will be required. On the MS side, RNA fragmentation via conventional collision-based methods like HCD not only induces the sequential backbone cleavage, but also unfavourable base loss, which eliminates sequence information and complicates spectrum interpretation 17 ( Figure S5E ). Nevertheless, sequencing of oligonucleotides as long as 100 nucleotides has been achieved with specialized fragmentation techniques, which will be instrumental to sequencing peptide-RNA crosslinks as well 62 . With the rising size of crosslinked molecules, MS methodology developed for the analysis of intact proteins or labile modifications might become necessary to prevent premature RNA fragmentation 63 , 64 . When aiming to detect crosslinks with longer RNA chains, the diversity of peptide-RNA combinations can quickly exceed the acquisition capacity of current MS instrumentation. Offline fractionation strategies or the adoption of ultra-fast MS acquisition methods may be a path forward to achieving transcriptome-wide analytical depth. Finally, no software supports RNA sequencing of photo-crosslinked hybrids at the moment. However, our dual-MS2 strategy has demonstrated that peptide and RNA sequencing can be separated already on the acquisition level, facilitating analysis of either one with specialized software for peptides or oligonucleotides. With these issues addressed, we anticipate that MS analysis of peptide-RNA crosslinks will soon map thousands of protein-RNA interaction sites with single amino acid and nucleotide resolution across the proteome and transcriptome in a single MS run. PepR-MS establishes the biochemical foundation for purifying long-chain peptide-RNA crosslinks, overcoming a long-standing bottleneck in the field and paving the way for bilateral analysis of protein-nucleic acid interactions by MS. Methods Cell culture MCF7 (human, female breast adenocarcinoma) was obtained from ATCC. Cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, PAN-Biotech GmbH) supplemented with 10% fetal bovine serum (FBS, PAN-Biotech GmbH), in a 37°C incubator with a CO 2 concentration of 5%. The cells were seeded in 150 mm diameter tissue culture dish (Sarstedt) with 20 mL culture medium at 2 million cells per dish. After three days of culture, cells were labeled for 18 h with 100 µM 4-SU (Biosynth) by adding from a 100 mM stock in DMSO directly to the culture medium. RNA digest spike-in experiments RNA was extracted from 4SU-labeled MCF7 cells with TRI reagent (Sigma Aldrich) according to the manufacturer’s instructions. After isopropanol precipitation, RNA pellets were dissolved in 50 mM Tris-HCl (pH 7.5) and adjusted to a concentration of 1 µg/µL, as determined by NanoDrop UV spectroscopy (Thermo Fisher Scientific). For 100 µg RNA (in 100 µL), residual proteins were digested by adding 20 µL proteinase K (NEB) at 20 µg/µL and incubating at 55°C for 15 min. The sample was cooled to room temperature, then 400 µL RNA Lysis Buffer (Zymo research, Quick-RNA Miniprep Kit) and 500 µL ethanol were added, mixed by inversion, purified over spin columns according to the manufacturer’s instructions and eluted with three times 100 µL ultrapure water. To 250 µL eluate, 5 µL 500 mM ammonium bicarbonate (final 10 mM), 2 µL RNase A (0.5 µg/µL, Thermo Fisher Scientific) and 2 µL RNase T1 (0.5 µg/µL, Sigma-Aldrich) were added and incubated at 37°C overnight. After RNA digestion, 0 (vehicle control), 0.5 or 5 µg RNA fragments as determined by NanoDrop were added to 150 ng HeLa peptide standard (in-house) in triplicates and the samples were lyophilized in a SpeedVac (Thermo Fisher Scientific) for LC-MS/MS analysis. Romidepsin treatment To each dish of 4SU labeled MCF7 cells, 200 µL romidepsin stock (1 mM in DMSO; final 10 µM romidepsin and 1% DMSO) or 200 µL DMSO (vehicle) were added to the culture medium and incubated at 37°C for 6 h. After the treatment, the cells were irradiated as described below. Per replicate three dishes were combined. UV crosslinking For bulb-based UV crosslinking, culture medium was removed from 4SU-labeled MCF7 cells, which were washed with 10 mL ice-cold Dulbecco’s phosphate buffered saline (PBS, with calcium chloride and magnesium chloride, Sigma-Aldrich), covered with 20 mL ice-cold PBS and kept on ice during irradiation. Photo-crosslinking was performed in a conventional UV crosslinker (Biolink, Vilber) with 365 nm bulbs for 5, 10, 100 or 500 s (for UV irradiation timeline samples, Figure S2B ) or to a total irradiation energy of 200 mJ/cm 2 (∼50 s, for workflow comparison samples, Figures 2B - 2E ). Cells were scraped into the ice-cold PBS they were irradiated in and pelleted in 3-dish batches by centrifugation at 500 g for 5 min for further processing. For LED-based UV crosslinking, cells were transferred directly from the incubator to the high-intensity irradiation device UVEN 18 and irradiated in their original culture medium with 365 nm LEDs for 1, 3, 5 or 10 s (for UV irradiation timeline samples, Figure S2B ) or 5 s (all other peptide-RNA crosslink samples if not specified above). Cells were washed twice with 5 mL ice-cold PBS, scraped into 15 mL ice-cold PBS and pelleted in 3-dish batches by centrifugation at 500 g for 5 min for further processing. PepR-MS workflow for peptide-RNA crosslink isolation Cell pellets from three 150 mm dishes were lysed in 3 mL TRI reagent by pipetting until completely homogeneous. Lysates were combined with 600 µL chloroform, mixed by inversion and centrifuged at 3000 g for 20 min at 4°C. The interphase was transferred to a fresh tube, dissolved in 1 mL TRI reagent by pipetting until completely homogeneous, combined with 200 µL chloroform, mixed by inversion and centrifuged at 12000 g for 10 min at 4°C. The aqueous and organic phases were removed, and the interphase was disintegrated in recovery buffer (10 mM Tris-HCl, 1 mM EDTA and 1% SDS, pH=7.4) until completely dissolved. For every 1 mL of dissolved sample, 100 µL 5 M NaCl and 1 mL isopropanol were added, mixed by inversion and incubated at -20°C overnight to precipitate RNA. Samples were centrifuged at 17000 g for 30 min at -11°C. The supernatant was discarded, and the pellets were washed with 1 mL 70% ethanol and combined into one tube for a 3-dish sample. Residual ethanol was removed to completion. Combined 3-dish pellets were hydrated with 760 µL ultrapure water, detached from the tube wall by gentle pipetting and incubated for 15 min on ice. DNA was digested by adding 90 µL 10× DNase I buffer (NEB), 2 µL RNasin ribonuclease inhibitor (Promega) and 50 µL 2000 units/mL DNase I (NEB), followed by incubation for 1 h at 37°C with shaking at 500 rpm. Protein was digested by adding 5 µL 20% SDS, 5 µL 1 M DTT and 60 µL 1 µg/µL trypsin (Roche) and incubating for 3 h at 37°C with shaking at 500 rpm. For workflow comparison experiment ( Figure 2B - 2E ), two 3-dish samples were pooled and split equally to minimize technical variation up to this step. The ‘without RNA fragment removal’ sample in the workflow comparison experiment (upper panel of Figure 2B - 2E ) was topped off with ultrapure water to 1 mL final volume and directly processed by silica column purification as described below. For all other peptide-RNA crosslink isolations, 100 µL 20% SDS was added to each trypsin-digested 3-dish sample, which was incubated for 15 min at 60°C with shaking at 500 rpm. RNA was precipitated by adding 100 µL 5 M NaCl and 1 mL isopropanol, mixing by inversion and incubating at -20°C for at least 30 min. Precipitated samples were centrifuged at 17000 g for 30 min at -11°C. The supernatant was discarded, and pellets were washed with 1 mL 70% ethanol. For all 3-dish samples (all samples unless specified as 12-dish below), pellets were resuspended in 1 mL ultrapure water. For all 12-dish samples (peptide-RNA crosslink sample for ID comparison in Figure 2F - 2I and Figure S2C , tunable-length samples in Figure 3 and Figure 4 , and sequential sequencing sample in Figure 5 ), four 3-dish pellets were combined and resuspended in 2 mL ultrapure water. To each mL of resuspended sample, 3.5 mL RLT buffer (Qiagen, RNeasy Midi Kit) was added and mixed by inversion. The mixture was incubated at 85°C for 15 min with shaking at 500 rpm and cooled to room temperature before adding 2.5 mL ethanol. Each 3-dish (total volume 7 mL) or 12-dish sample (total volume 14 mL) was loaded to one RNeasy Midi column in 3.5 mL increments, washed with 8 mL (3-dish samples) or 12 mL (12-dish samples) RW1 buffer and 9 mL RPE buffer, and eluted with 1 mL (3-dish samples) or 6 mL (12-dish samples) ultrapure water. Eluates were heated to 85°C for 1 min with shaking at 500 rpm. RNA was precipitated by adding 100 µL 5 M NaCl and 1 mL isopropanol to each 1 mL eluate, mixing by inversion and incubating at -20°C overnight. Precipitated samples were centrifuged at 17000 g for 30 min at -11°C. The supernatant was discarded, and the pellets were washed with 1 mL 70% ethanol, then resuspended with 100 µL (3-dish) or 400 µL (12-dish) 50 mM Tris-HCl (pH 7.5) and heated to 85°C for 1 min with shaking at 500 rpm. RNA amounts were determined by NanoDrop. For each 100 µg RNA, 2 µL RNase A (0.5 µg/µL) and 2 µL RNase T1 (0.5 µg/µL) were added and incubated for 16 h at 37°C with shaking at 500 rpm, unless specified otherwise. For tuning the RNA length with different nucleases, all samples were pooled and split equally into four groups (a-d below) and supplemented with 20 µL 100 mM MgCl 2 (final 5 mM). For these samples, each 100 µg RNA was incubated with a) 2 µL RNase A (0.5 µg/µL), 2 µL RNase T1 (0.5 µg/µL) and 2 µL benzonase (0.1 µg/µL; GENIUS nuclease, Santa Cruz Biotechnology); b) 2 µL RNase A (0.5 µg/µL); c) 2 µL RNase T1 (0.5 µg/µL); and d) 2 µL benzonase (0.1 µg/µL) for 16 h at 37°C with shaking at 500 rpm. The ‘original workflow’ sample in the workflow comparison experiment (upper panel of Figure 2B - 2E ) was adjusted to 150 µL with water and formic acid to a final concentration of 0.1% formic acid and loaded onto a self-packed C18 StageTip 65 (five discs, Sigma Aldrich). Tips were equilibrated with 0.1% formic acid. After loading, tips were washed four times with 200 µL 50 mM ammonium formate (pH 10) in water. Peptides were eluted with 150 µL 50 mM ammonium formate (pH 10) in 35% acetonitrile. For all other samples, volumes were adjusted to 150 µL (3-dish) or 600 µL (12-dish) with acetonitrile and formic acid to final concentrations of 30% acetonitrile and 0.1% formic acid, and samples were centrifuged at 20000 g for 5 min at room temperature. Ninety percent volume of each sample was transferred to a fresh tube and loaded onto a self-packed SCX StageTip (two discs, Sigma Aldrich). Tips were equilibrated with 0.1% formic acid in 30% acetonitrile. After loading, tips were washed four times with 200 µL 0.1% formic acid in 30% acetonitrile. Peptides were eluted with 150 µL 100 mM ammonium formate (pH 6.5) in 50% acetonitrile. Finally, all StageTip eluates were lyophilized in a SpeedVac for LC-MS/MS analysis. LC-MS/MS analysis All LC-MS/MS samples were measured on an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific) connected to a Dionex UltiMate 3000 RSLCnano system (Thermo Fisher Scientific) equipped with an in-house packed trap column (75 μm x 2 cm) packed with 5 μm C18 resin (Reprosil PUR AQ - Dr. Maisch) and an analytical column (75 μm x 70 cm, heated to 55°C) packed with 3 μm C18 resin (Reprosil PUR AQ - Dr. Maisch). Solvent A consisted of 0.1% formic acid and 5% DMSO in water; solvent B consisted of 0.1% formic acid and 5% DMSO in acetonitrile; solvent C consisted of 0.1% formic acid in water. Orbitrap resolution values are reported relative to m/z 200. LC-MS/MS for RNA digest spike-in experiments and peptide-RNA crosslink Lyophilized HeLa peptide standard with or without RNA digest spike-in and peptide-RNA crosslink samples (except for the sample for sequential sequencing in MS) were reconstituted in 12 µL solvent C and 10 μL were injected per run. After injection, the sample was transferred onto the trap column and then washed with C at 5 μL/min for 10 min before conveying them to the analytical column using a gradient of solvent A and solvent B. Separation was performed on an 80 min (12-dish peptide-RNA crosslink sample for ID comparison in Figure 2F - 2I and Figure S2C ) or a 50 min (HeLa peptide standard and all other peptide-RNA crosslink samples) linear gradient with a flow rate of 300 nL/min from 4% solvent B to 32% solvent B. Nano source voltage was 2100 V, ion transfer tube temperature 275°C. Detection occurred with data-dependent acquisition using an OTOT method and a cycle time of 2 s. Full MS scans (MS1) were acquired at a resolution of 60000, scan range 360-1300, RF lens 50%, normalized AGC target 100% and maximum injection time 50 ms. Based on the MS1s, precursors were targeted for MS2 scans if the charge was between 2 and 6, the monoisotopic peak determination (MIPS) was peptide, and the intensity exceeded 2.5E4. MS2 isolation occurred with a quadrupole isolation window of 1.7 Th and fragmentation with 30% HCD collision energy. MS2 spectra were acquired at a resolution of 15000 (HeLa peptide standard) or 30000 (peptide-RNA crosslink samples), first mass 100 Th, normalized AGC target 300% (HeLa peptide standard) or 100% (peptide-RNA crosslink samples), and maximum injection time 25 ms (HeLa peptide standard) or 50 ms (peptide-RNA crosslink samples). Dynamic exclusion was enabled with an exclusion duration of 20 s following precursor fragmentation. Collision energy ramp for oligonucleotides and peptides Twenty-four ng RNA oligo (5’-UAGUAG-3’, dissolved in 10 µL 0.1% formic acid), 100 fmol PROCAL 58 (dissolved in 1 µL 0.1% formic acid) or 50 ng HeLa peptide standard (dissolved in 1 µL 0.1% formic acid) were injected per run. After injection, the sample was transferred onto the trap column and then washed with C at 5 µL/min for 5 min before conveying them to the analytical column using a gradient of solvent A and solvent B. Separation was performed on an 9 min (RNA oligo) linear gradient from 4% solvent B to 20% solvent B or a 22 min (PROCAL and HeLa peptide standard) with a flow rate of 300 nL/min. Nano source voltage was 2100 V, ion transfer tube temperature 275°C. Detection occurred with data-dependent acquisition using an OTOT method and a cycle time of 2 s. For RNA oligo, MS1 acquisition used quadrupole-isolated targeted selected ion monitoring (tSIM) over the m/z range 1000-2000 Th at a resolution of 60000, RF lens 100%, normalized AGC target 200% and maximum injection time 50 ms. Based on the MS1s, precursors were targeted for MS2 scans if the charge was between 1 and 6. MS2 isolation occurred with a quadrupole isolation window of 1.3 Th and HCD collision energy was stepped from 10% to 30% in 2% increments. MS2 spectra were acquired at a resolution of 30000, first mass 100 Th, normalized AGC target 300%, and maximum injection time 100 ms. Dynamic exclusion was enabled with an exclusion duration of 10 s following precursor fragmentation. For PROCAL and HeLa peptide standard, full MS scans (MS1) were acquired at a resolution of 60000, scan range 360-1300, RF lens 50%, normalized AGC target 100% and maximum injection time 50 ms. Based on the MS1s, precursors were targeted for MS2 scans if the charge was between 2 and 6, the MIPS was peptide, and the intensity exceeded 2.5E4. MS2 isolation occurred with a quadrupole isolation window of 1.7 Th and HCD collision energy was stepped from 10% to 30% in 2% increments. MS2 spectra were acquired at a resolution of 15000, first mass 100 Th, normalized AGC target 300%, and maximum injection time 25 ms. Dynamic exclusion was enabled with an exclusion duration of 20 s following precursor fragmentation. Dual-MS2 method for sequential sequencing of peptide-RNA crosslinks Lyophilized peptide-RNA crosslinks produced with benzonase-digestion from twelve 150 mm dishes of MCF7 cells were reconstituted in 24 µL solvent C and 10 µL (approximately 6-dish input) were injected. After injection, the sample was transferred onto the trap column and then washed with C at 5 µL/min for 10 min before conveying them to the analytical column using a gradient of solvent A and solvent B. Separation was performed on a 50 min linear gradient with a flow rate of 300 nL/min from 4% solvent B to 32% solvent B. Nano source voltage was 2100 V, ion transfer tube temperature 275°C. Detection occurred with data-dependent acquisition using an OTOT method and a cycle time of 2 s. Full MS scans (MS1) were acquired at a resolution of 60000, scan range 360-1300, RF lens 50%, normalized AGC target 100% and maximum injection time 50 ms. Based on the MS1s, each precursor was targeted for two dependent MS2 scans if the charge was between 2 and 6, the MIPS was peptide, and the intensity exceeded 2.5E4. MS2 isolation occurred with a quadrupole isolation window of 1.7 Th and HCD collision energy was 30% for the first MS2 and 20% for the second. All MS2 spectra were acquired at a resolution of 30000, first mass 100 Th, normalized AGC target 100%, and maximum injection time 50 ms. Dynamic exclusion was enabled with an exclusion duration of 20 s following precursor fragmentation. Data analysis Database search All data was searched against the human UniProt database 66 (SwissProt reviewed, from December 2022, 20,434 entries). MaxQuant 67 (version 2.2.0.0) was used for searching HeLa peptide standards with RNA spike-ins, using all parameters at default value. All other data was searched with MSFragger 68 (version 4.1) within FragPipe (version 22.0) for peptide-RNA crosslink identification and quantification, together with IonQuant 69 (version 1.10.27) and Philosopher 70 (version 5.1.1). MS2 spectra were searched against the human UniProt database supplemented with common contaminants and a 50% decoy set using the built-in ‘Add decoys’ function in FragPipe. Default parameters for the built-in workflow ‘XRNAX-MassOffset’ were applied unless stated otherwise. Enzymatic specificity was set to ‘stricttrypsin’ with up to 2 missed cleavages allowed. Peptide length was set to 2 to 50 amino acids with a peptide mass range of 50-5000 Da. Mass tolerances were set to 10 ppm for precursor and 20 ppm for fragment, with isotope error correction of 0/1/2. For all pepR-MS samples, no fixed modification was set and variable modifications included oxidation (M) and protein N-terminal acetylation. For search of representative MS data from previous studies, carbamidomethyl (C) was set for fixed modification or no modification was set depending on their workflow (for details see parameter files deposited to MassIVE); variable modifications were set to oxidation (M) and protein N-terminal acetylation. Number of maximum variable modifications on a peptide was set to 3. Mass offset searching was enabled with modification allowing for all amino acids and a mass offset list customized for each workflow (pepR-MS or from previous studies). In general, for all pepR-MS MS data, mass offsets were generated by searching representative files with all possible mass offsets up to one nucleotide and the rarely identified mass offsets were filtered out. Based on the most identified mass offsets up to one nucleotide (0, 94.0167, 288.0147, 306.0253, 322.0025, and 340.013), consider all possible masses of RNA chain up to three (default) or four (only for the sequential sequencing sample) nucleotides. For representative MS data from previous studies, a similar procedure was performed to generate mass offsets with up to three nucleotides for each raw file from Kramer et al., 2014 6 and Trendel et al., 2019 9 , only considering uridine as the crosslinking base. For Bae et al. 15 , 16 mass offsets were directly adapted from the search parameters used in the original study. All mass offsets can be found in the search parameter files deposited to MassIVE. Labile-mode searching 71 was enabled for all mass offsets with remainder fragment masses of one nucleotide (for three-nucleotide searches, excluding 0) or mass offsets for single base (for HF-digested files). PSM validation used PeptideProphet and protein inference used ProteinProphet with default parameters for offset search. Quantification was performed with IonQuant using LFQ (MaxLFQ calculated with min. 2 ions) and match-between-runs set to off. NuXL 34 within OpenMS 33 (version 3.0.0-pre-HEAD) was used in addition to MSFragger to identify crosslinks with RNA chains longer than four nucleotides in samples digested with different nuclease combinations ( Figure 4 ) and benzonase-digested samples for sequential sequencing ( Figure 5 ). Raw files were converted to mzML format using MSConvert 72 (version 3.0.23011-b947b9f) by peak picking for all MS levels. MS2 spectra were searched against the human UniProt database. The search was performed with presets ‘RNA-UV Extended (4SU)’ up to five nucleotides for samples digested with different nuclease combinations and ‘report top 1 hit’, and up to six nucleotides for sequential sequencing and ‘report top 4 hits’, with NuXL scoring at slow mode. Variable modification is oxidation (M). Peptides with a length of 6 to 30 amino acids and maximal two variable modifications and two missed cleavages were allowed. Filters included ‘autotune’ ‘idfilter’ and ‘filter_pc_mass_error’. Statistical Testing All proteomic data was processed in R (version 4.3.0) within RStudio (version 2024.12.0). The count of identified crosslinks was defined as the number of unique modified peptides crosslinked to unique RNA mass offsets at specific amino acid positions in the peptide sequence (e.g. peptide PEPTIDEK crosslinked to an RNA of mass 306 at position P3). The count of identified crosslinking sites was defined as the number of unique crosslinking amino acid positions within the full-length protein sequence, regardless of the crosslinked RNA (e.g. G1158 in the protein DHX9). For differential analysis of crosslinks under romidepsin treatment, data was filtered to include only crosslinks identified in at least two replicates in either condition. Statistical analysis was performed using a linear model with empirical Bayes moderation in the R package limma 73 , with multiple testing correction with the Benjamini-Hochberg method. Regulated crosslinks were identified by the thresholds adj. p 2. Romidepsin-regulated acetylation and phosphorylation sites were retrieved from Chang et al. 2024 20 . Regulated phosphorylation or acylation sites were defined as potentially influencing RNA interactions if located within 10 amino acids of a crosslinking site. The coefficient of variation (CV) was calculated based on MaxLFQ intensities of crosslinks quantified in all four replicates under the same condition. For the analysis of crosslinked amino acids ( Figure 4 ), frequencies of crosslinked amino acids were calculated based on crosslink counts (i.e., frequency = number of crosslinks involving a given amino acid at the crosslinking site / total number of crosslinks). Statistical enrichment or depletion was tested using one-sided Fisher’s exact tests based on site counts, comparing amino acid frequencies at crosslinking sites to their overall frequencies in the protein sequences. P-values were corrected for multiple testing using the Benjamini–Hochberg procedure, considering both enrichment and depletion tests. Gene ontology (GO) term analysis was performed using the DAVID 74 (version 6.8) functional annotation tool, using a deep MCF7 full proteome as background 19 . RNA sequence motif generation Region-resolved RNA sequence motifs were derived from all PSMs identified by either NuXL or MSFragger that mapped to the same protein region, defined as PSMs with the identical peptide sequence and any peptides whose sequence overlaps the original by more than five amino acids. For each PSM, all possible permutations of the RNA chain were enumerated, with 4SU (‘X’) at the crosslinking site (position 0). Positions with unresolved nucleotide identity were annotated as ‘O’ ( Figure S3A ). The probability of ‘X’ at position 0 was fixed at 1, and the probability of each nucleotide (not ‘X’ or ‘O’) from the crosslinked RNA chain appearing at the relative position 𝑝 (𝑝 = −𝑘 + 1, …, 0, …, 𝑘 − 1, 𝑘 is the total length of the RNA chain) was calculated by: The probability of unresolved nucleotides ‘O’ at position 𝑝 was: The probability corresponding to unresolved nucleotides ‘O’ was then equally distributed to A, U, C, or G to create a position-specific sequence motif for each PSM. Motifs from all PSMs assigned to a given protein region were combined by their average to generate the final region-resolved sequence motifs. Protein domain annotation and analysis Domain annotations for crosslinked proteins were obtained from InterPro 75 (January 2025). Disordered regions were annotated using mobiDB 76 (January 2025), predicted with the IUPred3 long disorder model 77 . The list of human ribosomal proteins was downloaded from UniProt (189 entries, January 2025). To generate a viable AlphaFold 47 prediction of a generic RRM domain ( Figure 4E ), the Pfam consensus sequence needed to be extended on both sides. Only the consensus sequence extracted from the Pfam profile hidden Markov model (HMM) appeared truncated and lacked some secondary structure elements upon AlphaFold 47 prediction. To extend the profile HMM, the seed alignment of 68 sequences of the Pfam entry PF00076 was downloaded. These sequences were mapped back to their original protein sequences and extended at both the N and C-termini by five amino acids. The extended sequences were aligned to the original profile HMM using the hmmalign function in HMMER 78 (version 3.3) and the new alignment was used to build an extended HMM using hmmbuild. Crosslink annotations for RRMs were refined by searching all crosslinked proteins with the extended HMM using hmmsearch. Sequences of RRM domains annotated in this step were realigned with hmmalign, and crosslinks were mapped to this updated alignment for Figure 4E . The extended HMM was then used to generate sequence logos for the RRM, and its consensus sequence alongside the RNA sequence 5’-CAUC-3’ was provided as input for AlphaFold 3 47 structure prediction. Positions of secondary structures from the predicted RRM consensus ( Figures 4E and 4G ) were extracted using DSSP 4 79 . Data visualization Schematics in Figure 1 , 2 , S2, and S3 were partially created with Biorender. Data was plotted using the R packages ggplot2 80 (version 3.5.2) and ggseqlogo 81 (version 0.2). Protein structures were visualized in UCSF ChimeraX 82 (version 1.7.1). Chemical structures were drawn with Chemdraw (version 21.0.0). Figures were composed in Adobe Illustrator 2024. Data availability Proteomic data and search results have been deposited in the MassIVE database under the identifier MSV000097992. Author contributions Conceptualization, J.T., B.K.; methodology, S.S., J.T.; investigation, S.S.; writing—original draft, S.S., J.T., B.K.; resources, B.K.; data curation, S.S., J.T.; writing—review and editing, S.S., J.T., B.K.; visualization, S.S.; supervision, J.T., B.K.; project administration, J.T.; funding acquisition, J.T., B.K.. Competing interests B.K. is a co-founder and shareholder of MSAID. He has no operational role in the company. Acknowledgements We thank all members of the Kuster laboratory for continuous support and discussion. We gratefully acknowledge funding by the German Research Council (DFG) supporting this work (project number 325871075 (SFB1309-B02) and project number 492625837). Funder Information Declared Deutsche Forschungsgemeinschaft, https://ror.org/018mejw64 , 325871075 , 492625837 References 1. ↵ Liao , J. Y. et al. RBPWorld for exploring functions and disease associations of RNA-binding proteins across species . Nucleic Acids Res 53 , D220 – D232 ( 2025 ). OpenUrl CrossRef PubMed 2. ↵ Ule , J. et al. CLIP Identifies Nova-Regulated RNA Networks in the Brain . Science (1979) 302 , 1212 – 1215 ( 2003 ). OpenUrl Abstract / FREE Full Text 3. Hafner , M. et al. Transcriptome-wide Identification of RNA-Binding Protein and MicroRNA Target Sites by PAR-CLIP . Cell 141 , 129 – 141 ( 2010 ). OpenUrl CrossRef PubMed Web of Science 4. Schueler , M. et al. Differential protein occupancy profiling of the mRNA transcriptome . Genome Biol 15 , 1 – 17 ( 2014 ). OpenUrl CrossRef PubMed 5. ↵ Trendel , J. , Boileau , E. , Jochem , M. , Dieterich , C. & Krijgsveld , J . PEPseq quantifies transcriptome-wide changes in protein occupancy and reveals selective translational repression after translational stress . Nucleic Acids Res 51 , e79 – e79 ( 2023 ). OpenUrl PubMed 6. ↵ Kramer , K. et al. Photo-cross-linking and high-resolution mass spectrometry for assignment of RNA-binding sites in RNA-binding proteins . Nat Methods 11 , 1064 – 1070 ( 2014 ). OpenUrl CrossRef PubMed 7. ↵ Panhale , A. et al. CAPRI enables comparison of evolutionarily conserved RNA interacting regions . Nat Commun 10 , 2682 ( 2019 ). OpenUrl CrossRef PubMed 8. Queiroz , R. M. L. et al. Comprehensive identification of RNA–protein interactions in any organism using orthogonal organic phase separation (OOPS) . Nat Biotechnol 37 , 169 – 178 ( 2019 ). OpenUrl CrossRef PubMed 9. ↵ Trendel , J. et al. The Human RNA-Binding Proteome and Its Dynamics during Translational Arrest . Cell 176 , 391 – 403 .e19 ( 2019 ). OpenUrl CrossRef PubMed 10. ↵ Shchepachev , V. et al. Defining the RNA interactome by total RNA-associated protein purification . Mol Syst Biol 15 , ( 2019 ). 11. ↵ Castello , A. et al. Comprehensive Identification of RNA-Binding Domains in Human Cells . Mol Cell 63 , 696 – 710 ( 2016 ). OpenUrl CrossRef PubMed 12. ↵ Mullari , M. , Lyon , D. , Jensen , L. J. & Nielsen , M. L . Specifying RNA-Binding Regions in Proteins by Peptide Cross-Linking and Affinity Purification . J Proteome Res 16 , 2762 – 2772 ( 2017 ). OpenUrl CrossRef PubMed 13. ↵ He , C. et al. High-Resolution Mapping of RNA-Binding Regions in the Nuclear Proteome of Embryonic Stem Cells . Mol Cell 64 , 416 – 430 ( 2016 ). OpenUrl CrossRef PubMed 14. ↵ Heindel , A. J. et al. Chemoproteomic capture of RNA binding activity in living cells . Nat Commun 14 , 6282 ( 2023 ). OpenUrl CrossRef PubMed 15. ↵ Bae , J. W. , Kwon , S. C. , Na , Y. , Kim , V. N. & Kim , J. S . Chemical RNA digestion enables robust RNA-binding site mapping at single amino acid resolution . Nat Struct Mol Biol 27 , 678 – 682 ( 2020 ). OpenUrl CrossRef PubMed 16. ↵ Bae , J. W. , Kim , S. , Kim , V. N. & Kim , J. S . Photoactivatable ribonucleosides mark base-specific RNA-binding sites . Nat Commun 12 , ( 2021 ). 17. ↵ Riml , C. , Glasner , H. , Rodgers , M. T. , Micura , R. & Breuker , K . On the mechanism of RNA phosphodiester backbone cleavage in the absence of solvent . Nucleic Acids Res 43 , 5171 – 5181 ( 2015 ). OpenUrl CrossRef PubMed 18. ↵ Trendel , J. et al. Enhanced photo-crosslinking in living cells with high-intensity longwave ultraviolet light . Preprint at doi: 10.1101/2025.08.20.671286 ( 2025 ). OpenUrl Abstract / FREE Full Text 19. ↵ Trendel , J. et al. The human proteome with direct physical access to DNA . Cell 188 , ( 2025 ). 20. ↵ Chang , Y.-C. et al. Decrypting lysine deacetylase inhibitor action and protein modifications by dose-resolved proteomics . Cell Rep 43 , 114272 ( 2024 ). 21. ↵ Sharma , S. et al. Acetylation-Dependent Control of Global Poly(A) RNA Degradation by CBP/p300 and HDAC1/2 . Mol Cell 63 , 927 – 938 ( 2016 ). OpenUrl CrossRef PubMed 22. ↵ Poetz , F. et al. Control of immediate early gene expression by CPEB4-repressor complex-mediated mRNA degradation . Genome Biology 2022 23:1 23 , 1 – 35 ( 2022 ). OpenUrl CrossRef PubMed 23. ↵ Aratani , S. et al. The nuclear import of RNA helicase A is mediated by importin-α3 . Biochem Biophys Res Commun 340 , 125 – 133 ( 2006 ). OpenUrl CrossRef PubMed Web of Science 24. ↵ Calame , D. G. et al. Monoallelic variation in DHX9, the gene encoding the DExH-box helicase DHX9, underlies neurodevelopment disorders and Charcot-Marie-Tooth disease . Am J Hum Genet 110 , 1394 – 1413 ( 2023 ). OpenUrl CrossRef PubMed 25. ↵ Su , C. et al. Tudor staphylococcal nuclease (Tudor-SN), a novel regulator facilitating G1/S phase transition, acting as a co-activator of E2F-1 in cell cycle regulation . Journal of Biological Chemistry 290 , 7208 – 7220 ( 2015 ). OpenUrl Abstract / FREE Full Text 26. ↵ Li , C. L. , Yang , W. Z. , Chen , Y. P. & Yuan , H. S . Structural and functional insights into human Tudor-SN, a key component linking RNA interference and editing . Nucleic Acids Res 36 , 3579 – 3589 ( 2008 ). OpenUrl CrossRef PubMed Web of Science 27. ↵ Gutierrez-Beltran , E. , Denisenko , T. V. , Zhivotovsky , B. & Bozhkov , P. V . Tudor staphylococcal nuclease: biochemistry and functions . Cell Death & Differentiation 2016 23:11 23 , 1739 – 1748 ( 2016 ). OpenUrl CrossRef PubMed 28. ↵ Välineva , T. , Yang , J. & Silvennoinen , O . Characterization of RNA helicase A as component of STAT6-dependent enhanceosome . Nucleic Acids Res 34 , 3938 – 3946 ( 2006 ). OpenUrl CrossRef PubMed Web of Science 29. ↵ Sridharan , S. et al. Systematic discovery of biomolecular condensate-specific protein phosphorylation . Nature Chemical Biology 2022 18:10 18 , 1104 – 1114 ( 2022 ). OpenUrl PubMed 30. ↵ Haubold , B. , Pierstorff , N. , Möller , F. & Wiehe , T . Genome comparison without alignment using shortest unique substrings . BMC Bioinformatics 6 , 123 ( 2005 ). 31. ↵ Singh , R. & Valcárcel , J . Building specificity with nonspecific RNA-binding proteins . Nature Structural & Molecular Biology 2005 12:8 12 , 645 – 653 ( 2005 ). OpenUrl PubMed 32. ↵ Lambert , N. et al. RNA Bind-n-Seq: Quantitative Assessment of the Sequence and Structural Binding Specificity of RNA Binding Proteins . Mol Cell 54 , 887 – 900 ( 2014 ). OpenUrl CrossRef PubMed Web of Science 33. ↵ Röst , H. L. et al. OpenMS: a flexible open-source software platform for mass spectrometry data analysis . Nature Methods 2016 13:9 13 , 741 – 748 ( 2016 ). OpenUrl PubMed 34. ↵ Welp , L. M. et al. Chemical crosslinking extends and complements UV crosslinking in analysis of RNA/DNA nucleic acid–protein interaction sites by mass spectrometry . Nucleic Acids Res 53 , 13 – 14 ( 2025 ). OpenUrl CrossRef 35. ↵ Cléry , A. et al. Structure of SRSF1 RRM1 bound to RNA reveals an unexpected bimodal mode of interaction and explains its involvement in SMN1 exon7 splicing . Nature Communications 2021 12:1 12 , 1 – 12 ( 2021 ). OpenUrl PubMed 36. ↵ Slišković , I. , Eich , H. & Muller-McNicoll , M . Exploring the multifunctionality of SR proteins . Biochem Soc Trans 50 , 187 – 198 ( 2022 ). OpenUrl CrossRef PubMed 37. ↵ Görlach , M. , Wittekind , M. , Beckman , R. A. , Mueller , L. & Dreyfuss , G . Interaction of the RNA-binding domain of the hnRNP C proteins with RNA . EMBO J 11 , 3289 – 3295 ( 1992 ). OpenUrl PubMed Web of Science 38. Görlach , M. , Burd , C. G. & Dreyfuss , G . The determinants of RNA-binding specificity of the heterogeneous nuclear ribonucleoprotein C proteins . Journal of Biological Chemistry 269 , 23074 – 23078 ( 1994 ). OpenUrl Abstract / FREE Full Text 39. König , J. et al. iCLIP reveals the function of hnRNP particles in splicing at individual nucleotide resolution . Nat Struct Mol Biol 17 , 909 ( 2010 ). 40. ↵ Cieniková , Z. , Damberger , F. F. , Hall , J. , Allain , F. H. T. & Maris , C . Structural and Mechanistic Insights into Poly(uridine) Tract Recognition by the hnRNP C RNA Recognition Motif . J Am Chem Soc 136 , 14536 – 14544 ( 2014 ). OpenUrl CrossRef PubMed 41. ↵ Knott , G. J. et al. Structural basis of dimerization and nucleic acid binding of human DBHS proteins NONO and PSPC1 . Nucleic Acids Res 50 , 522 – 535 ( 2022 ). OpenUrl CrossRef PubMed 42. ↵ Passon , D. M. et al. Structure of the heterodimer of human NONO and paraspeckle protein component 1 and analysis of its role in subnuclear body formation . Proc Natl Acad Sci U S A 109 , 4846 – 4850 ( 2012 ). OpenUrl Abstract / FREE Full Text 43. ↵ Knott , G. J. , Bond , C. S. & Fox , A. H . The DBHS proteins SFPQ, NONO and PSPC1: a multipurpose molecular scaffold . Nucleic Acids Res 44 , 3989 ( 2016 ). OpenUrl CrossRef PubMed 44. ↵ Knörlein , A. et al. Nucleotide-amino acid π-stacking interactions initiate photo cross-linking in RNA-protein complexes . Nat Commun 13 , 2719 ( 2022 ). OpenUrl CrossRef PubMed 45. ↵ Romero , P. et al. Sequence Complexity of Disordered Protein . Proteins: Structure, Function, and Bioinformatics 42 , 1 – 139 ( 2000 ). OpenUrl 46. ↵ Williams , R. M. et al. The protein non-folding problem: amino acid determinants of intrinsic order and disorder . Pac Symp Biocomput 6 , 89 – 100 ( 2001 ). OpenUrl 47. ↵ Abramson , J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3 . Nature 2024 630:8016 630 , 493 – 500 ( 2024 ). OpenUrl CrossRef PubMed 48. ↵ Hargous , Y. et al. Molecular basis of RNA recognition and TAP binding by the SR proteins SRp20 and 9G8 . EMBO Journal 25 , 5126 – 5137 ( 2006 ). OpenUrl CrossRef PubMed Web of Science 49. Maris , C. , Dominguez , C. & Allain , F. H. T . The RNA recognition motif, a plastic RNA-binding platform to regulate post-transcriptional gene expression . FEBS Journal 272 , 2118 – 2131 ( 2005 ). OpenUrl PubMed 50. Oberstrass , F. C. et al. Structure of PTB Bound to RNA: Specific Binding and Implications for Splicing Regulation . Science (1979) 309 , 2054 – 2057 ( 2005 ). OpenUrl Abstract / FREE Full Text 51. ↵ Morgan , C. E. et al. The First Crystal Structure of the UP1 Domain of hnRNP A1 Bound to RNA Reveals a New Look for an Old RNA Binding Protein . J Mol Biol 427 , 3241 – 3257 ( 2015 ). OpenUrl CrossRef PubMed 52. ↵ Peng , Z. et al. A creature with a hundred waggly tails: intrinsically disordered proteins in the ribosome . Cell Mol Life Sci 71 , 1477 ( 2013 ). OpenUrl PubMed 53. ↵ Kumar , M. et al. ELM—the Eukaryotic Linear Motif resource—2024 update . Nucleic Acids Res 52 , D442 – D455 ( 2024 ). OpenUrl CrossRef PubMed 54. ↵ Sridharan , S. et al. Systematic discovery of biomolecular condensate-specific protein phosphorylation . Nature Chemical Biology 2022 18:10 18 , 1104 – 1114 ( 2022 ). OpenUrl PubMed 55. ↵ Okuwaki , M. , Tsujimoto , M. & Nagata , K . The RNA binding activity of a ribosome biogenesis factor, nucleophosmin/B23, is modulated by phosphorylation with a cell cycle-dependent kinase and by association with its subtype . Mol Biol Cell 13 , 2016 – 2030 ( 2002 ). OpenUrl Abstract / FREE Full Text 56. ↵ Hisaoka , M. , Nagata , K. & Okuwaki , M . Intrinsically disordered regions of nucleophosmin/B23 regulate its RNA binding activity through their inter- and intra-molecular association . Nucleic Acids Res 42 , 1180 – 1195 ( 2014 ). OpenUrl CrossRef PubMed 57. ↵ Jensen , O. N. , Kulkarni , S. , Aldrich , J. V. & Barofsky , D. F . Characterization of Peptide-Oligonucleotide Heteroconjugates by Mass Spectrometry . Nucleic Acids Res 24 , 3866 – 3872 ( 1996 ). OpenUrl CrossRef PubMed Web of Science 58. ↵ Zolg , D. P. et al. PROCAL: A Set of 40 Peptide Standards for Retention Time Indexing, Column Performance Monitoring, and Collision Energy Calibration . Proteomics 17 , ( 2017 ). 59. ↵ Khatter , H. , Myasnikov , A. G. , Natchiar , S. K. & Klaholz , B. P . Structure of the human 80S ribosome . Nature 2015 520:7549 520 , 640 – 645 ( 2015 ). OpenUrl CrossRef PubMed 60. ↵ Dorn , G. et al. Structural modeling of protein-RNA complexes using crosslinking of segmentally isotope-labeled RNA and MS/MS . Nat Methods 14 , 487 – 490 ( 2017 ). OpenUrl CrossRef PubMed 61. ↵ Sarnowski , C. P. et al. A highly sensitive protein-RNA cross-linking mass spectrometry workflow with enhanced structural modeling potential . Nucleic Acids Res 53 , ( 2025 ). 62. ↵ Lanzillotti , M. B. & Brodbelt , J. S . Progress in Tandem Mass Spectrometry Data Analysis for Nucleic Acids . Mass Spectrom Rev ( 2025 ) doi: 10.1002/MAS.21923 . OpenUrl CrossRef 63. ↵ Kilpatrick , L. E. & Kilpatrick , E. L . Optimizing High-Resolution Mass Spectrometry for the Identification of Low-Abundance Post-Translational Modifications of Intact Proteins . J Proteome Res 16 , 3255 ( 2017 ). OpenUrl PubMed 64. ↵ Tamara , S. , Den Boer , M. A. & Heck , A. J. R . High-Resolution Native Mass Spectrometry . Chem Rev 122 , 7269 – 7326 ( 2022 ). OpenUrl CrossRef PubMed 65. ↵ Rappsilber , J. , Mann , M. & Ishihama , Y . Protocol for micro-purification, enrichment, pre-fractionation and storage of peptides for proteomics using StageTips . Nat Protoc 2 , 1896 – 1906 ( 2007 ). OpenUrl CrossRef PubMed Web of Science 66. ↵ Bateman , A. et al. UniProt: the Universal Protein Knowledgebase in 2025 . Nucleic Acids Res 53 , D609 – D617 ( 2025 ). OpenUrl CrossRef PubMed 67. ↵ Cox , J. & Mann , M . MaxQuant enables high peptide identification rates, individualized p . p.b.-range mass accuracies and proteome-wide protein quantification. Nat Biotechnol 26 , 1367 – 1372 ( 2008 ). OpenUrl PubMed 68. ↵ Kong , A. T. , Leprevost , F. V , Avtonomov , D. M. , Mellacheruvu , D. & Nesvizhskii , A. I . MSFragger: ultrafast and comprehensive peptide identification in mass spectrometry–based proteomics . Nat Methods 14 , 513 – 520 ( 2017 ). OpenUrl CrossRef PubMed 69. ↵ Yu , F. , Haynes , S. E. & Nesvizhskii , A. I . IonQuant Enables Accurate and Sensitive Label-Free Quantification With FDR-Controlled Match-Between-Runs . Molecular & Cellular Proteomics 20 , 100077 ( 2021 ). 70. ↵ da Veiga Leprevost , F. , et al. Philosopher: a versatile toolkit for shotgun proteomics data analysis . Nat Methods 17 , 869 – 870 ( 2020 ). OpenUrl CrossRef PubMed 71. ↵ Polasky , D. A. et al. MSFragger-Labile: A Flexible Method to Improve Labile PTM Analysis in Proteomics . Molecular and Cellular Proteomics 22 , 100538 ( 2023 ). 72. ↵ Chambers , M. C. et al. A cross-platform toolkit for mass spectrometry and proteomics . Nat Biotechnol 30 , 918 – 920 ( 2012 ). OpenUrl CrossRef PubMed 73. ↵ Ritchie , M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies . Nucleic Acids Res 43 , e47 – e47 ( 2015 ). OpenUrl CrossRef PubMed 74. ↵ Sherman , B. T . et al. DAVID: a web server for functional enrichment analysis and functional annotation of gene lists (2021 update) . Nucleic Acids Res 50 , W216 – W221 ( 2022 ). OpenUrl CrossRef PubMed 75. ↵ Blum , M. et al. InterPro: the protein sequence classification resource in 2025 . Nucleic Acids Res 53 , D444 – D456 ( 2025 ). OpenUrl CrossRef PubMed 76. ↵ Piovesan , D. et al. MOBIDB in 2025: integrating ensemble properties and function annotations for intrinsically disordered proteins . Nucleic Acids Res 53 , D495 – D503 ( 2025 ). OpenUrl CrossRef PubMed 77. ↵ Erdős , G. , Pajkos , M. & Dosztányi , Z . IUPred3: prediction of protein disorder enhanced with unambiguous experimental annotation and visualization of evolutionary conservation . Nucleic Acids Res 49 , W297 – W303 ( 2021 ). OpenUrl CrossRef PubMed 78. ↵ Eddy , S. R. HMMER 3.3. http://hmmer.org/ ( 2023 ). 79. ↵ Hekkelman , M. L. , Salmoral , D. Á. , Perrakis , A. & Joosten , R. P . DSSP 4: FAIR annotation of protein secondary structure . Protein Science 34 , ( 2025 ). 80. ↵ Wickham , Hadley . Ggplot2: Elegant Graphics for Data Analysis . ( Springer , 2016 ). 81. ↵ Wagih , O. ggseqlogo: a versatile R package for drawing sequence logos . Bioinformatics 33 , 3645 – 3647 ( 2017 ). OpenUrl CrossRef PubMed 82. ↵ Meng , E. C. et al. UCSF ChimeraX: Tools for structure building and analysis . Protein Science 32 , ( 2023 ). View the discussion thread. Back to top Previous Next Posted September 07, 2025. Download PDF Supplementary Material Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Peptide-RNA photo-crosslinks with tunable RNA chain map protein-RNA interfaces Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Peptide-RNA photo-crosslinks with tunable RNA chain map protein-RNA interfaces Shuyao Sha , Bernhard Kuster , Jakob Trendel bioRxiv 2025.09.05.674461; doi: https://doi.org/10.1101/2025.09.05.674461 Share This Article: Copy Citation Tools Peptide-RNA photo-crosslinks with tunable RNA chain map protein-RNA interfaces Shuyao Sha , Bernhard Kuster , Jakob Trendel bioRxiv 2025.09.05.674461; doi: https://doi.org/10.1101/2025.09.05.674461 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Molecular Biology Subject Areas All Articles Animal Behavior and Cognition (7618) Biochemistry (17633) Bioengineering (13856) Bioinformatics (41841) Biophysics (21399) Cancer Biology (18529) Cell Biology (25422) Clinical Trials (138) Developmental Biology (13352) Ecology (19860) Epidemiology (2067) Evolutionary Biology (24282) Genetics (15582) Genomics (22462) Immunology (17700) Microbiology (40295) Molecular Biology (17140) Neuroscience (88419) Paleontology (666) Pathology (2823) Pharmacology and Toxicology (4813) Physiology (7632) Plant Biology (15107) Scientific Communication and Education (2042) Synthetic Biology (4284) Systems Biology (9808) Zoology (2267)