The protein corona formed on the surface of polystyrene nanoparticles (PSNPs) was fully characterized using cryo-transmission electron microscopy (cryo-TEM), dynamic light scattering (DLS), zeta potential, and analyzed through both BUP and TDP MS (Fig. 2a-c). All the generated BUP and TDP data are listed in Supporting Data 1. Ensuring that the corona is free from significant aggregation or protein contamination is essential for accurate characterization of the protein-nanoparticle interactions.³¹ Cryo-TEM images demonstrated highly monodispersed, protein corona-coated PSNPs, confirming the successful formation of a uniform and pure corona layer (Fig. 2b). DLS and zeta potential measurements conducted before and after corona formation showed consistent results indicating successful coating: the nanoparticle size increased, reflecting the presence of the protein corona, while the surface charge became less negative post-coating (Fig. 2c). These observations are in full agreement with reported literature findings, supporting the reproducibility and reliability of the corona formation process^32–39.

To achieve a more robust and comprehensive understanding of how integrating BUP and TDP enhances the accuracy and reliability of proteoform characterization for protein corona, we need to have a large protein corona proteoform dataset. To produce this proteoform dataset, we analyzed a spectrum of protein coronas from various human plasma samples, for example, three samples from healthy individuals, five samples from patients with grade I breast cancer, and four samples from patients with grade II breast cancer. The diversified proteome profiles of human plasma samples from various individuals and health conditions help improve the number of proteoform identifications from protein coronas¹². The proteoform profiles of protein coronas from diverse individuals could also reflect the biological variability associated with personalized and disease-specific factors⁴⁰. We also employed two different measurement approaches, capillary zone electrophoresis-tandem mass spectrometry (CZE-MS/MS) and reversed-phase liquid chromatography (RPLC)-MS/MS (Fig. 2a), to boost the number of proteoform identifications from protein coronas, because these two approaches have been well documented for complementary peptide/proteoform identification from complex proteomes^25,41–47.CZE-MS/MS identified 2,272 proteoforms and 283 proteoform families—approximately 34% more proteoforms (2,272 vs. 1,692) and 50% more families (283 vs. 189) than RPLC-MS/MS. The relatively low overlap of proteoforms between the two methods highlights their strong complementarity in enhancing the depth of corona proteoform analysis, Fig. 3a. By collectively analyzing protein coronas from 12 human plasma samples, we identified a total of 3,503 proteoforms corresponding to 344 genes, Fig. 3b. Interestingly, the protein corona proteoform profiles of the three types of human plasma samples (healthy control, grade I breast cancer, and grade II breast cancer) are substantially different, evidenced by the low proteoform overlaps among the three sample types, Fig. S1.

While TDP enabled the detection of intact proteoforms in the protein corona, it alone was insufficient for comprehensive PTM characterization, primarily due to incomplete backbone cleavages that limited precise PTM localization. To address this, we integrated TDP with two BUP experiments, Fig. 2a. In the first BUP experiments, one-third of each corona sample was digested and analyzed by RPLC-MS/MS in triplicate. In this experiment, we identified an average of 390 protein groups and 2,645 peptide groups per sample (Fig. 3d), totaling 588 unique proteins and 4,899 unique peptides across all samples (Fig. 3d). The protein mass from BUP is up to 600 kDa and the TDP data only covers proteoforms smaller than 30 kDa, Fig. 3c, which represents another technical challenge of TDP regarding large proteoform identification. In the second BUP experiment, we aim to create a much larger peptide dataset to cover more PTM information for better interpretation of TDP data. We pooled the leftover peptide materials from all 12 human plasma samples to produce a more complex peptide mixture and employed high-pH RPLC fractionation followed by nanoflow RPLC-MS/MS to analyze the sample. To maximize the PTM information, we utilized an open-search approach with MSFragger⁴⁸. We identified 4570 protein groups, 45790 peptides, and 23632 peptides containing modifications, e.g., glycosylation, phosphorylation, acetylation, oxidation, and deamidation. The number of protein IDs in this study represents one of the largest human plasma proteome datasets in one study and is more than 150% higher than that from previous polystyrene NP-based protein corona studies^49,50. The large number of peptides with PTMs allows us to establish a PTM library for the PSNP-based protein corona. We also performed another database search using Proteome Discoverer (PD2.2, SEQUEST HT) and identified 4504 protein groups, 35543 peptides, and 3933 peptides with PTMs. The number of peptides with PTMs is much smaller compared to MSFragger because we only specified several specific PTMs (i.e., oxidation, acetylation, methylation, succinylation, and phosphorylation) in the PD search. We then integrated the BUP and TDP datasets using PTM-TBA to enhance the characterization quality of proteoforms, particularly in terms of annotation and localization of PTMs. We mainly used the MSFragger BUP data for this purpose.

Utilizing this integrated pipeline, we successfully matched the BUP PTM/mass-shift data (MSFragger) with the TDP mass-shift data for hundreds of proteoforms—471 proteoforms from the CZE-MS/MS dataset (representing 35.9% of the 1,312 proteoforms containing mass shifts) and 331 proteoforms from the RPLC-MS/MS dataset (34.5% of the 958 proteoforms with mass shifts), Fig. 4.

The matched proteoform and peptide information are listed in the Supporting Information. The BUP and TDP combination approach allows us to confirm or determine some common PTMs on proteoforms, e.g., oxidation, multiple oxidation combinations, deamidation, acetylation, phosphorylation, and lysine (K) deletion, Fig. 4. Many mass shifts in the identified proteoforms cannot be matched with the BUP data regarding PTMs because those mass shifts could be due to the combinations of different PTMs, and the current version of PTM-TBA software cannot handle this situation, which will be one focus of the future development of the software.

Figure 5 shows four examples of enhancing proteoform characterization quality by the combination of BUP and TDP. We observed a proteoform derived from myosin-9 (MYH9) carrying a + 79.96 Da mass shift. The integrated TDP–BUP analysis identified this modification as serine phosphorylation, supported by matching phosphopeptides detected in the bottom-up dataset (Fig. 5d). In another case, a prominent corona protein, —the major protein component of high-density lipoprotein (HDL) known for its protective roles against cardiovascular disease—exhibited a − 128.06 Da mass loss. Bottom-up sequencing revealed this to be a lysine deletion (Fig. 5c). Furthermore, an additional APOA1 proteoform displayed a + 42 Da mass shift. Without the combined analysis, this subtle PTM could have remained ambiguous; the bottom-up data confirmed it as lysine acetylation (Fig. 5a). Lysine acetylation is a well-established regulatory PTM that modulates protein function, interactions, and localization, underscoring the functional relevance of this modification in the protein corona environment. Finally, TDP revealed a proteoform from apolipoprotein F (APOF) with a + 48.07 Da mass shift, which, in conjunction with BUP data, was characterized as triple oxidation (Fig. 5b). Figures S2-S4 illustrate additional examples, showing the improved determination and localization of modifications on proteoforms of Transthyretin (TTR) and apolipoprotein A-I (APOA1).

We further studied the proteoform profile differences of protein corona of human plasma samples from healthy controls and breast cancer patients (Grades I and II). Label-free quantification enabled measurement of proteoform abundances across groups (healthy vs. Grade I vs. Grade II). Differential expression analysis revealed differentially expressed proteoforms associated with disease progression: 115 proteoforms (from 23 genes) in the RPLC–MS/MS dataset (Fig. S5) and 31 proteoforms (from 10 genes) in the CZE–MS/MS dataset (Fig. S6). Those groups of differentially expressed proteoforms clearly separate the various disease conditions, documenting the potential of TDP-based protein corona analysis for disease diagnosis. The combination of TDP and BUP also improved the characterization of the differentially expressed proteoforms, Fig. S5. A notable case was an apolipoprotein C-II (APOC2) proteoform, markedly enriched in Grade II samples compared to Grade I and healthy controls. Top-down analysis showed a + 16 Da mass shift, consistent with single oxygen addition, and bottom-up sequencing confirmed methionine oxidation (methionine sulfoxide) at a defined site. Methionine oxidation is a hallmark of oxidative stress⁵¹, and the enrichment of this oxidized APOC2 proteoform in Grade II patients likely reflects the elevated oxidative environment of advanced cancer, with possible implications for APOC2’s role in lipid metabolism and corona interactions. Another example involved an apolipoprotein B-100 (APOB) proteoform, abundant in healthy samples but depleted in both patient groups. This proteoform carried a + 31.98 Da shift, identified as dihydroxylation, which was localized to a specific APOB region by bottom-up analysis. The loss of this modified APOB proteoform in cancer patients underscores how PTM-defined proteoforms can distinguish health from disease within the plasma corona.

Overall, our results demonstrate that the integration of TDP and BUP strategies significantly enhances our ability to accurately characterize proteoforms and their PTMs within complex protein corona. This comprehensive approach will advance the field of nanomedicine by providing an accurate proteome landscape in protein corona and offering critical insights into how specific PTMs may influence protein behavior, surface affinity, and nanoparticle interactions, thereby advancing our understanding of proteoform diversity in disease contexts.

This study pioneers the integration of BUP and TDP data for the accurate characterization of proteoform landscape in protein corona. The novel approach markedly advances the precise characterization of proteoforms and their PTMs (i.e., types and localizations) within the protein corona. By combining the strengths of both approaches—TDP providing intact proteoform information and bottom-up offering detailed PTM localization—we achieve a level of resolution and confidence unattainable by either method alone. The development of the PTM-TBA pipeline further enhances data integration, enabling accurate PTM annotation and site-specific localization across complex biological samples. Our findings highlight the critical influence of PTMs on protein–nanoparticle interactions and highlight the importance of proteoform-level analysis in nanomedicine research. This comprehensive methodology also enables precise localization of modifications and revealing proteoform diversity associated with disease states. The observed differences in PTM abundances across healthy and breast cancer samples demonstrate the potential of proteoform profiling in biomarker discovery and personalized nanomedicine applications. Ultimately, this comprehensive characterization approach offers valuable insights into nanoparticle biodistribution, biosystem interactions, and proteoform-based biomarker discovery, paving the way for improved design and application of nanomedicines with enhanced safety and efficacy.