Search

Background: We aimed to establish a novel diagnostic model for kidney diseases by combining artificial intelligence with complete mass spectrum information from urinary proteomics. Methods: We enrolled 134 patients (IgA nephropathy, membranous nephropathy, and diabetic kidney disease) and 68 healthy participants as controls, with a total of 610,102 mass spectra from their urinary proteomic profiles. The training data set (80%) was used to create a diagnostic model using XGBoost, random forest (RF), a support vector machine (SVM), and artificial neural networks (ANNs). The diagnostic accuracy was evaluated using a confusion matrix with a test dataset (20%). We also constructed receiver operating-characteristic, Lorenz, and gain curves to evaluate the diagnostic model. Results: Compared with the RF, SVM, and ANNs, the modified XGBoost model, called Kidney Disease Classifier (KDClassifier), showed the best performance. The accuracy of the XGBoost diagnostic model was 96.03%. The area under the curve of the extreme gradient boosting (XGBoost) model was 0.952 (95% confidence interval, 0.9307–0.9733). The Kolmogorov-Smirnov (KS) value of the Lorenz curve was 0.8514. The Lorenz and gain curves showed the strong robustness of the developed model. Conclusion: The KDClassifier achieved high accuracy and robustness and thus provides a potential tool for the classification of kidney diseases. Keywords: Kidney disease classification, urinary proteomics, machine learning algorithm, diagnosis, artificial intelligence

Immunoglobulin G (IgG) molecules modulate an immune response. However, site-specific N-glycosylation signatures of plasma IgG in patients with chronic kidney disease (CKD) remain unclear. This study aimed to propose a novel method to explore the N-glycosylation pattern of IgG and to compare it with reported methods. We separated human plasma IgG from 58 healthy controls (HC) and 111 patients with CKD. Purified IgG molecules were digested by trypsin. Tryptic peptides without enrichment of intact N-glycopeptides were analyzed using a combination of electron-transfer/higher-energy collisional dissociation (EThcD) and stepped collision energy/higher-energy collisional dissociation (sceHCD) mass spectrometry (EThcD-sceHCD-MS/MS). This resulted in higher spectral quality, more informative fragment ions, higher Byonic score, and nearly twice the depth of intact N-glycopeptide identification than sceHCD or EThcD alone. Site-specific N-glycosylation mapping revealed that intact N-glycopeptides were differentially expressed in HC and CKD patients; thus, it can be a diagnostic tool. This study provides a method for the determination of glycosylation patterns in CKD and a framework for understanding the role of IgG in the pathophysiology of CKD. Data are available via ProteomeXchange with identifier PXD027174.

The discovery of novel non-invasive biomarkers for discriminating between prostate carcinoma (PCa) patients and benign prostatic hyperplasia (BPH) patients is necessary to reduce the burden of biopsies, avoid overdiagnosis and improve quality of life. Previous studies suggest that abnormal glycosylation of immunoglobulin gamma molecules (IgGs) is strongly associated with immunological diseases and prostate diseases. Hence, characterizing N-linked intact glycopeptides of IgGs that correspond to the N-glycan structure with specific site information might enable a better understanding of the molecular pathogenesis and discovery of novel signatures in preoperative discrimination of BPH from PCa. In this study, we profiled N-linked intact glycopeptides of purified IgGs from 51 PCa patients and 45 BPH patients by our developed N-glycoproteomic method using hydrophilic interaction liquid chromatography enrichment coupled with high resolution LC-MS/MS. The quantitative analysis of the N-linked intact glycopeptides using pGlyco 2.0 and MaxQuant software provided quantitative information on plasma IgG subclass-specific and site-specific N-glycosylation. As a result, we found four aberrantly expressed N-linked intact glycopeptides across different IgG subclasses. In particular, the N-glycopeptide IgG2-GP09 (EEQFNSTFR (H5N5S1)) was dramatically elevated in plasma from PCa patients, compared with that in BPH patients (PCa/BPH ratio = 5.74, p = 0.001). Additionally, the variations in these N-linked intact glycopeptide abundances were not caused by the changes in the IgG concentrations. Furthermore, IgG2-GP09 displayed a more powerful prediction capability (auROC = 0.702) for distinguishing PCa from BPH than the clinical index t-PSA (auROC = 0.681) when used alone or in combination with other indicators (auROC = 0.853). In conclusion, these abnormally expressed N-linked intact glycopeptides have potential for non-invasive monitoring and pre-stratification of prostate diseases.

N-glycoproteins are involved in various biological processes. Certain distinctive glycoforms on specific glycoproteins enhance the specificity and/or sensitivity of cancer diagnosis. Therefore, the characterization of plasma N-glycoproteome is essential for a new biomarker discovery. The absence of suitable analytical methods for in-depth and large-scale analyses of low-abundance plasma glycoproteins makes it challenging to investigate the role of glycosylation. In this study, we developed an integrated method termed Glyco-CPLL, which integrates combinatorial peptide ligand libraries, high-pH reversed-phase prefractionation, hydrophilic interaction chromatography, trypsin and PNGase F digestion, shotgun proteomics, and various analysis software (MaxQuant and pGlyco2.0) for the low-abundance plasma glycoproteomic profiling. Then, we utilized the method to perform a comparative study and to explore papillary thyroid carcinoma-related proteins and glycosylations with reference to healthy controls. Finally, a large and comprehensive human plasma N-glycoproteomic database was established, containing 786 proteins, 369 N-glycoproteins, 862 glycosites, 171 glycan compositions, and 1644 unique intact N-glycopeptides. Additionally, several low-abundance plasma glycoproteins were identified, including SVEP1 (∼0.54 ng/mL), F8 (∼0.83 ng/mL), and ADAMTS13 (∼1.2 ng/mL). These results suggest that this method will be useful for analyzing plasma intact glycopeptides in future studies. Besides, the Glyco-CPLL method has a great potential to be translated to clinical applications. Data are available via ProteomeXchange with identifier PXD016428.

Immunoglobulin G (IgG) molecules modulate an immune response. However, site-specific

BackgroundBy extracting the spectrum features from urinary proteomics based on an advanced mass spectrometer and machine learning algorithms, more accurate reporting results can be achieved for disease classification. We attempted to establish a novel diagnosis model of kidney diseases by combining machine learning with an extreme gradient boosting (XGBoost) algorithm with complete mass spectrum information from the urinary proteomics.MethodsWe enrolled 134 patients (including those with IgA nephropathy, membranous nephropathy, and diabetic kidney disease) and 68 healthy participants as a control, and for training and validation of the diagnostic model, applied a total of 610,102 mass spectra from their urinary proteomics produced using high-resolution mass spectrometry. We divided the mass spectrum data into a training dataset (80%) and a validation dataset (20%). The training dataset was directly used to create a diagnosis model using XGBoost, random forest (RF), a support vector machine (SVM), and artificial neural networks (ANNs). The diagnostic accuracy was evaluated using a confusion matrix. We also constructed the receiver operating-characteristic, Lorenz, and gain curves to evaluate the diagnosis model.ResultsCompared with RF, the SVM, and ANNs, the modified XGBoost model, called a Kidney Disease Classifier (KDClassifier), showed the best performance. The accuracy of the diagnostic XGBoost model was 96.03% (CI = 95.17%-96.77%; Kapa = 0.943; McNemar’s Test, P value = 0.00027). The area under the curve of the XGBoost model was 0.952 (CI = 0.9307-0.9733). The Kolmogorov-Smirnov (KS) value of the Lorenz curve was 0.8514. The Lorenz and gain curves showed the strong robustness of the developed model.ConclusionsThis study presents the first XGBoost diagnosis model, i.e., the KDClassifier, combined with complete mass spectrum information from the urinary proteomics for distinguishing different kidney diseases. KDClassifier achieves a high accuracy and robustness, providing a potential tool for the classification of all types of kidney diseases.

The densely glycosylated spike (S) proteins that are highly exposed on the surface of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) facilitate viral attachment, entry, and membrane fusion. We have previously reported all the 22 N-glycosites and site-specific N-glycans in the S protein protomer. Herein, we report the comprehensive and precise site-specific O-glycosylation landscapes of SARS-CoV-2 S proteins, which were characterized using high-resolution mass spectrometry. Following digestion using trypsin and trypsin/Glu-C, and de-N-glycosylation using PNGase F, we determined the mucin-type (GalNAc-type) O-glycosylation pattern of S proteins, including unambiguous O-glycosites and the 6 most common O-glycans occupying them, via Byonic identification and manual validation. Finally, 43 O-glycosites were identified in the insect cell-expressed S protein. Most glycosites were modified by non-sialylated O-glycans such as HexNAc(1) and HexNAc(1)Hex(1). In contrast, 30 O-glycosites were identified in the human cell-expressed S protein S1 subunit. Most glycosites were modified by sialylated O-glycans such as HexNAc(1)Hex(1)NeuAc(1) and HexNAc(1)Hex(1)NeuAc(2). Our results are the first to reveal that the SARS-CoV-2 S protein is a mucin-type glycoprotein; clustered O-glycans often occur in the N- and the C-termini of the S protein, and the O-glycosite and O-glycan compositions vary with the host cell type. These site-specific O-glycosylation landscapes of the SARS-CoV-2 S protein are expected to provide novel insights into the viral binding mechanism and present a strategy for the development of vaccines and targeted drugs.

Native intact N-glycopeptide analysis can provide access to the comprehensive characteristics of N-glycan occupancy, including N-glycosites, N-glycan compositions, and N-glycoproteins for complex samples. The sample pre-processing method used for the analysis of intact N-glycopeptides usually depends on the enrichment of low abundance N-glycopeptides from a tryptic peptide mixture using hydrophilic substances before LC-MS/MS detection. However, the number of identified intact N-glycopeptides remains inadequate to achieve an in-depth profile of the N-glycosylation landscape. Here, we optimized the sample preparation workflow prior to LC-MS/MS analysis by systematically comparing different analytical methods, including the use of different sources of trypsin, combinations of different proteases, and different enrichment materials. Finally, we found that the combination of Trypsin (B)/Lys-C digestion and zwitterionic HILIC (Zic-HILIC) enrichment significantly improved the mass spectrometric characterization of intact N-glycopeptides, increasing the number of identified intact N-glycopeptides and offering better analytical reproducibility. Furthermore, the optimized workflow was applied to the analysis of intact N-glycopeptides in two-dimensional (2D) and three-dimensional (3D)-cultured breast cancer cells in vitro and xenografted tumors in mice. These results indicated that the same breast cancer cells, when cultured in different microenvironments, can show different N-glycosylation patterns. This study also provides an interesting comparison of the N-glycoproteome of breast cancer cells cultured in different growth conditions, indicating the important role of N-glycosylated proteins in cancer cell growth and the choice of the cell culture model for studies in tumor biology and drug evaluation.

Papillary thyroid carcinoma (PTC) is the most common type of thyroid cancer among women worldwide. It is confirmed mainly by fine-needle aspiration biopsy (FNAB), an invasive diagnostic method. The key proteins responsible for thyroid hormone biosynthesis are glycosylated. Hence, changes in site-specific glycosylation are associated with thyroid cancer. Integrated quantitative proteomic and glycoproteomic analyses of body fluids from patients with PTC may identify potential noninvasive biomarkers, improve diagnostic accuracy, and elucidate the basic mechanisms of tumor development. In the present study, we demonstrate an integrated, highly reproducible, rapid method involving body fluid proteome and glycoproteome analysis. Moreover, this method may quantitatively profile protein glycosylation. Intact N-glycopeptides from the urine and plasma of healthy controls (HC), PTC, and PTC with Hashimoto's thyroiditis (PHT) were enriched by hydrophilic interaction liquid chromatography. Sialic acid was removed from the N-glycopeptides with trifluoroacetic acid and heat. The desialo-N-glycopeptides were analyzed by HCD-MS/MS using stepped collision energies and several search engines for quantitative profiling. Ninety-two altered proteins and 134 intact N-glycopeptides were isolated from the plasma and urine samples of the three groups (90 samples from 15 subjects). To the best of our knowledge, this study is the first to compare the plasma and urinary proteomes and glycoproteomes of HC, PTC, and PHT. Moreover, we reveal a novel indicator (ratio of fucosylated to nonfucosylated N-glycopeptide or F/NF) through desialo-N-glycopeptide analysis. These differently expressed glycoproteins and F/NF may serve as biomarkers contributing to clinical cancer diagnostics and could be used to improve diagnostic accuracy noninvasively.

The nairoviruses include assorted tick-borne bunyaviruses that are emerging as causative agents of infectious diseases among humans and animals. As negative-sense single-stranded RNA (−ssRNA) viruses, nairoviruses encode nucleoprotein (NP) that encapsidates the genomic RNA and further forms ribonucleoprotein (RNP) complex with viral RNA-dependent RNA polymerase (RdRp). We previously revealed that the monomeric NP encoded by Crimean-Congo hemorrhagic fever virus (CCHFV) presents a racket-shaped structure and shows unusual DNA-specific endonuclease activity. To examine the structural and biological variation of nairovirus-encoded NPs, here, we systematically solved the crystal structures of NPs encoded by various nairoviruses, including Hazara virus (HAZV), Kupe virus (KUPV), and Erve virus (ERVEV). Combined with biochemical analysis, our results generate a clearer picture to aid in the understanding of the functional diversity of nairovirus-encoded NPs and the formation of nairovirus RNPs. IMPORTANCE Nairoviruses comprise several tick-borne bunyaviruses that are emerging as causative agents of infectious diseases among humans and animals; however, little is known of the nairovirus genome assembly and transcription mechanisms. Based on the previous study of CCHFV NP reported by different research groups, we systematically investigate here the structural and functional diversity among three different nairoviruses. This work provides important information on nairovirus nucleoprotein function and the formation of RNPs.

Because of its high infectivity and pathogenicity,Mycobacterium tuberculosisis a serious threat to human health. While the transcription-regulatory system ofM. tuberculosisremains incompletely understood, Rv0081, an essential regulatory hub, is known to mediate the initial response to hypoxia in the long-term survival ofM. tuberculosis. Here, the production, crystallization and initial X-ray crystallographic analysis of Rv0081 are reported. The crystals of Rv0081 belonged to space groupP62, with unit-cell parametersa= 67.48,b = 67.48,c = 40.84 Å, γ = 120°. The Matthews coefficient is 2.09 Å3 Da−1, assuming the presence of one molecule in the asymmetric unit, with a corresponding solvent content of 41.27%. Phasing of the native crystal form of Rv0081 was performed by molecular replacement. Currently, the structure has been refined to 2.00 Å resolution with anRworkof 25.99% and anRfreeof 30.88%.

Esophageal squamous cell carcinoma (ESCC) is the eighth most common cause of cancer-related death worldwide. However, previous genome-wide single nucleotide polymorphism association analyses have not explained the high heritability associated with ESCC. In this study, we performed genome-wide copy number variation (CNV) analysis on 128 discordant sibling pairs to identify novel genes that contribute to ESCC susceptibility. A total of 57 774 individual CNVs were identified, and an interactive network of common CNV-associated genes was constructed, which showed that several ABC transporter genes contain CNVs in ESCC patients. Independent validation of a CNV at 13q32.1 in 1048 northern Chinese Han subjects demonstrated that the amplification of ABCC4 significantly correlated with ESCC risk [odds ratio: 3.36 (1.65-7.93), P = 0.0013]. Immunohistochemistry staining suggested that high copy numbers correlated with increased protein levels. High expression of ABCC4 was an independent poor prognostic factor for ESCC [relative risk: 1.73 (1.10-2.73), P = 0.0181]. The CNV region showed strong enhancer activity. Furthermore, inhibition of ABCC4 protein in ESCC cells decreased cell proliferation and motility via the inhibition of COX-2, PGE2 receptors and c-Myc expression; AKT, extracellular signal-regulated kinase and cAMP response element-binding protein phosphorylation; and β-catenin nuclear translocation in ESCC cells. In conclusion, the CNV at 13q32.1 is associated with ESCC susceptibility, and a gene within this locus, ABCC4, activates the oncogenic pathways in ESCC and thus facilitates cancer cell development and progression. A direct genetic contribution of ESCC risk through CNV common variants was determined in this study, and ABCC4 might therefore have predictive and therapeutic potential for ESCC.

Background: Esophageal squamous cell carcinoma (ESCC) is the major lethal upper gastrointestinal tract cancer in China. The highly regional and family aggregative incidence of ESCC in China suggests that genetic susceptibility contributed to pathogenesis. Copy number variation (CNV) is an important form of gene structure variations and has been associated with multiple complex diseases, such as malignant tumors. This study aimed to identify the common CNVs correlated with ESCC susceptibility and demonstrate the functional significances. Method: A genome-wide associated study was conducted on 128 discordant sibling pairs of ESCC, which were native residents in Yangquan area, Shan Xi Province of China. The genomic DNA of peripheral blood leukocytes were extracted and genotyping was detected using the Illumina HumanHap 610Q BeadChips. After stringent quality control filtering, the CNVs-penotype association analysis were performed using the paired Hidden Markov Model calling algorithm in the Partek Genomic Suite with settings of length above 1kb and minimum 10 consecutive probes. Another 223 Northern Chinese Han people (111 ESCC patients and 112 health controls) were chosen to validate the CNVs using quantitative real-time PCR assays. Then the paraffin-embedded tissue samples of 16 ESCC patients with known genotyping data were collected for immunohistochemistry. Last, the highest frequency of variation region was cloned into luciferase reporter vectors to investigate its biological significance in ESCC cells. Results: A total of 57,774 individual CNVs covering all autosomes were identified. Of them, 13q32.1 was identified as a critical enrichment region, and it mainly was involved in one candidate gene. Conditional logistic regression test showed that this region was significantly associated with ESCC with the odd ratio being 65.289 (95%CI: 11.177-381.366; P Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 5133. doi:1538-7445.AM2012-5133