Han-Yin Yang (Presenter)
University of Washington
Bio: I am in my forth year as a PhD candidate in Dr. MacCoss Lab at University of Washington. My research interests are mainly focus on developing mass spectrometry methods for proteomics research and clinical applications.
Authorship: Han-Yin Yang(1), James G. Bollinger(1), Dao-Fu Dai(2), Andrew N. Hoofnagle(3), Christine C. Wu(4), Kelly D. Smith(2), Michael J. MacCoss(1)
(1) Department of Genome Sciences, (2) Department of Pathology, and (3) Department of Lab Medicine, University of Washington, Seattle, WA. (4) Stratus Biosciences, Seattle, WA
Laser-capture microdissection (LMD) used in conjunction with tandem mass spectrometry (MS/MS) has been a superior alternative to the previously applied immunohistochemistry assay. However, there are several aspects of current approach can be reconsidered. Here, we applied a systematic sampling MS/MS strategy to unbiased quantify amyloidosis peptides in different samples. We aim to find a subset of amyloid relate peptides that has best discriminate power for amyloidosis diagnosis and sub-typing using FFPE sample. Preliminary results show that we could successfully type and distinguish amyloidosis and normal cases base on the observed intensities of selected peptides.
Amyloidosis is group of diseases that are characterized by abnormal protein aggregation in various tissues. There are more than thirty structurally diverse proteins can form the beta-sheets observed in amyloid deposits. Different amyloidgenic proteins are represented in the various sub-types of amyloidosis pathology. Since each type of amyloidosis requires different treatment/intervention strategies, sub-typing needs to be performed after the initial diagnosis. Currently this is accomplished by observing significant amounts of the various amyloidgenic proteins in a formalin-fixed paraffin-embedded (FFPE) tissue biopsy. Laser-capture microdissection (LMD) used in conjunction with tandem mass spectrometry (MS/MS) has been successfully applied in the sub-typing of amyloidosis. Although the current LMD-MS/MS method is a superior alternative to the previously applied immunohistochemistry assay, there are several aspects of the approach that must be reconsidered in order to improve the reliability of the diagnostic platform. For instance, current platform quantifies proteins using the semi-random sampling procedure of data-dependent acquisition (DDA), which biases toward high abundant signal and introduces additional variance for diagnosis. In addition, peptides are the signal detected in MS/MS, different peptides from a single protein can behave very differently in MS/MS experiment. Perform quantification of protein level requires additional step of inferring proteins from identified peptides, which has been known a difficult problem in MS/MS experiment and could complicate diagnosis pipeline.
To address above mentioned issues, we quantify peptides with integrated chromatographic peak area from spectral data acquired using a systematic sampling strategy, data-independent acquisition (DIA). Many of amyloid related peptides can be observed in both amyloidosis and normal cases, but show different intensity distributions between amyloidosis and normal or between different amyloid types. Instead of aggregating peptide intensity to protein level, we aim to find a subset of amyloid related peptides that has that has best discriminate power for amyloidosis diagnosis and sub-typing using FFPE sample.
One hundred forty-three LMD samples were collected from seventy-one amyloidosis patients and normal cases. We extract proteins using a heat-mediated antigen retrieval method. 15N Apolipoprotein AI (ApoA1) was added as internal standard to correct for variance from digestion and instrument status. The resulting mixture was reduced, alkylated, and digested with trypsin for eighteen hours. We acquired spectral data with DDA and DIA method on a Q-Exactive HF mass spectrometer (Thermo) coupled to a nanoAcquity UPLC (Waters). Peptide spectral library is built based on peptide identification result from DDA dataset. With spectral library and DIA spectra data, we extract chromatographic peak area of peptides with Skyline software (MacCoss Lab). Extracted chromatographic peaks are further validated with mProphet algorithm.
With peptide quantification information and immunochemistry results of each case, we split all the LMD samples into training, validation, and testing datasets, in which training dataset is used to model peptide intensity distributions in amyloidosis and normal populations. We perform feature selection algorithms to find a combination of peptides that can yield best performance for identifying amyloidosis and sub-typing in validation dataset. We then estimated our model and typing accuracy with testing dataset.
To make sure the selected peptide from previous stage are not just reflecting the variance in experimental procedure and digestion, we perform time course digestion in FFPE samples and then only used the peptides reach digestion steady state in our protocol. In addition, post-digestion variation is also tested on selected peptides.
Several peptides from apolipoprotein E and serum amyloid P component were found to be the best indicators to confirm amyloidosis, which is consistent with previous studies. In addition, we found many peptides are enriched in specific amyloidosis types and useful for typing. In the validation dataset, most of cases are diagnosed and typed correctly base on the selected peptides. However, there are few cases could be assigned to multiple types and need to be discussed. Since current immunohistochemistry assay can only test one amyloidogenic protein at a time and stop testing when an abundant amyloidogenic protein is found, it could miss the possibility that multiple amyloidgenic proteins are involved in disease.
References & Acknowledgements:
IP Royalty: no
|Planning to mention or discuss specific products or technology of the company(ies) listed above:||