The Speicher laboratory is pursuing five major projects as well as additional collaborative projects in diverse biomedical research areas. Most projects utilize proteomics and mass spectrometry as major tools. These projects are briefly summarized below, and in greater detail in the following sections.
The first project uses proteomics, metabolomics and systems biology to identify pathways and networks implicated in tumor progression, and response to therapies as well as identify and characterize cancer biomarkers associated with ovarian cancer. This includes use of ovarian cancer cell lines, in vitro, short-term organ cultures of ovarian cancer tissue removed at time of surgery, targeted mass spectrometry (MS) based validation of biomarkers in patient sera, and development of higher throughput immunoMS quantitation methods.
The second project takes a proteomics-based systems biology approach to explore melanoma tumor progression and resistance to therapy, as well as the role of autophagy in these processes and in secretion of diagnostically useful protein biomarkers.
The third project project uses state-of-the-art proteomics and computational methods to identify plasma biomarkers of cardiotoxicity that is induced by cancer therapies in breast cancer patients.
The fourth project uses state-of-the-art proteomics approaches similar to Project 3 to identify plasma biomarkers that distinguish ectopic pregnancy from normal intrauterine pregnancy and non-viable intrauterine pregnancy.
The fifth project uses chemical crosslinking, high resolution MS and molecular modeling to determine structures of large protein complexes as well as to probe biologically associated protein dynamics including large changes in conformation and changes in protein-protein interactions.
One goal of this project is to discover and validate novel protein biomarkers of human ovarian cancer that can improve early detection of the disease as well as clinical management after diagnosis, and a second goal is to better understand tumor development and therapy resistance mechanisms in order to identify new therapeutic targets and companion diagnostics for this disease. Ovarian cancer is the leading cause of death from gynecological cancers in the United States. More than 24,000 women are diagnosed with ovarian cancer every year, and approximately 15,000 die of this disease annually in the USA alone. Ovarian cancers are biologically very heterogeneous with common epithelial tumors consisting of serous, endometrioid, mucinous, clear cell, and undifferentiated sub-types. The serous sub-type tumors account for about half of the epithelial tumors and generally occur in women between 40 and 60 years of age. They are usually highly aggressive and account for the greatest number of ovarian cancer-related deaths.
The five-year survival for invasive epithelial ovarian cancer is about 90% when the disease is confined to the ovaries, but is only about 30% when the cancer has spread to other organs. Even high-grade serous tumors do well if diagnosis occurs when the tumor is confined to the ovary. However, about 75% of ovarian cancers are not diagnosed until after the cancer has spread, primarily because early-stage tumors are generally asymptomatic.
There is no effective molecular test for ovarian cancer at present. So far, the best plasma biomarker that detects ovarian cancer prior to symptoms is CA125, but CA125 suffers from a large number of false positives and false negatives, which prevents it from being used as a routine screen for ovarian cancer. CA125 is commonly used to monitor progression of the disease after initial diagnosis but only about 60% of patients have tumors that are strongly CA125 positive. Hence, one goal is to identify new, better, molecular biomarkers for ovarian cancer that would form the basis for a minimally invasive blood test for early diagnosis as well as improved clinical management of the disease after initial diagnosis.
In pursuit of our first goal, we identified several hundred candidate ovarian cancer biomarkers using a xenograft model coupled with in-depth proteome analysis of the resulting chimeric (human/mouse) plasma to identify human proteins shed by the tumors into the blood. Multiplexed quantitative mass spectrometry assays using multiple reaction monitoring (MRM) were subsequently used for initial laboratory scale validation of approximately 40 high priority biomarkers. Approximately half of these biomarkers were significantly higher in ovarian cancer patients compared with normal donors or those with benign disease. Some of the most promising biomarkers were further tested using immunohistochemistry of tissue microarrays comprised of patient tumor or normal tissues, and a several proteins were identified that react strongly with all ovarian cancer subtypes and, importantly, complement the reactivity of CA125. The utility of these and other biomarkers to detect all cancer subtypes was confirmed by conducting proteome analyses of proteins shed by ovarian cancer tumors (tumor secretomes) using fresh surgical specimens. One current focus of this project is to finalize a panel of the most promising biomarkers to be tested using plasma from an independent patient cohort. Another part of this initiative is to develop high affinity antibodies for the target proteins and to develop a high throughput multiplexed immunoMS quantitative assay. The biological roles of several of these biomarkers on tumor progression and in response to therapy are also being evaluated.
A second major goal of our ovarian cancer project involves a proteomics and metabolomics based systems biology approach to the study of tumor progression and therapy resistance. These efforts have focused on clear cell ovarian cancer and the role of the ARID1A protein in this ovarian cancer subtype. ARID1A is mutated with loss of function in over 50% of clear cell ovarian cancers and across all cancers, it is the most frequently mutated epigenetic factor. Multiple proteome analyses of cells with and without functional ARID1A showed that the mevalonate and glycogen synthesis pathways were the most significantly affected cellular pathways in response to ARID1A status. Interesting, regulation of the mevalonate pathway occurs at the post-transcriptional level as mRNA levels for most enzymes in the pathway were not appreciably affected by ARID1A status, whereas most enzyme levels in the pathway changed significantly. Consistent with these studies, a recent Danish population study showed that blocking the mevalonate pathway by statin usage was a significant risk factor for development of clear cell ovarian cancer but not other ovarian cancer subtypes. One current focus of this project is to determine which metabolic pathways downstream from the mevalonate pathway affect growth and progression of clear cell ovarian tumors.
Malignant melanoma is one of the most aggressive forms of cancer, with a median survival of 6-10 months upon onset of metastasis, and it is one of the few cancers where the incidence in the general population continues to increase. Dramatic recent improvements in patient care have occurred using targeted therapies and immunotherapy. But for most patients, benefits are transient and hence long term patient survival rates remain low and additional therapeutic options are needed.
Several related aspects of melanoma progression and therapy resistance are being pursued. Systems biology analyses of protein profile changes associated with tumor progression and acquisition of therapy resistance are being used to gain novel insights into key mechanisms associated with metastatic and therapy resistant phenotypes. A major recent focus has been the role of autophagy, which is a common mechanism of therapy resistance that also correlates with melanoma aggressiveness. Clinical trials are currently underway involving combinations of anticancer therapies and hydroxychloroquine, an autophagy inhibitor. However, these efforts are limited by the lack of predictive biomarkers to select patients most likely to respond to autophagy targeting therapies and the lack of markers to monitor effects on autophagy levels during treatment using minimally invasive methods.
Current goals include identification of predictive and pharmacodynamic plasma biomarkers to assist in clinical management of patients receiving autophagy inhibitors and identification of new therapeutic targets. Proteome analyses of supernatants from 3D cultures of melanoma cells with high and low autophagy have led to the identification of a number of promising candidate biomarkers. Several of these biomarkers have been validated in additional cell lines and in a small group of patients where the biomarkers appear to correlate with tumor autophagy levels. Future efforts will expand the evaluation of candidate biomarkers in patients as well as pursue an unbiased proteomics approach to identify global changes in posttranslational modifications and protein levels related to autophagy and autophagy.
Millions of breast cancer survivors are at risk of developing cardiotoxicity resulting from therapeutic treatments, and biomarkers for predicting cardiotoxicity are urgently needed for these patients. Trastuzumab (Herceptin®) is used widely to treat HER2+ breast cancer, which has resulted in important survival gains, but treatment involves significant risk of cardiovascular morbidity and mortality. When it is used in combination with doxorubicin, it results in left ventricular (LV) dysfunction in 18% of treated individuals and severe, symptomatic heart failure (HF) in 2-4% of patients. Hence, there is a critical need for minimally invasive biomarkers for early identification of subclinical cardiac dysfunction that would: 1) enable earlier patient treatment with cardioprotective strategies; 2) prevent interruption of necessary cancer therapy; and 3) reduce morbidity and mortality. Use of either drug alone, also has substantial risk of inducing cardiotoxicity.
We recently conducted several in-depth profiling comparisons of human plasma proteomes using longitudinally collected plasma samples from case-control patients to discover candidate biomarkers that identify breast cancer patients at increased risk for therapy-induced cardiotoxicity prior to appearance of clinical symptoms. We anticipated that most biomarkers would be diagnostic, that is, the level of the biomarker in the blood of patients would either increase or decrease over the time period when the patient is developing cardiac dysfunction. Surprising, the most promising biomarkers are actually predictive, that is, prior to beginning therapy, the biomarkers could stratify patients into high and low risk groups. The most promising candidate biomarkers are being validated in pre-treatment plasma from additional patients as well as patients receiving related therapy treatment regimens. A working hypothesis is that a multi-protein panel of circulating biomarkers can identify patients with cancer therapy-induced cardiotoxicity earlier than conventional clinical methods. When available, ELISA assays are used for biomarker quantitation. When ELISA assays are not available, multiplexed MRM-MS assays are developed and used.
Ectopic Pregnancy (EP) occurs in about 1-2% of pregnant women and may compromise a woman’s health and future fertility. It is a leading cause of maternal mortality and morbidity accounting for 6% of pregnancy deaths due to a rupture of the fallopian tube with resulting intraperitoneal bleeding. Most patients present before tube rupture with nonspecific symptoms of abdominal pain and/or vaginal bleeding but these symptoms are neither sufficiently sensitive nor specific, and some women remain asymptomatic. If diagnosed early, EP can be effectively treated with little risk to the patient but current diagnostic methods, transvaginal ultrasound and serial quantitative serum human chorionic gonadotropin concentrations are inconclusive in up to 40% of patients. Also, the diagnosis is currently cumbersome requiring multiple office visits, serial blood tests for up to 6 weeks, multiple ultrasound examinations, and surgical procedures such as uterine curettage and laparoscopy.
The goal of this project is to develop more effective, minimally invasive blood tests that can reliably distinguish EP at an early stage from nonviable intrauterine pregnancy (spontaneous abortion or SAB) or from normal intrauterine pregnancy (IUP). The proteomic strategies are similar to those described in project 3. In-depth quantitative comparisons of plasma from patients with EP, SAB and IUP were used to identify a number of novel biomarkers and several specific isoforms of previously known biomarker protein families. These results also confirmed several previously known biomarkers. Followup validation of the most promising biomarkers are being pursued in an independent patient cohort to identify a panel of biomarkers with clinical utility. Because robust sandwich ELISA are not available for most of the new biomarker candidates, a targeted quantitative MS method will be utilized to multiplex biomarker validation. An advantage of the targeted MS approach is that it can reliably distinguish between closely related isoforms that may be present in the plasma.
Future efforts will be directed to producing multiple highly specific, high-affinity monoclonal antibodies to each of the best biomarkers based on the targeted MS results. These antibodies will be used to develop a multiplexed biomarker ELISA for clinical studies because the expectation is that an optimal clinical diagnostic assay will consist of a panel of multiple biomarkers rather than a single protein.
Protein-protein interactions and protein conformational changes are key drivers of most biological processes. Detailed understanding of these processes can be greatly aided by structures of the responsible protein complexes and conformational changes associated with protein function. Crystallography and NMR can provide very high resolution structures, and indeed, representative structures of most protein folds have been determined using these powerful techniques. However, not all proteins can be crystallized and the difficulty of obtaining crystal structures increases rapidly with protein size. Structural mass spectrometry (MS) methods such as chemical crosslinking coupled with MS (CX-MS) are highly complementary to crystallography and NMR. Identification of inter-chain crosslinked peptides can define previously unknown protein-protein interactions and distance constraints defined by intra-chain as well as inter-chain can be used to determine protein structures using structural modeling programs.
We developed new methods for identifying zero-length crosslinks because they provide the most precise distance constraints and are therefore the best crosslinks for defining direct protein-protein interactions and for verifying and refining structural models. The method utilizes label-free LC-MS/MS comparisons of control and crosslinked samples as well as software we developed specifically for zero-length crosslinks. We used this method to determine conformational changes in an antioxidant enzyme, Prdx6, that were associated with its enzyme mechanism and large conformational changes involved in the interconversion of spectrin dimers and tetramers. Spectrin is an essential 1 million Dalton actin-bridging tetramer in the red cell membrane skeleton protein. Structures of the head region of wild type spectrin dimers and tetramers as well as dimers containing mutations responsible for hereditary hemolytic anemias were also determined using this approach. We subsequently improved the method and began applying it to intact red cell membranes. These data were used to determine the structure of full length anion channel dimers as well as map interactions with ankryin, a protein that both stabilizes anion exchanger tetramers and links spectrin to the lipid bilayer. Mapping of other protein-protein interactions showed that most interaction regions defined by biochemical studies using protein fragments severely under-estimate the interaction interface.
One future goal is to further increase the density of detected high confidence crosslinked peptides by improving the chemistry, optimizing the MS data acquisition method, and refining the software and data scoring. This optimized method will be applied to further elucidation of structures of the red cell membrane including obtaining an experimentally verified high resolution structure of the anion exchanger tetramer: ankyrin complex and a structure of the full length spectrin tetramer. Another goal is to apply this method to chromatin remodeling complexes such as the SWI/SNF complex and changes associated with ARID1A mutations in ovarian cancer and melanoma.