Autophagy is a process in which cells recycle their damaged or unwanted organelles. The process of macroautophagy encompasses a variety of selective autophagy mechanisms such as ribophagy, mitophagy, and lipophagy - each of which has a different specific target for degradation. These processes are initiated for multiple reasons, incuding ATP starvation and stem cell differentiation. During erythropoiesis, the process of hematopoietic stem cells differentiating into red blood cells, selective autophagy is known to be used to clear mitochondria in the terminal stages. Doulatov lab has recently identified the selective expression of BNIP3, a homolog of the known mitophagy regulator BNIP3 (NIX), in early erythroid progenitors. These preliminary data suggest that there may be an unnappreciated role for mitophagy in the earlier stages of erythropoiesis in addition to its late stage role. To better understand how mitophagy is involved at this time point, the level of expression of proteins BNIP3, BNIP3L (NIX), and another known mitophagy protein PINK1 will be evaluated at days 1, 3, 5, 7, 9, 11, and 15 of erythropoiesis by western blot. The expression of these proteins will be analyzed in conjunction with mitochondiral mass of the cells, which has been observed to rapidly decline following a sudden burst in mass at day three of erythropoiesis. We hypothesize that autophagy is responsible for this dynamic behavior, and investigating the expression of these proteins may implicate which specific mitophagy pathway is being used. After determining these expression patterns, we can use lentiviruses expressing shRNAs to seperately reduce and overexpress autophagy proteins in order to manipulate mitochondrial mass during erythropoiesis. Exploring the relationship between selective autophagy and mitochondrial dynamics during normal human erythropoiesis may eventually provide insight into pathologies in which erythropoiesis is impaired.

Cystic fibrosis-related diabetes (CFRD) complicates up to 50% of adult cases with CF. However, its pathogenesis is poorly understood. One thing that has been shown, however, is that amyloid deposition occurs in the islets of 60% of subjects with CFRD, which is known, from data in other forms of diabetes, to be toxic to insulin-producing islet β-cells. Currently existing animal models do not reproduce this feature of human CFRD, since islet amyloid deposition requires the expression of an amyloidogenic form of the islet peptide islet amyloid polypeptide (IAPP). To address this gap in the field, our group is generating a CF mouse model that expresses the human amyloidogenic form of islet amyloid polypeptide (hIAPP). To do this, we have generated a double transgenic mouse from a male homozygous for the CF mutation and a female hemizygote for the hIAPP transgene. Mice will be followed for up to 12 months, or until hyperglycemia develops. During this time, body weights and non-fasted blood glucose levels will be measured monthly. When hyperglycemia is identified, animals will be euthanized in order to determine if islet amyloid is present in the pancreas. We are expecting to find that mice bearing the CF mutation and expressing the hIAPP transgene will develop increased and/or accelerated amyloid deposition in their islets as compared to the control mice generated by the cross. In addition to this, we expect that these mice will also show increased glucose levels (hyperglycemia) and decreased β-cell mass. There are currently 16 mice on study, and data collection is ongoing, with additional animals being enrolled approximately every two weeks. The goal of this study is to generate a new model of CFRD, which better reflects the human disease state, and can be used to develop and test improved therapies for human CFRD.

Alzheimer’s disease (AD) is a debilitating neurodegenerative disease that currently affects >5 million individuals. We have previously demonstrated genetic and neuropathology-based variation across cognitively-defined AD subgroups. Here we sought to determine whether patterns of vascular risk factors differed across cognitively-defined subgroups. We used data from the Adult Changes in Thought (ACT) Study and the Religious Orders Study/Memory and Aging Project (ROS/MAP). These are prospective cohort studies of individuals who are dementia-free at enrollment. We used cognitive data at the time of AD diagnosis to determine scores for memory, visuospatial abilities, language, and executive functioning. We used these scores to determine each individual’s average cognition and then domain-specific deficits below that individual average. We used these data to define six groups: no domain with a substantial relative impairment; isolated substantial relative impairment in memory, in visuospatial functioning, in language, or in executive functioning, and multiple domains with substantial relative impairments. We evaluated risk factors for these six groups from self-report and/or diagnosis data on vascular conditions, including diabetes, stroke, hypertension, atrial fibrillation, cholesterol levels, and coronary artery disease. We used multinomial logistic regression models with the no prominent domain group as the reference. To account for multiple comparisons, we present tests of the null that each risk factor is unrelated to each subgroup, and an omnibus test of the association between each risk factor and any subgroup. The proportion of individuals in each subgroup differ between the two studies. Our results suggest possible differences in risk for specific cognitive domain deficits associated with stroke history, HDL cholesterol, and treated diabetes. Subsequent work will attempt to understand possible mechanisms behind these associations through literature review. These results support additional efforts to further understand cognitively-defined AD subgroups.

QRS-like artifacts are present in ECG recordings during CPR but have not been described previously. We coin these artifacts compression induced complexes (CICs). They prevent accurate real-time cardiac rhythm characterization during CPR compressions. Additionally, we hypothesize that CICs may be an early prognostic tool for return of rhythm (ROR), return of spontaneous circulation (ROSC) and an indicator of cardiac health. The objective is to characterize and describe these QRS-like complexes quantitatively and create an algorithm that distinguishes CICs from native cardiac activity. CIC characteristics that are being measured are a) amplitude, b) sharpness, and c) temporal relationship with CPR compressions. Custom software was used to analyze the ECG data from King County Emergency Services with CICs annotated by expert review. 50 cases were used to determine the range of features specific to CICs, to include: a) amplitude as measured at CIC peaks, b) sharpness as quantified by measuring the peak of the second derivate at CIC locations and wavelet transform analysis to measure the range of frequencies present in CICs, c) the temporal relationship between CICs and CPR compressions as measured using peak finding algorithms and comparing the delay between the impedance peaks from CPR compressions and CIC peaks. Ten CICs from each case for each feature were analyzed using standard deviations to define quantitative parameters for a CIC detection algorithm. After development, this algorithm was tested to determine the percent of CIC complexes that are detected in a test set. It was also used to examine the association of CICs with ROR and ROSC. This algorithm can potentially aid in advising pre-hospital care tailored to the patient’s cardiac condition.

Antibiotic resistance, drug cytotoxicity, and lengthy treatment regimens are major barriers to ending the tuberculosis epidemic. Host-directed therapies (HDTs) that avoid direct targeting of Mycobacterium tuberculosis (Mtb) while enhancing macrophage clearance of the bacteria may overcome such challenges. Because host protein kinases (PKs) regulate multiple innate immune signaling pathways involved with Mtb clearance, we hypothesized that host PKs are HDT candidates. We investigated a group of ATP analogues with specific activity against parasite serine/threonine protein kinases called bumped kinase inhibitors (BKIs). BKIs inhibit few mammalian PKs, so they offer HDT potential while avoiding broad activity that could result in toxicity. To assess the effect of BKIs on Mtb growth within monocyte-derived macrophages (MDMs), MDMs from healthy donors were infected with a luminescent Mtb reporter strain (Mtb-lux) in the presence or absence of BKIs and daily luminescence readings were recorded. We found three BKIs that limit Mtb intracellular replication in MDMs at a final reaction concentration of 10-20 µM. Interestingly, none of these compounds inhibited Mtb-lux growth in 7H9 broth culture. To examine possible mechanisms, we measured the effect of the BKIs on pro-inflammatory cytokine release in MDMs in response to various stimuli including Toll-like receptor agonists, virulent Mtb, and the attenuated vaccine strain Mycobacterium bovis BCG. In response to all of these stimuli, the BKIs potently inhibited TNF secretion and moderately inhibited IL6 secretion by ELISA. In contrast, the mRNA expression level of antimicrobial peptide cathelicidin was maintained in the presence of the BKI tested. Since pro-inflammatory cytokine induction and cathelicidin expression are generally host-protective in early Mtb infection, it is unlikely that the effect of the BKIs on these pathways are responsible for inhibiting intracellular Mtb growth. We are currently evaluating the effect of BKI treatment on other antimicrobial pathways that restrict Mtb infection, including phagosomal uptake and autophagy.

Kidneys are responsible for regulating blood glucose levels in the body through the reabsorption of sugars in the proximal tubule of the nephron. Issues with glucose absorption can lead to diabetic nephropathies where too much glucose is excreted. To study kidney glucose absorption, the lab made kidney organoids, which are miniature stem cell-derived in vitro organs that mimic the structure and function of kidneys. We used induced pluripotent stem cells (iPSCs), a type of stem cell that can be generated directly from one’s own adult stem cells. In a previous experiment, the lab found glucose absorption was present through the Sodium Glucose Co-Transporter 2 (SGLT2). The lab wanted to see if this was the only channel responsible for glucose absorption. Using the Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) system, we knocked out SGLT1 and SGLT2 separately and also knocked out Glucose Transporters (GLUT1 and GLUT2) separately in the WTC11 kidney stem cell line.  We selected a guide RNA corresponding with each glucose transporter in the cell membrane of kidney organoids.  Then RNA-guided Cas9 created the mutation by breaking the sequence for each glucose transporter in a specific site of the genome.  We verified our mutations were done correctly by sending our results to a DNA sequencing company.  We expect to find mutant lines will have less glucose absorption than our control and hope to learn which transporters are more active.  We are further using this method to create double and quadruple knockout lines with multiple glucose transporters knocked out at once.  We hope to learn which transporters are most responsible for glucose absorption, and in the future, can assess when certain transporters are used over others, which will open the doors for disease models for diabetes not only in kidneys but also in other organs.

Treatment options for kidney disease are limited to dialysis and transplantation, which are of limited efficacy and availability. Human pluripotent stem cells (hPSCs) can be used to generate immunocompatible kidney organoids, which contain the major proximal structures of the nephron, including podocytes, proximal tubules, distal tubules, and endothelial cells. They can be made on-demand to study kidney disease and regeneration. Unfortunately, these organoids lack a collecting duct system, a crucial component of the kidney. To address this need, we have developed a protocol to differentiate hPSCs into ureteric bud (UB) cells, which are the precursors of collecting duct cells. First, we identified markers for the collecting duct lineage in developing kidneys. Immunofluorescence analysis of human kidney tissue revealed co-expression of cytokeratin 8 and Dolichos biflorus agglutinin (DBA) in the collecting ducts. These markers were not expressed in the proximal tubules. Next, we investigated the ability of hPSCs to express these markers. Undifferentiated hPSCs were treated with a concentration gradient of small molecules known to induce kidney lineage differentiation, including CHIR99021 and glial cell line-derived neurotropic factor (GDNF); their effects were jointly and individually examined. Immunofluorescence was used to characterize the resulting cells, which were initially stained with DBA and Lotus tetragonolobus lectin (LTL), a proximal tubule marker. We found that CHIR99021 alone was sufficient to produce DBA positive and LTL negative cells in vitro. Notably, the DBA levels in these cells were lower than observed in kidney tissue, indicating that the structures we obtained were not yet fully mature. These studies provide the first evidence that collecting duct cells can be generated from hPSCs. Optimization of this differentiation protocol will yield kidney organoid structures that include the collecting duct system, which will make them ideal for kidney disease modeling and regenerative medicine approaches to reduce the need for kidney transplants.

Cystic Fibrosis (CF) is a genetic disorder that is characterized by recurrent pulmonary infections and a progressive decline in lung function caused by increased inflammation. While the host immune response can help fight these infections, an overzealous inflammatory response in the lungs may be harmful. Inflammation can be induced by multimeric proteins called inflammasomes, which catalyze the maturation of pro-inflammatory cytokines and pyroptosis. Our laboratory is interested in investigating genetic variants that alter the response of the NLRP3 inflammasome in human CF patients as a means of identifying potential targets for host-directed therapies. Preliminary data from the EPIC study has identified two genetic variants in CF children that is associated with colonization by P. aeruginosa, a common bacteria in the lung that can induce deleterious inflammation. We hypothesize that genetic variants in the NLRP3 inflammasome will increase the inflammatory response to bacterial infection and will lead to worsened pulmonary outcomes in patients with CF. First, we used CRISPR/Cas9 and gene editing techniques to knock out the NLRP3 gene in human cell lines. We then inserted NLRP3 genetic variants and stimulate the NLRP3 inflammasome using Pathogen associated molecular patterns such as Nigericin and ATP. The extent of inflammation was measured using ELISA (Enzyme Linked Immunosorbent Assay), and results were compared among NLRP3 genetic variants and wild type controls. The ultimate goal of this investigation is to learn how genetic variants are associated with the clinical outcomes of CF patients. Future studies can involve harnessing host-directed therapies to target specific inflammatory pathways in order to reduce harmful inflammation in CF. Finding the relationship between the NLRP3 inflammasome and CF is critical to generating an effective treatment for CF.

Primary cilia are antenna-like structures on the surface of diverse cell types. Cilia are important for signal transduction and processes such as cell division and development, but their precise role in the cell remains poorly understood. Defects in cilia result in a wide spectrum of genetic disorders called ciliopathies, which commonly involve polycystic kidney disease (PKD), the formation of fluid-filled sacs in the kidney which eventually result in kidney failure. To understand the role of cilia in these diseases, we generated human pluripotent stem cells (hPSCs) with defects in cilia formation for the first time. We used the CRISPR-Cas9 genome editing system to introduce mutations in two different genes that are required for cilia formation, kinesin family member 3A (KIF3A) and kinesin family member 3B (KIF3B). We isolated four independent cells lines with indel mutations in exon 3 of the KIF3A gene and two in exon 2 of KIF3B. The introduced mutations led to frameshifts and protein truncations. Immunoblot analysis confirmed that no full-length protein was produced in these mutant cells. The cells were stained with acetylated alpha-tubulin, a marker of cilia. No cilia were detected in the mutant hPSCs, while an average of 48% cilia were detected in controls of identical genetic background. Surprisingly, the absence of cilia did not alter hPSC pluripotency, self-renewal, amniotic cavity formation, or growth. Further, we differentiated these cell lines into kidney organoids. The cilia deficient kidney organoids formed cysts reminiscent of PKD, while the isogenic controls did not. In conclusion, we have generated human cells and organoids lacking an entire organelle, the primary cilium, and used these to conclusively link cilia to PKD. These cilia deficient hPSC lines can be used to study cilia function, to model various ciliopathies, and to screen for drugs that can ameliorate the disease phenotypes associated with them.

Impella devices provide temporary mechanical circulatory support (MCS) by means of a catheter consisting of a left ventricular (LV) inlet, a microaxial pump within the catheter, and an outflow port in the proximal aorta. The recommended position of the inlet is 4-4.5cm below the aortic valve annulus. Following initial placement, the device is prone to migration, potentially leading to ineffective support, hemolysis, ventricular arrhythmia, or mitral regurgitation. Positioning is typically monitored by echocardiography. However, the Impella device has a bend between the inflow and outflow ports, and traditional 2D-imaging may not be reliable in assessing the location of the device, as single plane 2D windows may not include both the inflow port and the aortic annulus, leading to foreshortening of the cannula and under-measurement of the distance. We hypothesized that 3D echocardiography provides more accurate characterization of the Impella cannula location. We analyzed 25 echocardiograms of patients with Impella devices placed in LV which had analyzable 2D and 3D images. Measurements of the distance from the aortic annulus to the inflow port were made according to recommendations from established measurement guidelines. Three sets of measurements were made at end systole and at end diastole. Full volume 3D datasets acquired from the parasternal window were acquired an analyzed using Qlab software to overcome possible foreshortening of the Impella cannula. In addition, the 3D dataset was used to measure the angle deviation of the cannula from the long axis of the left ventricular outflow tract. There was no recognizable benefit of 3D techniques in measuring aortic annulus to Impella inflow port distance. A larger sample size may be necessary to detect a significant difference. 3D imaging may have a benefit in observing other relationships, such as the mitral apparatus, and bears ongoing investigation.

Our project is on urothelial urinary bladder cancer, one of the more common types of cancers. Our lab focuses on protein synthesis control of ARID1A, a known tumor suppressor in the context of urothelial urinary bladder cancer. Our initial workup led us to believe that ARID1A loss is not important for tumor initiation but may be important for tumor progression. To prove this, we used CRE recombinase editing tool to knock out ARID1A after treatment with BBN, a known carcinogen that causes bladder cancer. The results confirmed our hypothesis. In our investigation of this mouse model, we discovered an unexpected relationship between ARID1A loss and lowered protein synthesis. There is a growing body of literature showing that high levels of protein synthesis are required for transformation. As such, we conducted an experiment with rpL24+/- mouse model to test this hypothesis. We found that rpL24+/- mice showed a significant delay in cancer initiation compared to wild type mice when treated with BBN, thus establishing the role of protein synthesis in cell transformation. Our focus now is to figure out the relationship between ARID1A loss and protein synthesis. Based on preliminary results that we have conducted, I hypothesize that ARID1A loss and decreased protein synthesis leads to a weakened cell state which makes them vulnerable to carcinogen induced DNA damage and cell death. This was shown by measuring DNA damage, apoptosis, and cell proliferation as cellular markers of cancer and then comparing them with protein synthesis levels in ARID1A knock out (KO) cells and WT cells upon short treatments with BBN. We used both a mouse model and in vitro model to address this hypothesis.

Atherosclerosis is a disease in which cholesterol accumulates in the walls of arteries, limits blood flow, and causes heart attacks and strokes. Atherosclerosis could be prevented or reversed by expressing therapeutic genes in the artery wall. Gene therapy for atherosclerosis will require high-level therapeutic gene expression for two reasons: 1) to increase efficacy; and 2) to allow use of lower vector doses, thereby minimizing vector-related toxicity. Our goal is to increase expression of an apolipoprotein A-I (apoA-I) transgene, which both prevents and reverses atherosclerosis in rabbits. We used an in silico bioinformatics approach to identify endothelial cell-specific cis-regulatory modules (CRMs). CRMs are short enhancer regions of DNA that are expected – when introduced into endothelial cells – to increase transcriptional activity of a nearby gene. We identified 11 CRMs in the CDH5, EFEMP1, THBS1, and VWF genes. We hypothesize that insertion of these CRMs into a helper-dependent adenoviral vector encoding apoA-I will increase apoA-I expression in endothelial cells. We also hypothesize that the genomic regions surrounding these CRMs contain sequences that, along with the CRMs, could increase transgene expression above levels obtained with the CRMs alone. Our adenoviral vectors have a larger capacity for the insertion of DNA sequences than other gene-therapy vectors. This large capacity allows us to evaluate CRMs in their native genomic context (i.e., with several kilobases of neighboring DNA included). We cloned large CRM-containing regions of the 4 genes listed above and we are inserting the apoA-I gene into the translational start sites of these genes. We will test vectors containing individual CRMs and vectors containing the larger CRM-containing regions in cultured endothelial cells (in vitro) and in rabbit carotid arteries (in vivo) to identify CRM-containing constructs that express the highest apoA-I levels. This project may provide higher-expressing vectors for use in gene therapy for atherosclerosis.

Microglia are the resident immune cells of the CNS and are hypothesized to influence aging in the brain. Like somatic cells, microglia can be replaced by self-renewal. Recently, some studies have suggested that new microglia derive from asymmetric cell division of a progenitor population. Microglia progenitor cells have been difficult to study due to a lack of specific molecular markers of this population. However, the Garden lab has recently identified novel candidate markers. We hypothesize that in neurodegenerative disorders associated with advanced age, microglia progenitor senescence may contribute to disease pathology. To efficiently study the senescence of microglia progenitors, we turned to neonatal mixed glia cultures, in which the presence of microglia progenitors has long been inferred. In these cultures, microglia are harvested from cells floating above a monolayer culture of mixed neonatal glial cells. The size of each microglia harvest generally decreases with successive harvests. This suggests that microglia progenitors in the attached monolayer may become senescent after multiple rounds of the cell cycle, leading to stagnation in the generation of new floating microglia. We evaluated microglia progenitor senescence in neonatal mixed-glia cultures by labeling with BrdU, a thymidine analog taken up by proliferating cells and remaining in their daughters. Microglia harvested from these cultures weekly were assessed for BrdU incorporation using flow cytometry and immunofluorescent microscopy. We co-labeled floating microglia and dissociated monolayer mixed glia cultures with antibodies directed against a progenitor marker (CD133), a microglia marker (Iba1), and BrdU. The attached mixed-glia cell layer was also labeled for SA-ß-Gal, an indicator of cellular senescence. Progenitor senescence will be detected by a decrease in CD133/BrdU-positive cells and an increase in CD133/ SA-ß-Gal positive cells.

Autosomal dominant polycystic kidney disease (ADPKD) is a genetic disorder in which fluid-filled cysts form in the kidneys and other organs. Mutations in the PKD1 and PKD2 genes, encoding the polycystin-1 (PC1) and polycystin-2 (PC2) proteins respectively, result in ADPKD. The underlying mechanisms that connect these mutations to ADPKD pathogenesis are poorly understood. Therefore, there is a great need to understand the mechanism of this disease and find an effective treatment. Our laboratory has developed an ADPKD cellular model by differentiating human pluripotent stem cells (hPSC) with mutations in PKD1 or PKD2 into kidney organoids. We are using this model to screen drugs that potentially induce or inhibit cyst formation in a dish. The cyclic AMP signaling pathway is hypothesized to contribute to cystogenesis in PKD. To test this, PKD mutant hPSC lines and isogenic controls were differentiated into kidney organoids and treated with forskolin, a strong cAMP agonist. Images of live organoids in adherent cultures were acquired in intervals and quantified for cyst formation using the cell counter feature of ImageJ. Organoids were subsequently fixed, stained, and imaged with fluorescent microscopy. Forskolin, induced a rapid and dose-dependent swelling in kidney organoids derived from both PKD hPSCs and isogenic controls. After removal of the drug, the tubules returned to its original form. These studies show that chemical modifiers of cystogenesis can be successfully tested using this system. Surprisingly, cyclic AMP does not appear to have effects that are specific to PKD. We are currently testing other drugs and culture conditions for cyst-inducing or –inhibiting effects. Interestingly, PKD kidney organoids form cysts in suspension cultures at significant higher rates than adherent cultures, revealing a role for the microenvironment in cystogenesis. These findings help us gain insight into the disease pathophysiology, which will ultimately guide us to discoveries of better treatments for PKD.

The majority of adults have acquired cytomegalovirus (CMV), a human herpes virus transmitted by close personal contact, during life. Primary infection is usually asymptomatic in otherwise healthy persons, and is followed by life-long latent infection. Immunosuppression (from conditions such as HIV or transplant) predisposes to severe and symptomatic disease resulting either from primary infection or reactivation of latent infection, and CMV infection is a major cause of morbidity and mortality in organ transplant recipients. Antiviral agents are commonly used for prevention/suppression of CMV in transplant patients, but their use is limited by toxicities and emergence of resistance. Despite the use of antiviral agents as prophylaxis, CMV disease continues to occur, either as breakthrough infection in patients receiving prophylaxis, or after prophylaxis has been discontinued. The incidence, timing, clinical characteristics, and risk factors for CMV disease in heart transplant patients receiving antiviral prophylaxis has not been well-characterized. The goals of the present study are to retrospectively characterize the incidence, timing, clinical characteristics, and risk factors for CMV disease in a large cohort of heart transplant recipients at the University of Washington Transplant Program. We will use descriptive analyses for incidence and timing of CMV disease and multivariate logistic regression and Cox-proportional hazard models to assess risk factors. The cohort consists of 333 consecutive orthotopic heart transplant recipients between 1 July 2005 through 30 September 2016, with at least 1 year of follow-up. Overall incidence of Cytomegalovirus disease as well as by Donor+/Recipient- and Recipient+ serologies is respectively 10.5%, 27.9%, and 5.2% at a median of 10 months post transplant. 68.6% of patients who acquired disease had D+/R- serology, and the most common manifestations were Syndrome, Colitis, Pneumonia and Gastritis. Analysis of optimum prophylactic course length and associations between Cytomegalovirus and Cardiac Allograft Vasculopathy are in progress.

Alzheimer's disease (AD) is a neurodegenerative illness responsible for 60-80% of dementia cases in the United States. Studies have shown that the formation of two primary pathologies, amyloid plaques and tau tangles, are present prior to neuronal death that results in this loss of memory and worsened motor function. Our research focuses on understanding the role of tau in disease progression. Tau stabilizes microtubules in the cell, which function to maintain cell structure and assist in intracellular transport. To successfully regulate microtubules, tau is modulated by site-specific phosphorylation. Evidence points specifically to the hyperphosphorylation of tau by certain kinases in playing a key role in neurodegenerative dieases such as AD. One kinase that has been identified as a tau-phosphorylating agent is tau-tubulin kinase 2 (TTBK2). To understand the role of TTBK2 kinase activity in the context of tau, we created double transgenic C. elegans lines expressing both human tau and TTBK2. We assessed behavioral abnormalities in our subsequent populations using locomotor assays of the homozygous crosses. From the data that was generated, we observed a significant difference in the movement of the double transgenic tau/TTBK2 strains as compared to their controls. The crosses moved substantially less than the other strains, and exhibited a more uncoordinated phenotype. These results support the hypothesis that the abnormal phosphorylation of tau by TTBK2 results in worsened motor control and general health of affected individuals. Additional testing of protein levels within the double transgenic lines will enable us to determine the mechanisms underlying this effect, and could eventually lead to information that might inform treatments counteracting the activity of kinases such as TTBK2 involved in hyperphosphorylation of tau proteins. Furthermore, this research has the potential to aid in finding a cure for a currently incurable disease and providing a hopeful future for Alzheimer's-affected individuals.

Heart failure (HF) has a myriad of negative effects in humans, including fatigue, weakness and reduced ability to exercise. Understanding the mechanisms that lead to this weakened state is critical in order to know how to treat it. There is recent suggestion that HF leads to decreased skeletal muscle force generation and exercise capacity. However, the mechanism responsible for this process is unknown and there are no current pharmacological treatments that target the skeletal muscle. My project aims to use a mouse model of HF called D230N, which have a mutation in the sarcomeric gene tropomyosin that results in progressive systolic dysfunction similar to what is seen in human HF. I will use the D230N mice to test the hypothesis that progressive cardiac dysfunction in a rodent model of HF leads to decreased exercise capacity, reduced skeletal muscle function and structural changes in the skeletal muscle that mimics the changes seen in human HF. I have taken preliminary measurements of 9 month old D230 and WT mice. I used a treadmill apparatus to measure exercise capacity of the mice, which showed that the D230N exercised for a shorter duration compared to WT mice. To measure the maximum force generated in the gastrocnemius muscle I used in vivo muscle stimulations and normalized to the muscle weight. I observed that the D230N generated lower muscle force than WT mice. Based on these results, I propose a longitudinal study to determine how skeletal muscle dysfunction changes as the disease progresses by taking these measurements in mice ranging from 2-12 months. Once we establish that there is skeletal muscle dysfunction in this rodent model of HF, we will move towards more in depth studies to fully understand the underlying mechanism.

Store-and-forward teldermatology (SFT) has grown in use nationwide to increase accessibility of dermatological care especially towards underserved populations. It has demonstrated its cost-effective services by minimizing avoidable referrals to dermatologists and granting quicker overall access to dermatologists for initial interventions. However, due to transportation and other logistics, there are concerns that SFT may have significantly longer delays and lengths of treatment of malignant melanoma than face-to-face (FTF) dermatology. Melanoma is the most lethal type of skin cancer which makes it imperative to seek intervention as soon as possible. We conducted a descriptive comparison of melanoma discovery and timeline for treatment among veterans whose melanomas were discovered and treated through face-to-face and store-and-forward teledermatology. We tested the hypothesis that SFTD is non-inferior to FTF dermatology in discovery and treatment of primary cutaneous melanoma. We ascertained chart data for Veterans diagnosed with melanoma between January 1, 2006 through December 31, 2013 from VA corporate data warehouse (CDW), Veterans Integrated Service Network 20 (VISN 20) and and Computerized Patient Record System (CPRS). To determine possible opportunities to discover a melanoma, and lengths of each episode of care, we retrospectively analyzed 300 cases of melanomas 365 days prior to the date of their biopsy to the date of its definitive procedure. The mean days between the earliest opportunity to discover and discovery, discovery to biopsy, and biopsy to definitive procedure was 90, 39, and 41 days respectively for melanoma cases seen by FTF dermatology and 44, 50, 40 days respectively for melanoma cases seen by SFTD. With no biological significance between the timelines of care, SFTD is non-inferior to FTF dermatology in timeliness to treat melanoma. 

Oncolytic adenoviruses such as serotype 5 adenoviruses (Ad5) selectively infect and destroy cancer cells and are widely used in preclinical and clinical trials for gene therapy. However, Ad5 has many downsides including preexisting anti-Ad5 immunity in humans, and the Ad5 receptor (CAR) is known to be down regulated in many aggressive tumors. Species B adenoviruses which use a different receptor than CAR are an alternative which shows great promise. These vectors avoid the immune response against Ad5 and show different cell tropism. Our lab has identified desmoglein2 (DSG2) as the main receptor for a group of species B adenoviruses, including Ad3. DSG2 is an epithelial junction protein and is highly expressed in malignant tumors. During Ad3 infection, binding of Ad3 fiber to DSG2 triggers the transient opening of these intercellular junctions and thus supports the lateral spread of the progeny virus. Our lab has recently developed an Ad3 fiber knob mutant (Y250F) which has an enhanced affinity to DSG2. We have shown that an increased affinity is correlated with a stronger ability to open these junctions. In this study, we developed an Ad3 based oncolytic vector containing fibers with enhanced affinity to DSG2 (> 400x higher than wild type Ad3.) We developed this vector using PCR site-directed mutagenesis and cloning the recombinant gene in E. coli. By inserting our gene into a Ad3 plasmid it allowed us to transfect the plasmid into human embryonic kidney cells (HEK 293 cells) and to amplify and recover the mutant oncolytic virus. We are working on in vitro experiments and in vivo tumor xenograft models to test our hypothesis that the oncolytic virus with an enhanced affinity to DSG2 will be translated to a significant improvement to the spreading of progeny virus, and thus a significant improvement to the effectiveness of oncolytic virotherapy.

Atherosclerotic vascular disease and its resulting myocardial infarction (MI) is the leading cause of death in subjects with diabetes. Additionally, patients with a history of a MI are more likely to experience a recurrent MI, however the mechanism is not yet understood. Previous work from our laboratory, and others, has demonstrated that diabetes accelerates atherosclerosis partly by affecting monocytes and macrophages, resulting in increased inflammation and macrophage accumulation in the artery wall. Work from others has shown that experimental MI can also accelerate atherosclerosis in non-diabetic mice by increasing the number of blood monocytes, which enter the artery wall to become inflammatory macrophages. Thus, we wanted to ask if diabetes and experimental MI have synergistic effects on macrophage accumulation in the artery wall. Experimental MI, or sham surgeries, were carried out in an atherosclerotic-prone, transgenic mouse model of type 1 diabetes that relies on viral mimicry (Ldlr-/-Gp+) and in non-diabetic controls. At the end of the study, cross sections of the dissected brachiocephalic artery were used to quantify macrophage accumulation by measuring areas of Mac-2 immunoreactivity on sections adjacent to the largest cross-sectional lesion. If macrophage accumulation is synergistically increased with diabetes and experimental MI, attempts will be made to understand the underlying mechanism. Atherosclerotic macrophage accumulation is a dynamic process that involves monocyte recruitment and local proliferation. Macrophage proliferation and monocyte recruitment will be analyzed by staining with proliferation marker Ki67 and estimating the number of previously labeled monocytes that enter the atherosclerotic lesion, respectively. To mechanistically test the role of monocyte recruitment on macrophage accumulation, Mac-2 staining will be quantified in mice (diabetic and non-diabetic with and without MI) deficient in a key adhesion molecule (CD49D) required for monocyte adhesion to activated endothelium. These results should yield important information on potential synergistic effects of diabetes and MI.

Rapamycin is a drug that has shown to extend lifespan across many model organisms including laboratory mice. Although the drug’s effect on extending lifespan has been investigated, it’s effect on transport of metabolic hormones across the blood-brain barrier (BBB), including insulin, are unknown. Rapamycin extends life by impacting whole body metabolism by acting as a caloric restriction mimetic inhibiting a cellular signaling molecule, mammalian target of rapamycin (mTOR). Insulin transportation into the brain via the BBB has been seen to increase during periods of starvation. Therefore, we wanted to investigate whether treatment of rapamycin would increase insulin transport across the BBB into the brain by inducing a 'starvation state'. I gave IP injections of 8 mg of rapamycin per kg body weight to two-month old male CD-1 mice for two weeks daily. A control group was given IP injections of vehicle for the same duration and amounts. At the end of the two weeks, I gave the mice IV injections of radioactive insulin and albumin (a vascular space marker) that circulated for various times (1-10 min) before collecting blood samples and performing whole brain removals. The radioactivity in the serum and whole brain was measured. The rate of uptake of the radioactive insulin into the brain was calculated and compared between chronic rapamycin treatment and control groups. Two-week rapamycin treatment showed elevated insulin levels in whole brain tissue when compared to the control group suggesting that rapamycin may have an effect on insulin transport.

Early antibiotic administration is a critical intervention in sepsis; however, identifying infection in the emergency department can be challenging. We hypothesize that emergency clinicians over-treat for infection overall, even when non-infectious etiologies are felt most likely. In two academic Emergency Departments (EDs), we prospectively investigated clinician gestalt – the use of historical facts and physical examination findings to recognize disease patterns and make a clinical decision – for infection and antibiotic administration to assess 1) the frequency in which clinicians administer antibiotics to patients felt most likely to not have infection, and 2) the number of patients felt to have infection and received antibiotics, who ultimately did not have infection. We enrolled subjects at both the University of Washington and Harborview Medical Centers. We included patients with signs of critical illness, defined as 1) two or more systemic inflammatory response syndrome criteria and organ dysfunction, 2) systolic blood pressure < 90 mmHg, and/or serum lactate levels greater than 4.0 mmol/L. At the time of admission, attending physicians were surveyed regarding the etiology of illness (infectious or non-infectious). Patient demographics, medical history, and ED interventions, including antibiotics, were abstracted by chart review. Ultimate determination of diagnosis was adjudicated by chart review by an attending emergency physician. We enrolled 405 patients, 191 (47.2%) with infection by final adjudication. Physicians identified 199 (49.1%) as most likely non-infectious, of whom 79/199 (39.7%) received antibiotics. Of these, 53/79 (67.1%) did not have the infectious etiologies. Of the 206 patients whom physicians suspected to most likely have infection, 185/206 (89.9%) received antibiotics, and 37/185 (20%) were adjudicated to non-infectious etiologies. Emergency clinicians mitigate diagnostic uncertainty by administering antibiotics. Improving the identification of infection among critically ill emergency patients may be a meaningful target for antibiotic stewardship efforts.

Diabetes is increasingly prevalent in the world, and there are currently over 29 million people in the United States with diabetes. Type 1 diabetic patients depend on insulin injections to maintain their blood insulin within a certain level so that their cells are getting enough sugar for energy production. Type 2 diabetic patients mostly rely on diet treatments and combinations of insulin and other drug treatments to avoid hyperglycemia. Moreover, type 2 diabetes often leads to different complications such as cardiovascular disease, blindness, and kidney failure. The goal of our study is to investigate insulin effects on the activation of macrophages, which are immune cells with important roles in inflammation. Since a pro-inflammatory state of macrophages can enhance insulin resistance and lead to cardiovascular diseases, it is crucial to understand if insulin changes the macrophage activation states that can cause off target effects leading to other medical complications. The experiments for this project are performed using RAW 264.7 cells, an immortal cell line derived from mouse macrophages. A dose experiment of insulin stimulation was performed to test if phosphorylated STAT1 (pro-inflammatory or M1) and phosphorylated STAT6 (anti-inflammatory or M2) were detected via western blot. Additional experiments determined the intensity of the macrophage response by measuring downstream signaling proteins (PI3K or IRS2) via immunoprecipitation and western blot. Also, RNA extraction and qPCR were performed to see if the gene transcription of macrophage polarization is activated by the above transcription factors. For example,  TNFα and IFNγ were measured for M1 polarization and IL-4 and IL-10 were measured for M2 polarization. Finally, NF-κB, another transcription factor was measured to determine if phosphorylation occurred, which should lead to M1 gene transcription. Together, these experiments revealed whether insulin affects key pro-inflammatory and anti-inflammatory signaling pathways in macrophages.

Oncology trials often entail high-stakes interventions where the potential for morbidity and for fatal side effects, as well as for life-prolongation or cure, intensify bioethical issues related to informed consent. These challenges are further compounded in multistage randomized trials, which are increasingly prevalent in oncology. We sought to elucidate the major challenges in achieving true patient understanding of informed consent documents in high-stakes oncology trials in general and the best consent practices for multistage randomized trials in particular. We queried the PubMed database for original studies published on 1-1-1990 to 4-1-2017 that focused on readability, quality, simplicity, complexity or length of consent documents, sickness level of participants, or interventions and enhancements that influence informed consent. Additional articles were identified through comprehensive bibliographic review. Twenty-four articles were retained as the basis for the narrative review; 18 enrolled participants and 6 examined samples of consent documents. This body of literature provides useful information about the pitfalls of the informed consent process as currently practiced in most high-stakes interventional oncologic trials. The understanding level of study participants is examined as impacted by the readability of documents, the complexity of the documents, the length of the forms and the “what’s at stake” nature of oncology trials. The length of the informed consent documents has increased from 338 in 1987 to 3982 in 2010 and patient understanding has been shown to be inversely proportional to the page count of the documents. Additionally, patients’ belief in the “therapeutic myth” also has a negative effect on true patient understanding. The unique challenges of obtaining consent in multistage trials are noted and examined. Though no studies specifically addressed the additional problems posed by multistage randomizations (such as change in risk-benefit ratio due to time-varying treatment responses or organ toxicities), the findings are likely applicable and especially relevant in that context. 

There are several approaches to treating Acute Myeloid Leukemia (AML) relapse after an allogeneic stem cell transplant. We were curious about the efficacy of the different types of treatments commonly deployed for patients with post-allograft AML relapse. Eligible patients included those who underwent allogeneic stem cell transplant for a diagnosis of AML and then relapsed between 1/1/2005 and 12/31/2015. We abstracted information about their treatments, GVHD status, immunosuppressive drugs, response (ie relapse or remission) events and vital status from their medical records. Among 120 patients, we found a median number of 2.6 and a mean of 2.9 relapse-directed lines of treatment, while 21% received no treatment. The types of treatment deployed were tightly correlated with the patient’s malignant blast count and the time that had elapsed since first transplant. There were 74 induction chemotherapy attempts, 64 hypomethylating therapy prescriptions, 19 donor lymphocyte infusions (DLI), and 16 second transplants, for example. Ongoing work includes completing data entry on the remaining 186 subjects, and analyzing the relative efficacy of the different treatment options. Importantly, we found that patients whose relapse presented with very low-level disease burden (<5% malignant blasts) still had a dismal prognosis, with only 20% being alive in remission at last follow-up. By contrast, patients relapsing with >5% blasts had an even worse prognosis – only 7% alive in remission at last follow-up. Closer surveillance for post-transplant relapse might catch more patients relapsing at low disease burdens. Initiating treatment when the disease burden is low may be more effective. Our ongoing work will identify the best sequences of treatments (leading to the longest relapse free survival), taking into account the blast burden, time elapsed since transplant, and other predictive factors.
 

The neuropathology of Alzheimer's Disease (AD) is characterized by deposition of amyloid plaques and neurofibrillary tangles of phosphorylated tau, but little is known about the extracellular matrix (ECM) that associates with plaques and tangles during progression of AD pathology. Brain ECM is abundant in Hyaluronan (HA), a non-sulfated glycosaminoglycan that is synthesized by HA synthases (HASes 1-3) at high molecular weight (MW), but readily degraded by hyaluronidases (HYALs 1-3) into low MW neuroinflammatory forms. HA size is stabilized by the presence of chondroitin sulfate proteoglycans (CSPGs) and other HA-binding proteins, such as the tenascins and TSG-6. HA deposition has been previously noted in regions of amyloid plaques, neurofibrillary tangles, and neuroinflammation. We propose that progression of AD neuropathology is associated with accumulation of HA as well as mediators that promote HA degradation into low MW neuroinflammatory forms. We examined gray matter (GM) parietal cortex of brains with severe (n=23, 12M, 11F), moderate (n=18, 11M, 7F), and no AD neuropathology (n=21, 13M, 8F). Histology and RT-PCR were utilized to examine for HA accumulation, HA co-localization with β-amyloid and phosphorylated tau, and mediators of HA degradation and stabilization (HASes 1-3, HYALs 1-3, CSPGs, tenascin C/R, and TSG-6). We found that AD neuropathology severity correlated significantly with HA accumulation. Compared to control brains, GM of advanced AD neuropathology samples showed an increase in expression of HAS2 and HYALs 1-3, and HA-associated ECM including CSPGs, Tenascin C/R, and TSG-6. In summary, HA and mediators of both HA accumulation, degradation and stabilization significantly increase with AD neuropathology. We propose that HA accumulation promotes progression of AD neuropathology by providing a substrate for low MW HA, which induces subsequent inflammation.

Human cytomegalovirus (HCMV) is a member of the Herpesvirus family. Most of the US population has been exposed to this virus, but it is only deadly in immunocompromised individuals. When a human cell becomes infected, one of the earliest mechanisms of anti-viral defense is the activation of protein kinase R (PKR). Active PKR phosphorylates the translation initiation factor, eIF2α, shutting down most protein synthesis in the cell. HCMV circumvents cellular shutdown by expressing protein TRS1, which binds to PKR and inhibits it. Another cellular kinase, the heme-regulated inhibitor (HRI), also phosphorylates eIF2α upon activation by reactive oxygen species. Because HRI and PKR are similar in structure, I hypothesize that HRI activation during HCMV infection is inhibited by TRS1. I infected PKR-knockout human fibroblasts with HCMV and then activated HRI expression by inducing production of reactive oxygen species. These cells also have a luciferase gene, expression of which is induced upon eIF2α phosphorylation. I measured the abundance of phosphorylated eIF2α (eIF2α-P) and luciferase activity in the lysate of infected cells in which HRI has been activated. I expect that TRS1 will bind and inhibit HRI, therefore eIF2α-P levels and luciferase activity will be low compared to cells lacking TRS1. These results reveal that HCMV replication is sensitive to HRI activation in reactive oxygen-rich inflammatory tissues and provides insights into the evolutionary history and structural similarities between HRI and PKR.

Primary sclerosing cholangitis (PSC) is a rare liver disease associated with inflammatory bowel disease (IBD) that causes strictures of the biliary tree and could result in the need for liver transplantation. One disease mechanism is upregulation of the mucosal vascular addressin cell adhesion molecule 1 (MAdCAM-1) protein and subsequent increased lymphocyte migration to the biliary epithelium. The purpose of this research project is to compare the levels of enzymes (especially alkaline phosphatase, ALP) that represent surrogate markers of liver inflammation in PSC patients before and after infusion of vedolizumab (VDZ), a drug to treat IBD that interferes with this lymphocyte migration. We conducted a retrospective observational study to collect information on PSC patients from their electronic medical records and enter it into Excel. Information collected before and after VDZ infusion included: demographics, medical and social history, laboratory blood tests, information from first clinical encounters, and histology and radiology reports (e.g. ERCP, CT, MRI, endoscopies). Subsequently, we wrote a script in R to import the dataset, visualize the enzyme levels before and after VDZ infusion, and perform statistical inference using a matched-pairs t-test to determine if there was a significant difference in mean ALP levels in PSC patients before and after VDZ infusion. We obtained the following results (n = 13, α = 0.05): mean ALP difference = 10.67 U/L, t = 0.23, df = 12, p = 0.82, 95% CI = (-88.41, 109.76). Since our p-value was greater than 0.05 (and our confidence interval contained the value 0), we failed to reject our null hypothesis and could not conclude that there was a significant difference in mean ALP levels before and after VDZ infusion. Thus, more information is needed to determine if VDZ may have a beneficial effect on liver function.

Lipopolysaccharide (LPS) is a bacterial membrane surface antigen that potently stimulates the immune system. In response to LPS, immune cells release cytokines and chemokines that propagate the immune response. LPS and cytokines can increase in the circulation during infections, which may cause disruption of the blood-brain barrier (BBB). BBB disruption results in leakage of circulating factors into the brain, which can damage neurons. Because there are known sex differences in the immune response, we aimed to determine in mice whether sex and strain affect LPS-induced BBB disruption, and assess cytokine expression patterns in blood and brain that may contribute to sex or strain differences. To test this, we injected different strains of male and female mice, C57 and CD-1, with three different LPS doses or with saline. BBB disruption and the levels of 23 cytokines were evaluated 28 hours later. We found that only the highest dose of LPS induced significant BBB disruption in all sexes and strains. The severity of BBB disruption to the highest LPS dose significantly differed between sexes and strains, and ranked CD-1 male < CD-1 female < C57 male < C57 female. Two-way ANOVA revealed significant sex and strain effects on BBB disruption in the high-dose LPS group, with 14.83 % of the variance attributable to sex, and 24.46% attributable to strain. Cytokines showed distinct profiles in brain and blood among sex and strain following LPS treatment. In blood, interleukin-10, keratinocyte chemoattractant, and interleukin-12(p40) trends in sex and strain paralleled those found for BBB disruption, and IL-12(p70) showed an opposite trend. Our results indicate that the severity of BBB disruption to LPS is partially dependent on sex and strain, with mouse strain having a greater effect. Future directions are warranted to determine whether cytokines identified in this study are causal in the BBB response to LPS.

ELISA (enzyme-linked immunosorbent assay) is a widely used antibody-based assay in research and clinical laboratories to quantify levels of antigens. Typically, it takes an overnight incubation step, followed by several hands-on hours the next day. In many settings, including clinical labs, more robust and faster ELISA is highly desirable. Many assays, including ELISA, are done in 96-well plates. Ultrasound (US) mixing and cavitation enhances binding rates of antibodies to antigens but, surprisingly, US has not been incorporated into ELISA. We set out to test if PIXUL can be used to enhance binding kinetics, reproducibility and sensitivity of ELISA. PIXUL is an ultrasound instrument made up of miniature transducers designed to treat samples with high-intensity focused ultrasound (HIFU) in 96-well microplates. PIXUL provides the flexibility to vary different US parameters including peak amplitude, pulse length, pulse repetition frequency (PRF) and treatment time. Our preliminary data showed that PIXUL treatment shortened the overall time to 3 hours, and eliminated the overnight incubation step but the sensitivity remains lower compared to the standard protocol. These results suggest that US can speed up the ELISA assays but additional optimizations are required to improve the sensitivity PIXUL-ELISA.

Although syphilis is often labeled as a disease of the past, this infection is still a significant public health concern in both developing countries, where most cases occur, and developed ones, including the United States, where syphilis has been resurgent for over a decade. Public health initiatives aimed at syphilis control would greatly benefit from the availability of a vaccine that could confer complete immunological protection against Treponema pallidum (T. pallidum), the syphilis agent. Modern pre-clinical vaccination/challenge trials focused on the use of recombinant T. pallidum antigens and although were effective in conferring partial protection against syphilis, complete protection was never achieved. A possible explanation for this result is that these immunizations were performed with antigens lacking their natural three-dimensional conformation. This, in turn, might have abrogated production of antibodies to conformational epitopes of these proteins and diminished their protectivity. To evaluate whether administration of antigens in their proper conformation could improve current vaccine formulations, we expressed T. pallidum antigens in a bacterial surrogate system that utilizes a lab-derived strain of Borrelia burgdorferi (B. burgdorferi), a spirochete genetically and structurally related to T. pallidum but engineered to be non-pathogenic. We used B. burgdorferi to express selected T. pallidum antigens that have been shown in the past to be partially protective and then injected these B. burgdorferi strains into New Zealand white rabbits to induce humoral and cellular immunity against the T. pallidum proteins. Following immunization, rabbits were challenged with infectious T. pallidum to evaluate the efficacy of this new vaccine design. Immunized animals developed a specific antibody response to the T. pallidum antigens, but were only partially protected after challenge. We showed that our innovative vaccine design is very effective in inducing an immune response to T. pallidum antigens and will be a useful tool to test additional vaccine candidates.

Researchers often deal with massive longitudinal data sets. Currently-available tools visualize large amounts of longitudinal data but are very rigid or require strict formatting of the data to process it and translate it into visuals. We created an online longitudinal data visualization tool that can be customized by its user in nearly every way. We aimed to visualize trends in sequentially-applied anti-cancer therapies in a cohort of patients from the Fred Hutchinson Cancer Research Center who received cell transplantation for leukemia, but later relapsed. The researchers wanted to visualize multiple disease statuses in parallel to the treatment-and-response trajectories. My objective was to make this tool as customizable as possible, so that it could be used by other researchers with different longitudinal data. I first created a prototype using JavaFX that allowed the user to construct customizable swimmer plots. My design allowed the user to customize visuals such as the colors, shapes, and scale of events in the swimmer plot. To further improve the dynamic features and efficiency of this tool, I approached the developers of Oncoscape, an open source JavaScript software that allows for data visualization and exploratory data analysis, to refine and incorporate my tool into the Oncoscape suite. The current iteration of our tool provides unparalleled flexibility to accommodate any longitudinal data set. It allows researchers to select subsets of subjects based on multiple shared characteristics, allow simultaneous visualization of two or more sequences per subject, and allow researchers to zoom in to a granular level of detail for each subject. This tool will provide doctors a means to quickly study which treatments were administered in which order to similar, historical patients, and their associated outcomes. Thus, by disseminating this tool, we will help doctors make better therapeutic decisions during key moments in an individual patient’s treatment course.