Exploration of differentiability in a proton computed tomography simulation framework.

Objective.Gradient-based optimization using algorithmic derivatives can be a useful technique to improve engineering designs with respect to a computer-implemented objective function. Likewise, uncertainty quantification through computer simulations can be carried out by means of derivatives of the computer simulation. However, the effectiveness of these techniques depends on how 'well-linearizable' the software is. In this study, we assess how promising derivative information of a typical proton computed tomography (pCT) scan computer simulation is for the aforementioned applications.Approach.This study is mainly based on numerical experiments, in which we repeatedly evaluate three representative computational steps with perturbed input values. We support our observations with a review of the algorithmic steps and arithmetic operations performed by the software, using debugging techniques.Main results.The model-based iterative reconstruction (MBIR) subprocedure (at the end of the software pipeline) and the Monte Carlo (MC) simulation (at the beginning) were piecewise differentiable. However, the observed high density and magnitude of jumps was likely to preclude most meaningful uses of the derivatives. Jumps in the MBIR function arose from the discrete computation of the set of voxels intersected by a proton path, and could be reduced in magnitude by a 'fuzzy voxels' approach. The investigated jumps in the MC function arose from local changes in the control flow that affected the amount of consumed random numbers. The tracking algorithm solves an inherently non-differentiable problem.Significance.Besides the technical challenges of merely applying AD to existing software projects, the MC and MBIR codes must be adapted to compute smoother functions. For the MBIR code, we presented one possible approach for this while for the MC code, this will be subject to further research. For the tracking subprocedure, further research on surrogate models is necessary.

Uncertainty-aware spot rejection rate as quality metric for proton therapy using a digital tracking calorimeter.

Objective.Proton therapy is highly sensitive to range uncertainties due to the nature of the dose deposition of charged particles. To ensure treatment quality, range verification methods can be used to verify that the individual spots in a pencil beam scanning treatment fraction match the treatment plan. This study introduces a novel metric for proton therapy quality control based on uncertainties in range verification of individual spots.Approach.We employ uncertainty-aware deep neural networks to predict the Bragg peak depth in an anthropomorphic phantom based on secondary charged particle detection in a silicon pixel telescope designed for proton computed tomography. The subsequently predicted Bragg peak positions, along with their uncertainties, are compared to the treatment plan, rejecting spots which are predicted to be outside the 95% confidence interval. The such-produced spot rejection rate presents a metric for the quality of the treatment fraction.Main results.The introduced spot rejection rate metric is shown to be well-defined for range predictors with well-calibrated uncertainties. Using this method, treatment errors in the form of lateral shifts can be detected down to 1 mm after around 1400 treated spots with spot intensities of 1 × 107protons. The range verification model used in this metric predicts the Bragg peak depth to a mean absolute error of 1.107 ± 0.015 mm.Significance.Uncertainty-aware machine learning has potential applications in proton therapy quality control. This work presents the foundation for future developments in this area.

Investigating particle track topology for range telescopes in particle radiography using convolutional neural networks.

BACKGROUND: Proton computed tomography (pCT) and radiography (pRad) are proposed modalities for improved treatment plan accuracy and in situ treatment validation in proton therapy. The pCT system of the Bergen pCT collaboration is able to handle very high particle intensities by means of track reconstruction. However, incorrectly reconstructed and secondary tracks degrade the image quality. We have investigated whether a convolutional neural network (CNN)-based filter is able to improve the image quality. MATERIAL AND METHODS: The CNN was trained by simulation and reconstruction of tens of millions of proton and helium tracks. The CNN filter was then compared to simple energy loss threshold methods using the Area Under the Receiver Operating Characteristics curve (AUROC), and by comparing the image quality and Water Equivalent Path Length (WEPL) error of proton and helium radiographs filtered with the same methods. RESULTS: The CNN method led to a considerable improvement of the AUROC, from 74.3% to 97.5% with protons and from 94.2% to 99.5% with helium. The CNN filtering reduced the WEPL error in the helium radiograph from 1.03 mm to 0.93 mm while no improvement was seen in the CNN filtered pRads. CONCLUSION: The CNN improved the filtering of proton and helium tracks. Only in the helium radiograph did this lead to improved image quality.

Esophageal Diameters in Children Correlated to Body Weight.

INTRODUCTION: The development of stenoses after correction of an esophageal atresia or acid and lye burn of the esophagus are well-known problems in pediatric surgery. Currently, stenoses are treated in the majority of cases by repeated balloon dilatations. The diameter of the balloons used is not standardized; standard curves do not exist. The aim of this study was to evaluate the diameter of the esophagus correlated to the body weight of the children as measured in upper gastrointestinal (GI) studies to answer the important question to what extent a stenosis should be dilated. MATERIALS AND METHODS: Within the time period from 2011 through 2016, 60 patients with upper GI studies were selected. Evaluations were blinded to two different examiners. The diameters were measured under maximum contrast filling between the second and third rib (cranial point of measurement) and between the seventh and eighth rib (caudal point of measurement). For both, the anteroposterior and lateral aspect was examined. The diameter was calculated as the arithmetic average of both measurements within one level. The diameters were correlated to the weight of the children. RESULTS: All children (n = 38) within the 3rd to 97th weight percentile were analyzed. Linear correlation and coefficients of 0.67 at the cranial point and 0.70 at the caudal point were found. Mean diameter at the cranial point of measurement was 6.75 mm at the lowest weight (2.6 kg) and 14 mm at 74 kg. Mean weight of these children (standard deviation [SD]) was 25.3 (18.8) kg and median age was 7 years. Within weight groups (0-10 kg; 10-20 kg; 20-35 kg; 35-50 kg; >50 kg), we calculated SD and two side tested critical 95% confidence interval for all measurements (n = 74). CONCLUSION: Although the variation in measurements is considerable, this evaluation gives a reliable hint to which extent esophageal stenoses should be dilated in relation to the body weight. To the best of our knowledge, this is the first investigation to evaluate the diameter of the esophagus in children in relation to the body weight.

Molecular basis of substrate recognition and degradation by human presequence protease.

Human presequence protease (hPreP) is an M16 metalloprotease localized in mitochondria. There, hPreP facilitates proteostasis by utilizing an ∼13,300-Å(3) catalytic chamber to degrade a diverse array of potentially toxic peptides, including mitochondrial presequences and ß-amyloid (Aß), the latter of which contributes to Alzheimer disease pathogenesis. Here, we report crystal structures for hPreP alone and in complex with Aß, which show that hPreP uses size exclusion and charge complementation for substrate recognition. These structures also reveal hPreP-specific features that permit a diverse array of peptides, with distinct distributions of charged and hydrophobic residues, to be specifically captured, cleaved, and have their amyloidogenic features destroyed. SAXS analysis demonstrates that hPreP in solution exists in dynamic equilibrium between closed and open states, with the former being preferred. Furthermore, Aß binding induces the closed state and hPreP dimerization. Together, these data reveal the molecular basis for flexible yet specific substrate recognition and degradation by hPreP.

Transglutaminase and polyamination of tubulin: posttranslational modification for stabilizing axonal microtubules.

Neuronal microtubules support intracellular transport, facilitate axon growth, and form a basis for neuronal morphology. While microtubules in nonneuronal cells are depolymerized by cold, Ca(2+), or antimitotic drugs, neuronal microtubules are unusually stable. Such stability is important for normal axon growth and maintenance, while hyperstability may compromise neuronal function in aging and degeneration. Though mechanisms for stability are unclear, studies suggest that stable microtubules contain biochemically distinct tubulins that are more basic than conventional tubulins. Transglutaminase-catalyzed posttranslational incorporation of polyamines is one of the few modifications of intracellular proteins that add positive charges. Here we show that neuronal tubulin can be polyaminated by transglutaminase. Endogenous brain transglutaminase-catalyzed polyaminated tubulins have the biochemical characteristics of neuronal stable microtubules. Inhibiting polyamine synthesis or transglutaminase activity significantly decreases microtubule stability in vitro and in vivo. Together, these findings suggest that transglutaminase-catalyzed polyamination of tubulins stabilizes microtubules essential for unique neuronal structures and functions.

Polymerization of MIP-1 chemokine (CCL3 and CCL4) and clearance of MIP-1 by insulin-degrading enzyme.

Macrophage inflammatory protein-1 (MIP-1), MIP-1α (CCL3) and MIP-1ß (CCL4) are chemokines crucial for immune responses towards infection and inflammation. Both MIP-1α and MIP-1ß form high-molecular-weight aggregates. Our crystal structures reveal that MIP-1 aggregation is a polymerization process and human MIP-1α and MIP-1ß form rod-shaped, double-helical polymers. Biophysical analyses and mathematical modelling show that MIP-1 reversibly forms a polydisperse distribution of rod-shaped polymers in solution. Polymerization buries receptor-binding sites of MIP-1α, thus depolymerization mutations enhance MIP-1α to arrest monocytes onto activated human endothelium. However, same depolymerization mutations render MIP-1α ineffective in mouse peritoneal cell recruitment. Mathematical modelling reveals that, for a long-range chemotaxis of MIP-1, polymerization could protect MIP-1 from proteases that selectively degrade monomeric MIP-1. Insulin-degrading enzyme (IDE) is identified as such a protease and decreased expression of IDE leads to elevated MIP-1 levels in microglial cells. Our structural and proteomic studies offer a molecular basis for selective degradation of MIP-1. The regulated MIP-1 polymerization and selective inactivation of MIP-1 monomers by IDE could aid in controlling the MIP-1 chemotactic gradient for immune surveillance.

Molecular basis for the recognition and cleavages of IGF-II, TGF-alpha, and amylin by human insulin-degrading enzyme.

Insulin-degrading enzyme (IDE) is involved in the clearance of many bioactive peptide substrates, including insulin and amyloid-beta, peptides vital to the development of diabetes and Alzheimer's disease, respectively. IDE can also rapidly degrade hormones that are held together by intramolecular disulfide bond(s) without their reduction. Furthermore, IDE exhibits a remarkable ability to preferentially degrade structurally similar peptides such as the selective degradation of insulin-like growth factor (IGF)-II and transforming growth factor-alpha (TGF-alpha) over IGF-I and epidermal growth factor, respectively. Here, we used high-accuracy mass spectrometry to identify the cleavage sites of human IGF-II, TGF-alpha, amylin, reduced amylin, and amyloid-beta by human IDE. We also determined the structures of human IDE-IGF-II and IDE-TGF-alpha at 2.3 A and IDE-amylin at 2.9 A. We found that IDE cleaves its substrates at multiple sites in a biased stochastic manner. Furthermore, the presence of a disulfide bond in amylin allows IDE to cut at an additional site in the middle of the peptide (amino acids 18-19). Our amylin-bound IDE structure offers insight into how the structural constraint from a disulfide bond in amylin can alter IDE cleavage sites. Together with NMR structures of amylin and the IGF and epidermal growth factor families, our work also reveals the structural basis of how the high dipole moment of substrates complements the charge distribution of the IDE catalytic chamber for the substrate selectivity. In addition, we show how the ability of substrates to properly anchor their N-terminus to the exosite of IDE and undergo a conformational switch upon binding to the catalytic chamber of IDE can also contribute to the selective degradation of structurally related growth factors.

Protective role of Cys-178 against the inactivation and oligomerization of human insulin-degrading enzyme by oxidation and nitrosylation.

Insulin-degrading enzyme (IDE), a 110-kDa metalloendopeptidase, hydrolyzes several physiologically relevant peptides, including insulin and amyloid-beta (Abeta). Human IDE has 13 cysteines and is inhibited by hydrogen peroxide and S-nitrosoglutathione (GSNO), donors of reactive oxygen and nitrogen species, respectively. Here, we report that the oxidative burst of BV-2 microglial cells leads to oxidation or nitrosylation of secreted IDE, leading to the reduced activity. Hydrogen peroxide and GSNO treatment of IDE reduces the V(max) for Abeta degradation, increases IDE oligomerization, and decreases IDE thermostability. Additionally, this inhibitory response of IDE is substrate-dependent, biphasic for Abeta degradation but monophasic for a shorter bradykinin-mimetic substrate. Our mutational analysis of IDE and peptide mass fingerprinting of GSNO-treated IDE using Fourier transform-ion cyclotron resonance mass spectrometer reveal a surprising interplay of Cys-178 with Cys-110 and Cys-819 for catalytic activity and with Cys-789 and Cys-966 for oligomerization. Cys-110 is near the zinc-binding catalytic center and is normally buried. The oxidation and nitrosylation of Cys-819 allow Cys-110 to be oxidized or nitrosylated, leading to complete inactivation of IDE. Cys-789 is spatially adjacent to Cys-966, and their nitrosylation and oxidation together trigger the oligomerization and inhibition of IDE. Interestingly, the Cys-178 modification buffers the inhibition caused by Cys-819 modification and prevents the oxidation or nitrosylation of Cys-110. The Cys-178 modification can also prevent the oligomerization-mediated inhibition. Thus, IDE can be intricately regulated by reactive oxygen or nitrogen species. The structure of IDE reveals the molecular basis for the long distance interactions of these cysteines and how they regulate IDE function.

Molecular basis of catalytic chamber-assisted unfolding and cleavage of human insulin by human insulin-degrading enzyme.

Insulin is a hormone vital for glucose homeostasis, and insulin-degrading enzyme (IDE) plays a key role in its clearance. IDE exhibits a remarkable specificity to degrade insulin without breaking the disulfide bonds that hold the insulin A and B chains together. Using Fourier transform ion cyclotron resonance (FTICR) mass spectrometry to obtain high mass accuracy, and electron capture dissociation (ECD) to selectively break the disulfide bonds in gas phase fragmentation, we determined the cleavage sites and composition of human insulin fragments generated by human IDE. Our time-dependent analysis of IDE-digested insulin fragments reveals that IDE is highly processive in its initial cleavage at the middle of both the insulin A and B chains. This ensures that IDE effectively splits insulin into inactive N- and C-terminal halves without breaking the disulfide bonds. To understand the molecular basis of the recognition and unfolding of insulin by IDE, we determined a 2.6-A resolution insulin-bound IDE structure. Our structure reveals that IDE forms an enclosed catalytic chamber that completely engulfs and intimately interacts with a partially unfolded insulin molecule. This structure also highlights how the unique size, shape, charge distribution, and exosite of the IDE catalytic chamber contribute to its high affinity ( approximately 100 nm) for insulin. In addition, this structure shows how IDE utilizes the interaction of its exosite with the N terminus of the insulin A chain as well as other properties of the catalytic chamber to guide the unfolding of insulin and allowing for the processive cleavages.

Signaling through CD43 regulates CD4 T-cell trafficking.

The mucin-like protein CD43 is excluded from the immune synapse, and regulates T-cell proliferation as well as T-cell migration. While the CD43 cytoplasmic domain is necessary for regulation of T-cell activation and proliferation, the mechanism via which CD43 regulates trafficking is not well defined. To investigate whether CD43 phosphorylation regulates its function in T cells, we used tandem mass spectrometry and identified Ser76 in murine CD43 as a previously unidentified site of basal phosphorylation. Interestingly, mutation of this single serine to alanine greatly diminishes T-cell trafficking to the lymph node, while CD43 exclusion and CD43-mediated regulation of T-cell proliferation remain intact. Furthermore, the CD43 extracellular domain was also required for T-cell trafficking, providing a hitherto unknown function for the extracellular domain, and suggesting that the extracellular domain may be required to transduce signals via the cytoplasmic domain. These data reveal a novel mechanism by which CD43 regulates T-cell function, and suggest that CD43 functions as a signaling molecule, sensing extracellular cues and transducing intracellular signals that modulate T-cell function.

Optimizing thiophosphorylation in the presence of competing phosphorylation with MALDI-TOF-MS detection.

Thiophosphorylation provides a metabolically stable, chemically reactive phosphorylation analogue for analyzing the phosphoproteome in vitro and in vivo. We developed a MALDI-TOF-MS based assay for optimizing thiophosphopeptide production by a kinase even in the presence of Mg(2+) and ATP. We found that Abl kinase thiophosphorylation rates can be "rescued" using Mn(2+) in the presence of Mg(2+). Under our ideal conditions, titration of Mn(2+) and ATPgammaS in the presence of Mg(2+) allowed relatively rapid, highly specific thiophosphorylation by Abl tyrosine kinase, both as purified enzyme and in complex cell extracts.