Investigating visual navigation using spiking neural network models of the insect mushroom bodies.

Ants are capable of learning long visually guided foraging routes with limited neural resources. The visual scene memory needed for this behaviour is mediated by the mushroom bodies; an insect brain region important for learning and memory. In a visual navigation context, the mushroom bodies are theorised to act as familiarity detectors, guiding ants to views that are similar to those previously learned when first travelling along a foraging route. Evidence from behavioural experiments, computational studies and brain lesions all support this idea. Here we further investigate the role of mushroom bodies in visual navigation with a spiking neural network model learning complex natural scenes. By implementing these networks in GeNN-a library for building GPU accelerated spiking neural networks-we were able to test these models offline on an image database representing navigation through a complex outdoor natural environment, and also online embodied on a robot. The mushroom body model successfully learnt a large series of visual scenes (400 scenes corresponding to a 27 m route) and used these memories to choose accurate heading directions during route recapitulation in both complex environments. Through analysing our model's Kenyon cell (KC) activity, we were able to demonstrate that KC activity is directly related to the respective novelty of input images. Through conducting a parameter search we found that there is a non-linear dependence between optimal KC to visual projection neuron (VPN) connection sparsity and the length of time the model is presented with an image stimulus. The parameter search also showed training the model on lower proportions of a route generally produced better accuracy when testing on the entire route. We embodied the mushroom body model and comparator visual navigation algorithms on a Quanser Q-car robot with all processing running on an Nvidia Jetson TX2. On a 6.5 m route, the mushroom body model had a mean distance to training route (error) of 0.144 ± 0.088 m over 5 trials, which was performance comparable to standard visual-only navigation algorithms. Thus, we have demonstrated that a biologically plausible model of the ant mushroom body can navigate complex environments both in simulation and the real world. Understanding the neural basis of this behaviour will provide insight into how neural circuits are tuned to rapidly learn behaviourally relevant information from complex environments and provide inspiration for creating bio-mimetic computer/robotic systems that can learn rapidly with low energy requirements.

Click Chemistry: Reaction Rates and Their Suitability for Biomedical Applications.

Click chemistry has become a commonly used synthetic method due to the simplicity, efficiency, and high selectivity of this class of chemical reactions. Since their initial discovery, further click chemistry methods have been identified and added to the toolbox of click chemistry reactions for biomedical applications. However, selecting the most suitable reaction for a specific application is often challenging, as multiple factors must be considered, including selectivity, reactivity, biocompatibility, and stability. Thus, this review provides an overview of the benefits and limitations of well-established click chemistry reactions with a particular focus on the importance of considering reaction rates, an often overlooked criterion with little available guidance. The importance of understanding each click chemistry reaction beyond simply the reaction speed is discussed comprehensively with reference to recent biomedical research which utilized click chemistry. This review aims to provide a practical resource for researchers to guide the selection of click chemistry classes for different biomedical applications.

Synthesis and preclinical evaluation of a 89Zr-labelled human single chain antibody for non-invasive detection of hepatic myofibroblasts in acute liver injury.

Synaptophysin is expressed on fibrogenic hepatic myofibroblasts. C1-3 is a single chain human antibody (scAb) that binds specifically to synaptophysin on hepatic myofibroblasts, providing a targeting vector for novel in vivo imaging agents of chronic liver disease. C1-3 and a negative control scAb, CSBD9, were radiolabelled with zirconium-89 via desferrioxamine chelation to enable non-invasive molecular imaging with positron emission tomography (PET). DFO-scAb conjugates were characterised by gel electrophoresis (SDS-PAGE) and MALDI-TOF spectrometry, and 89Zr-labelled with high radiolabelling efficiency (99%). [89Zr]Zr-DFO-C1-3 exhibited high in vitro stability (> 99%) in mouse and human sera over 3 days at 25 and 37 °C. Activated hepatic myofibroblasts incubated with [89Zr]Zr-DFO-C1-3 displayed significantly higher internalised activity (59.46%, P = 0.001) compared to the [89Zr]Zr-DFO-CSBD9 control, indicating synaptophysin-mediated uptake and high binding specificity of [89Zr]Zr-DFO-C1-3. Mice with CCl4-induced acute liver damage exhibited significantly higher liver uptake of [89Zr]Zr-DFO-C1-3, compared to controls, confirmed by both Cerenkov imaging and ex vivo gamma counting (4.41 ± 0.19%ID/g, P < 0.0001). CCl4-induced liver damage and the number of hepatic myofibroblasts was confirmed by αSMA staining of liver sections. These findings indicate that [89Zr]Zr-DFO-C1-3 has promising utility as a PET imaging agent for non-invasive detection of hepatic myofibroblasts following acute liver injury.

Synthesis and In Vitro Evaluation of a HER2-Specific ImmunoSCIFI Probe.

Secondary Cerenkov-induced fluorescence imaging (SCIFI) is an emerging biomedical optical imaging modality that leverages Cerenkov luminescence, primarily generated by ß-emitting radioisotopes, to excite fluorophores that offer near-infrared emissions with optimal tissue penetrance. Dual-functionalized immunoconjugates composed of an antibody, a near-infrared fluorophore, and a ß-emitting radioisotope have potential utility as novel SCIFI constructs with high specificity for molecular biomarkers of disease. Here, we report the synthesis and characterization of [89Zr]Zr-DFO-trastuzumab-BOD665, a self-excitatory HER2-specific "immunoSCIFI" probe capable of yielding near-infrared fluorescence in situ without external excitation. The penetration depth of the SCIFI signal was measured in hemoglobin-infused optical tissue phantoms that indicated a 2.05-fold increase compared to 89Zr-generated Cerenkov luminescence. Additionally, the binding specificity of the immunoSCIFI probe for HER2 was evaluated in a cellular assay that showed significantly higher binding to SKBR3 (high HER2 expression) relative to MDA-MB-468 (low HER2) breast cancer cells based on measurements of total flux in the near-infrared region with external excitation blocked. Taken together, the results of this study indicate the potential utility of immunoSCIFI constructs for interrogation of molecular biomarkers of disease.

Descending neurons of the hoverfly respond to pursuits of artificial targets.

Many animals use motion vision information to control dynamic behaviors. Predatory animals, for example, show an exquisite ability to detect rapidly moving prey, followed by pursuit and capture. Such target detection is not only used by predators but is also important in conspecific interactions, such as for male hoverflies defending their territories against conspecific intruders. Visual target detection is believed to be subserved by specialized target-tuned neurons found in a range of species, including vertebrates and arthropods. However, how these target-tuned neurons respond to actual pursuit trajectories is currently not well understood. To redress this, we recorded extracellularly from target-selective descending neurons (TSDNs) in male Eristalis tenax hoverflies. We show that they have dorso-frontal receptive fields with a preferred direction up and away from the visual midline. We reconstructed visual flow fields as experienced during pursuits of artificial targets (black beads). We recorded TSDN responses to six reconstructed pursuits and found that each neuron responded consistently at remarkably specific time points but that these time points differed between neurons. We found that the observed spike probability was correlated with the spike probability predicted from each neuron's receptive field and size tuning. Interestingly, however, the overall response rate was low, with individual neurons responding to only a small part of each reconstructed pursuit. In contrast, the TSDN population responded to substantially larger proportions of the pursuits but with lower probability. This large variation between neurons could be useful if different neurons control different parts of the behavioral output.

The influence of degree of labelling upon cellular internalisation of antibody-cell penetrating peptide conjugates.

Antibody-based agents are increasingly used as therapeutics and imaging agents, yet are generally restricted to cell surface targets due to inefficient cellular internalisation and endosomal entrapment. Enhanced cell membrane translocation of antibodies can be achieved by the covalent attachment of cell-penetrating peptides, including the HIV-1-derived transactivator of transcription (TAT) peptide. This study evaluated the cellular internalisation properties of five anti-HER2 Herceptin-TAT conjugates with degrees of TAT labelling (DOLTAT) ranging from one to five. Herceptin-TAT conjugates were synthesised via a strain-promoted alkyne-azide cycloaddition reaction, characterised by UV-vis spectroscopy, MALDI-TOF, and gel electrophoresis, then radiolabelled with zirconium-89 to permit measurement of cellular internalisation by gamma counting. [89Zr]Zr-DFO-Her-TAT(0-5) conjugates were isolated in high radiochemical purity (>99%) and exhibited high stability in murine and human serum over 7 days at 37 °C. Significant increases in cellular internalisation were observed for [89Zr]Zr-DFO-Her-TAT conjugates with DOLTAT values of 2 and above in SKBR3 (high HER2) cells over 48 h, in contrast to low-level non-specific uptake in MDA-MB-468 (low HER2) cells that did not increase over time. Notably, [89Zr]Zr-DFO-Her-TAT conjugates with DOLTAT of 3, 4, and 5 reached uptake values in SKBR3 cells of 5, 6, and 8% of the applied dose at 48 h respectively, representing 9, 10, 14-fold increases relative to the TAT-free control conjugate, [89Zr]Zr-DFO-Her-TAT(0).

In situ excitation of BODIPY fluorophores by 89Zr-generated Cerenkov luminescence.

Secondary Cerenkov-induced fluorescence imaging (SCIFI) is an emerging optical imaging technology that affords high signal-to-noise images by utilising radionuclide-generated Cerenkov luminescence to excite fluorescent probes. BODIPY dyes offer attractive properties for SCIFI, including high quantum yields and photochemical stability, yet their utility in this application in combination with clinically relevant ß+-emitting radioisotopes remains largely unexplored. In this report, the fluorescence properties of three meso-substituted BODIPY analogues have been assessed in combination with the positron emitter zirconium-89. Most notably, SCIFI data acquired over 7 days shows the BODIPY scaffold remain largely inert to radiolysis, indicating the promising utility of this fluorophore class in SCIFI applications.

Synthesis and In Vivo Evaluation of a Site-specifically Labeled Radioimmunoconjugate for Dual-Modal (PET/NIRF) Imaging of MT1-MMP in Sarcomas.

Bone sarcomas are devastating primary bone cancers that mostly affect children, young adults, and the elderly. These aggressive tumors are associated with poor survival, and surgery remains the mainstay of treatment. Surgical planning is increasingly informed by positron emission tomography (PET), and tumor margin identification during surgery is aided by near-infrared fluorescence (NIRF) imaging, yet these investigations are confounded by probes that lack specificity for sarcoma biomarkers. We report the development of a dual-modal (PET/NIRF) immunoconjugate ([89Zr]Zr-DFO-anti-MT1-MMP-IRDye800CW) that targets MT1-MMP, a matrix metalloproteinase overexpressed in high-grade sarcomas. [89Zr]Zr-DFO-anti-MT1-MMP-IRDye800CW was synthesized via site-specific chemoenzymatic glycan modification, characterized, and isolated in high specific activity and radiochemical purity. Saturation binding and immunoreactivity assays indicated only minor perturbation of binding properties. A novel mouse model of dedifferentiated chondrosarcoma based on intrafemoral inoculation of HT1080 WT or KO cells (high and low MT1-MMP expression, respectively) was used to evaluate target binding and biodistribution. Fluorescence and Cerenkov luminescence images of [89Zr]Zr-DFO-anti-MT1-MMP-IRDye800CW showed preferential uptake in HT1080 WT tumors. Ex vivo gamma counting revealed that uptake in MT1-MMP-positive tumors was significantly higher than that in control groups. Taken together, [89Zr]Zr-DFO-anti-MT1-MMP-IRDye800CW is a promising dual-modal sarcoma imaging agent for pre-operative surgical planning and intraoperative surgical guidance.

PyGeNN: A Python Library for GPU-Enhanced Neural Networks.

More than half of the Top 10 supercomputing sites worldwide use GPU accelerators and they are becoming ubiquitous in workstations and edge computing devices. GeNN is a C++ library for generating efficient spiking neural network simulation code for GPUs. However, until now, the full flexibility of GeNN could only be harnessed by writing model descriptions and simulation code in C++. Here we present PyGeNN, a Python package which exposes all of GeNN's functionality to Python with minimal overhead. This provides an alternative, arguably more user-friendly, way of using GeNN and allows modelers to use GeNN within the growing Python-based machine learning and computational neuroscience ecosystems. In addition, we demonstrate that, in both Python and C++ GeNN simulations, the overheads of recording spiking data can strongly affect runtimes and show how a new spike recording system can reduce these overheads by up to 10×. Using the new recording system, we demonstrate that by using PyGeNN on a modern GPU, we can simulate a full-scale model of a cortical column faster even than real-time neuromorphic systems. Finally, we show that long simulations of a smaller model with complex stimuli and a custom three-factor learning rule defined in PyGeNN can be simulated almost two orders of magnitude faster than real-time.

Larger GPU-accelerated brain simulations with procedural connectivity.

Simulations are an important tool for investigating brain function but large models are needed to faithfully reproduce the statistics and dynamics of brain activity. Simulating large spiking neural network models has, until now, needed so much memory for storing synaptic connections that it required high performance computer systems. Here, we present an alternative simulation method we call 'procedural connectivity' where connectivity and synaptic weights are generated 'on the fly' instead of stored and retrieved from memory. This method is particularly well suited for use on graphical processing units (GPUs)-which are a common fixture in many workstations. Using procedural connectivity and an additional GPU code generation optimization, we can simulate a recent model of the macaque visual cortex with 4.13 × 106 neurons and 24.2 × 109 synapses on a single GPU-a significant step forward in making large-scale brain modeling accessible to more researchers.

89Zr-PET imaging of DNA double-strand breaks for the early monitoring of response following α- and ß-particle radioimmunotherapy in a mouse model of pancreatic ductal adenocarcinoma.

Rationale: The evaluation of early treatment response is critical for patient prognosis and treatment planning. When the current methods rely on invasive protocols that evaluate the expression of DNA damage markers on patient biopsy samples, we aim to evaluate a non-invasive PET imaging approach to monitor the early expression of the phosphorylated histone γH2AX in the context of pancreatic cancer targeted radionuclide therapy. Pancreatic ductal adenocarcinoma has a poor patient prognosis due to the absence of curative treatment for patients with advanced disease. There is therefore a critical need for the fast clinical translation of new therapeutic options. In line with these observations, our group has been focusing on the development of radiotheranostic agents based on a fully human monoclonal antibody (5B1) with exceptional affinity for CA19.9, an antigen overexpressed in PDAC. Two on-going clinical trials resulted from these efforts, one with 89Zr (diagnosis) and one with 177Lu (ß-particle therapy). More recently, we successfully developed and evaluated in PDAC mouse models a targeted α-therapy strategy with high clinical translation potential. We aim to expedite the clinical translation of the developed radioimmunotherapy approaches by investigating the early therapeutic response and effect of radiation therapy in a PDAC mouse model via PET imaging. Methods: Mice bearing BxPC3 tumor xenografts were treated with α- and ß-particle pretargeted radioimmunotherapy (PRIT), external beam radiotherapy (EBRT), or sham-treated (vehicle). The phosphorylated histone γH2AX produced as a response to DNA double strand breaks was quantified with the PET radiotracer, [89Zr]Zr-DFO-anti-γH2AX-TAT. Results: PET imaging studies in BxPC3 PDAC mouse models demonstrated increased uptake of [89Zr]Zr-DFO-anti-γH2AX-TAT (6.29 ± 0.15 %IA/g) following ß-PRIT in BxPC3 PDAC xenografts as compared to the saline control group (4.58 ± 0.76 %IA/g) and EBRT control group (5.93 ± 0.76 %IA/g). Similarly, significantly higher uptake of [89Zr]Zr-DFO-anti-γH2AX-TAT was observed in tumors of the 225Ac-PRIT and EBRT (10 Gy) cohorts (7.37 ± 1.23 and 6.80 ± 1.24 %IA/g, respectively) compared to the negative control cohort (5.08 ± 0.95 %IA/g). Ex vivo γH2AX immunohistochemistry and immunofluorescence analysis correlated with in vivo89Zr-anti-γH2AX PET/CT imaging with increased γH2AX positive cell and γH2AX foci per cell in the treated cohorts. When α-PRIT resulted in prolonged overall survival of treated animals (107.5 days) as compared to ß-PRIT (73.0 days), no evidence of difference in [89Zr]Zr-DFO-anti-γH2AX-TAT uptake at the tumor site was observed, highlighting that DNA damage is not the sole radiobiology paradigm and that off-targeted (bystander) effects should be considered. Conclusions: PET imaging studies with [89Zr]Zr-DFO-anti-γH2AX-TAT following α- and ß-particle PRIT in a BxPC3 PDAC subcutaneous xenograft mouse model allowed the monitoring of tumor radiobiological response to treatment.

Event-Based Computation for Touch Localization Based on Precise Spike Timing.

Precise spike timing and temporal coding are used extensively within the nervous system of insects and in the sensory periphery of higher order animals. However, conventional Artificial Neural Networks (ANNs) and machine learning algorithms cannot take advantage of this coding strategy, due to their rate-based representation of signals. Even in the case of artificial Spiking Neural Networks (SNNs), identifying applications where temporal coding outperforms the rate coding strategies of ANNs is still an open challenge. Neuromorphic sensory-processing systems provide an ideal context for exploring the potential advantages of temporal coding, as they are able to efficiently extract the information required to cluster or classify spatio-temporal activity patterns from relative spike timing. Here we propose a neuromorphic model inspired by the sand scorpion to explore the benefits of temporal coding, and validate it in an event-based sensory-processing task. The task consists in localizing a target using only the relative spike timing of eight spatially-separated vibration sensors. We propose two different approaches in which the SNNs learns to cluster spatio-temporal patterns in an unsupervised manner and we demonstrate how the task can be solved both analytically and through numerical simulation of multiple SNN models. We argue that the models presented are optimal for spatio-temporal pattern classification using precise spike timing in a task that could be used as a standard benchmark for evaluating event-based sensory processing models based on temporal coding.

Early Detection in a Mouse Model of Pancreatic Cancer by Imaging DNA Damage Response Signaling.

Despite its widespread use in oncology, the PET radiotracer 18F-FDG is ineffective for improving early detection of pancreatic ductal adenocarcinoma (PDAC). An alternative strategy for early detection of pancreatic cancer involves visualization of high-grade pancreatic intraepithelial neoplasias (PanIN-3s), generally regarded as the noninvasive precursors of PDAC. The DNA damage response is known to be hyperactivated in late-stage PanINs. Therefore, we investigated whether the SPECT imaging agent 111In-anti-γH2AX-TAT allows visualization of the DNA damage repair marker γH2AX in PanIN-3s in an engineered mouse model of PDAC, to facilitate early detection of PDAC. Methods: Genetically engineered KPC (KRasLSL.G12D/+; p53LSL.R172H/+; PdxCre) mice were imaged with 18F-FDG and 111In-anti-γH2AX-TAT. The presence of PanIN/PDAC as visualized by histologic examination was compared with autoradiography and immunofluorescence. Separately, the survival of KPC mice imaged with 111In-anti-γH2AX-TAT was evaluated. Results: In KPC mouse pancreata, γH2AX expression was increased in high-grade PanINs but not in PDAC, corroborating earlier results obtained from human pancreas sections. Uptake of 111In-anti-γH2AX-TAT, but not 111In-IgG-TAT or 18F-FDG, within the pancreas correlated positively with the age of KPC mice, which correlated with the number of high-grade PanINs. 111In-anti-γH2AX-TAT localizes preferentially in high-grade PanIN lesions but not in established PDAC. Younger, non-tumor-bearing KPC mice that show uptake of 111In-anti-γH2AX-TAT in the pancreas survive for a significantly shorter time than mice with physiologic 111In-anti-γH2AX-TAT uptake. Conclusion:111In-anti-γH2AX-TAT imaging allows noninvasive detection of DNA damage repair signaling upregulation in preinvasive PanIN lesions and is a promising new tool to aid in the early detection and staging of pancreatic cancer.

Probing the limits of Q-tag bioconjugation of antibodies.

Site-selective labelling of antibodies (Abs) can circumvent problems from heterogeneity of conventional conjugation. Here, we evaluate the industrially-applied chemoenzymatic 'Q-tag' strategy based on transglutaminase-mediated (TGase) amide-bond formation in the generation of 89Zr-radiolabelled antibody conjugates. We show that, despite previously suggested high regioselectivity of TGases, in the anti-Her2 Ab Herceptin™ more precise native MS indicates only 70-80% functionalization at the target site (Q298H), in competition with modification at other sites, such as Q3H critically close to the CDR1 region.

Dual-isotope imaging allows in vivo immunohistochemistry using radiolabelled antibodies in tumours.

While radiolabelled antibodies have found great utility as PET and SPECT imaging agents in oncological investigations, a notable shortcoming of these agents is their propensity to accumulate non-specifically within tumour tissue. The degree of this non-specific contribution to overall tumour uptake is highly variable and can ultimately lead to false conclusions. Therefore, in an effort to obtain a reliable measure of inter-individual differences in non-specific tumour uptake of radiolabelled antibodies, we demonstrate that the use of dual-isotope imaging overcomes this issue, enables true quantification of epitope expression levels, and allows non-invasive in vivo immunohistochemistry. The approach involves co-administration of (i) an antigen-targeting antibody labelled with zirconium-89 (89Zr), and (ii) an isotype-matched non-specific control IgG antibody labelled with indium-111 (111In). As an example, the anti-HER2 antibody trastuzumab was radiolabelled with 89Zr, and co-administered intravenously together with its 111In-labelled non-specific counterpart to mice bearing human breast cancer xenografts with differing HER2 expression levels (MDA-MB-468 [HER2-negative], MDA-MB-231 [low-HER2], MDA-MB-231/H2N [medium-HER2], and SKBR3 [high-HER2]). Simultaneous PET/SPECT imaging using a MILabs Vector4 small animal scanner revealed stark differences in the intratumoural distribution of [89Zr]Zr-trastuzumab and [111In]In-IgG, highlighting regions of HER2-mediated uptake and non-specific uptake, respectively. Normalisation of the tumour uptake values and tumour-to-blood ratios obtained with [89Zr]Zr-trastuzumab against those obtained with [111In]In-IgG yielded values which were most strongly correlated (R = 0.94; P = 0.02) with HER2 expression levels for each breast cancer type determined by Western blot and in vitro saturation binding assays, but not non-normalised uptake values. Normalised intratumoural distribution of [89Zr]Zr-trastuzumab correlated well with intratumoural heterogeneity HER2 expression.

Targeting phosphatidylserine for radionuclide-based molecular imaging of apoptosis.

One major characteristic of programmed cell death (apoptosis) results in the increased expression of phosphatidylserine (PS) on the outer membrane of dying cells. Consequently, PS represents an excellent target for non-invasive imaging of apoptosis by single-photon emission computed tomography (SPECT) and positron emission tomography (PET). Annexin V is a 36 kDa protein which binds with high affinity to PS in the presence of Ca2+ ions. This makes radiolabeled annexins valuable apoptosis imaging agents for clinical and biomedical research applications for monitoring apoptosis in vivo. However, the use of radiolabeled annexin V for in vivo imaging of cell death has been met with a variety of challenges which have prevented its translation into the clinic. These difficulties include: complicated and time-consuming radiolabeling procedures, sub-optimal biodistribution, inadequate pharmacokinetics leading to poor tumour-to-blood contrast ratios, reliance upon Ca2+ concentrations in vivo, low tumor tissue penetration, and an incomplete understanding of what constitutes the best imaging protocol following induction of apoptosis. Therefore, new concepts and improved strategies for the development of PS-binding radiotracers are needed. Radiolabeled PS-binding peptides and various Zn(II) complexes as phosphate chemosensors offer an innovative strategy for radionuclide-based molecular imaging of apoptosis with PET and SPECT. Radiolabeled peptides and Zn(II) complexes provide several advantages over annexin V including better pharmacokinetics due to their smaller size, better availability, simpler synthesis and radiolabeling strategies as well as facilitated tissue penetration due to their smaller size and faster blood clearance profile allowing for optimized image contrast. In addition, peptides can be structurally modified to improve metabolic stability along with other pharmacokinetic and pharmacodynamic properties. The present review will summarize the current status of radiolabeled annexins, peptides and Zn(II) complexes developed as radiotracers for imaging apoptosis through targeting PS utilizing PET and SPECT imaging.

Translational molecular imaging in exocrine pancreatic cancer.

Effective treatment for pancreatic cancer remains challenging, particularly the treatment of pancreatic ductal adenocarcinoma (PDAC), which makes up more than 95% of all pancreatic cancers. Late diagnosis and failure of chemotherapy and radiotherapy are all too common, and many patients die soon after diagnosis. Here, we make the case for the increased use of molecular imaging in PDAC preclinical research and in patient management.

Neuromodulated Synaptic Plasticity on the SpiNNaker Neuromorphic System.

SpiNNaker is a digital neuromorphic architecture, designed specifically for the low power simulation of large-scale spiking neural networks at speeds close to biological real-time. Unlike other neuromorphic systems, SpiNNaker allows users to develop their own neuron and synapse models as well as specify arbitrary connectivity. As a result SpiNNaker has proved to be a powerful tool for studying different neuron models as well as synaptic plasticity-believed to be one of the main mechanisms behind learning and memory in the brain. A number of Spike-Timing-Dependent-Plasticity(STDP) rules have already been implemented on SpiNNaker and have been shown to be capable of solving various learning tasks in real-time. However, while STDP is an important biological theory of learning, it is a form of Hebbian or unsupervised learning and therefore does not explain behaviors that depend on feedback from the environment. Instead, learning rules based on neuromodulated STDP (three-factor learning rules) have been shown to be capable of solving reinforcement learning tasks in a biologically plausible manner. In this paper we demonstrate for the first time how a model of three-factor STDP, with the third-factor representing spikes from dopaminergic neurons, can be implemented on the SpiNNaker neuromorphic system. Using this learning rule we first show how reward and punishment signals can be delivered to a single synapse before going on to demonstrate it in a larger network which solves the credit assignment problem in a Pavlovian conditioning experiment. Because of its extra complexity, we find that our three-factor learning rule requires approximately 2× as much processing time as the existing SpiNNaker STDP learning rules. However, we show that it is still possible to run our Pavlovian conditioning model with up to 1 × 104 neurons in real-time, opening up new research opportunities for modeling behavioral learning on SpiNNaker.

Imaging of Claudin-4 in Pancreatic Ductal Adenocarcinoma Using a Radiolabelled Anti-Claudin-4 Monoclonal Antibody.

PURPOSE: Despite its widespread use, the positron emission tomography (PET) radiotracer 2-deoxy-2-[18F]fluoro-D-glucose ([18F]FDG) has been shown in clinical settings to be ineffective for improving early diagnosis of pancreatic ductal adenocarcinoma (PDAC). A promising biomarker for PDAC detection is the tight junction protein claudin-4. The purpose of this study was to evaluate a new single-photon emission computed tomography (SPECT) imaging agent, [111In]anti-claudin-4 mAb, with regard to its ability to allow visualisation of claudin-4 in a xenograft and a genetically engineered mouse model of PDAC. PROCEDURES: The ability of [111In]anti-claudin-4 mAb to selectively target claudin-4 was assessed using two human xenograft tumour models with differential claudin-4 status in mice. [111In]anti-claudin-4 mAb was also used to detect PDAC development in genetically engineered KPC mice. The PDAC status of these mice was confirmed with [18F]FDG-PET, magnetic resonance imaging (MRI), histology, and immunofluorescence microscopy. RESULTS: High uptake of [111In]anti-claudin-4 mAb was observed in PDAC xenografts in mice, reaching 16.9 ± 4.5 % of injected dose per gram (% ID/g) at 72 h post-injection. This uptake was mediated specifically by the expression of claudin-4. Uptake of [111In]anti-claudin-4 mAb also enabled clear visualisation of spontaneous PDAC formation in KPC mice. CONCLUSIONS: [111In]anti-claudin-4 mAb allows non-invasive detection of claudin-4 upregulation during development of PDAC and could potentially be used to aid in the early detection and characterisation of this malignancy.