Abatacept is a CTLA-4-Ig fusion protein that binds to the costimulatory ligands CD80 and CD86 and blocks their interaction with the CD28 and CTLA-4 receptors expressed by T cells, therefore inhibiting T cell activation and function. Abatacept has shown clinical efficacy in treating some autoimmune diseases but has failed to show clinical benefit in other autoimmune conditions. The reasons for these disparate results are not clear and warrant further investigation of abatacept’s mode of action. Longitudinal specimens from the Immune Tolerance Network's A Cooperative Clinical Study of Abatacept in Multiple Sclerosis trial were used to examine the effects of abatacept treatment on the frequency and transcriptional profile of specific T cell populations in peripheral blood. We found that the relative abundance of CD4+ T follicular helper (Tfh) cells and regulatory T cells was selectively decreased in participants following abatacept treatment. Within both cell types, abatacept reduced the proportion of activated cells expressing CD38 and ICOS and was associated with decreased expression of genes that regulate cell-cycle and chromatin dynamics during cell proliferation, thereby linking changes in costimulatory signaling to impaired activation, proliferation, and decreased abundance. All cellular and molecular changes were reversed following termination of abatacept treatment. These data expand upon the mechanism of action of abatacept reported in other autoimmune diseases and identify new transcriptional targets of CD28-mediated costimulatory signaling in human regulatory T and Tfh cells, further informing on its potential use in diseases associated with dysregulated Tfh activity.

Human activity recognition (HAR) is essential to many contextaware applications in mobile and ubiquitous computing. A human's physical activity can be decomposed into a sequence of simple actions or body movements, corresponding to what we denote as mid-level features. Such mid-level features ("leg up, " 'leg down, " "leg still, " ...), which we contrast to high-level activities ("walking, " "sitting, " ...) and low-level features (raw sensor readings), can be developed manually. While proven to be effective, this manual approach is not scalable and relies heavily on human domain expertise. In this paper, we address this limitation by proposing a machine learning method, At-triNet, based on deep belief networks. Our AttriNet method automatically constructs mid-level features and outperforms baseline approaches. Interestingly, we show in experiments that some of the features learned by AttriNet highly correlate with manually defined features. This result demonstrates the potential of using deep learning techniques for learning mid-level features that are semantically meaningful, as a replacement to handcrafted features. Generally, this empirical finding provides an improved understanding of deep learning methods for HAR. CCS CONCEPTS • Computing methodologies → Artificial intelligence; Activity recognition and understanding.

Human Activity Recognition (HAR) is a prominent application in mobile computing and Internet of Things (IoT) that aims to detect human activities based on multimodal sensor signals generated as a result of diverse body movements. Human physical activities are typically composed of simple actions (such as “arm up”, “arm down”, “arm curl”, etc.), referred to as semantic features. Such abstract semantic features, in contrast to high-level activities (“walking”, “sitting”, etc.) and low-level signals (raw sensor readings), can be developed manually to assist activity recognition. Although effective, this manual approach relies heavily on human domain expertise and is not scalable. In this paper, we address this limitation by proposing a machine learning method, SemNet, based on deep belief networks. SemNet automatically constructs semantic features representative of the axial bodily movements. Experimental results show that SemNet outperforms baseline approaches and is capable of learning features that highly correlate with manually defined semantic attributes. Furthermore, our experiments using a different model, namely deep convolutional LSTM, on household activities illustrate the broader applicability of semantic attribute interpretation to diverse deep neural network approaches. These empirical results not only demonstrate that such a deep learning technique is semantically meaningful and superior to its handcrafted counterpart, but also provides a better understanding of the deep learning methods that are used for Human Activity Recognition.

Background: Preclinical studies indicate that cannabidiol (CBD), the primary nonaddictive component of cannabis, has a wide range of reported pharmacological effects such as analgesic and anxiolytic actions; however, the exact mechanisms of action for these effects have not been examined in chronic osteoarthritis (OA). Similar to other chronic pain syndromes, OA pain can have a significant affective component characterized by mood changes. Serotonin (5-HT) is a neurotransmitter implicated in pain, depression, and anxiety. Pain is often in comorbidity with mood and anxiety disorders in patients with OA. Since primary actions of CBD are analgesic and anxiolytic, in this first in vivo positron emission tomography (PET) imaging study, we investigate the interaction of CBD with serotonin 5-HT1A receptor via a combination of in vivo neuroimaging and behavioral studies in a well-validated OA animal model. Methods: The first aim of this study was to evaluate the target involvement, including the evaluation of modulation by acute administration of CBD, or a specific target antagonist/agonist intervention, in control animals. The brain 5-HT1A activity/availability was assessed via in vivo dynamic PET imaging (up to 60 min) using a selective 5-HT1A radioligand ([18F]MeFWAY). Tracer bindings of 17 ROIs were evaluated based on averaged SUVR values over the last 10 min using CB as the reference region. We subsequently examined the neurochemical and behavioral alterations in OA animals (induction with monosodium iodoacetate (MIA) injection), as compared to control animals, via neuroimaging and behavioral assessment. Further, we examined the effects of repeated low-dose CBD treatment on mechanical allodynia (von Frey tests) and anxiety-like (light/dark box tests, L/D), depressive-like (forced swim tests, FST) behaviors in OA animals, as compared to after vehicle treatment. Results: The tracer binding was significantly reduced in control animals after an acute dose of CBD administered intravenously (1.0 mg/kg, i.v.), as compared to that for baseline. This binding specificity to 5-HT1A was further confirmed by a similar reduction of tracer binding when a specific 5-HT1A antagonist WAY1006235 was used (0.3 mg/kg, i.v.). Mice subjected to the MIA-induced OA for 13-20 days showed a decreased 5-HT1A tracer binding (25% to 41%), consistent with the notion that 5-HT1A plays a role in the modulation of pain in OA. Repeated treatment with CBD administered subcutaneously (5 mg/kg/day, s.c., for 16 days after OA induction) increased 5-HT1A tracer binding, while no significant improvement was observed after vehicle. A trend of increased anxiety or depressive-like behavior in the light/dark box or forced swim tests after OA induction, and a decrease in those behaviors after repeated low-dose CBD treatment, are consistent with the anxiolytic action of CBD through 5HT1A receptor activation. There appeared to be a sex difference: females seem to be less responsive at the baseline towards pain stimuli, while being more sensitive to CBD treatment. Conclusion: This first in vivo PET imaging study in an OA animal model has provided evidence for the interaction of CBD with the serotonin 5-HT1A receptor. Behavioral studies with more pharmacological interventions to support the target involvement are needed to further confirm these critical findings.

No abstract