Although biomedical applications of carbon nanotubes have been intensively studied in recent years, its sister, graphene, has been rarely explored in biomedicine. In this work, for the first time we study the in vivo behaviors of nanographene sheets (NGS) with polyethylene glycol (PEG) coating by a fluorescent labeling method. In vivo fluorescence imaging reveals surprisingly high tumor uptake of NGS in several xenograft tumor mouse models. Distinctive from PEGylated carbon nanotubes, PEGylated NGS shows several interesting in vivo behaviors including highly efficient tumor passive targeting and relatively low retention in reticuloendothelial systems. We then utilize the strong optical absorbance of NGS in the near-infrared (NIR) region for in vivo photothermal therapy, achieving ultraefficient tumor ablation after intravenous administration of NGS and low-power NIR laser irradiation on the tumor. Furthermore, no obvious side effect of PEGylated NGS is noted for the injected mice by histology, blood chemistry, and complete blood panel analysis in our pilot toxicity study. Although a lot more efforts are required to further understand the in vivo behaviors and the long-term toxicology of this new type of nanomaterials, our work is the first success of using carbon nanomaterials for efficient in vivo photothermal therapy by intravenous administration and suggests the great promise of graphene in biomedical applications, such as cancer treatment.

Figure 17. Optimization of graphene-based photothermal agents. (a) AFM images of different graphene derivatives: while nGO-PEG and nRGO-PEG showed similar ultrasmall sizes at about 20−30 nm, the size of RGO-PEG was about 60−70 nm. Insets are photos of the respective solutions. RGO-PEG and nRGO-PEG showed much enhanced optical absorbance as compared to nGO-PEG. (b) The blood circulation of GO derivatives measured by collecting blood from mice iv injected with 125 I labeled nGO-PEG, nRGO-PEG, and RGO-PEG at various time points (n = 3). (c) The biodistribution of GO derivatives in 4T1 tumor-bearing mice 2 days after injection. The radioactivities in tissue and blood samples were determined by a gamma counter. (d) The 4T1 tumor growth curves of mice after various treatments indicated. The laser irradiation was conducted at the power density of 0.15 W/cm 2 for 5 min. (e) Survival of tumor-bearing mice after various treatments indicated. Reprinted with permission from ref 169.

Owing to their unique physical and chemical properties, graphene and its derivatives such as graphene oxide (GO), reduced graphene oxide (RGO) and GO-nanocomposites have attracted tremendous interest in many different fields including biomedicine in recent years. With every atom exposed on its surface, single-layered graphene shows ultra-high surface area available for efficient molecular loading and bioconjugation, and has been widely explored as novel nano-carriers for drug and gene delivery. Utilizing the intrinsic near-infrared (NIR) optical absorbance, in vivo graphene-based photothermal therapy has been realized, achieving excellent anti-tumor therapeutic efficacy in animal experiments. A variety of inorganic nanoparticles can be grown on the surface of nano-graphene, obtaining functional graphene-based nanocomposites with interesting optical and magnetic properties useful for multi-modal imaging and imaging-guided cancer therapy. Moreover, significant efforts have also been devoted to study the behaviors and toxicology of functionalized nano-graphene in animals. It has been uncovered that both surface chemistry and sizes play key roles in controlling the biodistribution, excretion, and toxicity of nano-graphene. Biocompatibly coated nano-graphene with ultra-small sizes can be cleared out from body after systemic administration, without rendering noticeable toxicity to the treated mice. In this review article, we will summarize the latest progress in this rapidly growing field, and discuss future prospects and challenges of using graphene-based materials for theranostic applications.

The mechanistic target of rapamycin (mTOR) pathway integrates diverse environmental inputs, including immune signals and metabolic cues, to direct T cell fate decisions1. Activation of mTOR, comprised of mTORC1 and mTORC2 complexes, delivers an obligatory signal for proper activation and differentiation of effector CD4+ T cells2,3, whereas in the regulatory T cell (Treg) compartment, the Akt-mTOR axis is widely acknowledged as a crucial negative regulator of Treg de novo differentiation4–8 and population expansion9. However, whether mTOR signaling affects the homeostasis and function of Tregs remains largely unexplored. Here we show that mTORC1 signaling is a pivotal positive determinant of Treg function. Tregs have elevated steady-state mTORC1 activity compared to naïve T cells. Signals via T cell receptor (TCR) and IL-2 provide major inputs for mTORC1 activation, which in turn programs suppressive function of Tregs. Disruption of mTORC1 through Treg-specific deletion of the essential component Raptor leads to a profound loss of Treg suppressive activity in vivo and development of a fatal early-onset inflammatory disorder. Mechanistically, Raptor/mTORC1 signaling in Tregs promotes cholesterol/lipid metabolism, with the mevalonate pathway particularly important for coordinating Treg proliferation and upregulation of suppressive molecules CTLA-4 and ICOS to establish Treg functional competency. In contrast, mTORC1 does not directly impact the expression of Foxp3 or anti- and pro-inflammatory cytokines in Tregs, suggesting a non-conventional mechanism for Treg functional regulation. Lastly, we provide evidence that mTORC1 maintains Treg function partly through inhibiting the mTORC2 pathway. Our results demonstrate that mTORC1 acts as a fundamental ‘rheostat’ in Tregs to link immunological signals from TCR and IL-2 to lipogenic pathways and functional fitness, and highlight a central role of metabolic programming of Treg suppressive activity in immune homeostasis and tolerance.

Graphene has emerged as interesting nanomaterials with promising applications in a range of fields including biomedicine. In this work, for the first time we study the long-term in vivo biodistribution of (125)I-labeled nanographene sheets (NGS) functionalized with polyethylene glycol (PEG) and systematically examine the potential toxicity of graphene over time. Our results show that PEGylated NGS mainly accumulate in the reticuloendothelial system (RES) including liver and spleen after intravenous administration and can be gradually cleared, likely by both renal and fecal excretion. PEGylated NGS do not cause appreciable toxicity at our tested dose (20 mg/kg) to the treated mice in a period of 3 months as evidenced by blood biochemistry, hematological analysis, and histological examinations. Our work greatly encourages further studies of graphene for biomedical applications.

In this work, a nanoscale reduced graphene oxide-iron oxide nanoparticle (RGO-IONP) complex is noncovalently functionalized with polyethylene glycol (PEG), obtaining a RGO-IONP-PEG nanocomposite with excellent physiological stability, strong NIR optical absorbance, and superparamagnetic properties. Using this theranostic nanoprobe, in-vivo triple modal fluorescence, photoacoustic, and magnetic resonance imaging are carried out, uncovering high passive tumor targeting, which is further used for effective photothermal ablation of tumors in mice.

The stringent requirements for low-latency and privacy of the emerging high-stake applications with intelligent devices such as drones and smart vehicles make the cloud computing inapplicable in these scenarios. Instead, edge machine learning becomes increasingly attractive for performing training and inference directly at network edges without sending data to a centralized data center. This stimulates a nascent field termed as federated learning for training a machine learning model on computation, storage, energy and bandwidth limited mobile devices in a distributed manner. To preserve data privacy and address the issues of unbalanced and non-IID data points across different devices, the federated averaging algorithm has been proposed for global model aggregation by computing the weighted average of locally updated model at each selected device. However, the limited communication bandwidth becomes the main bottleneck for aggregating the locally computed updates. We thus propose a novel over-the-air computation based approach for fast global model aggregation via exploring the superposition property of a wireless multiple-access channel. This is achieved by joint device selection and beamforming design, which is modeled as a sparse and low-rank optimization problem to support efficient algorithms design. To achieve this goal, we provide a differenceof-convex-functions (DC) representation for the sparse and lowrank function to enhance sparsity and accurately detect the fixed-rank constraint in the procedure of device selection. A DC algorithm is further developed to solve the resulting DC program with global convergence guarantees. The algorithmic advantages and admirable performance of the proposed methodologies are demonstrated through extensive numerical results.

We present Spider, a large-scale, complex and cross-domain semantic parsing and textto-SQL dataset annotated by 11 college students. It consists of 10,181 questions and 5,693 unique complex SQL queries on 200 databases with multiple tables, covering 138 different domains. We define a new complex and cross-domain semantic parsing and textto-SQL task where different complex SQL queries and databases appear in train and test sets. In this way, the task requires the model to generalize well to both new SQL queries and new database schemas. Spider is distinct from most of the previous semantic parsing tasks because they all use a single database and the exact same programs in the train set and the test set. We experiment with various state-of-the-art models and the best model achieves only 12.4% exact matching accuracy on a database split setting. This shows that Spider presents a strong challenge for future research. Our dataset and task are publicly available at https://yale-lily. github.io/spider.