Clinical and laboratory findings in scrub typhus associated Guillain-Barré syndrome in South Korea.

BACKGROUND AND AIMS: Scrub typhus is an endemic disease in the fall season that occurs in a limited number of places known as the Tsutsugamushi Triangle. Peripheral neuropathy is a common complication of scrub typhus. Herein, we encountered several patients with ascending paralysis after scrub typhus infection, who were diagnosed with Guillain-Barré syndrome (GBS). We aimed to investigate the clinical and laboratory characteristics of patients who developed GBS after scrub typhus. METHODS: Patients were retrospectively recruited from six nationwide tertiary centers in South Korea from January 2017 to December 2021. Patients who had been clinically diagnosed with GBS and confirmed to have scrub typhus via laboratory examination and/or the presence of an eschar before the onset of acute limb paralysis were included. The GBS-associated clinical and electrophysiological characteristics, outcomes, and scrub typhus-associated features were collected. RESULTS: Of the seven enrolled patients, six were female and one was male. The median time from scrub typhus infection to the onset of limb weakness was 6 (range: 2-14) days. All patients had eschar on their bodies. Four patients (57.1%) were admitted to the intensive care unit and received artificial ventilation for respiratory distress. At 6 months, the median GBS disability score was 2 (range, 1-4) points. INTERPRETATION: Patients with scrub typhus-associated GBS have a severe clinical presentation and require intensive treatment with additional immunotherapies. Therefore, GBS should be included in the differential diagnosis when peripheral neuropathies develop during scrub typhus treatment. Notably, scrub typhus is associated to GBS.

Prediction model for mild cognitive impairment in patients with type 2 diabetes using the autonomic function test.

OBJECTIVE: Type 2 diabetes mellitus (T2DM) is a risk factor for cognitive impairment, and reduced heart rate variability (HRV) has been correlated with cognitive impairment in elderly individuals. This study investigated risk factors and validated a predictive model for mild cognitive impairment (MCI) in patients with T2DM using an autonomic function test. METHODS: Patients with T2DM, 50-85 years of age, who attended the diabetes clinic at Gyeongsang National University Hospital between March 2018 and December 2019, were included. A total of 201 patients had been screened; we enrolled 124 patients according to the inclusion and exclusion criteria in this study. Cognitive function was assessed using the Montreal Cognitive Assessment-Korean version (MOCA-K); MCI was defined as a total MOCA-K score ≤ 23. Risk factors for MCI in patients with T2DM, including demographic- and diabetes-related factors, and autonomic function test results, were analyzed. Based on multivariate logistic regression, a nomogram was developed as a prediction model for MCI. RESULTS: Thirty-nine of 124 patients were diagnosed with MCI. Age, education, and decreased cardiovagal function were associated with a high risk for MCI, with cardiovagal function exerting the greatest influence. However, diabetes-related factors, such as glycemic control, duration of diabetes, or medications, were not associated with the risk for MCI. The nomogram demonstrated excellent discrimination (area under the curve, 0.832) and was well calibrated. CONCLUSION: Approximately one-third of patients had MCI; as such, carefully evaluating cognitive function in elderly T2DM patients with reduced HRV is important to prevent progression to dementia.

NGS data vectorization, clustering, and finding key codons in SARS-CoV-2 variations.

The rapid global spread and dissemination of SARS-CoV-2 has provided the virus with numerous opportunities to develop several variants. Thus, it is critical to determine the degree of the variations and in which part of the virus those variations occurred. Therefore, in this study, methods that could be used to vectorize the sequence data, perform clustering analysis, and visualize the results were proposed using machine learning methods. To conduct this study, a total of 224,073 cases of SARS-CoV-2 sequence data were collected through NCBI and GISAID, and the data were visualized using dimensionality reduction and clustering analysis models such as T-SNE and DBSCAN. The SARS-CoV-2 virus, which was first detected, was distinguished from different variations, including Omicron and Delta, in the cluster results. Furthermore, it was possible to examine which codon changes in the spike protein caused the variants to be distinguished using feature importance extraction models such as Random Forest or Shapely Value. The proposed method has the advantage of being able to analyse and visualize a large amount of data at once compared to the existing tree-based sequence data analysis. The proposed method was able to identify and visualize significant changes between the SARS-CoV-2 virus, which was first detected in Wuhan, China, in December 2019, and the newly formed mutant virus group. As a result of clustering analysis using sequence data, it was possible to confirm the formation of clusters among various variants in a two-dimensional graph, and by extracting the importance of variables, it was possible to confirm which codon changes played a major role in distinguishing variants. Furthermore, since the proposed method can handle a variety of data sequences, it can be used for all kinds of diseases, including influenza and SARS-CoV-2. Therefore, the proposed method has the potential to become widely used for the effective analysis of disease variations.

Reflex and Spontaneous Movements in Pediatric Patients with Brain Death.

BACKGROUND: Reflex and spontaneous movements are not uncommon in brain death patients. However, most studies have been conducted in adults, while reports in infants and children are rare. Thus, we aimed to evaluate the frequency and characteristics of these movements in pediatric patients declared as brain death. METHODS: Records of pediatric patients who were diagnosed as brain death from 15 hospitals in the Yeongnam region, South Korea, between January 2013 and September 2016 were analyzed. All body movements in patients who met the criteria for brain death as established by the Korea Medical Association were assessed by medical doctors and trained organ transplant coordinators. The frequency and characteristics of these movements were identified. Additionally, the demographic and clinical factors of the brain death patients with and without these movements were compared. RESULTS: A total of 31 patients who met the criteria for brain death were enrolled. Seven patients (22.6%) showed either reflex or spontaneous movements; six of them (85.7%) showed reflex movements only, and one patient (14.3%) showed both types of movements. The most common types of reflex movements were the flexor/extensor plantar response and isolated finger jerk. Four of seven patients (57.1%) showed a single movement pattern, while three (42.9%) showed two different movement patterns. CONCLUSION: It is essential for physicians who perform pediatric brain death examinations to recognize the frequency and characteristics of reflex and spontaneous movements, and this article may help in the accurate and prompt diagnosis of brain death.

Production of Prosaikogenin F, Prosaikogenin G, Saikogenin F and Saikogenin G by the Recombinant Enzymatic Hydrolysis of Saikosaponin and their Anti-Cancer Effect.

The saponins of Bupleurum falcatum L., saikosaponins, are the major components responsible for its pharmacological and biological activities. However, the anti-cancer effects of prosaikogenin and saikogenin, which are glycoside hydrolyzed saikosaponins, are still unknown due to its rarity in plants. In this study, we applied two recombinant glycoside hydrolases that exhibit glycoside cleavage activity with saikosaponins. The two enzymes, BglPm and BglLk, were cloned from Paenibacillus mucilaginosus and Lactobacillus koreensis, and exhibited good activity between 30-37 °C and pH 6.5-7.0. Saikosaponin A and D were purified and obtained from the crude B. falcatum L. extract using preparative high performance liquid chromatography technique. Saikosaponin A and D were converted into saikogenin F via prosaikogenin F, and saikogenin G via prosaikogenin G using enzyme transformation with high ß-glycosidase activity. The two saikogenin and two prosaikogenin compounds were purified using a silica column to obtain 78.1, 62.4, 8.3, and 7.5 mg of prosaikogenin F, prosaikogenin G, saikogenin F, and saikogenin G, respectively, each with 98% purity. The anti-cancer effect of the six highly purified saikosaponins was investigated in the human colon cancer cell line HCT 116. The results suggested that saikosaponins and prosaikogenins markedly inhibit the growth of the cancer cell line. Thus, this enzymatic technology could significantly improve the production of saponin metabolites of B. falcatum L.

New Non-Fullerene Acceptor with Extended Conjugation of Cyclopenta [2,1-b:3,4-b'] Dithiophene for Organic Solar Cells.

Herein, we design and characterize 9-heterocyclic ring non-fullerene acceptors (NFAs) with the extended backbone of indacenodithiophene by cyclopenta [2,1-b:3,4-b'] dithiophene (CPDT). The planar conjugated CPDT donor enhances absorption by reducing vibronic transition and charge transport. Developed NFAs with different end groups shows maximum absorption at approximately 790-850 nm in film. Because of the electronegative nature of the end-group, the corresponding acceptors showed deeper LUMO energy levels and red-shifted ultraviolet absorption. We investigate the crystallinity, film morphology, surface energy, and electronic as well as photovoltaic performance. The organic photovoltaic cells using novel NFAs with the halogen end groups fluorine or chlorine demonstrate better charge collection and faster exciton dissociation than photovoltaic cells using NFAs with methyl or lacking a substituent. Photovoltaic devices constructed from m-Me-ITIC with various end groups deliver power conversion efficiencies of 3.6-11.8%.

Simultaneously Passivating Cation and Anion Defects in Metal Halide Perovskite Solar Cells Using a Zwitterionic Amino Acid Additive.

Ionic defects (e.g., organic cations and halide anions), preferably residing along grain boundaries (GBs) and on perovskite film surfaces, are known to be a major source of the notorious environmental instability of perovskite solar cells (PeSCs). Although passivating ionic defects is desirable, previous approaches using Lewis base or acid molecules as additives suppress only the negatively or positively charged defects, thus leaving oppositely charged defects. In this work, both the cationic and anionic defects inside methyl ammonium lead tri-iodide (MAPbI3 ) are simultaneously passivated by introducing a zwitterionic form of the amino acid, L-alanine, into the precursor solution as an additive. L-alanine has both positive (NH3+ ) and negative (COO- ) functional groups at a specific solvent pH, thereby passivating both the cation and anion defects in MAPbI3 . The addition of L-alanine increases the grain size of the perovskite crystals and lengthens the charge carrier lifetime (τ > 1 µs), leading to improved power conversion efficiencies (PCEs) of 20.3% (from 18.3% without an additive) for small-area (4.64 mm2 ) devices and 15.6% (from 13.5%) for large-area submodules (9.06 cm2 ). More importantly, the authors' approach also significantly enhances the shelf storage and photoirradiation stabilities of PeSCs.

Development of carbapenem-based fluorogenic probes for the clinical screening of carbapenemase-producing bacteria.

This report describes the synthesis of a library of fluorogenic carbapenemase substrates consisting of carbapenem derivatives, fluorescence dyes, and active cleavable linkers and their evaluation for specifically detecting carbapenemase-producing organisms (CPOs). We synthesized a series of compounds having three different types of linkers such as benzyl ether, carbamate, and amine using hydroxymethyl carbapenem 7a and hydroxyallyl carbapenem 7b as key intermediates. Probe 1b exhibited high stability and a prompt turn-on fluorescence signal upon hydrolysis by carbapenemases. In particular, the screening of clinical samples indicated that the probe 1b exhibited excellent selectivity to the CPOs over other ß-lactamases or non-carbapenemase producing bacteria, which may be of clinical use for the rapid and accurate detection of CPOs for timely diagnosis and treatment.

Weekly ILI patient ratio change prediction using news articles with support vector machine.

BACKGROUND: Influenza continues to pose a serious threat to human health worldwide. For this reason, detecting influenza infection patterns is critical. However, as the epidemic spread of influenza occurs sporadically and rapidly, it is not easy to estimate the future variance of influenza virus infection. Furthermore, accumulating influenza related data is not easy, because the type of data that is associated with influenza is very limited. For these reasons, identifying useful data and building a prediction model with these data are necessary steps toward predicting if the number of patients will increase or decrease. On the Internet, numerous press releases are published every day that reflect currently pending issues. RESULTS: In this research, we collected Internet articles related to infectious diseases from the Centre for Health Protection (CHP), which is maintained the by Hong Kong Department of Health, to see if news text data could be used to predict the spread of influenza. In total, 7769 articles related to infectious diseases published from 2004 January to 2018 January were collected. We evaluated the predictive ability of article text data from the period of 2013-2018 for each of the weekly time horizons. The support vector machine (SVM) model was used for prediction in order to examine the use of information embedded in the web articles and detect the pattern of influenza spread variance. The prediction result using news text data with SVM exhibited a mean accuracy of 86.7 % on predicting whether weekly ILI patient ratio would increase or decrease, and a root mean square error of 0.611 on estimating the weekly ILI patient ratio. CONCLUSIONS: In order to remedy the problems of conventional data, using news articles can be a suitable choice, because they can help estimate if ILI patient ratio will increase or decrease as well as how many patients will be affected, as shown in the result of research. Thus, advancements in research on using news articles for influenza prediction should continue to be pursed, as the result showed acceptable performance as compared to existing influenza prediction researches.

Performance of a Novel Fluorogenic Assay for Detection of Carbapenemase-Producing Enterobacteriaceae from Bacterial Colonies and Directly from Positive Blood Cultures.

Rapid and accurate detection of carbapenemase-producing Enterobacteriaceae (CPE) is critical for appropriate treatment and infection control. We compared a rapid fluorogenic assay using a carbapenem-based fluorogenic probe with other phenotypic assays: modified carbapenem inactivation method (mCIM), Carba NP test (CNP), and carbapenemase inhibition test (CIT). A total of 217 characterized isolates of Enterobacteriaceae were included as follows: 63 CPE; 48 non-carbapenemase-producing carbapenem-resistant Enterobacteriaceae (non-CP-CRE); 53 extended-spectrum ß-lactamase producers; and 53 third-generation-cephalosporin-susceptible isolates. The fluorogenic assay using bacterial colonies (Fluore-C) was conducted by lysing the isolates followed by centrifugation and mixing the supernatant with fluorogenic probe. In addition, for the fluorogenic assay using spiked blood culture bottles (Fluore-Direct), pellets were obtained via the saponin preparation method, which can directly identify the pathogens using matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS). The fluorescence signal was measured over 50 min using a fluorometer. The fluorescent signal of CPE was significantly higher than that of non-CPE in both Fluore-C (median relative fluorescence units [RFU] [range], 5,814 [240 to 32,009] versus 804 [36 to 2,480], respectively; P < 0.0001) and Fluore-Direct (median RFU [range], 10,355 [1,689 to 31,463] versus 1,068 [428 to 2,155], respectively; P < 0.0001) tests. Overall, positive and negative percent agreements of Fluore-C, mCIM, CNP, CIT, and Fluore-Direct were 100% and 98.7%, 98.3% and 97.5%, 88.1% and 100%, 96.4% and 98.7%, and 98.3% and 98.1%, respectively. The relatively lower positive percent agreement (PPA) of CNP was mainly observed in OXA-type CPE. The fluorogenic assay showed excellent performance with bacterial colonies and also directly from positive blood cultures. We included many non-CP-CRE isolates for strict evaluation. The fluorogenic assay will be a useful tool for clinical microbiology laboratories.

Synthesis and Biological Evaluation of Disubstituted Pyrimidines as Selective 5-HT2C Agonists.

Here, we describe the synthesis of disubstituted pyrimidine derivatives and their biological evaluation as selective 5-HT2C agonists. To improve selectivity for 5-HT2C over other subtypes, we synthesized two series of disubstituted pyrimidines with fluorophenylalkoxy groups at either the 5-position or 4-position and varying cyclic amines at the 2-position. The in vitro cell-based assay and binding assay identified compounds 10a and 10f as potent 5-HT2C agonists. Further studies on selectivity to 5-HT subtypes and drug-like properties indicated that 2,4-disubstituted pyrimidine 10a showed a highly agonistic effect on the 5-HT2C receptor, with excellent selectivity, as well as exceptional drug-like properties, including high plasma and microsomal stability, along with low CYP inhibition. Thus, pyrimidine 10a could be considered a viable lead compound as a 5-HT2C selective agonist.

Cell-Penetrating Peptide-Patchy Deformable Polymeric Nanovehicles with Enhanced Cellular Uptake and Transdermal Delivery.

We herein propose a polymeric nanovehicle system that has the ability to remarkably improve cellular uptake and transdermal delivery. Cell-penetrating peptide-patchy deformable polymeric nanovehicles were fabricated by tailored coassembly of amphiphilic poly(ethylene oxide)- block-poly(ε-caprolactone) (PEO- b-PCL), mannosylerythritol lipid (MEL), and YGRKKRRQRRR-cysteamine (TAT)-linked MEL. Using X-ray diffraction, differential scanning calorimetry, and nuclear magnetic resonance analyses, we revealed that the incorporation of MEL having an asymmetric alkyl chain configuration was responsible for the deformable phase property of the vehicles. We also discovered that the nanovehicles were mutually attracted, exhibiting a gel-like fluid characteristic due to the dipole-dipole interaction between the hydroxyl group of MEL and the methoxy group of PEO- b-PCL. Coassembly of TAT-linked MEL with the deformable nanovehicles significantly enhanced cellular uptake due to macropinocytosis and caveolae-/lipid raft-mediated endocytosis. Furthermore, the in vivo skin penetration test revealed that our TAT-patchy deformable nanovehicles remarkably improved transdermal delivery efficiency.

Identification of Optically Active Pyrimidine Derivatives as Selective 5-HT2C Modulators.

A series of pyrimidine derivatives 4a-i were synthesized and evaluated for their binding affinities towards 5-HT2C receptors. With regard to designed molecules 4a-i, the influence of the size of alkyl ether and the absolute configuration of a stereogenic center on the 5-HT2C binding affinity and selectivity was studied. The most promising diasteromeric mixtures 4d and 4e were selected in the initial radioligand binding assay and they were further synthesized as optically active forms starting from optically active alcohols 5d and 5e, prepared by an enzymatic kinetic resolution. Pyrimidine analogue (R,R)-4e displayed an excellent 5-HT2C binding affinity with good selectivity values against a broad range of other 5-HT receptor subtypes.

Synthesis and biological evaluation of 2-substituted quinoline 6-carboxamides as potential mGluR1 antagonists for the treatment of neuropathic pain.

A series of 2-amino and 2-methoxy quinoline-6-carboxamide derivatives have been synthesized and their metabotropic glutamate receptor type 1 (mGluR1) antagonistic activities were evaluated in a functional cell-based assay. The compound 13c showed the highest potency with IC50 value of 2.16 µM against mGluR1. Finally, in vivo evaluation of 13c in the rat spinal nerve ligation (SNL) model exhibited weak analgesic effects with regard to both mechanical allodynia and cold allodynia.

Overcoming the Interfacial Photocatalytic Degradation of Nonfullerene Acceptor-Based Organic Photovoltaics by Introducing a UV-A-Insensitive Titanium Suboxide Layer.

Although recent dramatic advances in power conversion efficiencies (PCEs) have resulted in values over 19%, the poor photostability of organic photovoltaics (OPVs) has been a serious bottleneck to their commercialization. The photocatalytic effect, which is caused by incident ultraviolet-A (UV-A, 320-400 nm) light in the most commonly used zinc oxide (ZnOX) electron transport layer (ETL), significantly deteriorates the photostability of OPVs. In this work, we develop a new and facile method to enhance the photostability of nonfullerene acceptor-based OPVs by introducing UV-A-insensitive titanium suboxide (TiOX) ETL. Through an in-depth analysis of mass information at the interface between the ETL and photoactive layer, we confirm that the UV-A-insensitive TiOX suppresses the photocatalytic effect. The resulting device employing the TiOX ETL shows excellent photostability, obtaining 80% of the initial PCE for up to 200 h under 1 sun illumination, which is 10 times longer than that of the conventional ZnOX system (19 h).

Photopically Transparent Organic Solar Cells with Tungsten Oxide-Based Multilayer Electrodes.

The tailoring of the average photopic transmittance (APT) of transparent organic solar cells (T-OSCs) has been the greatest challenge in building-integrated photovoltaic applications for future smart solar windows to regulate indoor brightness, maintain a human circadian rhythm, and positively impact human emotions by allowing the observation of the external environment. However, a notorious trade-off exists between the APT and power conversion efficiency (PCE) of T-OSCs, mainly due to the absence of highly conductive and transparent top electrodes, which are a key building block determining the PCE and APT. Herein, we demonstrate a new tungsten oxide (WO3)-based multilayer as a highly conductive and transparent top electrode that provides an excellent APT while maintaining a high PCE in T-OSCs. With the assistance of optical simulation based on a transfer matrix method to calculate the optimum thicknesses of the multilayer electrodes, we achieve the best-performing T-OSC with a PCE of 7.0% and a full device APT of 46.7%, resulting in a high light utilization efficiency of 3.27%, which is superior to that of T-OSCs based on the same photoactive system. Furthermore, superior thermal stability at 85 °C in an N2 atmosphere is observed in WO3-based T-OSCs, maintaining 98% of the initial PCE after about 231 h. Our findings provide new insights into the development of T-OSCs with high efficiency and transparency.

New Ternary Blend Strategy Based on a Vertically Self-Assembled Passivation Layer Enabling Efficient and Photostable Inverted Organic Solar Cells.

Herein, a new ternary strategy to fabricate efficient and photostable inverted organic photovoltaics (OPVs) is introduced by combining a bulk heterojunction (BHJ) blend and a fullerene self-assembled monolayer (C60 -SAM). Time-of-flight secondary-ion mass spectrometry - analysis reveals that the ternary blend is vertically phase separated with the C60 -SAM at the bottom and the BHJ on top. The average power conversion efficiency - of OPVs based on the ternary system is improved from 14.9% to 15.6% by C60 -SAM addition, mostly due to increased current density (Jsc ) and fill factor -. It is found that the C60 -SAM encourages the BHJ to make more face-on molecular orientation because grazing incidence wide-angle X-ray scattering - data show an increased face-on/edge-on orientation ratio in the ternary blend. Light-intensity dependent Jsc data and charge carrier lifetime analysis indicate suppressed bimolecular recombination and a longer charge carrier lifetime in the ternary system, resulting in the enhancement of OPV performance. Moreover, it is demonstrated that device photostability in the ternary blend is enhanced due to the vertically self-assembled C60 -SAM that successfully passivates the ZnO surface and protects BHJ layer from the UV-induced photocatalytic reactions of the ZnO. These results suggest a new perspective to improve both performance and photostability of OPVs using a facial ternary method.

Efficient and Stable Quasi-2D Ruddlesden-Popper Perovskite Solar Cells by Tailoring Crystal Orientation and Passivating Surface Defects.

Solar cells (PSCs) with quasi-2D Ruddlesden-Popper perovskites (RPP) exhibit greater environmental stability than 3D perovskites; however, the low power conversion efficiency (PCE) caused by anisotropic crystal orientations and defect sites in the bulk RPP materials limit future commercialization. Herein, a simple post-treatment is reported for the top surfaces of RPP thin films (RPP composition of PEA2 MA4 Pb5 I16 = 5) in which zwitterionic n-tert-butyl-α-phenylnitrone (PBN) is used as the passivation material. The PBN molecules passivate the surface and grain boundary defects in the RPP and simultaneously induce vertical direction crystal orientations of the RPPs, which lead to efficient charge transport in the RPP photoactive materials. With this surface engineering methodology, the optimized devices exhibit a remarkably enhanced PCE of 20.05% as compared with the devices without PBN (≈17.53%) and excellent long-term operational stability with 88% retention of the initial PCE under continuous 1-sun irradiation for over 1000 h. The proposed passivation strategy provides new insights into the development of efficient and stable RPP-based PSCs.

Serum neurofilament and glial fibrillary acidic protein in idiopathic and seropositive transverse myelitis.

BACKGROUND: Serum levels of neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP) reflect the disease activity and disability in central nervous system (CNS) demyelinating diseases. However, the clinical significance of NfL and GFAP in idiopathic transverse myelitis (iTM), an inflammatory spinal cord disease with unknown underlying causes, remains unclear. This study aimed to investigate NfL and GFAP levels in iTM and their association with the clinical parameters compared with those in TM with disease-specific antibodies such as anti-aquaporin 4 or myelin oligodendrocyte glycoprotein antibodies (sTM). METHODS: We collected serum and clinical data of 365 patients with CNS inflammatory diseases from 12 hospitals. The serum NfL and GFAP levels were measured in patients with iTM (n = 37) and sTM (n = 39) using ultrasensitive single-molecule array assays. Regression analysis was performed to investigate the associations between serum levels of NfL and GFAP and the clinical parameters such as higher EDSS scores (EDSS ≥ 4.0). RESULTS: Mean NfL levels were not significantly different between iTM (50.29 pg/ml) and sTM (63.18 pg/ml) (p = 0.824). GFAP levels were significantly lower in iTM (112.34 pg/ml) than in sTM (3814.20 pg/ml) (p = 0.006). NfL levels correlated with expanded disability status scale (EDSS) scores in sTM (p = 0.001) but not in iTM (p = 0.824). Disease duration also correlated with higher EDSS scores in sTM (p = 0.017). CONCLUSION: NfL levels and disease duration correlated with EDSS scores in sTM, and GFAP levels could be a promising biomarker to differentiate iTM from sTM.

Two-Argument Activation Functions Learn Soft XOR Operations Like Cortical Neurons.

Neurons in the brain are complex machines with distinct functional compartments that interact nonlinearly. In contrast, neurons in artificial neural networks abstract away this complexity, typically down to a scalar activation function of a weighted sum of inputs. Here we emulate more biologically realistic neurons by learning canonical activation functions with two input arguments, analogous to basal and apical dendrites. We use a network-in-network architecture where each neuron is modeled as a multilayer perceptron with two inputs and a single output. This inner perceptron is shared by all units in the outer network. Remarkably, the resultant nonlinearities often produce soft XOR functions, consistent with recent experimental observations about interactions between inputs in human cortical neurons. When hyperparameters are optimized, networks with these nonlinearities learn faster and perform better than conventional ReLU nonlinearities with matched parameter counts, and they are more robust to natural and adversarial perturbations.