Chemical mutagenesis of a GPCR ligand: Detoxifying “inflammo-attraction” to direct therapeutic stem cell migration

A transplanted stem cell’s engagement with a pathologic niche is the first step in its restoring homeostasis to that site. Inflammatory chemokines are constitutively produced in such a niche; their binding to receptors on the stem cell helps direct that cell’s “pathotropism.” Neural stem cells (NSCs), which express CXCR4, migrate to sites of CNS injury or degeneration in part because astrocytes and vasculature produce the inflammatory chemokine CXCL12. Binding of CXCL12 to CXCR4 (a G protein-coupled receptor, GPCR) triggers repair processes within the NSC. Although a tool directing NSCs to where needed has been long-sought, one would not inject this chemokine in vivo because undesirable inflammation also follows CXCL12–CXCR4 coupling. Alternatively, we chemically “mutated” CXCL12, creating a CXCR4 agonist that contained a strong pure binding motif linked to a signaling motif devoid of sequences responsible for synthetic functions. This synthetic dual-moity CXCR4 agonist not only elicited more extensive and persistent human NSC migration and distribution than did native CXCL 12, but induced no host inflammation (or other adverse effects); rather, there was predominantly reparative gene expression. When co-administered with transplanted human induced pluripotent stem cell-derived hNSCs in a mouse model of a prototypical neurodegenerative disease, the agonist enhanced migration, dissemination, and integration of donor-derived cells into the diseased cerebral cortex (including as electrophysiologically-active cortical neurons) where their secreted cross-corrective enzyme mediated a therapeutic impact unachieved by cells alone. Such a “designer” cytokine receptor-agonist peptide illustrates that treatments can be controlled and optimized by exploiting fundamental stem cell properties (e.g., “inflammo-attraction”).

A transplanted stem cell’s engagement with a pathologic niche is the first step in cell-mediated restoration of homeostasis to that region, whether by cell replacement, protection, gene delivery, milieu alteration, toxin neutralization, or remodeling (1–4). Not surprisingly, the more host terrain covered by the stem cells, the greater their impact. We and others found that a propensity for neural stem cells (NSCs) to home in vivo to acutely injured or actively degenerating central nervous system (CNS) regions—a property called “pathotropism” (1–12), now viewed as central to stem cell biology—is undergirded, at least in part, by the presence of chemokine receptors on the NSC surface, enabling them to follow concentration gradients of inflammatory cytokines constitutively elaborated by pathogenic processes and expressed by reactive astrocytes and injured vascular endothelium within the pathologic niche (5–9). This engagement of NSC receptors was first described for the prototypical chemokine receptor CXCR4 (C-X-C chemokine receptor type 4; also known as fusin or cluster of differentiation-184 [CD184]) and its unique natural cognate agonist ligand, the inflammatory chemokine CXCL12 (C-X-C motif chemokine ligand-12; also known as stromal cell-derived factor 1α [SDF-1α]) (5), but has since been described for many chemokine receptor-agonist pairings (6–9). Chemokine receptors belong to a superfamily that is characterized by seven transmembrane GDP-binding protein-coupled receptors (GPCRs) (13–21). In addition to their role in mediating inflammatory reactions and immune responses (22, 23), these receptors and their agonists are components of the regulatory axes for hematopoiesis and organogenesis in other systems (21, 24). Therefore, it is not surprising that binding of CXCL12 to CXCR4 mediates not only an inflammatory response, but also triggers within the NSC a series of intracellular processes associated with migration (as well as proliferation, differentiation, survival, and, during early brain development, proper neuronal lamination) (10).A tool directing therapeutic NSCs to where they are needed has long been sought in regenerative medicine (11, 12). While it was appealing to contemplate electively directing reparative NSCs to any desired area by emulating this chemoattractive property through the targeted injection of exogenous recombinant inflammatory cytokines, it ultimately seemed inadvisable to risk increasing toxicity in brains already characterized by excessive and usually inimical inflammation from neurotraumatic or neurodegenerative processes. However, the notion of engaging the homing function of these NSC-borne receptors without triggering that receptor’s undesirable downstream inflammatory signaling [particularly given that the NSCs themselves can exert a therapeutic antiinflammatory action in the diseased region (1, 2)] seemed a promising heretofore unexplored “workaround.”There had already been an impetus to examine the structure–function relationships of CXCR4, known to be the entry route into cells for HIV-1, in order to create CXCR4 antagonists that block viral infection (25–30). Antagonists of CXCR4 were also devised to forestall hematopoietic stem cells from homing to the bone marrow, hence prolonging their presence in the peripheral blood (31) to treat blood dyscrasias. An agonist, however, particularly one with discrete and selective actions, had not been contemplated. In other words, if CXCL12 could be stripped of its undesirable actions while preserving its tropic activity, an ideal chemoattractant would be derived.Based on the concept that CXCR4’s functions are conveyed by two distinct molecular “pockets”—one mediating binding (i.e., allowing a ligand to engage CXCR4) and the other mediating signaling (i.e., enabling a ligand, after binding, to trigger CXCR4-mediated intracellular cascades that promote not only inflammation but also migration) (13–18)—we performed chemical mutagenesis that should optimize binding while narrowing the spectrum of signaling. We created a simplified de novo peptide agonist of CXCR4 that contained a strong pure binding motif derived from CXCR4’s strongest ligand, viral macrophage inflammatory protein-II (vMIP-II) and linked it to a truncated signaling motif (only 8 amino acid residues) derived from the N terminus of native CXCL12 (19, 20). This synthetic dual-moiety CXCR4 agonist, which is devoid of a large portion of CXCL12’s native sequence (presumably responsible for undesired functions such as inflammation) not only elicited (with great specificity) more extensive and long-lasting human NSC (hNSC) migration and distribution than native CXCL12 (overcoming migratory barriers), but induced no host inflammation (or other adverse effects). Furthermore, because all of the amino acids in the binding motif were in a D-chirality, rendering the peptide resistant to enzymatic degradation, persistence of this benign synthetic agonist in vivo was prolonged. The hNSC’s gene ontology expression profile was predominantly reparative in contrast to inflammatory as promoted by native CXCL12. When coadministered with transplanted human induced pluripotent stem cell (hiPSC)-derived hNSCs (hiPSC derivatives are now known to have muted migration) in a mouse model of a prototypical neurodegenerative disease [the lethal neuropathic lysosomal storage disorder (LSD) Sandhoff disease (29), where hiPSC-hNSC migration is particularly limited], the synthetic agonist enhanced migration, dissemination, and integration of donor-derived cells into the diseased cortex (including as electrophysiologically active cortical neurons), where their secreted cross-corrective enzyme could mediate a histological and functional therapeutic impact in a manner unachieved by transplanting hiPSC-derived cells alone.In introducing such a “designer” cytokine receptor agonist, we hope to offer proof-of-concept that stem cell-mediated treatments can be controlled and optimized by exploiting fundamental stem cell properties (e.g., “inflammo-attraction”) to alter a niche and augment specific actions. Additionally, when agonists are strategically designed, the various functions of chemokine receptors (and likely other GCPRs) may be divorced. We demonstrate that such a strategy might be used safely and effectively to direct cells to needed regions and broaden their chimerism. We discuss the future implications and uses within the life sciences of such a chemical engineering approach.     

Natural gas shortages during the “coal-to-gas” transition in China have caused a large redistribution of air pollution in winter 2017

The Chinese “coal-to-gas” and “coal-to-electricity” strategies aim at reducing dispersed coal consumption and related air pollution by promoting the use of clean and low-carbon fuels in northern China. Here, we show that on top of meteorological influences, the effective emission mitigation measures achieved an average decrease of fine particulate matter (PM_2.5) concentrations of ∼14% in Beijing and surrounding areas (the “2+26” pilot cities) in winter 2017 compared to the same period of 2016, where the dispersed coal control measures contributed ∼60% of the total PM_2.5 reductions. However, the localized air quality improvement was accompanied by a contemporaneous ∼15% upsurge of PM_2.5 concentrations over large areas in southern China. We find that the pollution transfer that resulted from a shift in emissions was of a high likelihood caused by a natural gas shortage in the south due to the coal-to-gas transition in the north. The overall shortage of natural gas greatly jeopardized the air quality benefits of the coal-to-gas strategy in winter 2017 and reflects structural challenges and potential threats in China’s clean-energy transition.

The “airpocalypse” in China stems from a multitude of air pollutants (1–3) that are associated with significant climate and health effects (4–8). Eliminating the severe fine particulate matter (PM_2.5; particulate matter with a diameter smaller than 2.5 µm) pollution smog has been perceived as a national priority, with the establishment of a coal-consumption cap that required the share of coal in the national primary energy mix to drop to below 65% in 2017 by a transition to cleaner natural gas and nonfossil energy sources (9). Coal reduction is, however, the crux of China’s air pollution control (10). Fig. 1 shows the coal-control roadmap of China from 2010 through 2030. As of 2017, over half of the world’s coal consumption has occurred in China, accounting for ∼60% of the country’s primary energy consumption (11). After effective coal reductions from the power sector and key energy-intensive industries (phase I), the further coal control in China has been focusing on the reduction of dispersed coal use in residential and small industrial facilities (phase II). The dispersed coal (the so-called “Sanmei” in Chinese) refers to raw coal, usually a high-polluting fuel with high ash residue, burned in noncentralized combustion facilities without end-of-pipe air pollutant treatment. The residential dispersed coal combustion in vast rural areas has been estimated to be a major contributor to high PM_2.5 exposure and premature mortality in China (7, 12–14).Open in a separate windowFig. 1.The roadmap of coal control in China from 2010 to 2030. The light orange shaded area shows the share of coal in the primary energy mix for 2010–2030 (11) with three-phased coal controls (phase I: in the power section and key energy-intensive industries; phase II: toward dispersed coal reductions; and phase III: pertaining to clean-energy development). The black bars indicate the dispersed coal consumption for 2016–2018 and 2020 with source decomposition for sectoral changes (17). The doughnut charts show the primary energy structures of China in 2010, 2017, and 2030, respectively (11, 37). Note: Energy data for Hong Kong, Macau, and Taiwan are not included here.Since early 2017, a series of clean-heating actions have been implemented in Beijing and its neighboring provinces, especially in the “2+26” cities located along the air pollution transport channel of the Beijing–Tianjin–Hebei (BTH) region (Beijing, Tianjin, and 26 other cities in Hebei, Shanxi, Shandong, and Henan provinces) (15, 16). The major dispersed coal control measures include replacing traditional household coal-fired stoves with wall-mounted natural gas heaters (“coal-to-gas”) or electric stoves (“coal-to-electricity”) and eliminating the small industrial coal-fired steam boilers and construction materials industrial kilns (brick, ceramic, and lime industries) (17). The changes in coal consumption related to individual measures are given in Fig. 1.Combining the ground-based measurements, localized emission estimates, and chemical transport model simulations, we show that, although air pollution in the northern 2+26 pilot cities has been greatly improved by the emission mitigation measures, the coal-to-gas action in winter 2017 has caused a severe natural gas shortage in the rest of China (18, 19 and SI Appendix, section S1), which necessitated the use of more polluting alternative energy sources and led to a deteriorated air quality in the gas-shortage regions.     

From the Cover: A Middle Eocene lowland humid subtropical “Shangri-La” ecosystem in central Tibet

Tibet’s ancient topography and its role in climatic and biotic evolution remain speculative due to a paucity of quantitative surface-height measurements through time and space, and sparse fossil records. However, newly discovered fossils from a present elevation of ∼4,850 m in central Tibet improve substantially our knowledge of the ancient Tibetan environment. The 70 plant fossil taxa so far recovered include the first occurrences of several modern Asian lineages and represent a Middle Eocene (∼47 Mya) humid subtropical ecosystem. The fossils not only record the diverse composition of the ancient Tibetan biota, but also allow us to constrain the Middle Eocene land surface height in central Tibet to ∼1,500 ± 900 m, and quantify the prevailing thermal and hydrological regime. This “Shangri-La”–like ecosystem experienced monsoon seasonality with a mean annual temperature of ∼19 °C, and frosts were rare. It contained few Gondwanan taxa, yet was compositionally similar to contemporaneous floras in both North America and Europe. Our discovery quantifies a key part of Tibetan Paleogene topography and climate, and highlights the importance of Tibet in regard to the origin of modern Asian plant species and the evolution of global biodiversity.

The Tibetan Plateau, once thought of as entirely the product of the India–Eurasia collision, is known to have had significant complex relief before the arrival of India early in the Paleogene (1–3). This large region, spanning ∼2.5 million km², is an amalgam of tectonic terranes that impacted Asia long before India’s arrival (4, 5), with each accretion contributing orographic heterogeneity that likely impacted climate in complex ways. During the Paleogene, the Tibetan landscape comprised a high (>4 km) Gangdese mountain range along the southern margin of the Lhasa terrane (2), against which the Himalaya would later rise (6), and a Tanghula upland on the more northerly Qiangtang terrane (7). Separating the Lhasa and Qiangtang blocks is the east–west trending Banggong-Nujiang Suture (BNS), which today hosts several sedimentary basins (e.g., Bangor, Nyima, and Lunpola) where >4 km of Cenozoic sediments have accumulated (8). Although these sediments record the climatic and biotic evolution of central Tibet, their remoteness means fossil collections have been hitherto limited. Recently, we discovered a highly diverse fossil assemblage in the Bangor Basin. These fossils characterize a luxuriant seasonally wet and warm Shangri-La forest that once occupied a deep central Tibetan valley along the BNS, and provide a unique opportunity for understanding the evolutionary history of Asian biodiversity, as well as for quantifying the paleoenvironment of central Tibet.^*Details of the topographic evolution of Tibet are still unclear despite decades of investigation (4, 5). Isotopic compositions of carbonates recovered from sediments in some parts of central Tibet have been interpreted in terms of high (>4 km) Paleogene elevations and aridity (9, 10), but those same successions have yielded isolated mammal (11), fish (12), plant (13–18), and biomarker remains (19) more indicative of a low (≤3-km) humid environment, but how low is poorly quantified. Given the complex assembly of Tibet, it is difficult to explain how a plateau might have formed so early and then remained as a surface of low relief during subsequent compression from India (20). Recent evidence from a climate model-mediated interpretation of palm fossils constrains the BNS elevation to below 2.3 km in the Late Paleogene (16), but more precise paleoelevation estimates are required. Further fossil discoveries, especially from earlier in the BNS sedimentary records, would document better the evolution of the Tibetan biota, as well as informing our understanding of the elevation and climate in an area that now occupies the center of the Tibetan Plateau.Our work shows that the BNS hosted a diverse subtropical ecosystem at ∼47 Ma, and this means the area must have been both low and humid. The diversity of the fossil flora allows us to 1) document floristic links to other parts of the Northern Hemisphere, 2) characterize the prevailing paleoclimate, and 3) quantify the elevation at which the vegetation grew. We propose that the “high and dry” central Tibet inferred from some isotope paleoaltimetry (9, 10) reflects a “phantom” elevated paleosurface (20) because fractionation over the bounding mountains allowed only isotopically light moist air to enter the valley, giving a false indication of a high elevation (21).     

An engineered 4-1BBL fusion protein with “activity on demand”

Engineered cytokines are gaining importance in cancer therapy, but these products are often limited by toxicity, especially at early time points after intravenous administration. 4-1BB is a member of the tumor necrosis factor receptor superfamily, which has been considered as a target for therapeutic strategies with agonistic antibodies or using its cognate cytokine ligand, 4-1BBL. Here we describe the engineering of an antibody fusion protein, termed F8-4-1BBL, that does not exhibit cytokine activity in solution but regains biological activity on antigen binding. F8-4-1BBL bound specifically to its cognate antigen, the alternatively spliced EDA domain of fibronectin, and selectively localized to tumors in vivo, as evidenced by quantitative biodistribution experiments. The product promoted a potent antitumor activity in various mouse models of cancer without apparent toxicity at the doses used. F8-4-1BBL represents a prototype for antibody-cytokine fusion proteins, which conditionally display “activity on demand” properties at the site of disease on antigen binding and reduce toxicity to normal tissues.

Cytokines are immunomodulatory proteins that have been considered for pharmaceutical applications in the treatment of cancer patients (1–3) and other types of disease (2). There is a growing interest in the use of engineered cytokine products as anticancer drugs, capable of boosting the action of T cells and natural killer (NK) cells against tumors (3, 4), alone or in combination with immune checkpoint inhibitors (3, 5–7).Recombinant cytokine products on the market include interleukin-2 (IL-2) (Proleukin) (8, 9), IL-11 (Neumega) (10, 11), tumor necrosis factor (TNF; Beromun) (12), interferon (IFN)-α (Roferon A, Intron A) (13, 14), IFN-β (Avonex, Rebif, Betaseron) (15, 16), IFN-γ (Actimmune) (17), granulocyte colony-stimulating factor (Neupogen) (18), and granulocyte macrophage colony-stimulating factor (Leukine) (19, 20). The recommended dose is typically very low (often <1 mg/d) (21–23), as cytokines may exert biological activity in the subnanomolar concentration range (24). Various strategies have been proposed to develop cytokine products with improved therapeutic index. Protein PEGylation or Fc fusions may lead to prolonged circulation time in the bloodstream, allowing the administration of low doses of active payload (25, 26). In some implementations, cleavable polyethylene glycol polymers may be considered, yielding prodrugs that regain activity at later time points (27). Alternatively, tumor-homing antibody fusions have been developed, since the preferential concentration of cytokine payloads at the tumor site has been shown in preclinical models to potentiate therapeutic activity, helping spare normal tissues (28–34). Various antibody-cytokine fusions are currently being investigated in clinical trials for the treatment of cancer and of chronic inflammatory conditions (reviewed in refs. 2, 33, 35–37).Antibody-cytokine fusions display biological activity immediately after injection into patients, which may lead to unwanted toxicity and prevent escalation to therapeutically active dosage regimens (9, 22, 38). In the case of proinflammatory payloads (e.g., IL-2, IL-12, TNF-α), common side effects include hypotension, nausea, and vomiting, as well as flu-like symptoms (24, 39–42). These side effects typically disappear when the cytokine concentration drops below a critical threshold, thus providing a rationale for slow-infusion administration procedures (43). It would be highly desirable to generate antibody-cytokine fusion proteins with excellent tumor-targeting properties and with “activity on demand”— biological activity that is conditionally gained on antigen binding at the site of disease, helping spare normal tissues.Here we describe a fusion protein consisting of the F8 antibody specific to the alternatively spliced extra domain A (EDA) of fibronectin (44, 45) and of murine 4-1BBL, which did not exhibit cytokine activity in solution but could regain potent biological activity on antigen binding. The antigen (EDA+ fibronectin) is conserved from mouse to man (46), is virtually undetectable in normal adult tissues (with the exception of the placenta, endometrium, and some vessels in the ovaries), but is expressed in the majority of human malignancies (44, 45, 47, 48). 4-1BBL, a member of the TNF superfamily (49), is expressed on antigen-presenting cells (50, 51) and binds to its receptor, 4-1BB, which is up-regulated on activated cytotoxic T cells (52), activated dendritic cells (52), activated NK and NKT cells (53), and regulatory T cells (54). Signaling through 4-1BB on cytotoxic T cells protects them from activation-induced cell death and skews the cells toward a more memory-like phenotype (55, 56).We engineered nine formats of the F8-4-1BBL fusion protein, one of which exhibited superior performance in quantitative biodistribution studies and conditional gain of cytokine activity on antigen binding. The antigen-dependent reconstitution of the biological activity of the immunostimulatory payload represents an example of an antibody fusion protein with “activity on demand.” The fusion protein was potently active against different types of cancer without apparent toxicity at the doses used. The EDA of fibronectin is a particularly attractive antigen for cancer therapy in view of its high selectivity, stability, and abundant expression in most tumor types (44, 45, 47, 48).     

Characterizing the “fungal shunt”: Parasitic fungi on diatoms affect carbon flow and bacterial communities in aquatic microbial food webs

Microbial interactions in aquatic environments profoundly affect global biogeochemical cycles, but the role of microparasites has been largely overlooked. Using a model pathosystem, we studied hitherto cryptic interactions between microparasitic fungi (chytrid Rhizophydiales), their diatom host Asterionella, and cell-associated and free-living bacteria. We analyzed the effect of fungal infections on microbial abundances, bacterial taxonomy, cell-to-cell carbon transfer, and cell-specific nitrate-based growth using microscopy (e.g., fluorescence in situ hybridization), 16S rRNA gene amplicon sequencing, and secondary ion mass spectrometry. Bacterial abundances were 2 to 4 times higher on individual fungal-infected diatoms compared to healthy diatoms, particularly involving Burkholderiales. Furthermore, taxonomic compositions of both diatom-associated and free-living bacteria were significantly different between noninfected and fungal-infected cocultures. The fungal microparasite, including diatom-associated sporangia and free-swimming zoospores, derived ∼100% of their carbon content from the diatom. By comparison, transfer efficiencies of photosynthetic carbon were lower to diatom-associated bacteria (67 to 98%), with a high cell-to-cell variability, and even lower to free-living bacteria (32%). Likewise, nitrate-based growth for the diatom and fungi was synchronized and faster than for diatom-associated and free-living bacteria. In a natural lacustrine system, where infection prevalence reached 54%, we calculated that 20% of the total diatom-derived photosynthetic carbon was shunted to the parasitic fungi, which can be grazed by zooplankton, thereby accelerating carbon transfer to higher trophic levels and bypassing the microbial loop. The herein termed “fungal shunt” can thus significantly modify the fate of photosynthetic carbon and the nature of phytoplankton–bacteria interactions, with implications for diverse pelagic food webs and global biogeochemical cycles.

Parasitism is one of the most common consumer strategies on Earth (1–3). Recently, it has also been identified as one of the dominating interactions within the planktonic interactome (4, 5), and yet parasites remain poorly considered in analyses of trophic interactions and element cycling in aquatic systems (6, 7). The foundation of trophic interactions in plankton communities is set by single-cell phytoplankton, which contributes almost half of the world’s primary production (8). According to our common understanding, the newly fixed carbon (C) is channeled either through the microbial loop, classical food web, or viral shunt, which supports the growth of heterotrophic bacteria and nanoflagellates, zooplankton and higher trophic levels, or viruses, respectively (9). However, fungi, particularly fungal microparasites, are rarely considered as contributors to C and nutrient cycling, although they are present and active in diverse aquatic environments (10–12).Members of the fungal division Chytridiomycota, referred to as chytrids, can thrive as microparasites on phytoplankton cells in freshwater (11, 13, 14) and marine systems (15–17), infecting up to 90% of the phytoplankton host population (18–21). A recent concept, called mycoloop, describes parasitic chytrids as an integral part of aquatic food webs (22). Energy and organic matter are thereby transferred from large, often inedible phytoplankton to chytrid zoospores, which are consumed by zooplankton (23–27). Hence, parasitic chytrids establish a novel trophic link between phytoplankton and zooplankton. Our understanding of element cycling and microbial interactions during chytrid epidemics, however, remains sparse. For instance, the cell-to-cell C transfer from single phytoplankton cells to their directly associated chytrids has not been quantified to date. Moreover, the relationship between parasitic chytrids and heterotrophic bacteria is largely undescribed.Phytoplankton cells release substantial amounts of dissolved organic C (DOC) (28), whereby up to 50% of photosynthetic C is consumed as DOC by bacteria (29–32). Thus, bacterial communities are intimately linked to phytoplankton abundances and production (33). Phytoplankton–bacteria interactions are particularly strong within the phycosphere, the region immediately surrounding individual phytoplankton cells (33–35), where nutrient concentrations are several-folds higher compared to the ambient water (36, 37), and nutrient assimilation rates of phytoplankton-associated bacteria are at least twice as fast as those of their free-living counterparts (38, 39). Importantly, parasitic chytrids may distort these phytoplankton–bacteria interactions within and outside the phycosphere since they modulate substrate and nutrient availabilities and presumably also bacterial activity and community composition. The effects of this distortion are virtually unresolved, but the few available data indicate that chytrid infections alter the composition and concentration of DOC (40), while abundances of free-living bacteria increase (25, 40) or remain unchanged (24).To disentangle phytoplankton–fungi–bacteria interactions at a microspatial single-cell scale—the scale at which phytoplankton, fungi, and bacteria intimately interact—we used one of the few existing model pathosystems, composed of the freshwater diatom Asterionella formosa, the chytrid Rhizophydiales sp., and coenriched populations of heterotrophic bacteria. Our methodology included dual stable-isotope incubations (¹³C-bicarbonate and ¹⁵N-nitrate), single-cell–resolution secondary ion mass spectrometry (SIMS) (IMS 1280 and NanoSIMS 50L), 16S rRNA gene/16S rRNA sequencing, microscopy (e.g., fluorescence in situ hybridization [FISH]), and nutrient analyses. We particularly focused on the initial C transfer from the phytoplankton host to parasitic chytrids, which we term the “fungal shunt,” as part of the mycoloop. The objectives were twofold: 1) quantifying the transfer of photosynthetic C from phytoplankton cells to infectious chytrids, cell-associated bacteria, and free-living bacteria and 2) characterizing the effect of parasitic fungi on bacterial populations, considering bacterial abundances, bacterial–diatom attachment, single-cell activity rates, and community composition. The obtained data challenge the common perception of aquatic microbial food webs by demonstrating the significant role that parasitic fungi can play in microbial community structure, interactions, and element cycling during phytoplankton growth.     

Rules of contact inhibition of locomotion for cells on suspended nanofibers

Contact inhibition of locomotion (CIL), in which cells repolarize and move away from contact, is now established as a fundamental driving force in development, repair, and disease biology. Much of what we know of CIL stems from studies on two-dimensional (2D) substrates that do not provide an essential biophysical cue—the curvature of extracellular matrix fibers. We discover rules controlling outcomes of cell–cell collisions on suspended nanofibers and show them to be profoundly different from the stereotyped CIL behavior on 2D substrates. Two approaching cells attached to a single fiber do not repolarize upon contact but rather usually migrate past one another. Fiber geometry modulates this behavior; when cells attach to two fibers, reducing their freedom to reorient, only one cell repolarizes on contact, leading to the cell pair migrating as a single unit. CIL outcomes also change when one cell has recently divided and moves with high speed—cells more frequently walk past each other. Our computational model of CIL in fiber geometries reproduces the core qualitative results of the experiments robustly to model parameters. Our model shows that the increased speed of postdivision cells may be sufficient to explain their increased walk-past rate. We also identify cell–cell adhesion as a key mediator of collision outcomes. Our results suggest that characterizing cell–cell interactions on flat substrates, channels, or micropatterns is not sufficient to predict interactions in a matrix—the geometry of the fiber can generate entirely new behaviors.

Cell migration is an essential component of various physiological processes such as morphogenesis, wound healing, and metastasis (1). Cell–cell interactions in which cell–cell contact reorients cell polarity are necessary for the correct function of many developmental events (2). One of the earliest such interactions known was termed “contact inhibition of locomotion” (CIL) by Abercombie and Heaysman over five decades ago in chick fibroblasts cultured on flat two-dimensional (2D) substrates (2–4). In CIL, two approaching cells isolated from the rest of the cell population first make contact, followed by protrusion inhibition at the site of contact, which leads to cell repolarization through formation of new protrusions away from the site of contact. Subsequently, cells migrate away from each other in the direction of newly formed protrusions (1). This sequence can, however, be altered in specific conditions such as metastasis in which a loss of CIL allows malignant cells to invade fibroblast cultures—this is a loss of CIL between different cell types (heterotypic CIL) (4, 5). Recent work has also begun to identify the molecular players that initiate and regulate CIL, including Rac activity, microtubules, Eph/Ephrin binding, and E- and N-cadherin expression (6–10).CIL is most commonly studied and analyzed on flat 2D substrates using several invasion and collision assays (2, 3, 11). By contrast, cells traveling in matrix in vivo are constrained to move along narrow fibers. A common shortcoming in featureless 2D assays is thus the inability to study CIL under natural constraints (11–13). Recently, micropatterned substrates have been used to understand restricted motility, developing one-dimensional (1D) collision assays where cell migration is constrained to straight lines, allowing for a greater occurrence of cell–cell collisions to quantify rates and outcomes of different types of cell–cell interactions (11, 13–15). These interactions do not necessarily resemble the stereotyped CIL behavior. Broadly, experiments and simulations (16–18) have observed the following: 1) the classical stereotype of CIL with two cells contacting head-on, with both cells repolarizing (referred to as “reversal” or “mutual CIL”); 2) after a head-on collision, only one cell reverses (“training” or “nonmutual CIL”); and 3) cells manage to crawl past or over one another, exchanging positions (“walk past” or “sliding”). Within the well-studied neural-crest cell explants, walk past is extremely rare (11), but it can occur in epithelial cells, especially in those that have been metastatically transformed or that have decreased E-cadherin expression (15).Both 2D substrates and micropatterned stripes provide controllable and reproducible environments but neither fully models the details of in vivo native cellular environments, which consist of extracellular matrices (ECM) of fibrous proteins, with these fibers having different radii. Our earlier in vitro recapitulation of the effects of fiber curvature showed that both protrusive and migratory behavior is sensitive to fiber diameter (19–21). Furthermore, we have shown that suspended, flat 2D ribbons do not capture the protrusive behavior observed on suspended round fibers (19); thus, we wanted to inquire if the CIL rules developed on 1D collision and 2D assays extend to contextually relevant fibrous environments. To understand CIL in fibrous environments that mimic native ECM, we use suspended and aligned nanofiber networks to study CIL behavior in NIH/3T3 fibroblast cell–cell pairs exhibiting two distinct elongated morphologies: spindle, attached to a single fiber and parallel cuboidal, attached to two fibers (22). We further investigate the effect of cell division on CIL by studying the encounters of cells that have recently divided (daughter cells) with other cells; these recently divided cells are much faster, consistent with earlier work (23). Our work allows us to determine the types and rates of cell–cell contact outcomes—the “rules of CIL”—in a biologically relevant system with a controlled geometry. These rules are radically different from the known stereotypical behavior in 2D assays, but the essential features of these rules emerge robustly from a minimal computational model of CIL in confined geometries.     

Prespacers formed during primed adaptation associate with the Cas1–Cas2 adaptation complex and the Cas3 interference nuclease–helicase

For Type I CRISPR-Cas systems, a mode of CRISPR adaptation named priming has been described. Priming allows specific and highly efficient acquisition of new spacers from DNA recognized (primed) by the Cascade-crRNA (CRISPR RNA) effector complex. Recognition of the priming protospacer by Cascade-crRNA serves as a signal for engaging the Cas3 nuclease–helicase required for both interference and primed adaptation, suggesting the existence of a primed adaptation complex (PAC) containing the Cas1–Cas2 adaptation integrase and Cas3. To detect this complex in vivo, we here performed chromatin immunoprecipitation with Cas3-specific and Cas1-specific antibodies using cells undergoing primed adaptation. We found that prespacers are bound by both Cas1 (presumably, as part of the Cas1–Cas2 integrase) and Cas3, implying direct physical association of the interference and adaptation machineries as part of PAC.

CRISPR-Cas systems of adaptive immunity provide prokaryotes with resistance against bacteriophages and plasmids (1–4). They consist of CRISPR DNA arrays and cas genes. Functionally, CRISPR defense can be subdivided into the interference and adaptation steps. The interference step involves specific recognition of regions in foreign nucleic acids, named protospacers, based on their complementarity to CRISPR arrays spacers followed by their destruction (5). The CRISPR adaptation step leads to integration of new spacers into the array (6, 7), forming inheritable memory that allows the entire lineage of cells derived from a founder that acquired a particular spacer to do away with genetic invaders carrying matching protospacers (8).Both interference and adaptation can be subdivided into multiple steps. For interference to occur, the CRISPR array is transcribed from a promoter located in the upstream leader region. The resulting pre-CRISPR RNA (pre-crRNA) is processed into short CRISPR RNAs (crRNAs), each containing a spacer flanked by repeat fragments (9). Individual crRNAs are bound by Cas proteins forming the effector complex, which is capable of recognizing sequences complementary to the spacer part of crRNA (10). Upon protospacer recognition, the target is destroyed either by a protein component of the effector complex or by additional recruitable Cas nucleases (3, 11–14). In a well-studied Type I-E CRISPR-Cas system of Escherichia coli, the effector comprises a multisubunit Cascade protein complex bound to a crRNA (11, 12, 15). The complementary interaction of Cascade-bound crRNA with a target protospacer leads to localized protospacer DNA melting and formation of an R-loop complex, where the crRNA spacer is annealed to the protospacer “target” strand, while the opposing “nontarget” strand is displaced and is present in a single-stranded form (16, 17). To avoid potentially suicidal recognition of CRISPR array spacers from which crRNAs originate, target recognition and R-loop complex formation require, in addition to complementarity with the crRNA spacer, the presence of a three-nucleotide long PAM (protospacer adjacent motif) preceding the protospacer (15, 18, 19). For E. coli type I-E system, the consensus PAM sequence is 5′-AAG-3′ on the nontarget strand. Some other trinucleotides also allow target recognition, though with decreased efficiency (15, 20). Below, we will refer to consensus PAM as “PAM^AAG.” The Cas3 nuclease-helicase is recruited to the R-loop complex and is responsible for target destruction (21–24). Cas3 first introduces a single-stranded break in the nontarget protospacer strand 11 to 15 nucleotides downstream of the PAM on the nontarget strand (25). Next, Cas3 unwinds and cleaves DNA in the 3′-5′ direction from the PAM (26–29). In vitro, Cas3-dependent degradation of DNA at the other side of the protospacer was also detected (16). Bidirectional Cas3-dependent degradation of DNA was also detected in vivo (30). The details of Cas3 “molecular gymnastics” required for such bidirectional destruction of DNA around the R-loop complex are not known.The main proteins of CRISPR adaptation are Cas1 and Cas2. In vitro, these proteins interact with each other, and the resulting complex is capable of inserting spacer-sized fragments in substrate DNA molecules containing at least one CRISPR repeat and a repeat-proximal leader region (31, 32). In the course of spacer integration, the Cas1–Cas2 complex first catalyzes a direct nucleophilic attack by the 3′-OH end of the incoming spacer at a phosphodiester bond between the leader and the first repeat in the top CRISPR strand (32, 33). This reaction proceeds via concurrent cleavage of the leader-repeat junction and covalent joining of one spacer strand to the 5′ end of the repeat. Subsequently, the 3′-OH on the second spacer strand attacks the phosphodiester bond at the repeat-spacer junction in the bottom CRISPR strand leading to full integration (32, 33). As a result, an intermediate with the newly incorporated spacer flanked by single-stranded repeat sequences is formed (32, 34). The gaps are filled in by a DNA polymerase, possibly DNA polymerase I (35).When overexpressed, E. coli Cas1 and Cas2 can integrate new spacers into the array in the absence of other Cas proteins (7, 36). During such “naive” adaptation, ∼50% of newly acquired spacers are selected from sequences flanked by the 5′-AAG-3′ trinucleotide, that is, consensus interference-proficient PAM^AAG. It thus follows that at least 50% of spacers acquired by Cas1 and Cas2 alone will be defensive during the interference step. The adaptation process must be tightly controlled, activated in the presence of the infecting mobile genetic elements, and directed toward foreign DNA, for otherwise, spacers acquired from host DNA will lead to suicidal self-interference. The details of the activation of CRISPR adaptation upon the entry of foreign DNA into the cell remain elusive. Some data indicate that active replication and/or a small size of phage or plasmid DNA may be responsible for a preferential selection of spacers from these molecules compared to selection of self-targeting spacers from host chromosomes (19). In addition, DNA repair/recombination signals present in host DNA, but lacking in foreign DNA may also increase the bias of the adaptation machinery to the latter (37).The bias of spacer acquisition machinery toward foreign DNA does not have to be significant, for acquisition of a self-targeting spacer by an infected cell will lead to the demise of such a cell in an act of altruism that would help control the spread of the infectious agent through the population. In contrast, acquisition of interference-proficient spacers from foreign DNA may allow the infected cell to survive, clear the infection, and endow its progeny with inheritable resistance—clearly an advantageous trait.To overcome CRISPR resistance, viruses and plasmids accumulate “escaper” mutations in the targeted protospacer or its PAM (36, 38). Given that the acquisition of protective spacers in infected cells is likely to be a rare event and the ease with which escaper mutations accumulate, the complex multistage CRISPR defense could become costly and ineffective (39). To increase the efficiency of CRISPR defense and counter the spread of mobile genetic elements with escaper mutations, CRISPR-Cas systems have evolved a specialized mode of spacer acquisition referred to as “primed adaptation” or “priming” (36, 40–47). Unlike the naive adaptation, in Type I CRISPR-Cas systems, priming requires, in addition to Cas1 and Cas2, a Cascade charged with crRNA recognizing the foreign target and the Cas3 nuclease–helicase. Spacers acquired during priming originate almost exclusively from DNA located in cis with the protospacer initially recognized by the effector complex (referred to hereafter as the “priming protospacer” or “PPS”). Furthermore, 90% or more of spacers acquired during priming by the I-E system of E. coli originate from protospacers with PAM^AAG and are therefore capable of efficient interference. Another hallmark of primed adaptation is the following: spacers acquired from DNA located at different sides of the PPS map to opposite DNA strands. The mapping of spacers acquired during naive adaptation shows no strand bias (48). Thus, the strand bias of spacers acquired during priming is probably related to Cas3 nuclease activity; however, exact details are lacking.The overall yield of spacers acquired during priming is increased when the PPS is imperfectly matched with a Cascade-bound crRNA spacer or when the PAM of the PPS is suboptimal (49). Thus, escaper protospacers serve as PPS, and priming initiated by inefficient recognition of such protospacers allows cells to quickly update their immunological memory by specific and efficient acquisition of additional interference-proficient spacers from mobile genetic elements that accumulated escaper mutations to earlier acquired spacers.The exact molecular mechanism of primed adaptation is not fully understood. Clearly, it should involve tight coordination between suboptimal interference against escaper targets and the spacer acquisition process. The DNA fragments produced by Cas3, a nuclease responsible for target degradation during interference, may feed primed adaptation, directly or indirectly, providing a functional link between the interference and adaptation arms of the CRISPR-Cas response. Based on results of in vitro experiments, it has been proposed that Cas3-generated degradation products may be used as substrates for the generation of prespacers (50)—DNA fragments that can be incorporated by the Cas1–Cas2 complex into arrays. However, no Cas3-generated products were detected in cells undergoing interference only, suggesting that Cas3 may degrade DNA to very short, subspacer length products (30). On the other hand, mutations abolishing the Cas3 nuclease activity lead to very little primed adaptation, indicating that priming requires the Cas3 nuclease activity (51). A possible way out from this impasse would be the existence of a “priming complex” that includes both Cas1–Cas2 and Cas3 and is responsible for the generation of prespacers by the Cas1–Cas2 complex from DNA along which Cas3 moves. Single-molecule analysis supports the existence of such a complex and even suggests that PPS-bound Cascade may be part of the priming complex (52). Here, we show that both Cas1–Cas2 and Cas3 associate with the same set of prespacers in cells undergoing primed adaptation, functionally linking CRISPR interference and adaptation machineries during priming. We also investigate the phenomenon of strand bias of spacer acquisition during priming and show that this bias does not depend on the orientation of PPS.     

From the Cover: Discovery of a body-wide photosensory array that matures in an adult-like animal and mediates eye–brain-independent movement and arousal

The ability to respond to light has profoundly shaped life. Animals with eyes overwhelmingly rely on their visual circuits for mediating light-induced coordinated movements. Building on previously reported behaviors, we report the discovery of an organized, eye-independent (extraocular), body-wide photosensory framework that allows even a head-removed animal to move like an intact animal. Despite possessing sensitive cerebral eyes and a centralized brain that controls most behaviors, head-removed planarians show acute, coordinated ultraviolet-A (UV-A) aversive phototaxis. We find this eye–brain-independent phototaxis is mediated by two noncanonical rhabdomeric opsins, the first known function for this newly classified opsin-clade. We uncover a unique array of dual-opsin–expressing photoreceptor cells that line the periphery of animal body, are proximal to a body-wide nerve net, and mediate UV-A phototaxis by engaging multiple modes of locomotion. Unlike embryonically developing cerebral eyes that are functional when animals hatch, the body-wide photosensory array matures postembryonically in “adult-like animals.” Notably, apart from head-removed phototaxis, the body-wide, extraocular sensory organization also impacts physiology of intact animals. Low-dose UV-A, but not visible light (ocular-stimulus), is able to arouse intact worms that have naturally cycled to an inactive/rest-like state. This wavelength selective, low-light arousal of resting animals is noncanonical-opsin dependent but eye independent. Our discovery of an autonomous, multifunctional, late-maturing, organized body-wide photosensory system establishes a paradigm in sensory biology and evolution of light sensing.

Light sensing has independently evolved multiple times and has profoundly shaped life. The ability to process light information in distinct ways and respond to a changing light environment can dramatically shape physiology and fitness of life forms. Movement, triggered by light, is one of the most fundamental responses in nature (1). Among metazoans, a wide variety of animals are known to show coordinated motion in response to light stimuli. So far, this is overwhelmingly known to be mediated through the animal eyes. In fact, eye-driven light sensing and taxis has been extensively studied across phyla. Interestingly, motion in metazoans can also be mediated through eye-independent or extraocular (EO) photoreception (2–5). However, our conceptual and mechanistic grasp on how coordinated movement can be triggered and controlled through EO light-sensing systems is extremely limited. Moreover, the few prominent examples of EO phototaxis have all been reported in life forms/developmental stages completely lacking eyes or possessing only rudimentary eyes (2, 5). Almost nothing is known about sensitive EO light-sensing systems capable of triggering coordinated motion that may coexist with sensitive eyes in a single organism.There are intriguing reports of photoreceptor molecules that are expressed in locations other than conventional eyes, including in unusual structures seen in polyclad flatworms, clitellate segmented worms, crustaceans, cephalopods, and fishes. However, the functions of such structures remain elusive (6–13). A single organism may indeed possess multiple, independent light-responsive systems, both eye based as well as eye independent (13–17), but the functions rarely overlap. “Nonvisual”/EO sensory systems like pineal glands and deep-brain photoreceptors across vertebrates and retinal ganglion cells in mammals (18–21) have been reported. However, these sensory systems are generally known to perform “nonvisual” functions like maintaining circadian rhythms and modulating behavior (22–25). Here, we report an EO phototactic network that can independently trigger coordinated movement just like what the eye-based (ocular) system can, while also having its own distinctive role even when the eyes are present.Do highly sensitive EO phototactic systems coexist and function in life forms that have prominent eye-based networks as well? How would such a system operate? What would be the mechanistic framework and the functional consequence of such an eye-independent light-sensory system? Planarian flatworms offer a fascinating opportunity to explore such a paradigm. Planarians are highly light aversive and have well-developed ocular cerebral eyes (eyes connected to a centralized ganglion) that process light stimuli and guide behavior like feeding, escape, and predation (26–30). In fact, the planarian ocular network is highly sensitive and capable of surprisingly complex processing (17). These eye-mediated behaviors are reliant on an organized, cephalized bilobed brain, a prominent example of a “primitive” brain in evolution (17, 31, 32). Indeed, the brain is required for most locomotive behaviors including thermotaxis, chemotaxis, and eye-mediated phototaxis including the ability to discriminate closely related light stimuli, shown by these animals (17, 32–34). However, planarians also show dramatic, eye–brain-independent light-induced movements (17, 35). Even after sudden decapitation (removal of both eyes and brain), worms are able to acutely respond to ultraviolet-A (UV-A) light and show seemingly coordinated movement away from light (17, 35). While such eye–brain-independent behavior has long fascinated biologists, almost nothing is known about how this dramatic behavior is mediated (17, 35–39). It is also not clear what would be the physiological role of such an acutely sensitive EO sensory network capable of triggering coordinated movement, especially since planarians do have a well-developed ocular network.Here, we show how such an acute response to light is mediated by an organism removed of its “primary” light-sensory organ and brain. We report the discovery of photoreceptor molecules as well as a widespread but organized network of photoreceptor cells that are required for this acute eye-independent UV-A light response. Intriguingly, this entire multiscale sensory system from photoreceptors to the network of cells arises and matures postembryonically, in an “adult-like” organism. This developmental trajectory is distinct from that of the cerebral eyes, which develop embryonically. We also demonstrate that while both the eyes as well the EO network led to coordinated movements relying on the same locomotion machinery, the physiological consequences of the ocular and EO sensory responses can be divergent. Intact planarians periodically go into “sleep-like” resting phases, in which their activity diminishes and sensory perception reduces (40). Strikingly, we find that the EO sensory system in these intact animals can override the natural “rest”-activity cycles and is able to acutely photoactivate and arouse even resting worms. This is distinct from the ocular network that becomes dormant during the “rest phase.” Our work illustrates an unprecedented level of organization and complexity in form and function of an acutely sensitive EO light-sensory system that matures and functions in parallel to the ocular network.     

Increasing risk of another Cape Town “Day Zero” drought in the 21st century

Three consecutive dry winters (2015–2017) in southwestern South Africa (SSA) resulted in the Cape Town “Day Zero” drought in early 2018. The contribution of anthropogenic global warming to this prolonged rainfall deficit has previously been evaluated through observations and climate models. However, model adequacy and insufficient horizontal resolution make it difficult to precisely quantify the changing likelihood of extreme droughts, given the small regional scale. Here, we use a high-resolution large ensemble to estimate the contribution of anthropogenic climate change to the probability of occurrence of multiyear SSA rainfall deficits in past and future decades. We find that anthropogenic climate change increased the likelihood of the 2015–2017 rainfall deficit by a factor of five to six. The probability of such an event will increase from 0.7 to 25% by the year 2100 under an intermediate-emission scenario (Shared Socioeconomic Pathway 2-4.5 [SSP2-4.5]) and to 80% under a high-emission scenario (SSP5-8.5). These results highlight the strong sensitivity of the drought risk in SSA to future anthropogenic emissions.

The Day Zero Cape Town drought was one of the worst water crises ever experienced in a metropolitan area (1, 2). Droughts are a regular occurrence in southwestern South Africa (SSA), having occurred during the late 1920s, early 1970s, and, more recently, during 2003–2004 (Fig. 1 A and B). However, the extended winter (April to September [AMJJAS]) 3-y rainfall deficit (Fig. 1 A and B; SI Appendix, Fig. S1) which drove the 2015–2017 Cape Town drought (2–8) was exceptional over the last century (4, 9). Storage in reservoirs supplying water to 3.7 million people in the Cape Town metropolitan area dropped to about 20% of capacity in May 2018. As a consequence, strict water-usage restrictions were implemented to delay water levels reaching 13.5%, the level at which much of the city’s municipal supply would have been disconnected (7), a scenario referred to as “Day Zero” by the municipal water authorities (7). Above-average winter rain over the rest of the 2018 austral winter allowed Cape Town to avoid the Day Zero scenario.Open in a separate windowFig. 1.(A) Mean 2015–2017 AMJJAS rainfall anomaly relative to 1921–1970. The dashed (continuous) line denotes negative anomalies beyond 1 (1.5) SD. (B) Time series of the observed (GPCC, blue; CRU, red) 3-y running mean AMJJAS WRI (Materials and Methods) from 1901 to 2017. The 2015–2017 mean is record-breaking over the period 1901–2017. (C–E) Mean 1921–1970 AMJJAS rainfall (millimeters per month) in observations (GPCC) (C), SPEAR_MED (D), and SPEAR_LO (E). The red lines encircle the area receiving at least 65% of the total annual rainfall during AMJJAS used to define WRI. (F) Monthly WRI in observations and models. Comparison of SPEAR_MED with SPEAR_LO shows how an enhanced resolution is key to capture finer-scale regional details of winter rainfall in the relatively small SSA Mediterranean region.While poor water-management practices and infrastructure deficiencies worsened the crisis (10, 11), the 2015–2017 rainfall deficit was the main driver of the drought (5). To facilitate the improvement of water-management practices and the infrastructure necessary to make the system more resilient, it is critical to first determine how likely a meteorological drought like the one in 2015–2017 might be in the coming decades. Increased aridity is expected in most of southern Africa (12–14) as a consequence of the Hadley Cell poleward expansion (4, 15–18) and southward shift of the Southern Hemisphere jet stream (19). Second, the risk of more extreme droughts should be quantified to understand the potential for emerging risks that could make a Day Zero event in Cape Town unavoidable.Previous work (5) has suggested that the Day Zero drought may have been made 1.4 to 6.4 times more likely over the last century due to +1 K of global warming, with the risk expected to scale linearly with one additional degree of warming. Such estimates make use of statistical models of the probability distribution’s tail (e.g., the generalized extreme value) applied to observations and previous-generation [i.e., as those participating to the Coupled Model Intercomparison Project Phase 3 (CMIP3) (20) and 5 (21)] climate models. CMIP3 and CMIP5 models have been shown to have a systematically biased position of the Southern Hemisphere jet stream toward the Equator, due to insufficient horizontal resolution (19). This produces a large uncertainty in model projections of jet-stream shifts (22, 23), thus hindering realistic projections of Southern Hemisphere climate change. Furthermore, for hydroclimatic variables, a statistical extrapolation of the probability distribution’s tail might have inherent limitations in providing precise estimates of the event probability of future extreme events, although its precision profits from the use of large ensembles (24, 25).Large ensembles of comprehensive climate models provide thousands of years of data that allow direct construction of the underlying probability distribution of hydroclimatic extremes without relying on a hypothesized statistical model of extremes (25, 26). South African winter rains have high interannual and decadal variability due to El Niño–Southern Oscillation (27), the Southern Annular Mode (28), and interdecadal variability (29). A multidecade to multicentury record may be required to detect the emergence of statistically significant trends in regional precipitation extremes. A large ensemble is, thus, a powerful method to isolate, at the decadal timescale, internal natural variability (e.g., SI Appendix, Fig. S2) from the forced signal (30–32).     

A general model of hippocampal and dorsal striatal learning and decision making

Humans and other animals use multiple strategies for making decisions. Reinforcement-learning theory distinguishes between stimulus–response (model-free; MF) learning and deliberative (model-based; MB) planning. The spatial-navigation literature presents a parallel dichotomy between navigation strategies. In “response learning,” associated with the dorsolateral striatum (DLS), decisions are anchored to an egocentric reference frame. In “place learning,” associated with the hippocampus, decisions are anchored to an allocentric reference frame. Emerging evidence suggests that the contribution of hippocampus to place learning may also underlie its contribution to MB learning by representing relational structure in a cognitive map. Here, we introduce a computational model in which hippocampus subserves place and MB learning by learning a “successor representation” of relational structure between states; DLS implements model-free response learning by learning associations between actions and egocentric representations of landmarks; and action values from either system are weighted by the reliability of its predictions. We show that this model reproduces a range of seemingly disparate behavioral findings in spatial and nonspatial decision tasks and explains the effects of lesions to DLS and hippocampus on these tasks. Furthermore, modeling place cells as driven by boundaries explains the observation that, unlike navigation guided by landmarks, navigation guided by boundaries is robust to “blocking” by prior state–reward associations due to learned associations between place cells. Our model, originally shaped by detailed constraints in the spatial literature, successfully characterizes the hippocampal–striatal system as a general system for decision making via adaptive combination of stimulus–response learning and the use of a cognitive map.

Behavioral and neuroscientific studies suggest that animals can apply multiple strategies to the problem of maximizing future reward, referred to as the reinforcement-learning (RL) problem (1, 2). One strategy is to build a model of the environment that can be used to simulate the future to plan optimal actions (3) and the past for episodic memory (4–6). An alternative, model-free (MF) approach uses trial and error to estimate a direct mapping from the animal’s state to its expected future reward, which the agent caches and looks up at decision time (7, 8), potentially supporting procedural memory (9). This computation is thought to be carried out in the brain through prediction errors signaled by phasic dopamine responses (10). These strategies are associated with different tradeoffs (2). The model-based (MB) approach is powerful and flexible, but computationally expensive and, therefore, slow at decision time. MF methods, in contrast, enable rapid action selection, but these methods learn slowly and adapt poorly to changing environments. In addition to MF and MB methods, there are intermediate solutions that rely on learning useful representations that reduce burdens on the downstream RL process (11–13).In the spatial-memory literature, a distinction has been observed between “response learning” and “place learning” (14–16). When navigating to a previously visited location, response learning involves learning a sequence of actions, each of which depends on the preceding action or sensory cue (expressed in egocentric terms). For example, one might remember a sequence of left and right turns starting from a specific landmark. An alternative place-learning strategy involves learning a flexible internal representation of the spatial layout of the environment (expressed in allocentric terms). This “cognitive map” is thought to be supported by the hippocampal formation, where there are neurons tuned to place and heading direction (17–19). Spatial navigation using this map is flexible because it can be used with arbitrary starting locations and destinations, which need not be marked by immediate sensory cues.We posit that the distinction between place and response learning is analogous to that between MB and MF RL (20). Under this view, associative reinforcement is supported by the DLS (21, 22). Indeed, there is evidence from both rodents (23–25) and humans (26, 27) that spatial-response learning relies on the same basal ganglia structures that support MF RL. Evidence also suggests an analogy between MB reasoning and hippocampus (HPC)-based place learning (28, 29). However, this equivalence is not completely straightforward. For example, in rodents, multiple hippocampal lesion and inactivation studies failed to elicit an effect on action-outcome learning, a hallmark of MB planning (30–35). Nevertheless, there are indications that HPC might contribute to a different aspect of MB RL: namely, the representation of relational structure. Tasks that require memory of the relationships between stimuli do show dependence on HPC (36–42).Here, we formalize the perspective that hippocampal contributions to MB learning and place learning are the same, as are the dorsolateral striatal contributions to MF and response learning. In our model, HPC supports flexible behavior by representing the relational structure among different allocentric states, while dorsolateral striatum (DLS) supports associative reinforcement over egocentric sensory features. The model arbitrates between the use of these systems by weighting each system’s action values by the reliability of the system, as measured by a recent average of prediction errors, following Wan Lee et al. (43). We show that HPC and DLS maintain these roles across multiple task domains, including a range of spatial and nonspatial tasks. Our model can quantitatively explain a range of seemingly disparate findings, including the choice between place and response strategies in spatial navigation (23, 44) and choices on nonspatial multistep decision tasks (45, 46). Furthermore, it explains the puzzling finding that landmark-guided navigation is sensitive to the blocking effect, whereas boundary-guided navigation is not (27), and that these are supported by the DLS and HPC, respectively (26). Thus, different RL strategies that manage competing tradeoffs can explain a longstanding body of spatial navigation and decision-making literature under a unified model.     

Cortical ensembles selective for context

Neural processing of sensory information is strongly influenced by context. For instance, cortical responses are reduced to predictable stimuli, while responses are increased to novel stimuli that deviate from contextual regularities. Such bidirectional modulation based on preceding sensory context is likely a critical component or manifestation of attention, learning, and behavior, yet how it arises in cortical circuits remains unclear. Using volumetric two-photon calcium imaging and local field potentials in primary visual cortex (V1) from awake mice presented with visual “oddball” paradigms, we identify both reductions and augmentations of stimulus-evoked responses depending, on whether the stimulus was redundant or deviant, respectively. Interestingly, deviance-augmented responses were limited to a specific subset of neurons mostly in supragranular layers. These deviance-detecting cells were spatially intermixed with other visually responsive neurons and were functionally correlated, forming a neuronal ensemble. Optogenetic suppression of prefrontal inputs to V1 reduced the contextual selectivity of deviance-detecting ensembles, demonstrating a causal role for top-down inputs. The presence of specialized context-selective ensembles in primary sensory cortex, modulated by higher cortical areas, provides a circuit substrate for the brain’s construction and selection of prediction errors, computations which are key for survival and deficient in many psychiatric disorders.

In the mammalian brain, sensorineural processing is significantly influenced by context. For instance, sensory processing circuits tend to suppress processing of predictable or “redundant” stimuli (e.g., tree branches in a forest), while amplifying responses to contextually salient stimuli (like an airborne predator). This strategy conserves energy, while also allowing stimuli with potential behavioral or survival relevance to “stand out” and garner additional neuronal resources.In a healthy neocortex, neural responses to incident sensory stimuli are modulated by past experience on short (0.01 to 10 s) as well as long (>10 s) time scales. For instance, in a classic sensory “oddball” paradigm, repetition of a given stimulus at a rate of 0.1 to 2 Hz results in a phenomenon termed “stimulus specific adaptation” (SSA), wherein response magnitudes decrease rapidly to the repeated or “adapted” stimulus (1–3). In contrast, when a stimulus deviates from the established contextual regularities (e.g., probability of occurrence), a phenomenon termed “deviance detection” (DD) is observed, wherein responses are augmented beyond the typical magnitude observed in a neutral context (1).Theoretical work has sought to explain these phenomena in a “predictive coding” framework, wherein a generative model of the environment, embedded in an increasingly hierarchical cortical network, serves to suppress sensory cortical responses to predictable stimuli while allowing responses to contextually deviant stimuli (i.e., prediction errors) to propagate and update the model (4). Experimentally, these phenomena have been studied at either the single-neuron level in animals (1–3) or at the brain-wide level with electroencephalography (EEG) or magnetoencephalography (MEG) in humans (5, 6). Resulting work highlights a role of N-methyl-d-aspartate (NMDA) receptors (7) as well as gamma-Aminobutyric acid (GABA)-ergic interneurons for SSA (3) and DD (1).Still, much remains unclear about how DD and SSA are computed within cortical circuits (8). For instance, are response decreases or increases (to redundancy and deviance, respectively) expressed evenly within responsive neurons in a local cortical region, mirroring the gross-level readout of EEG event-related potentials, or is there some division of labor within a cortical region, with subsets of neurons expressing DD or “prediction error” (9)? A recent study in mouse primary visual cortex (V1) investigating sensory–motor mismatch suggests that a subset of layer 2/3 neurons respond selectively when visual stimuli do not match locomotor predictions (10). Whether and how this finding applies to sensory–sensory mismatch, that is, when incident stimuli do not match purely sensory-based predictions (i.e., as studied with the classic “oddball” paradigm) is unknown. Furthermore, whether and how such “prediction error” cells are selectively influenced by top-down inputs (i.e., backward projections) remains unknown as well. Such questions, and prediction error in general, carry major clinical significance since cortical SSA and DD are characteristically reduced in individuals with psychotic disorders (6, 11), potentially due to diminished integration of top-down modulation (12) or destabilized local ensembles in V1 (13–15).Utilizing a standard sensory “oddball” paradigm commonly used in clinical neuropsychiatry, we show with volumetric two-photon calcium imaging in awake mouse V1 that, among all V1 neurons responsive to a given visual stimulus, a subset (about a third) respond selectively when the stimulus is contextually deviant. This strong deviance preference remains stable across trials and subsequent experiments, while the rest of the visually driven neurons show largely absent contextual modulation. Such “deviance detectors” display high intragroup activity correlations even in the absence of direct visual stimulation, suggesting that they form functional neuronal ensembles. “Deviance detectors” included mainly excitatory neurons and were most prevalent in superficial layers, consistent with theoretical models of predictive coding (9). Further, cortico-cortical inputs from prefrontal regions supported context processing in V1 by selectively modulating the activity of “deviance detector” ensembles bidirectionally, enhancing their responses to contextually deviant stimuli while reducing their responses to predictable stimuli, suggesting a mediating role for local V1 inhibitory and disinhibitory circuitry (1, 8). This is consistent with a theorized role of both prefrontal cortical (PFC) and sensory cortices in mismatch negativity and schizophrenia-related psychopathology (16, 17).     

Cholinergic suppression of hippocampal sharp-wave ripples impairs working memory

Learning and memory are assumed to be supported by mechanisms that involve cholinergic transmission and hippocampal theta. Using G protein–coupled receptor-activation–based acetylcholine sensor (GRAB_ACh3.0) with a fiber-photometric fluorescence readout in mice, we found that cholinergic signaling in the hippocampus increased in parallel with theta/gamma power during walking and REM sleep, while ACh3.0 signal reached a minimum during hippocampal sharp-wave ripples (SPW-R). Unexpectedly, memory performance was impaired in a hippocampus-dependent spontaneous alternation task by selective optogenetic stimulation of medial septal cholinergic neurons when the stimulation was applied in the delay area but not in the central (choice) arm of the maze. Parallel with the decreased performance, optogenetic stimulation decreased the incidence of SPW-Rs. These findings suggest that septo–hippocampal interactions play a task-phase–dependent dual role in the maintenance of memory performance, including not only theta mechanisms but also SPW-Rs.

The neurotransmitter acetylcholine is thought to be critical for hippocampus-dependent declarative memories (1, 2). Reduction in cholinergic neurotransmission, either in Alzheimer’s disease or in experiments with cholinergic antagonists, such as scopolamine, impairs memory function (3–8). Acetylcholine may bring about its beneficial effects on memory encoding by enhancing theta rhythm oscillations, decreasing recurrent excitation, and increasing synaptic plasticity (9–11). Conversely, drugs which activate cholinergic receptors enhance learning and, therefore, are a neuropharmacological target for the treatment of memory deficits in Alzheimer’s disease (5, 12, 13).The contribution of cholinergic mechanisms in the acquisition of long-term memories and the role of the hippocampal–entorhinal–cortical interactions are well supported by experimental data (5, 12, 13). In addition, working memory or “short-term” memory is also supported by the hippocampal–entorhinal–prefrontal cortex (14–16). Working memory in humans is postulated to be a conscious process to “keep things in mind” transiently (16). In rodents, matching to sample task, spontaneous alternation between reward locations, and the radial maze task have been suggested to function as a homolog of working memory [“working memory like” (17)].Cholinergic activity is a critical requirement for working memory (18, 19) and for sustaining theta oscillations (10, 20–22). In support of this contention, theta–gamma coupling and gamma power are significantly higher in the choice arm of the maze, compared with those in the side arms where working memory is no longer needed for correct performance (23–26). It has long been hypothesized that working memory is maintained by persistent firing of neurons, which keep the presented items in a transient store in the prefrontal cortex and hippocampal–entorhinal system (27–31), although the exact mechanisms are debated (32–37). An alternative hypothesis holds that items of working memory are stored in theta-nested gamma cycles (38). Common in these models of working memory is the need for an active, cholinergic system–dependent mechanism (39–41). However, in spontaneous alternation tasks, the animals are not moving continuously during the delay, and theta oscillations are not sustained either. During the immobility epochs, theta is replaced by intermittent sharp-wave ripples (SPW-R), yet memory performance does not deteriorate. On the contrary, artificial blockade of SPW-Rs can impair memory performance (42, 43), and prolongation of SPW-Rs improves performance (44). Under the cholinergic hypothesis of working memory, such a result is unexpected.To address the relationship between cholinergic/theta versus SPW-R mechanism in spontaneous alternation, we used a G protein–coupled receptor-activation–based acetylcholine sensor (GRAB_ACh3.0) (45) to monitor acetylcholine (ACh) activity during memory performance in mice. In addition, we optogenetically enhanced cholinergic tone, which suppresses SPW-Rs by a different mechanism than electrically or optogenetically induced silencing of neurons in the hippocampus (43, 44). We show that cholinergic signaling in the hippocampus increases in parallel with theta power/score during walking and rapid eye movement (REM) sleep and reaches a transient minimum during SPW-Rs. Selective optogenetic stimulation of medial septal cholinergic neurons decreased the incidence of SPW-Rs during non-REM sleep (46–48), as well as during the delay epoch of a working memory task and impaired memory performance. These findings demonstrate that memory performance is supported by complementary theta and SPW-R mechanisms.     

The potential harms of the Tor anonymity network cluster disproportionately in free countries

The Tor anonymity network allows users to protect their privacy and circumvent censorship restrictions but also shields those distributing child abuse content, selling or buying illicit drugs, or sharing malware online. Using data collected from Tor entry nodes, we provide an estimation of the proportion of Tor network users that likely employ the network in putatively good or bad ways. Overall, on an average country/day, ∼6.7% of Tor network users connect to Onion/Hidden Services that are disproportionately used for illicit purposes. We also show that the likely balance of beneficial and malicious use of Tor is unevenly spread globally and systematically varies based upon a country’s political conditions. In particular, using Freedom House’s coding and terminological classifications, the proportion of often illicit Onion/Hidden Services use is more prevalent (∼7.8%) in “free” countries than in either “partially free” (∼6.7%) or “not free” regimes (∼4.8%).

Debate rages about the social utility of an anonymous portion of the global Internet accessible via the Tor network and colloquially known as the Dark Web (1).^* Although other similar tools exist, The Onion Router (Tor) is currently the largest anonymity network. Tor users can act as publishers of content by using the network to anonymously administer Onion/Hidden Services for the use of others. They can also use the Tor browser to anonymously read either these Onion/Hidden Services (i.e., sites with rendezvous points located internal to the Tor network) or to access Clear Web sites (1–5). With these diverse supply-side and demand-side functions (6), many point to the socially harmful uses of Tor as an anonymous platform for child abuse imagery sites (7, 8), illicit drug markets (9–13), gun sales (14, 15), and potential extremist content that has shifted to the Dark Web after extensive Clear Web content moderation efforts (16). Others emphasize its socially beneficial potential as a privacy-enhancing tool and censorship circumvention technology (17–22).Both sides of the debate illustrate genuine uses of the technology. Like any tool that is inherently dual use, questions abound about whether its benefits are worth the costs. Such questions have both net (i.e., do costs or benefits predominate) and distributional (i.e., how are the harms/benefits spread out) dimensions. Overall, a technology like the Tor anonymity network might do more harm than good. It may also be more harmful in some locations than others. Ultimately, these are empirical questions.In the case of the Tor anonymity network, our data provide clear, if probabilistic, answers to these questions. Our data show that in net terms, only a small fraction of Tor users employ the anonymity system for likely malicious purposes. On an average day during our sample period, for example, about 6.7% of Tor network clients globally use the network to connect to “Onion/Hidden Services” that are predominantly used for illicit and illegal activities, such as buying drugs, distributing malware, or consuming and sharing child abuse imagery content. To be sure, there some socially beneficial content on Onion/Hidden Services and plenty of troubling content on the Clear Web. However, substantial evidence has shown that the preponderance of Onion/Hidden Services traffic connects to illicit sites (7). With this important caveat in mind, our data also show that the distribution of potentially harmful and beneficial uses is uneven, clustering predominantly in politically free regimes. In particular, the average rate of likely malicious use of Tor in our data for countries coded by Freedom House as “not free” is just 4.8%. In countries coded as “free,” the percentage of users visiting Onion/Hidden Services as a proportion of total daily Tor use is nearly twice as much or ∼7.8%. These findings are robust to a different measure of political freedom and the inclusion of a variety of statistical controls. They also give rise to a number of important public policy challenges.     

Unveiling the “invisible” druggable conformations of GDP-bound inactive Ras

The prevalent view on whether Ras is druggable has gradually changed in the recent decade with the discovery of effective inhibitors binding to cryptic sites unseen in the native structures. Despite the promising advances, therapeutics development toward higher potency and specificity is challenged by the elusive nature of these binding pockets. Here we derive a conformational ensemble of guanosine diphosphate (GDP)-bound inactive Ras by integrating spin relaxation-validated atomistic simulation with NMR chemical shifts and residual dipolar couplings, which provides a quantitative delineation of the intrinsic dynamics up to the microsecond timescale. The experimentally informed ensemble unequivocally demonstrates the preformation of both surface-exposed and buried cryptic sites in Ras•GDP, advocating design of inhibition by targeting the transient druggable conformers that are invisible to conventional experimental methods. The viability of the ensemble-based rational design has been established by retrospective testing of the ability of the Ras•GDP ensemble to identify known ligands from decoys in virtual screening.

Situated in a central position of the complex intracellular signaling network, Ras proteins play critical roles in regulating cell growth, differentiation, migration and apoptosis through cycling between the guanosine diphosphate (GDP)-bound inactive and guanosine triphosphate (GTP)-bound active forms (1, 2). Aberrant signaling caused by oncogenic mutations in Ras that break this physiological balance can result in uncontrolled cell proliferation and ultimately the development of human malignancies (3, 4). Despite its well-established role in tumorigenesis and the extensive efforts to target this oncoprotein in past decades, clinically approved therapies remain unavailable. One obstacle to the development of anti-Ras drugs lies in the native structures of active and inactive Ras that lack apparently druggable pockets for high-affinity interactions with inhibitory compounds (5–7).Both the active and inactive forms of Ras, however, are inherently flexible, populating rare conformers distinct from the native structures and presenting alternative opportunities for drug discovery (8–11). For example, in GTP-bound active Ras, a major and minor state (termed states 2 and 1, respectively) coexist in solution and exchange on a millisecond timescale, with state 1 showing surface roughness unobserved in the major state (12–18). The direct visibility of state 1 in the one-dimensional ³¹P NMR spectra of active Ras largely facilitated its early discovery and characterization (12, 19). And the available mutants of H-Ras (e.g., T35A), or the homolog M-Ras, which predominantly assume the state 1 conformation, further promoted the atomic-resolution studies of its structure and internal dynamics, as well as the concomitant drug discovery efforts targeting this low-populated conformer (17, 18, 20).In comparison to the intensive studies on active Ras, research on the dynamics of GDP-bound inactive Ras has lagged far behind, presumably due to its high degree of spectral homogeneity with little sign of resonance splitting or exchange broadening at room temperature (21). The previously reported cryptic pockets for covalent and noncovalent inhibitors of Ras•GDP (22–24), which are unseen in the compound-free structure, nevertheless indicate that the inactive form is also structurally plastic. The recent relaxation-based NMR experiments carried out at low temperature successfully captured the intrinsic microsecond timescale motions in Ras•GDP, which map to regions that overlap with those rearranged on the binding of inhibitors (11). However, the structural information of the transiently formed excited state, in the form of chemical shifts, is not available from the relaxation measurements, owing to the fast exchange rate on the chemical shift timescale. Moreover, unlike the case of active Ras, there are no known mutations that can stabilize the excited state of Ras•GDP for investigations using conventional biophysical techniques. Thus far, the sparsely populated conformations of inactive Ras derived from its microsecond dynamics remain poorly understood, precluding structure-based rational drug discovery.To address these challenges, in this work we constructed a solution ensemble of Ras•GDP by integrating atomistic computer simulation with diverse NMR experimental parameters containing complementary information about the intrinsic protein motions on timescales from picoseconds to microseconds. This NMR-based ensemble well covers the slow dynamics as probed by spin relaxation and provides an atomic-resolution delineation of thermally accessible conformations, including those bearing surface or buried pockets similar to the cryptic pockets previously observed in the inhibitor-bound forms. The utility of the Ras•GDP ensemble in the development of inhibitors is demonstrated by ensemble-based virtual screening, which achieves an impressive level of enrichment of known binders.     

From the Cover: Overexpression of CD47 is associated with brain overgrowth and 16p11.2 deletion syndrome

Copy number variation (CNV) at the 16p11.2 locus is associated with neuropsychiatric disorders, such as autism spectrum disorder and schizophrenia. CNVs of the 16p gene can manifest in opposing head sizes. Carriers of 16p11.2 deletion tend to have macrocephaly (or brain enlargement), while those with 16p11.2 duplication frequently have microcephaly. Increases in both gray and white matter volume have been observed in brain imaging studies in 16p11.2 deletion carriers with macrocephaly. Here, we use human induced pluripotent stem cells (hiPSCs) derived from controls and subjects with 16p11.2 deletion and 16p11.2 duplication to understand the underlying mechanisms regulating brain overgrowth. To model both gray and white matter, we differentiated patient-derived iPSCs into neural progenitor cells (NPCs) and oligodendrocyte progenitor cells (OPCs). In both NPCs and OPCs, we show that CD47 (a “don’t eat me” signal) is overexpressed in the 16p11.2 deletion carriers contributing to reduced phagocytosis both in vitro and in vivo. Furthermore, 16p11.2 deletion NPCs and OPCs up-regulate cell surface expression of calreticulin (a prophagocytic “eat me” signal) and its binding sites, indicating that these cells should be phagocytosed but fail to be eliminated due to elevations in CD47. Treatment of 16p11.2 deletion NPCs and OPCs with an anti-CD47 antibody to block CD47 restores phagocytosis to control levels. While the CD47 pathway is commonly implicated in cancer progression, we document a role for CD47 in psychiatric disorders associated with brain overgrowth.

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by deficits in social interaction and communication. Copy number variation (CNV) at the 16p11.2 locus is associated with ASD (1–8). People who have 16p11.2 deletion syndrome tend to have larger head circumferences (macrocephaly), with disproportionate enlargement in both gray and white matter volume (8–13). Individuals with ASD and macrocephaly have more severe behavioral and cognitive problems and are less responsive to standard medical and therapeutic interventions than those with ASD and normal head circumferences (14). In addition, prior work has documented a very strong cross-sectional and temporal association between macrocephaly and ASD symptoms (8, 9, 11, 12, 14–17). These findings suggest that understanding the underlying mechanisms regulating macrocephaly could provide a window of opportunity for intervention or mitigation of symptoms.Here, we used patient-derived human induced pluripotent stem cells (hiPSCs) to interrogate the underlying mechanisms contributing to gray and white matter enlargement. We focused on individuals with intellectual disability (IQ < 70) or ASD associated with brain overgrowth in 16p11.2 deletion carriers. We differentiated the iPSCs into neural progenitor cells (NPCs) and oligodendrocyte progenitor cells (OPCs) and investigate the hypothesis that brain enlargement in 16p11.2 deletion carriers may be due to improper cellular elimination. Under normal conditions, classic “eat me” and “don’t eat me” signaling mechanisms associated with phagocytosis maintain cellular homeostasis across diverse tissue types (18, 19). CD47 (a “don’t eat me” signal) protects normal cells from getting cleared (18), but can become overexpressed in many types of cancer cells, preventing tumorigenic cells from getting engulfed or phagocytosed (20–22). In fact, CD47 plays an important role in many pathological disorders associated with an overproduction of cells and cell removal, including cancer (20–22), atherosclerosis (23), and fibrotic diseases (24). NPCs derived from iPSCs of autistic individuals with macrocephaly have increased proliferation relative to controls (25, 26). Therefore, we hypothesized that CD47 may be involved in these disorders.We find that CD47 is overexpressed in NPCs and OPCs derived from 16p11.2 deletion carriers, leading to reduced phagocytosis by macrophages and microglia. Furthermore, the 16p11.2 deletion NPCs and OPCs have increased cell surface expression of calreticulin (CRT, a prophagocytic “eat me” signal), indicating that these cells should be eliminated but are not due to high levels of CD47 (27). Importantly, treatment with a CD47 blocking antibody restores phagocytosis of 16p11.2 deletion NPCs and OPCs to control levels, particularly in 16p_del NPCs and OPCs that have increased cell surface expression of CRT, indicating that the changes in phagocytosis are mediated by cell surface expression of CD47. We thus identify a role for CD47 in 16p11.2 deletion syndrome and highlight the potential importance of blocking CD47 to promote clearance of unhealthy NPCs and OPCs in 16p11.2 deletion with brain overgrowth.     

“Soft” oxidative coupling of methane to ethylene: Mechanistic insights from combined experiment and theory

The oxidative coupling of methane to ethylene using gaseous disulfur (2CH₄ + S₂ → C₂H₄ + 2H₂S) as an oxidant (SOCM) proceeds with promising selectivity. Here, we report detailed experimental and theoretical studies that examine the mechanism for the conversion of CH₄ to C₂H₄ over an Fe₃O₄-derived FeS₂ catalyst achieving a promising ethylene selectivity of 33%. We compare and contrast these results with those for the highly exothermic oxidative coupling of methane (OCM) using O₂ (2CH₄ + O₂ → C₂H₄ + 2H₂O). SOCM kinetic/mechanistic analysis, along with density functional theory results, indicate that ethylene is produced as a primary product of methane activation, proceeding predominantly via CH₂ coupling over dimeric S–S moieties that bridge Fe surface sites, and to a lesser degree, on heavily sulfided mononuclear sites. In contrast to and unlike OCM, the overoxidized CS₂ by-product forms predominantly via CH₄ oxidation, rather than from C₂ products, through a series of C–H activation and S-addition steps at adsorbed sulfur sites on the FeS₂ surface. The experimental rates for methane conversion are first order in both CH₄ and S₂, consistent with the involvement of two S sites in the rate-determining methane C–H activation step, with a CD₄/CH₄ kinetic isotope effect of 1.78. The experimental apparent activation energy for methane conversion is 66 ± 8 kJ/mol, significantly lower than for CH₄ oxidative coupling with O₂. The computed methane activation barrier, rate orders, and kinetic isotope values are consistent with experiment. All evidence indicates that SOCM proceeds via a very different pathway than that of OCM.

The oxidative coupling of methane (OCM) with O₂ would seem to be a concise, direct route to convert methane, one of the most Earth-abundant carbon sources (1), to ethylene (2CH₄ + O₂ → C₂H₄ + 2H₂O), a key chemical intermediate (2, 3), and this process has been extensively studied (1, 4–19) since 1982 (20). Nevertheless, the widespread use of OCM is challenged by methane overoxidation to CO₂ and other oxygenates. Furthermore, the severe reaction conditions of nonoxidative pathways (2, 21–28) typically risk carbon deposition and catalyst deactivation (2, 21–26). In preliminary studies, we reported a 2CH₄ + S₂ → C₂H₄ + 2H₂S coupling process that moderates the methane overoxidation driving force using gaseous disulfur (S₂) as a “soft” oxidant (SOCM; Fig. 1A) (29). S₂ is isoelectronic with O₂, the major sulfur vapor species at 700 to 925 °C (30–32), and is a less aggressive oxidant than O₂ (33). In this scenario, elemental sulfur is recovered from the H₂S coproduct via the known Claus process (Fig. 1B) (30), in a cycle where sulfur mediates/moderates the high nonselective O₂ reactivity. SOCM achieved promising ethylene selectivity, raising intriguing mechanistic questions and the possibility of higher selectivity. Methane + S_2(g) ethylene selectivities near ∼20% are achieved over a PdS/ZrO₂ catalyst (29), and oxide precatalysts give selectivities near 33% (34).Open in a separate windowFig. 1.Energetic comparison between the oxidative coupling of methane with O₂ (OCM) and with S₂ (SOCM) and the pathway to recover elemental sulfur from H₂S. (A) Gibbs free energy of desired and overoxidation processes in OCM and SOCM at 800 and 1,050 °C. (B) Industrialized catalytic Claus process used to recover elemental sulfur from H₂S.Nevertheless, in contrast to extensive OCM (17, 35–39) and nonoxidative CH₄ coupling studies (40), far less is known about the SOCM reaction pathway. Post-SOCM X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), and elemental analysis (29, 34) indicate that the oxide precatalysts are predominantly sulfided. Density functional theory (DFT) analyses of molybdenum sulfide catalysts suggest that methane is activated at M–S or S–S sites to form surface-bound CH₃* species which dehydrogenate to form CH₂* (methylidene) species, which then couple to produce C₂H₄. It was proposed that CH₃* species can also desorb as methyl radicals which couple to form ethane (29). The overoxidation product, CS₂, was suggested to form via sulfur addition to methylidene surface intermediates (29).Kinetic, mechanistic, and theoretical analyses are needed to better understand the CH₄ conversion pathways to C₂H₄ and other products. In principle, there are two plausible pathways following methane activation: 1) H abstraction from adsorbed methyl species forms methylidene (CH₂*) and methylidyne (CH*) species then couple to C₂ products or undergo oxidation to CS₂ or 2) coupling of surface or gas phase methyl species form ethane, which then dehydrogenates to form ethylene or oxidizes to CS₂. For further SOCM optimization it is important to determine which pathways are operative, their relative rates, and the C₂ and CS₂ formation sites.Here we investigate SOCM pathways over a sulfided Fe₃O₄ precatalyst which affords C₂H₄ selectivities near 33%, complete oxide to sulfide conversion, minimal carbon deposition (coking), and 48-h SOCM stability at 950 °C (34). We first summarize SOCM phenomenology, followed by analysis of the Fe phases during sulfurization and SOCM. Next, kinetic/mechanistic studies focus on the methane and S₂ reaction orders, activation energetics, and isotope effects and probe the pathways governing C₂ vs. CS₂ formation. Complementary DFT calculations focus on reaction mechanisms, the active sites, and their role in product formation. The results are used in a microkinetic model to simulate reaction rates, apparent activation barriers, and reaction rate orders and to compare with experiment. Finally, SOCM and OCM are compared, revealing that they follow distinctly different pathways.     

Retrieval-constrained valuation: Toward prediction of open-ended decisions

Real-world decisions are often open ended, with goals, choice options, or evaluation criteria conceived by decision-makers themselves. Critically, the quality of decisions may heavily rely on the generation of options, as failure to generate promising options limits, or even eliminates, the opportunity for choosing them. This core aspect of problem structuring, however, is largely absent from classical models of decision-making, thereby restricting their predictive scope. Here, we take a step toward addressing this issue by developing a neurally inspired cognitive model of a class of ill-structured decisions in which choice options must be self-generated. Specifically, using a model in which semantic memory retrieval is assumed to constrain the set of options available during valuation, we generate highly accurate out-of-sample predictions of choices across multiple categories of goods. Our model significantly and substantially outperforms models that only account for valuation or retrieval in isolation or those that make alternative mechanistic assumptions regarding their interaction. Furthermore, using neuroimaging, we confirm our core assumption regarding the engagement of, and interaction between, semantic memory retrieval and valuation processes. Together, these results provide a neurally grounded and mechanistic account of decisions with self-generated options, representing a step toward unraveling cognitive mechanisms underlying adaptive decision-making in the real world.

Some decisions, such as choosing an entree at a restaurant, come with a menu of well-defined options and associated information that aid in their evaluation and selection. For many other decisions, such as how to spend one’s evening or which career path to choose, the space of potential options is less well-defined and may need to be generated by the decision-maker themself. More generally, option generation is part of a larger set of processes critical for a class of decisions, often referred to as “open-ended” or “ill-structured” problems (1–6), that are characterized by a lack of well-specified goals, alternatives, or evaluation criteria, among others.Despite their ubiquity, however, such decisions pose considerable difficulties for standard models of decision-making, as processes generating these features are typically considered outside the scope of traditional decision analysis (3–11). To address this ambiguity, researchers have turned to one of two broad strategies. The first, and arguably the most frequent one employed, involves imposing strong auxiliary assumptions about the option set (12). For example, if one is choosing a breakfast cereal, the option set comprises everything on the market. This strategy has the benefit of simplicity and is consistent with the invocation of “full rationality” in neoclassical economic theory.In contrast, the second strategy attempts to relax these assumptions by emphasizing the “constructed” nature of decisions (13). This approach encompasses processes that can occur prior to valuation of choice options, such as the generation (14–16) and consideration (17–19) of said options, as well as those that can occur in parallel, including heuristics (20, 21) that make use of “fast and frugal” rules that do not require explicit weighing of the relative costs and benefits of each option. However, despite important advances in our understanding of mechanisms underlying factors that influence option generation, much less attention has been paid to connecting these accounts with formal models capable of specifying the contents of the “internal menu” (10–12, 22–25). As a result, it remains challenging to make quantitative predictions about the effects of option generation on choice.Here, we develop a neurally grounded model capable of making highly accurate predictions of people’s decisions for a class of ill-structured environments in which options must be retrieved from memory. Specifically, drawing on Marr’s three levels of analysis (26, 27), we replace the strong auxiliary assumptions about contents of the option set in standard choice models with a quantitative model capturing “predecision processes,” thereby linking formal models of option generation and choice processes. Across a set of experiments that evaluate decisions incorporating real-world goods, we show that, at Marr’s computational/conceptual level, these decisions rely critically upon the interaction of processes involved in valuation and the retrieval of semantic knowledge, the aspect of human declarative memory that deals with general, culturally shared knowledge of meanings, facts, ideas, and concepts accumulated over the lifetime (28, 29). In particular, we draw upon the rich literature on memory factors in consumer decision-making (24, 25, 30), where such choices are conceptualized as the products of a multistage process (14, 16). At the algorithmic level, we build upon two well-established principles from the cognitive science and neuroscientific literatures on memory retrieval and valuation. First, retrieval of semantic knowledge is a probabilistic process governed by the associative principle (28, 31, 32). Second, given a set of options, choice is governed by subjective preferences over the options, commonly referred to as utility or value (7). At the implementation level, we demonstrate that these memory and valuation processes are indeed subserved by separable neurocognitive systems, which can be probed and characterized independently (29, 33).By connecting these literatures, which have developed largely independently, the resulting framework makes a number of testable predictions. First, output resulting from the interaction of retrieval and valuation processes can be predicted using outputs from each component process. To test this idea, we use behavioral responses from two distinct tasks capturing valuation and semantic retrieval processes, respectively, to predict behavior in a third task that is hypothesized to require both processes. Furthermore, the framework predicts two reasons that an option was not chosen: 1) the option was not preferred and 2) it was not successfully retrieved. Across a diverse array of real-world goods, we show that our computational model makes highly precise and accurate predictions of the likelihood that each option was chosen by incorporating the extent to which each option benefits from a failure to retrieve some other option or is passed over due to successful retrieval of another option. Finally, we present functional neuroimaging evidence that confirms our model’s fundamental predictions about the separable but interactive nature of the valuation and retrieval systems that underlie these decisions.     

From the Cover: Exploitative leaders incite intergroup warfare in a social mammal

Collective conflicts among humans are widespread, although often highly destructive. A classic explanation for the prevalence of such warfare in some human societies is leadership by self-serving individuals that reap the benefits of conflict while other members of society pay the costs. Here, we show that leadership of this kind can also explain the evolution of collective violence in certain animal societies. We first extend the classic hawk−dove model of the evolution of animal aggression to consider cases in which a subset of individuals within each group may initiate fights in which all group members become involved. We show that leadership of this kind, when combined with inequalities in the payoffs of fighting, can lead to the evolution of severe intergroup aggression, with negative consequences for population mean fitness. We test our model using long-term data from wild banded mongooses, a species characterized by frequent intergroup conflicts that have very different fitness consequences for male and female group members. The data show that aggressive encounters between groups are initiated by females, who gain fitness benefits from mating with extragroup males in the midst of battle, whereas the costs of fighting are borne chiefly by males. In line with the model predictions, the result is unusually severe levels of intergroup violence. Our findings suggest that the decoupling of leaders from the costs that they incite amplifies the destructive nature of intergroup conflict.

Humans are capable of astonishing feats of altruism and cooperation (1–3), but, at the same time, of violent and destructive conflicts (4–8). A key factor contributing to the latter may be that wars are often waged at the behest of leaders who do not share fully in the immediate risks of conflict, and stand to gain benefits in terms of resources and status that are not enjoyed by the majority of combatants (4, 9–11). Could such “warmongering” be a feature of animal conflicts too? Only recently have models of animal aggression begun to explore the impact of inequalities among combatants in collective conflict (12, 13), and the usual assumption of existing theory is that individuals who initiate intergroup conflicts also contribute most to group conflict effort and thereby confer fitness benefits on the rest of their group (a positive or “heroic” model of leadership) (14–17). Here, we explore the more sinister possibility that those who initiate conflict may actually harm their fellows in pursuit of their own interests by exposing them to the risks of conflict while contributing little to fighting themselves (a negative or “exploitative” model of leadership).