Convolutional neural networks improve species distribution modelling by capturing the spatial structure of the environment

Convolutional Neural Networks (CNNs) are statistical models suited for learning complex visual patterns. In the context of Species Distribution Models (SDM) and in line with predictions of landscape ecology and island biogeography, CNN could grasp how local landscape structure affects prediction of species occurrence in SDMs. The prediction can thus reflect the signatures of entangled ecological processes. Although previous machine-learning based SDMs can learn complex influences of environmental predictors, they cannot acknowledge the influence of environmental structure in local landscapes (hence denoted “punctual models”). In this study, we applied CNNs to a large dataset of plant occurrences in France (GBIF), on a large taxonomical scale, to predict ranked relative probability of species (by joint learning) to any geographical position. We examined the way local environmental landscapes improve prediction by performing alternative CNN models deprived of information on landscape heterogeneity and structure (“ablation experiments”). We found that the landscape structure around location crucially contributed to improve predictive performance of CNN-SDMs. CNN models can classify the predicted distributions of many species, as other joint modelling approaches, but they further prove efficient in identifying the influence of local environmental landscapes. CNN can then represent signatures of spatially structured environmental drivers. The prediction gain is noticeable for rare species, which open promising perspectives for biodiversity monitoring and conservation strategies. Therefore, the approach is of both theoretical and practical interest. We discuss the way to test hypotheses on the patterns learnt by CNN, which should be essential for further interpretation of the ecological processes at play.     

Macrophage-specific responses to human- and animal-adapted tubercle bacilli reveal pathogen and host factors driving multinucleated cell formation

The Mycobacterium tuberculosis complex (MTBC) is a group of related pathogens that cause tuberculosis (TB) in mammals. MTBC species are distinguished by their ability to sustain in distinct host populations. While Mycobacterium bovis (Mbv) sustains transmission cycles in cattle and wild animals and causes zoonotic TB, M. tuberculosis (Mtb) affects human populations and seldom causes disease in cattle. The host and pathogen determinants underlying host tropism between MTBC species are still unknown. Macrophages are the main host cell that encounters mycobacteria upon initial infection, and we hypothesised that early interactions between the macrophage and mycobacteria influence species-specific disease outcome. To identify factors that contribute to host tropism, we analysed blood-derived primary human and bovine macrophages (hMϕ or bMϕ, respectively) infected with Mbv and Mtb. We show that Mbv and Mtb reside in different cellular compartments and differentially replicate in hMϕ whereas both Mbv and Mtb efficiently replicate in bMϕ. Specifically, we show that out of the four infection combinations, only the infection of bMϕ with Mbv promoted the formation of multinucleated giant cells (MNGCs), a hallmark of tuberculous granulomas. Mechanistically, we demonstrate that both MPB70 from Mbv and extracellular vesicles released by Mbv-infected bMϕ promote macrophage multinucleation. Importantly, we extended our in vitro studies to show that granulomas from Mbv-infected but not Mtb-infected cattle contained higher numbers of MNGCs. Our findings implicate MNGC formation in the contrasting pathology between Mtb and Mbv for the bovine host and identify MPB70 from Mbv and extracellular vesicles from bMϕ as mediators of this process.     

Impact of manipulation of glycerol/diol dehydratase activity on intestinal microbiota ecology and metabolism

Glycerol/diol dehydratases (GDH) are enzymes that catalyse the production of propionate from 1,2-propanediol, and acrolein from glycerol. Acrolein reacts with dietary carcinogenic heterocyclic amines (HCA), reducing HCA mutagenicity, but is itself also an antimicrobial agent and toxicant. Gut microbial GDH activity has been suggested as an endogenous acrolein source; however, there is limited information on the potential of the intestinal microbiota to have GDH activity, and what impact it can have on the intestinal ecosystem and host health. We hypothesized that GDH activity of gut microbiota is determined by the abundance and distribution of GDH-active taxa and can be enhanced by supplementation of the GDH active Anaerobutyricum hallii, and tested this hypothesis combining quantitative profiling of gdh, model batch fermentations, microbiota manipulation, and kinetic modelling of acrolein formation. Our results suggest that GDH activity is a common trait of intestinal microbiota shared by a few taxa, which was dependent on overall gdh abundance. Anaerobutyricum hallii was identified as a key taxon in GDH metabolism, and its supplementation increased the rate of GDH activity and acrolein release, which enhanced the transformation of HCA and reduced fermentation activity. The findings of this first systematic study on acrolein release by intestinal microbiota indicate that dietary and microbial modulation might impact GDH activity, which may influence host health.     

Selection against immigrants in wild seabird populations

Immigration is a major demographic parameter shaping population dynamics and is an important driver of eco‐evolutionary patterns, but the fitness consequences for individuals following their settlement to a new population (immigrants) remain poorly tested in wild animal populations, particularly among long‐lived species. Here we show that immigrants have a lower fitness than residents in three wild seabird populations (wandering albatross Diomedea exulans, southern fulmar Fulmarus glacialoides, snow petrel Pagodroma nivea). Across all species and during a 32‐year period, immigrants made on average ?9 to 29% fewer breeding attempts, had 5–31% fewer fledglings, had 2–16% lower breeding success and produced 6–46% fewer recruits. Female immigration and male residency were also favored through differences in breeding performance. We provide evidence for selection against immigrants in wild populations of long‐lived species and our results are consistent with female‐biased dispersal in birds being driven by asymmetric limiting resources and the competitive ability of dispersers vs. non‐dispersers.     

FSHIP2 regulates epithelial cell polarity through its lipid product,which binds to Dlg1, a pathway subverted by hepatitis C virus core protein

The main targets of hepatitis C virus (HCV) are hepatocytes, the highly polarized cells of the liver, and all the steps of its life cycle are tightly dependent on host lipid metabolism. The interplay between polarity and lipid metabolism in HCV infection has been poorly investigated. Signaling lipids, such as phosphoinositides (PIs), play a vital role in polarity, which depends on the distribution and expression of PI kinases and PI phosphatases. In this study, we report that HCV core protein, expressed in Huh7 and Madin–Darby canine kidney (MDCK) cells, disrupts apicobasal polarity. This is associated with decreased expression of the polarity protein Dlg1 and the PI phosphatase SHIP2, which converts phosphatidylinositol 3,4,5-trisphosphate into phosphatidylinositol 4,5-bisphosphate (PtdIns(3,4)P2). SHIP2 is mainly localized at the basolateral membrane of polarized MDCK cells. In addition, PtdIns(3,4)P2 is able to bind to Dlg1. SHIP2 small interfering RNA or its catalytically dead mutant disrupts apicobasal polarity, similar to HCV core. In core-expressing cells, RhoA activity is inhibited, whereas Rac1 is activated. Of interest, SHIP2 expression rescues polarity, RhoA activation, and restricted core level in MDCK cells. We conclude that SHIP2 is an important regulator of polarity, which is subverted by HCV in epithelial cells. It is suggested that SHIP2 could be a promising target for anti-HCV treatment.     

Selecting Species for Passive and Active Riparian Restoration in Southern Mexico

In revegetation projects, distinguishing species that can be passively restored by natural regeneration from those requiring active restoration is not a trivial decision. We quantified tree species dominance (measured by an importance value index, IVI_i) and used abundance–size correlations to select those species suitable for passive and/or active restoration of disturbed riparian vegetation in the Lacandonia region, Southern Mexico. We sampled riparian vegetation in a 50 × 10–m transect in each of six reference (RE) and five disturbed (DE) riparian ecosystems. Those species representing more than 50% of total IVI in each ecosystem were selected, and Spearman rank correlation between abundance and diameter classes was calculated. For eight species, it was determined that passive restoration could be sufficient for their establishment. Another eight species could be transplanted by means of active restoration. Five species regenerate well in only one ecosystem type, suggesting that both restoration strategies could be used depending on the degree of degradation. Finally, two species were determined to not be suitable for restoration in the RE (based on the above selection criteria) and were not selected during this initial stage of our restoration project. The high number of tree species found in the RE suggests that the species pool for ecological restoration is large. However, sampling in both ecosystem types helped us reduce the number of species that requires active restoration. Restoration objectives must guide the selection of which methods to implement; in different conditions, other criteria such as dispersal syndrome or social value could be considered in the species selection.     

2′-and/or 3Y-Deoxy-β-L-pentofuranosyl Nucleoside Derivatives: Stereospecific Synthesis and Antiviral Activities

Abstract

Several L-enantiomers of nucleoside analogues were stereospecifically synthesized by a multi-step reaction from L-xylose and their antiviral properties were examined in vitro. Two of them, namely β-L-2′,3,′-dideoxycytidine (β-L-ddC) and its 5-fluoro derivative (β-L-FddC) were found to have potent anti-human immunodeficiency virus (HIV) and significant anti-hepatitis B virus (HBV) activities in cell cultures.