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.     

Breathing space: deoxygenation of aquatic environments can drive differential ecological impacts across biological invasion stages

The influence of climate change on the ecological impacts of invasive alien species (IAS) remains understudied, with deoxygenation of aquatic environments often-overlooked as a consequence of climate change. Here, we therefore assessed how oxygen saturation affects the ecological impact of a predatory invasive fish, the Ponto-Caspian round goby (Neogobius melanostomus), relative to a co-occurring endangered European native analogue, the bullhead (Cottus gobio) experiencing decline in the presence of the IAS. In individual trials and mesocosms, we assessed the effect of high, medium and low (90%, 60% and 30%) oxygen saturation on: (1) functional responses (FRs) of the IAS and native, i.e. per capita feeding rates; (2) the impact on prey populations exerted; and (3) how combined impacts of both fishes change over invasion stages (Pre-invasion, Arrival, Replacement, Proliferation). Both species showed Type II potentially destabilising FRs, but at low oxygen saturation, the invader had a significantly higher feeding rate than the native. Relative Impact Potential, combining fish per capita effects and population abundances, revealed that low oxygen saturation exacerbates the high relative impact of the invader. The Relative Total Impact Potential (RTIP), modelling both consumer species’ impacts on prey populations in a system, was consistently higher at low oxygen saturation and especially high during invader Proliferation. In the mesocosm experiment, low oxygen lowered RTIP where both species were present, but again the IAS retained high relative impact during Replacement and Proliferation stages at low oxygen. We also found evidence of multiple predator effects, principally antagonism. We highlight the threat posed to native communities by IAS alongside climate-related stressors, but note that solutions may be available to remedy hypoxia and potentially mitigate impacts across invasion stages.

     

Arabidopsis ERG28 Tethers the Sterol C4-Demethylation Complex to Prevent Accumulation of a Biosynthetic Intermediate That Interferes with Polar Auxin Transport

Sterols are vital for cellular functions and eukaryotic development because of their essential role as membrane constituents. Sterol biosynthetic intermediates (SBIs) represent a potential reservoir of signaling molecules in mammals and fungi, but little is known about their functions in plants. SBIs are derived from the sterol C4-demethylation enzyme complex that is tethered to the membrane by Ergosterol biosynthetic protein28 (ERG28). Here, using nonlethal loss-of-function strategies focused on Arabidopsis thaliana ERG28, we found that the previously undetected SBI 4-carboxy-4-methyl-24-methylenecycloartanol (CMMC) inhibits polar auxin transport (PAT), a key mechanism by which the phytohormone auxin regulates several aspects of plant growth, including development and responses to environmental factors. The induced accumulation of CMMC in Arabidopsis erg28 plants was associated with diagnostic hallmarks of altered PAT, including the differentiation of pin-like inflorescence, loss of apical dominance, leaf fusion, and reduced root growth. PAT inhibition by CMMC occurs in a brassinosteroid-independent manner. The data presented show that ERG28 is required for PAT in plants. Furthermore, it is accumulation of an atypical SBI that may act to negatively regulate PAT in plants. Hence, the sterol pathway offers further prospects for mining new target molecules that could regulate plant development.     

Conservative Treatment for Postprostatectomy Incontinence

Postprostatectomy incontinence (PPI) is a bothersome complication of radical prostatectomy. Although most men recover from PPI, some men continue to have persistent urinary incontinence. The initial management of persistent PPI usually consists of conservative measures such as pelvic floor muscle exercises. Surgical treatments are usually not entertained for men with urinary incontinence until conservative treatments have failed. This article discusses risk factors for PPI and various options for its treatment, including biofeedback and pharmacotherapy.Key words: Postprostatectomy incontinence, Radical prostatectomy, Pelvic floor muscle training, Pelvic floor stimulationPostprostatectomy incontinence (PPI) is a bothersome complication of radical prostatectomy (RP). Although most men recover from PPI, some men continue to have persistent urinary incontinence and roughly half of these men seek treatment.¹ It is important to understand the natural history of postprostatectomy urinary dysfunction prior to initiating treatment. Generally, the initial management of persistent PPI consists of conservative measures such as pelvic floor muscle exercises.     

Effects of the entomopathogenic fungus Neozygites tanajoae and the predatory mite Typhlodromalus aripo on cassava green mite densities: screenhouse experiments

To improve biocontrol of the cassava green mite, Mononychellus tanajoa, a series of screenhouse experiments were conducted on cassava to determine the effects of single and combined releases of the predatory mite Typhlodromalus aripo and the entomopathogenic fungus Neozygites tanajoae on the suppression of M. tanajoa. We showed that, separately or in combination, T. aripo and N. tanajoae significantly reduced M. tanajoa populations. Moreover, combined release of T. aripo and N. tanajoae on the same cassava plants improved M. tanajoa control. However, our data suggest between T. aripo and N. tanajoae an asymmetric competition that significantly affected N. tanajoae only. This interference is indirect, probably mediated by their common prey or host mite, M. tanajoa since N. tanajoae is not pathogenic to T. aripo. We conclude that adding N. tanajoae to T. aripo populations in Africa would not negatively affect T. aripo and would further promote biocontrol of M. tanajoa.     

Two Homologous EF-G Proteins from Pseudomonas aeruginosa Exhibit Distinct Functions

Genes encoding two proteins corresponding to elongation factor G (EF-G) were cloned from Pseudomonas aeruginosa. The proteins encoded by these genes are both members of the EFG I subfamily. The gene encoding one of the forms of EF-G is located in the str operon and the resulting protein is referred to as EF-G1A while the gene encoding the other form of EF-G is located in another part of the genome and the resulting protein is referred to as EF-G1B. These proteins were expressed and purified to 98% homogeneity. Sequence analysis indicated the two proteins are 90/84% similar/identical. In other organisms containing multiple forms of EF-G a lower degree of similarity is seen. When assayed in a poly(U)-directed poly-phenylalanine translation system, EF-G1B was 75-fold more active than EF-G1A. EF-G1A pre-incubate with ribosomes in the presence of the ribosome recycling factor (RRF) decreased polymerization of poly-phenylalanine upon addition of EF-G1B in poly(U)-directed translation suggesting a role for EF-G1A in uncoupling of the ribosome into its constituent subunits. Both forms of P. aeruginosa EF-G were active in ribosome dependent GTPase activity. The kinetic parameters (K _M) for the interaction of EF-G1A and EF-G1B with GTP were 85 and 70 μM, respectively. However, EF-G1B exhibited a 5-fold greater turnover number (observed k _cat) for the hydrolysis of GTP than EF-G1A; 0.2 s^-1 vs. 0.04 s^-1. These values resulted in specificity constants (k _cat ^obs/K _M) for EF-G1A and EF-G1B of 0.5 x 10³ s^-1 M^-1 and 3.0 x 10³ s^-1 M^-1, respectively. The antibiotic fusidic acid (FA) completely inhibited poly(U)-dependent protein synthesis containing P. aeruginosa EF-G1B, but the same protein synthesis system containing EF-G1A was not affected. Likewise, the activity of EF-G1B in ribosome dependent GTPase assays was completely inhibited by FA, while the activity of EF-G1A was not affected.     

Turnover of C99 is Controlled by a Crosstalk between ERAD and Ubiquitin-Independent Lysosomal Degradation in Human Neuroglioma Cells

Alzheimer’s disease (AD) is characterized by the buildup of amyloid-β peptides (Aβ) aggregates derived from proteolytic processing of the β-amyloid precursor protein (APP). Amyloidogenic cleavage of APP by β-secretase/BACE1 generates the C-terminal fragment C99/CTFβ that can be subsequently cleaved by γ-secretase to produce Aβ. Growing evidence indicates that high levels of C99/CTFβ are determinant for AD. Although it has been postulated that γ-secretase-independent pathways must control C99/CTFβ levels, the contribution of organelles with degradative functions, such as the endoplasmic reticulum (ER) or lysosomes, is unclear. In this report, we investigated the turnover and amyloidogenic processing of C99/CTFβ in human H4 neuroglioma cells, and found that C99/CTFβ is localized at the Golgi apparatus in contrast to APP, which is mostly found in endosomes. Conditions that localized C99/CTFβ to the ER resulted in its degradation in a proteasome-dependent manner that first required polyubiquitination, consistent with an active role of the ER associated degradation (ERAD) in this process. Furthermore, when proteasomal activity was inhibited C99/CTFβ was degraded in a chloroquine (CQ)-sensitive compartment, implicating lysosomes as alternative sites for its degradation. Our results highlight a crosstalk between degradation pathways within the ER and lysosomes to avoid protein accumulation and toxicity.     

An Inducible,Isogenic Cancer Cell Line System for Targeting the State of Mismatch Repair Deficiency

The DNA mismatch repair system (MMR) maintains genome stability through recognition and repair of single-base mismatches and small insertion-deletion loops. Inactivation of the MMR pathway causes microsatellite instability and the accumulation of genomic mutations that can cause or contribute to cancer. In fact, 10-20% of certain solid and hematologic cancers are MMR-deficient. MMR-deficient cancers do not respond to some standard of care chemotherapeutics because of presumed increased tolerance of DNA damage, highlighting the need for novel therapeutic drugs. Toward this goal, we generated isogenic cancer cell lines for direct comparison of MMR-proficient and MMR-deficient cells. We engineered NCI-H23 lung adenocarcinoma cells to contain a doxycycline-inducible shRNA designed to suppress the expression of the mismatch repair gene MLH1, and compared single cell subclones that were uninduced (MLH1-proficient) versus induced for the MLH1 shRNA (MLH1-deficient). Here we present the characterization of these MMR-inducible cell lines and validate a novel class of rhodium metalloinsertor compounds that differentially inhibit the proliferation of MMR-deficient cancer cells.     

Comparative Anatomy of Chromosomal Domains with Imprinted and Non-Imprinted Allele-Specific DNA Methylation

Allele-specific DNA methylation (ASM) is well studied in imprinted domains, but this type of epigenetic asymmetry is actually found more commonly at non-imprinted loci, where the ASM is dictated not by parent-of-origin but instead by the local haplotype. We identified loci with strong ASM in human tissues from methylation-sensitive SNP array data. Two index regions (bisulfite PCR amplicons), one between the C3orf27 and RPN1 genes in chromosome band 3q21 and the other near the VTRNA2-1 vault RNA in band 5q31, proved to be new examples of imprinted DMRs (maternal alleles methylated) while a third, between STEAP3 and C2orf76 in chromosome band 2q14, showed non-imprinted haplotype-dependent ASM. Using long-read bisulfite sequencing (bis-seq) in 8 human tissues we found that in all 3 domains the ASM is restricted to single differentially methylated regions (DMRs), each less than 2kb. The ASM in the C3orf27-RPN1 intergenic region was placenta-specific and associated with allele-specific expression of a long non-coding RNA. Strikingly, the discrete DMRs in all 3 regions overlap with binding sites for the insulator protein CTCF, which we found selectively bound to the unmethylated allele of the STEAP3-C2orf76 DMR. Methylation mapping in two additional genes with non-imprinted haplotype-dependent ASM, ELK3 and CYP2A7, showed that the CYP2A7 DMR also overlaps a CTCF site. Thus, two features of imprinted domains, highly localized DMRs and allele-specific insulator occupancy by CTCF, can also be found in chromosomal domains with non-imprinted ASM. Arguing for biological importance, our analysis of published whole genome bis-seq data from hES cells revealed multiple genome-wide association study (GWAS) peaks near CTCF binding sites with ASM.