An NK-like CAR T cell transition in CAR T cell dysfunction.

Chimeric antigen receptor (CAR) T cell therapy has achieved remarkable success in hematological malignancies but remains ineffective in solid tumors, due in part to CAR T cell exhaustion in the solid tumor microenvironment. To study dysfunction of mesothelin-redirected CAR T cells in pancreatic cancer, we establish a robust model of continuous antigen exposure that recapitulates hallmark features of T cell exhaustion and discover, both in vitro and in CAR T cell patients, that CAR dysregulation is associated with a CD8+ T-to-NK-like T cell transition. Furthermore, we identify a gene signature defining CAR and TCR dysregulation and transcription factors, including SOX4 and ID3 as key regulators of CAR T cell exhaustion. Our findings shed light on the plasticity of human CAR T cells and demonstrate that genetic downmodulation of ID3 and SOX4 expression can improve the efficacy of CAR T cell therapy in solid tumors by preventing or delaying CAR T cell dysfunction.

Plasmodium Infection Promotes Genomic Instability and AID-Dependent B Cell Lymphoma.

Chronic infection with Plasmodium falciparum was epidemiologically associated with endemic Burkitt's lymphoma, a mature B cell cancer characterized by chromosome translocation between the c-myc oncogene and Igh, over 50 years ago. Whether infection promotes B cell lymphoma, and if so by which mechanism, remains unknown. To investigate the relationship between parasitic disease and lymphomagenesis, we used Plasmodium chabaudi (Pc) to produce chronic malaria infection in mice. Pc induces prolonged expansion of germinal centers (GCs), unique compartments in which B cells undergo rapid clonal expansion and express activation-induced cytidine deaminase (AID), a DNA mutator. GC B cells elicited during Pc infection suffer widespread DNA damage, leading to chromosome translocations. Although infection does not change the overall rate, it modifies lymphomagenesis to favor mature B cell lymphomas that are AID dependent and show chromosome translocations. Thus, malaria infection favors mature B cell cancers by eliciting protracted AID expression in GC B cells. PAPERCLIP.

Potentiating adoptive cell therapy using synthetic IL-9 receptors.

Synthetic receptor signalling has the potential to endow adoptively transferred T cells with new functions that overcome major barriers in the treatment of solid tumours, including the need for conditioning chemotherapy1,2. Here we designed chimeric receptors that have an orthogonal IL-2 receptor extracellular domain (ECD) fused with the intracellular domain (ICD) of receptors for common γ-chain (γc) cytokines IL-4, IL-7, IL-9 and IL-21 such that the orthogonal IL-2 cytokine elicits the corresponding γc cytokine signal. Of these, T cells that signal through the chimeric orthogonal IL-2Rß-ECD-IL-9R-ICD (o9R) are distinguished by the concomitant activation of STAT1, STAT3 and STAT5 and assume characteristics of stem cell memory and effector T cells. Compared to o2R T cells, o9R T cells have superior anti-tumour efficacy in two recalcitrant syngeneic mouse solid tumour models of melanoma and pancreatic cancer and are effective even in the absence of conditioning lymphodepletion. Therefore, by repurposing IL-9R signalling using a chimeric orthogonal cytokine receptor, T cells gain new functions, and this results in improved anti-tumour activity for hard-to-treat solid tumours.

Enhancing chimeric antigen receptor T cell therapy by modulating the p53 signaling network with Δ133p53α.

Chimeric antigen receptor (CAR) T cell dysfunction is a major barrier to achieving lasting remission in hematologic cancers, especially in chronic lymphocytic leukemia (CLL). We have shown previously that Δ133p53α, an endogenous isoform of the human TP53 gene, decreases in expression with age in human T cells, and that reconstitution of Δ133p53α in poorly functional T cells can rescue proliferation [A. M. Mondal et al., J. Clin. Invest. 123, 5247-5257 (2013)]. Although Δ133p53α lacks a transactivation domain, it can form heterooligomers with full-length p53 and modulate the p53-mediated stress response [I. Horikawa et al., Cell Death Differ. 24, 1017-1028 (2017)]. Here, we show that constitutive expression of Δ133p53α potentiates the anti-tumor activity of CD19-directed CAR T cells and limits dysfunction under conditions of high tumor burden and metabolic stress. We demonstrate that Δ133p53α-expressing CAR T cells exhibit a robust metabolic phenotype, maintaining the ability to execute effector functions and continue proliferating under nutrient-limiting conditions, in part due to upregulation of critical biosynthetic processes and improved mitochondrial function. Importantly, we show that our strategy to constitutively express Δ133p53α improves the anti-tumor efficacy of CAR T cells generated from CLL patients that previously failed CAR T cell therapy. More broadly, our results point to the potential role of the p53-mediated stress response in limiting the prolonged antitumor functions required for complete tumor clearance in patients with high disease burden, suggesting that modulation of the p53 signaling network with Δ133p53α may represent a translationally viable strategy for improving CAR T cell therapy.

Assessing computational predictions of antimicrobial resistance phenotypes from microbial genomes.

The advent of rapid whole-genome sequencing has created new opportunities for computational prediction of antimicrobial resistance (AMR) phenotypes from genomic data. Both rule-based and machine learning (ML) approaches have been explored for this task, but systematic benchmarking is still needed. Here, we evaluated four state-of-the-art ML methods (Kover, PhenotypeSeeker, Seq2Geno2Pheno and Aytan-Aktug), an ML baseline and the rule-based ResFinder by training and testing each of them across 78 species-antibiotic datasets, using a rigorous benchmarking workflow that integrates three evaluation approaches, each paired with three distinct sample splitting methods. Our analysis revealed considerable variation in the performance across techniques and datasets. Whereas ML methods generally excelled for closely related strains, ResFinder excelled for handling divergent genomes. Overall, Kover most frequently ranked top among the ML approaches, followed by PhenotypeSeeker and Seq2Geno2Pheno. AMR phenotypes for antibiotic classes such as macrolides and sulfonamides were predicted with the highest accuracies. The quality of predictions varied substantially across species-antibiotic combinations, particularly for beta-lactams; across species, resistance phenotyping of the beta-lactams compound, aztreonam, amoxicillin/clavulanic acid, cefoxitin, ceftazidime and piperacillin/tazobactam, alongside tetracyclines demonstrated more variable performance than the other benchmarked antibiotics. By organism, Campylobacter jejuni and Enterococcus faecium phenotypes were more robustly predicted than those of Escherichia coli, Staphylococcus aureus, Salmonella enterica, Neisseria gonorrhoeae, Klebsiella pneumoniae, Pseudomonas aeruginosa, Acinetobacter baumannii, Streptococcus pneumoniae and Mycobacterium tuberculosis. In addition, our study provides software recommendations for each species-antibiotic combination. It furthermore highlights the need for optimization for robust clinical applications, particularly for strains that diverge substantially from those used for training.

Site-specific serology unveils cross-reactive monoclonal antibodies targeting influenza A hemagglutinin epitopes.

Efficient identification of human monoclonal antibodies targeting specific antigenic sites is pivotal for advancing vaccines and immunotherapies against infectious diseases and cancer. Existing screening techniques, however, limit our ability to discover monoclonal antibodies with desired specificity. In this study, we introduce a novel method, blocking of binding (BoB) enzyme-linked immunoassay (ELISA), enabling the detection of high-avidity human antibodies directed to defined epitopes. Leveraging BoB-ELISA, we analyzed the antibody response to known epitopes of influenza A hemagglutinin (HA) in the serum of vaccinated donors. Our findings revealed that serum antibodies targeting head epitopes were immunodominant, whereas antibodies against the stem epitope, although subdominant, were highly prevalent. Extending our analysis across multiple HA strains, we examined the cross-reactive antibody response targeting the stem epitope. Importantly, employing BoB-ELISA we identified donors harboring potent heterosubtypic antibodies targeting the HA stem. B-cell clonal analysis of these donors revealed three novel, genealogically independent monoclonal antibodies with broad cross-reactivity to multiple HAs. In summary, we demonstrated that BoB-ELISA is a sensitive technique for measuring B-cell epitope immunogenicity, enabling the identification of novel monoclonal antibodies with implications for enhanced vaccine development and immunotherapies.

Tricyclic 1,4-Diphosphinines: Local vs. Global Aromaticity.

1D and 2D NICS π . zz SOM ${{{\rm { NICS}}}_{{\rm { {\rm \pi}{}}}{\rm { .zz}}}^{{\rm { SOM}}}}$ (Sigma only model) calculations were performed on recently established tricyclic 1,4-diphosphinines as well as related benzene and pyrazine derivatives. The study was extended to evaluate the effect of the fused rings on the overall aromatic properties with a special focus on functional groups such as carbenes. The effect of non-aromatic heterocycles on the local ring current of the central ring is small, while aromatic heterocycles (e. g. NHC, imidazolium) lead to a global aromaticity. A higher sulfur content of the adjacent five-membered rings reduces the central ring current. The comparison to related tricyclic benzene and pyrazine derivatives showed that the 1,4-diphosphinine systems resemble more closely the situation in the benzene derivatives than the pyrazines. The effect of charged systems was studied using bis(TTF)-fused 1,4-diphosphinines and, according to NICS π . zz SOM ${{{\rm { NICS}}}_{{\rm { {\rm \pi}{}}}{\rm { .zz}}}^{{\rm { SOM}}}}$ values, the neutral form doesn't possess significant aromaticity but the tetracation resembles the global aromatic situation observed for other heterocycles in this study.

Thulium fiber laser vs. holmium laser enucleation of the prostate: results of a prospective randomized non-inferiority trial.

PURPOSE: Holmium laser enucleation of the prostate (HoLEP) represents the current standard procedure for size-independent surgical therapy of benign prostatic obstruction (BPO). With advent of the novel laser technology thulium fiber laser (TFL), we hypothesized that the functional outcome of TFL enucleation of the prostate (ThuFLEP) is non-inferior compared to HoLEP. METHODS: From October 2021 to October 2022, 150 patients with BPO were recruited for the prospective randomized trial in accordance with CONSORT. Stratified randomization into the arms ThuFLEP (n = 74) or HoLEP (n = 76) was carried out. The primary endpoint was non-inferior international prostate symptom score (IPSS) and quality of life (QoL) at three months after treatment. Secondary endpoints were rates of complications, peak flow, residual urine and operation times. RESULTS: Preoperative characteristics showed no significant differences. Overall IPSS and QoL improved from 21 to 8 and 4 to 1.5, respectively, after three months of follow-up. No statistically significant differences between ThuFLEP and HoLEP were observed regarding median postoperative IPSS (8.5 vs. 7, p > 0.9), QoL (1 vs. 2, p = 0.6), residual urine (48 vs. 30ml, p = 0.065) and peak flow (19 vs. 17ml/s, p > 0.9). Similarly, safety profile was comparable with no statistically significant differences regarding rate of major complications (5.3 vs. 5.4%, p = 0.5), laser hemostasis time (3 vs. 2min, p = 0.2), use of additive electric coagulation (74 vs. 87%, p = 0.06) or electric coagulation time (8 vs. 8min, p = 0.4). CONCLUSIONS: In this prospective, randomized trial ThuFLEP showed non-inferior results compared to HoLEP in terms of functional outcomes measured by IPSS and QoL as primary endpoint. TRIAL REGISTRATION NUMBER: DRKS00032699 (18.09.2023, retrospectively registered).

Hyperfine-resolved rotational spectroscopy of HCNH.

The rotational spectrum of the molecular ion HCNH+ is revisited using double-resonance spectroscopy in an ion trap apparatus, with six transitions measured between 74 and 445 GHz. Due to the cryogenic temperature of the trap, the hyperfine splittings caused by the 14N quadrupolar nucleus were resolved for transitions up to J = 4 ← 3, allowing for a refinement of the spectroscopic parameters previously reported, especially the quadrupole coupling constant eQq.

The weepy cry - short neural signal bursts in intraoperative neuromonitoring.

PURPOSE: This study aimed to establish an in-vitro alternative to existing in-vivo systems to analyze nerve dysfunction using continuous neuromonitoring (C-IONM). METHODS: Three hundred sixty-three recurrent laryngeal nerves (RLN) (N(pigs) = 304, N(cattle) = 59) from food industry cadavers were exposed by microsurgical dissection following euthanasia. After rinsing with Ringer's lactate, they were tempered at 22 °C. Signal evaluation using C-IONM was performed for 10 min at 2 min intervals, and traction forces of up to 2N were applied for a median time of 60 s. Based on their post-traumatic electrophysiological response, RLNs were classified into four groups: Group A: Amplitude ≥ 100%, Group B: loss of function (LOS) 0-25%, Group C: ≥ 25-50%, and Group D: > 50%. RESULTS: A viable in-vitro neuromonitoring system was established. The median post-traumatic amplitudes were 112%, 88%, 59%, and 9% in groups A, B, C, and D, respectively. A time-dependent further dynamic LOS was observed during the 10 min after cessation of strain. Surprisingly, following initial post-traumatic hyperconductivity, complete LOS occurred in up to 20% of the nerves in group A. The critical threshold for triggering LOS was 2N in all four groups, resulting in immediate paralysis of up to 51.4% of the nerves studied. CONCLUSION: Consistent with in-vivo studies, RLN exhibit significant intrinsic electrophysiological variability in response to tensile forces. Moreover, nerve damage progresses even after the complete cessation of strain. Up to 20% of nerves with transiently increased post-traumatic amplitudes above 100% developed complete LOS, which we termed the "weepy cry." This time-delayed response must be considered during the interpretation of C-IONM signals.

Rapid and accurate identification of ribosomal RNA sequences via deep learning.

Advances in transcriptomic and translatomic techniques enable in-depth studies of RNA activity profiles and RNA-based regulatory mechanisms. Ribosomal RNA (rRNA) sequences are highly abundant among cellular RNA, but if the target sequences do not include polyadenylation, these cannot be easily removed in library preparation, requiring their post-hoc removal with computational techniques to accelerate and improve downstream analyses. Here, we describe RiboDetector, a novel software based on a Bi-directional Long Short-Term Memory (BiLSTM) neural network, which rapidly and accurately identifies rRNA reads from transcriptomic, metagenomic, metatranscriptomic, noncoding RNA, and ribosome profiling sequence data. Compared with state-of-the-art approaches, RiboDetector produced at least six times fewer misclassifications on the benchmark datasets. Importantly, the few false positives of RiboDetector were not enriched in certain Gene Ontology (GO) terms, suggesting a low bias for downstream functional profiling. RiboDetector also demonstrated a remarkable generalizability for detecting novel rRNA sequences that are divergent from the training data with sequence identities of <90%. On a personal computer, RiboDetector processed 40M reads in less than 6 min, which was ∼50 times faster in GPU mode and ∼15 times in CPU mode than other methods. RiboDetector is available under a GPL v3.0 license at https://github.com/hzi-bifo/RiboDetector.

Shear Stress-Induced AMP-Activated Protein Kinase Modulation in Endothelial Cells: Its Role in Metabolic Adaptions and Cardiovascular Disease.

Endothelial cells (ECs) line the inner surface of all blood vessels and form a barrier that facilitates the controlled transfer of nutrients and oxygen from the circulatory system to surrounding tissues. Exposed to both laminar and turbulent blood flow, ECs are continuously subject to differential mechanical stimulation. It has been well established that the shear stress associated with laminar flow (LF) is atheroprotective, while shear stress in areas with turbulent flow (TF) correlates with EC dysfunction. Moreover, ECs show metabolic adaptions to physiological changes, such as metabolic shifts from quiescence to a proliferative state during angiogenesis. The AMP-activated protein kinase (AMPK) is at the center of these phenomena. AMPK has a central role as a metabolic sensor in several cell types. Moreover, in ECs, AMPK is mechanosensitive, linking mechanosensation with metabolic adaptions. Finally, recent studies indicate that AMPK dysregulation is at the center of cardiovascular disease (CVD) and that pharmacological targeting of AMPK is a promising and novel strategy to treat CVDs such as atherosclerosis or ischemic injury. In this review, we summarize the current knowledge relevant to this topic, with a focus on shear stress-induced AMPK modulation and its consequences for vascular health and disease.

Gas-Phase Infrared Action Spectroscopy of CH2Cl+ and CH3ClH+: Likely Protagonists in Chlorine Astrochemistry.

Two fundamental halocarbon ions, CH2Cl+ and CH3ClH+, were studied in the gas phase using the FELion 22-pole ion trap apparatus and the Free Electron Laser for Infrared eXperiments (FELIX) at Radboud University, Nijmegen (the Netherlands). The vibrational bands of a total of four isotopologs, CH235,37Cl+ and CH335,37ClH+, were observed in selected wavenumber regions between 500 and 2900 cm-1 and then spectroscopically assigned based on the results of anharmonic force field calculations performed at the CCSD(T) level of theory. As the infrared photodissociation spectroscopy scheme employed probes singly Ne-tagged weakly bound complexes, complementary quantum-chemical calculations of selected species were also performed. The impact of tagging on the vibrational spectra of CH2Cl+ and CH3ClH+ is found to be virtually negligible for most bands; for CH3ClH+-Ne, the observations suggest a proton-bound structural arrangement. The experimental band positions as well as the best estimate rotational molecular parameters given in this work provide a solid basis for future spectroscopic studies at high spectral resolutions.

Enantioselective Syntheses of Wickerols A and B.

The evolution of a successful strategy for the synthesis of the strained, cage-like antiviral diterpenoids wickerols A and B is described. Initial attempts to access the carbocyclic core were surprisingly challenging and in retrospect, presaged the many detours needed to ultimately arrive at the fully adorned wickerol architecture. In most cases, conditions to trigger desired outcomes with respect to both reactivity and stereochemistry were hard-won. The successful synthesis ultimately leveraged alkenes in virtually all productive bond-forming events. A series of conjugate addition reactions generated the fused tricyclic core, a Claisen rearrangement was used to install an otherwise unmanageable methyl-bearing stereogenic center, and a Prins cyclization closed the strained bridging ring. This final reaction proved enormously interesting because the strain of the ring system permitted diversion of the presumed initial Prins product into several different scaffolds.

Evaluating assembly and variant calling software for strain-resolved analysis of large DNA viruses.

Infection with human cytomegalovirus (HCMV) can cause severe complications in immunocompromised individuals and congenitally infected children. Characterizing heterogeneous viral populations and their evolution by high-throughput sequencing of clinical specimens requires the accurate assembly of individual strains or sequence variants and suitable variant calling methods. However, the performance of most methods has not been assessed for populations composed of low divergent viral strains with large genomes, such as HCMV. In an extensive benchmarking study, we evaluated 15 assemblers and 6 variant callers on 10 lab-generated benchmark data sets created with two different library preparation protocols, to identify best practices and challenges for analyzing such data. Most assemblers, especially metaSPAdes and IVA, performed well across a range of metrics in recovering abundant strains. However, only one, Savage, recovered low abundant strains and in a highly fragmented manner. Two variant callers, LoFreq and VarScan2, excelled across all strain abundances. Both shared a large fraction of false positive variant calls, which were strongly enriched in T to G changes in a 'G.G' context. The magnitude of this context-dependent systematic error is linked to the experimental protocol. We provide all benchmarking data, results and the entire benchmarking workflow named QuasiModo, Quasispecies Metric determination on omics, under the GNU General Public License v3.0 (https://github.com/hzi-bifo/Quasimodo), to enable full reproducibility and further benchmarking on these and other data.

High-resolution ro-vibrational and rotational spectroscopy of HC3O.

The ro-vibrational and pure rotational spectra of the linear ion HC3O+ have been investigated in a 4 K cryogenic ion trap instrument. For this, a novel action spectroscopic technique, called leak-out-spectroscopy (LOS, Schmid et al., J. Phys. Chem. A 2022, 126, 8111), has been utilized and characterized. In total, 45 ro-vibrational transitions within the fundamental band of the ν1 C-H stretching mode were measured with a band center at 3237.132 cm-1, as well as 34 lines from the combination band ν2 + ν4, and 41 lines tentatively identified as the combination band ν2 + ν5 + ν7, interleaved and resonant with ν1. Surprisingly, also two hot bands were detected despite the cryogenic operation temperature. Based on the novel action spectroscopy approach, a new double-resonance rotational measurement scheme was established, consisting of rotational excitation followed by vibrational excitation. Seven rotational transitions were observed between 89 and 180 GHz. Highly accurate spectroscopic parameters were extracted from a fit using all available data. In addition, a pulsed laser system has been employed to record a low resolution vibrational spectrum, in order to demonstrate the compatibility of such lasers with the LOS method.

Proposal and multicentric validation of a laparoscopic Roux-en-Y gastric bypass surgery ontology.

BACKGROUND: Phase and step annotation in surgical videos is a prerequisite for surgical scene understanding and for downstream tasks like intraoperative feedback or assistance. However, most ontologies are applied on small monocentric datasets and lack external validation. To overcome these limitations an ontology for phases and steps of laparoscopic Roux-en-Y gastric bypass (LRYGB) is proposed and validated on a multicentric dataset in terms of inter- and intra-rater reliability (inter-/intra-RR). METHODS: The proposed LRYGB ontology consists of 12 phase and 46 step definitions that are hierarchically structured. Two board certified surgeons (raters) with > 10 years of clinical experience applied the proposed ontology on two datasets: (1) StraBypass40 consists of 40 LRYGB videos from Nouvel Hôpital Civil, Strasbourg, France and (2) BernBypass70 consists of 70 LRYGB videos from Inselspital, Bern University Hospital, Bern, Switzerland. To assess inter-RR the two raters' annotations of ten randomly chosen videos from StraBypass40 and BernBypass70 each, were compared. To assess intra-RR ten randomly chosen videos were annotated twice by the same rater and annotations were compared. Inter-RR was calculated using Cohen's kappa. Additionally, for inter- and intra-RR accuracy, precision, recall, F1-score, and application dependent metrics were applied. RESULTS: The mean ± SD video duration was 108 ± 33 min and 75 ± 21 min in StraBypass40 and BernBypass70, respectively. The proposed ontology shows an inter-RR of 96.8 ± 2.7% for phases and 85.4 ± 6.0% for steps on StraBypass40 and 94.9 ± 5.8% for phases and 76.1 ± 13.9% for steps on BernBypass70. The overall Cohen's kappa of inter-RR was 95.9 ± 4.3% for phases and 80.8 ± 10.0% for steps. Intra-RR showed an accuracy of 98.4 ± 1.1% for phases and 88.1 ± 8.1% for steps. CONCLUSION: The proposed ontology shows an excellent inter- and intra-RR and should therefore be implemented routinely in phase and step annotation of LRYGB.

Tubb3 expression levels are sensitive to neuronal activity changes and determine microtubule growth and kinesin-mediated transport.

Microtubules are dynamic polymers of α/ß-tubulin. They regulate cell structure, cell division, cell migration, and intracellular transport. However, functional contributions of individual tubulin isotypes are incompletely understood. The neuron-specific ß-tubulin Tubb3 displays highest expression around early postnatal periods characterized by exuberant synaptogenesis. Although Tubb3 mutations are associated with neuronal disease, including abnormal inhibitory transmission and seizure activity in patients, molecular consequences of altered Tubb3 levels are largely unknown. Likewise, it is unclear whether neuronal activity triggers Tubb3 expression changes in neurons. In this study, we initially asked whether chemical protocols to induce long-term potentiation (cLTP) affect microtubule growth and the expression of individual tubulin isotypes. We found that growing microtubules and Tubb3 expression are sensitive to changes in neuronal activity and asked for consequences of Tubb3 downregulation in neurons. Our data revealed that reduced Tubb3 levels accelerated microtubule growth in axons and dendrites. Remarkably, Tubb3 knockdown induced a specific upregulation of Tubb4 gene expression, without changing other tubulin isotypes. We further found that Tubb3 downregulation reduces tubulin polyglutamylation, increases KIF5C motility and boosts the transport of its synaptic cargo N-Cadherin, which is known to regulate synaptogenesis and long-term potentiation. Due to the large number of tubulin isotypes, we developed and applied a computational model based on a Monte Carlo simulation to understand consequences of tubulin expression changes in silico. Together, our data suggest a feedback mechanism with neuronal activity regulating tubulin expression and consequently microtubule dynamics underlying the delivery of synaptic cargoes.

Archetypal landscapes for deep neural networks.

The predictive capabilities of deep neural networks (DNNs) continue to evolve to increasingly impressive levels. However, it is still unclear how training procedures for DNNs succeed in finding parameters that produce good results for such high-dimensional and nonconvex loss functions. In particular, we wish to understand why simple optimization schemes, such as stochastic gradient descent, do not end up trapped in local minima with high loss values that would not yield useful predictions. We explain the optimizability of DNNs by characterizing the local minima and transition states of the loss-function landscape (LFL) along with their connectivity. We show that the LFL of a DNN in the shallow network or data-abundant limit is funneled, and thus easy to optimize. Crucially, in the opposite low-data/deep limit, although the number of minima increases, the landscape is characterized by many minima with similar loss values separated by low barriers. This organization is different from the hierarchical landscapes of structural glass formers and explains why minimization procedures commonly employed by the machine-learning community can navigate the LFL successfully and reach low-lying solutions.