Daily Respiratory Research Analysis
Three impactful respiratory papers span discovery-to-clinic: a Cell study demonstrates an AI language model can generate target-specific paired heavy/light-chain human antibodies to SARS‑CoV‑2, H5N1, and RSV-A; a multicenter prospective cohort shows serum HMGB1 robustly diagnoses acute exacerbations of idiopathic pulmonary fibrosis and outperforms KL‑6; and a mouse-adapted Omicron BA.5 model reveals post‑acute subpleural lung fibrosis with tertiary lymphoid structures, enabling long COVID fibros
Summary
Three impactful respiratory papers span discovery-to-clinic: a Cell study demonstrates an AI language model can generate target-specific paired heavy/light-chain human antibodies to SARS‑CoV‑2, H5N1, and RSV-A; a multicenter prospective cohort shows serum HMGB1 robustly diagnoses acute exacerbations of idiopathic pulmonary fibrosis and outperforms KL‑6; and a mouse-adapted Omicron BA.5 model reveals post‑acute subpleural lung fibrosis with tertiary lymphoid structures, enabling long COVID fibrosis research and countermeasure testing.
Research Themes
- AI-enabled antibody discovery for respiratory viruses
- Biomarkers for acute exacerbation in fibrotic lung disease
- Mechanistic models of post-acute COVID-19 lung fibrosis
Selected Articles
1. Generation of antigen-specific paired-chain antibodies using large language models.
This study introduces MAGE, a protein language model that generates paired heavy/light-chain human antibodies with validated binding to SARS‑CoV‑2, H5N1, and RSV-A. The approach bypasses template dependence and demonstrates broad, target-specific antibody design capability.
Impact: Represents a step-change in biologics discovery by enabling rapid, de novo design of antigen-specific paired-chain antibodies against high-consequence respiratory pathogens. This can compress timelines for outbreak response and therapeutic development.
Clinical Implications: While not yet clinically validated, this platform could accelerate generation of therapeutic and prophylactic antibodies for emerging respiratory viruses (e.g., pandemic influenza, RSV, novel coronaviruses), enabling faster translation to trials.
Key Findings
- A sequence-based PLM (MAGE) generated paired heavy/light-chain human antibodies with experimental binding to SARS‑CoV‑2, H5N1, and RSV‑A.
- Antibody design was accomplished without a starting template, indicating de novo capability.
- Generated antibodies were novel and diverse, demonstrating breadth across multiple antigens.
Methodological Strengths
- Cross-pathogen experimental validation of binding specificity (SARS‑CoV‑2, H5N1, RSV‑A).
- Paired-chain (VH/VL) generation directly addresses antibody developability considerations.
Limitations
- Lacks in vivo neutralization and efficacy data; binding does not guarantee therapeutic activity.
- Manufacturability, stability, and immunogenicity were not assessed.
Future Directions: Prospectively validate neutralization and protection in relevant animal models and early-phase trials; integrate developability filters and multi-objective optimization (affinity, specificity, stability) into the LLM pipeline.
The traditional process of antibody discovery is limited by inefficiency, high costs, and low success rates. Recent approaches employing artificial intelligence (AI) have been developed to optimize existing antibodies and generate antibody sequences in a target-agnostic manner. In this work, we present MAGE (monoclonal antibody generator), a sequence-based protein language model (PLM) fine-tuned for the task of generating paired human variable heavy- and light-chain antibody sequences against targets of interest. We show that MAGE can generate novel and diverse antibody sequences with experimentally validated binding specificity against SARS-CoV-2, an emerging avian influenza H5N1, and respiratory syncytial virus A (RSV-A). MAGE represents a first-in-class model capable of designing human antibodies against multiple targets with no starting template.
2. Diagnostic utility of high-mobility group box 1 for acute exacerbations of idiopathic pulmonary fibrosis.
In a multicenter prospective cohort of 269 IPF patients with 779 HMGB1 measurements, serial serum HMGB1 showed high diagnostic accuracy for AE‑IPF (AUC >0.75) and outperformed KL‑6. This supports HMGB1 as a clinically actionable biomarker to trigger earlier evaluation and treatment.
Impact: Addresses a key unmet need by providing a robust serum biomarker for AE‑IPF, a high-mortality event where early recognition is critical. Outperforming KL‑6 could change diagnostic pathways.
Clinical Implications: Incorporating serial HMGB1 testing could enable earlier AE detection and triage, prompt initiation of supportive and antifibrotic strategies, and better risk stratification during follow‑up.
Key Findings
- Serial serum HMGB1 measurements achieved high diagnostic accuracy for AE‑IPF across four analytic approaches (AUC >0.75).
- HMGB1 outperformed KL‑6 in diagnosing AE‑IPF in head-to-head comparison.
- Prospective multicenter cohort with 269 patients and 505.6 person‑years of follow‑up recorded 46 AE events linked to HMGB1 levels.
Methodological Strengths
- Prospective, multicenter design with serial biomarker assessment.
- Direct comparison against a widely used biomarker (KL‑6) strengthens clinical interpretability.
Limitations
- Single-country (Japan) cohort may limit generalizability; assay standardization across platforms is needed.
- Cut-off optimization and integration into clinical decision pathways require external validation.
Future Directions: Validate HMGB1 thresholds internationally, assess combination with clinical/imaging predictors, and test whether HMGB1-guided care improves time-to-diagnosis and outcomes.
There are no established biomarkers for acute exacerbations in idiopathic pulmonary fibrosis (AE-IPF). The bronchoalveolar lavage fluid of patients with IPF notably has elevations in high-mobility group box 1 protein (HMGB1), a nuclear non-histone chromosomal protein that functions as an alarmin that perpetuates the inflammatory process. This study investigated the potential of serum HMGB1 levels as a diagnostic marker for AE-IPF. This prospective, multicenter, observational cohort study included 779 HMGB1 readings from 269 Japanese patients with IPF. Diagnostic performance was assessed using four different methods of recording serial HMGB1 measurements rather than relying on a simple comparison between stable IPF and AE-IPF. Additionally, KL-6 was measured in cases of stable IPF and AE-IPF. A comparative analysis for the usefulness of HMGB1 and KL-6 as biomarkers for AE-IPF was performed. The cohort accounted for 505.6 person-years, with a mean follow-up duration of 1.88 years. A total of 46 cases with AE were recorded with their corresponding HMGB1 levels. All four diagnostic methods examined had high diagnostic accuracy (area under the curve > 0.75). HMGB1 had significantly better diagnostic performance than KL-6. HMGB1 demonstrated high diagnostic utility in AE-IPF, which can be used to facilitate earlier diagnosis and treatment.
3. Mouse-adapted SARS-CoV-2 Omicron BA.5 infection induces post-acute lung fibrosis in BALB/c mice.
A mouse-adapted BA.5 virus causes acute disease and, in survivors, subpleural fibrosis with tertiary lymphoid structures up to 107 days, modeling post-acute lung sequelae. Prophylactic monoclonal antibodies protected against BA.5-induced lung disease, and sera showed variant-specific neutralization profiles.
Impact: Provides a needed in vivo platform to dissect mechanisms and test interventions for long COVID lung fibrosis, an urgent unmet need in respiratory medicine.
Clinical Implications: Although preclinical, this model enables evaluation of antifibrotic strategies, variant‑matched antibody prophylaxis, and immunomodulators targeting chronic lung sequelae.
Key Findings
- Mouse-adapted BA.5 infection induced subpleural lung fibrosis with tertiary lymphoid structures persisting to 107 days post-infection.
- Prophylactic administration of pre-clinical monoclonal antibodies conferred robust protection from BA.5-induced lung disease.
- Convalescent sera neutralized BA.5 strongly but showed reduced titers against early epidemic and XBB.1.5 variants.
Methodological Strengths
- Longitudinal in vivo characterization out to 107 days with histopathology and functional readouts.
- Intervention testing with monoclonal antibodies demonstrates model responsiveness.
Limitations
- Mouse adaptation and species differences limit direct extrapolation to humans.
- High-dose challenges and prophylaxis-focused testing may not mirror typical clinical scenarios.
Future Directions: Apply the model to evaluate antifibrotic and immunomodulatory therapies in therapeutic (post‑infection) settings and map cellular/molecular drivers of fibrosis and tertiary lymphoid organogenesis.
UNLABELLED: Following severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Omicron BA.1, subsequent Omicron sub-lineages have continued to emerge, challenging the development of intervention and prevention strategies, including monoclonal antibodies and vaccines. To better understand the pathogenic effects caused by Omicron BA.5 infection, we developed a mouse-adapted virus with overt disease burden in BALB/c mice. Acute disease was characterized by significant weight loss and lung dysfunction following high-dose challenges. In survivor animals that were followed through 107 days post-infection, subpleural fibrosis with associated tertiary lymphoid structures was noted. Serum from these mice demonstrated potent neutralization against BA.5, with substantially reduced neutralization titers against early epidemic, zoonotic, and more recent contemporary XBB.1.5 variants. Intervention with pre-clinical monoclonal antibodies revealed that robust protection from BA.5-induced lung disease was possible after prophylactic administration. Together, this model enables the investigation of therapeutic approaches for both acute and post-acute sequelae of COVID-19. IMPORTANCE: To best combat the evolving landscape of SARS-CoV-2 variants of interest and variants of concern, the development of effective small animal models is of critical importance. Herein, we describe the development of a model system in BALB/c mice to study the effects of SARS-CoV-2 BA.5 S gene in both acute and chronic disease manifestations. Intriguingly, we determined that fibrotic lung disease with tertiary lymphoid structures was a prominent feature in the lungs of mice that survived through the acute phase of infection. This is a prominent concern in human patients that survive the initial infection insult. As such, and most critically, the model system presented here provides researchers with an effective pathway in which long COVID manifestations and potential interventions can be studied.