Engineering a High-Throughput Microfluidic System: A Vascularized 3D Tumor Model for Quantitative Antibody-Drug Conjugate Screening = 고효율 혈관화 종양 칩 기반 항체-약물 접합체(ADC)의 효능 및 안전성 정량 평가

국문 초록 (Abstract)

의학 기술의 눈부신 발전에도 불구하고, 암은 여전히 전 세계적으로 주요 사망 원인이며 공중 보건의 심각한 문제로 남아 있습니다. 암 극복의 주요 장애물 중 하나는 신약 후보 물질이 임상 시험 단계에서 높은 실패율을 보이는 것입니다. 이러한 '죽음의 계곡' 현상은 주로 전통적인 전임상 모델의 한계에서 기인합니다. 2차원 세포 배양 및 동물 실험과 같은 기존 모델들은 복잡한 인체 종양 미세환경을 정확히 모사하지 못하고 종간 차이로 인해 임상에서의 약물 효능 및 안전성을 예측하는 데 어려움이 있습니다. 특히 항체-약물 접합체(ADC)와 같은 복잡한 바이오 의약품의 평가에 이러한 한계는 더욱 두드러집니다. ADC는 종양 이질성에 대응할 수 있는 유망한 치료 전략이지만, 생리적으로 적합한 환경 내에서 표적 효능과 잠재적인 종양 외 독성을 동시에 평가할 수 있는 정교한 평가 플랫폼을 필요로 합니다.
본 학위 논문은 포괄적인 전임상 평가를 위한 진보된 미세생리 시스템(MPS)을 개발하고 검증함으로써 이러한 문제점들을 극복하는 것을 목표로 합니다. 본 연구는 먼저 고처리량 혈관 표현형 및 단백체 분석 통합(VPT) 플랫폼 개발을 통해 기반 기술을 확립하였습니다. 이 시스템 내에 3차원 혈관신생 모델을 구축하여 약물에 의해 유도된 형태학적 변화와 기저의 단백체 변화를 기전적으로 연결할 수 있었습니다. 나아가, 혈관신생 과정에서 자가포식의 역할을 규명하기 위한 연구를 수행하여 단일 세포 수준에서의 전사체 분석을 통해 공간적 이질성을 밝혔습니다.
초기 단계에서 확립된 전략을 활용하여, ADC 평가의 미충족 수요를 해결하기 위해 특화된 고효율 MPS인 TRACE (ADC 독성 및 반응 분석) 플랫폼을 개발했습니다. 이 플랫폼은 관류화 된 인체 혈관 네트워크와 환자 유래 암세포(PDC)로 구축된 3차원 종양 미세환경을 통합합니다. 자동화된 종양 스페로이드 크기 분석을 통한 표적 효능과 새로운 혈관 안정성 점수를 이용한 종양 외 독성을 동시에 정량 평가하기 위한 지표를 확립하고 검증했습니다. TRACE 플랫폼은 표적 특이적인 ADC의 내재화, 3차원 침투 및 효능을 성공적으로 입증했습니다. 특히, 해당 위암 환자에게서 관찰된 임상적 약물 저항성 프로파일을 재현하여 플랫폼의 높은 신뢰성과 중개 연구 관련성을 검증했습니다. 또한, 칩 내에서 배양된 PDC의 유전체 분석을 통해 약물 배출 펌프 및 생존 경로의 상향 조절과 같은 환자 특이적 ADC 저항성을 유도하는 기저의 분자 기전을 규명할 수 있었습니다.
결론적으로, 본 연구에서 개발된 MPS 플랫폼들은 3차원 구조, 관류 가능한 혈관 구조, 인체 세포 환경 및 환자 이질성의 부재와 같은 기존 전임상 모델의 결정적인 한계점들을 해결하는 정확하고 인체 적합성이 높은 고효율 모델을 제공합니다. 효능, 독성 및 저항성 기전의 통합적 평가를 가능하게 함으로써, 본 연구는 전임상-임상 간극을 메우고 ADC 개발의 성공률을 향상시키는 체계적인 평가법을 제공함과 동시에 환자 맞춤 약물 반응 예측 기술의 발전에 크게 기여할 잠재력을 지닙니다.

다국어 초록 (Multilingual Abstract)

Cancer remains a preeminent cause of global mortality, presenting a persistent and formidable challenge to public health despite significant strides in medical technology and therapeutic interventions. A critical bottleneck in the oncological drug development pipeline is the precipitous attrition rate of novel candidates during clinical trials, often referred to as the "valley of death." This translational failure largely stems from the inherent limitations of conventional preclinical models. Traditional two-dimensional cell cultures and animal models often fail to recapitulate the intricate complexity of the human tumor microenvironment (TME), specifically the dynamic interaction between tumor cells and the vasculature and suffer from species-specific physiological discrepancies. These limitations are particularly pronounced in the evaluation of complex biologics such as antibody-drug conjugates (ADCs), which require a highly specific assessment of both targeted efficacy and potential off-target toxicity within a physiologically relevant context.
To address these critical unmet needs, this dissertation focuses on the engineering and validation of advanced microphysiological systems (MPS) designed to facilitate comprehensive and predictive preclinical evaluations. The research first establishes a high-throughput vascular phenotyping and proteomics/transcriptomics integration platform. Leveraging a 3D angiogenesis model, proteomic analysis was utilized to verify the mechanism of drug action within the engineered model and to interpret drug-induced angiogenic phenotypes at a molecular level. Furthermore, single-cell RNA sequencing was employed to unravel the mechanistic link between angiogenesis and autophagy, specifically elucidating the distinct functional roles of autophagy based on spatial location within the angiogenic sprouts. Ultimately, this integrative approach demonstrates the necessity of integrating molecular profiling with morphological data, providing comprehensive insights that transcend the limitations of conventional image-only analysis.
Building upon these foundational technologies, the dissertation presents the development of a specialized, high-efficiency MPS explicitly tailored for the rigorous screening of antibody-drug conjugates. By integrating a functional, perfusable three-dimensional vascular network co-cultured with tumor spheroids, this platform enables the simultaneous and distinct observation of on-target therapeutic efficacy and potential off-tumor vascular toxicity within a single unified system. Specifically, the study validated the fundamental pharmacodynamics of antibody-drug conjugates, successfully capturing target-specific internalization and the complex process of three-dimensional intratumoral penetration. A robust quantitative evaluation methodology was established to define the therapeutic window by rigorously comparing the extent of tumor regression against the severity of endothelial damage. Furthermore, the incorporation of patient-derived cells allowed for the recapitulation of individual clinical heterogeneity; notably, the system accurately reproduced the lack of response in drug-resistant patient samples, while subsequent transcriptomic analysis elucidated the molecular mechanisms driving this resistance. Ultimately, these findings underscore the clinical utility of the developed platform as a powerful preclinical tool capable of predicting patient-specific responses and optimizing complex therapeutic strategies.
Collectively, this research demonstrates that the proposed microfluidic platforms offer a robust and physiologically relevant alternative to traditional preclinical models. By providing a more accurate prediction of clinical outcomes, these systems hold the potential to significantly accelerate the development of novel cancer therapeutics and advance the realization of precision medicine.

목차 (Table of Contents)

• Abstract i
• Table of Contents iv
• List of Tables viii
• List of Figures ix
• Chapter 1. Introduction 1
• 1.1. Transition from conventional two-dimensional culture to microphysiological systems 4
• 1.2. Engineering and analytical limitations in current microphysiological systems 9
• 1.3. Engineering strategies for high-throughput standardization 14
• 1.4. Integrated approach for deep biological profiling 16
• 1.5. Physiological significance of the vasculature in therapeutic assessment 19
• 1.6. Research objectives and thesis outline 21
• Chapter 2. High-throughput Platform for Molecular Profiling of Angiogenesis 24
• 2.1. Introduction 24
• 2.2. Materials and Methods 28
• 2.2.1. Device design and fabrication 28
• 2.2.2. Cell preparation 29
• 2.2.3. Transfection of siRNA or LC3B-mCherry-GFP dual reporter 29
• 2.2.4. RNA isolation from ECs and quantitative real-time PCR (qRT-PCR) 32
• 2.2.5. Hydrogel and cell patterning 33
• 2.2.6. Drug treatment 34
• 2.2.7. Real-time imaging 35
• 2.2.8. Permeability assay 35
• 2.2.9. Transendothelial electrical resistance (TEER) measurement 36
• 2.2.10. Microbead assay 37
• 2.2.11. Immunofluorescence staining and confocal imaging 37
• 2.2.12. Sample preparation for liquid chromatography with tandem mass spectrometry (LC-MS/MS) 38
• 2.2.13. LC-MS/MS analysis 40
• 2.2.14. Protein identification and label-free quantification 41
• 2.2.15. Protein relative quantification 41
• 2.2.16. Pathway related proteins 42
• 2.2.17. Sample collection from angiogenesis-on-a-chip for single-cell RNA sequencing 43
• 2.2.18. Single-cell RNA sequencing 43
• 2.2.19. Data analysis in single-cell RNA sequencing 44
• 2.2.20. The data analysis of publicly available breast cancer EC data 48
• 2.2.21. Functional enrichment and network analysis 49
• 2.2.22. Image processing and quantitative analysis on morphology of angiogenic sprouts 49
• 2.2.23. Statistical analysis 51
• 2.3. Results and Discussion 52
• 2.3.1. Open microfluidic cells and hydrogels pattern design 52
• 2.3.2. Achieving highly reproducible and uniform cell monolayer with the VPT platform 56
• 2.3.3. High-throughput angiogenesis assay with high inter-experimental reproducibility 59
• 2.3.4. Quantitative morphological analysis of vascular sprouts 63
• 2.3.5. Distinctive angiogenic patterns revealed by the VPT platform across various angiogenic inhibitors 66
• 2.3.6. Investigating drug mechanisms via protein expression analysis in the VPT platform 71
• 2.3.7. Linking protein expression changes to cellular functions, including cell senescence and angiogenesis 75
• 2.3.8. Correlating proteomic alterations with cellular morphology on the VPT platform 83
• 2.3.9. Microfluidic angiogenesis-on-a-chip reveals high activation of autophagy near endpoints of angiogenic sprouts 87
• 2.3.10. scRNA-seq identifies tip- and stalk-like ECs and proliferating ECs 90
• 2.3.11. Autophagy shows spatially differential activation patterns in the basement and leading positions of sprouts 95
• 2.3.12. Knockdown of autophagy regulators in stalk-like cells causes severe reduction in the number of tip cells in vitro 102
• 2.4. Conclusion 110
• Chapter 3. Quantitative Analysis of ADC Efficacy and Safety 111
• 3.1. Introduction 111
• 3.2. Materials and Methods 114
• 3.2.1. Patient enrollment and ethical approval 114
• 3.2.2. Preparation of PDCs from primary solid tissue 114
• 3.2.3. Preparation of PDCs from ascites 115
• 3.2.4. PDC culture 116
• 3.2.5. Cell line culture 116
• 3.2.6. Fabrication of microfluidic device prototypes 117
• 3.2.7. Large-scale manufacturing of microfluidic devices 117
• 3.2.8. Tumor spheroid formation 118
• 3.2.9. Recapitulation of the 3D vascularized TME 118
• 3.2.10. Western blot analysis 119
• 3.2.11. Application of antibody-drug conjugates 120
• 3.2.12. Immunocytochemistry and image acquisition 121
• 3.2.13. Time-lapse ADC tracking 121
• 3.2.14. Quantification of morphological vessel network features 122
• 3.2.15. RNA isolation from on-chip cultures and qRT-PCR 123
• 3.2.16. Statistical analysis 125
• 3.3. Results and Discussion 126
• 3.3.1. Reconstruction of the tumor microenvironment with a 3D perfusable vascular network 126
• 3.3.2. Validation of target-specific internalization and 3D penetration of ADCs 128
• 3.3.3. A quantitative approach for on-target efficacy and off-tumor toxicity 131
• 3.3.4. Evaluation of ADC efficacy and safety in gastric cancer cell lines 134
• 3.3.5. Evaluation of ADC on-target efficacy in gastric patient-derived cell (PDC) models 137
• 3.3.6. Evaluation of ADC off-tumor toxicity in gastric PDC models 141
• 3.3.7. Mechanistic understanding of patient resistance 146
• 3.4. Conclusion 149
• Chapter 4. Concluding Remarks 151
• Bibliography 156
• 국문 초록 181