Multiplexed drug testing of micro-dissected tumors using a microfluidic platform with integrated electrochemical aptasensors

摘要：电化学生物传感器与微电子技术的集成为生物传感器的研究提供了一条独特的途径。 小型化，低成本的系统，将生物分子传感与数字计算，编程， 通信在这里，我们提出将基于电化学适体的传感器（“适体传感器”）整合到一个 微流控多孔平台，以实现多时间点、高度平行的细胞死亡和细胞因子读数 在药物治疗期间和之后从完整的肿瘤活检组织中分泌。我们的平台将允许收集 大量的分子数据是机器学习方法进行药物测试所需要的。 癌症药物测试-癌症药物开发和个性化肿瘤学的中心过程-通常 因为它通常依赖于细胞培养或动物的研究，而这些研究缺乏人类的参与。 肿瘤微环境（TME）。在每年进行临床试验的约1，000种药物中，只有<4%的癌症药物 通过安全性和有效性测试;超过一半的失败是由于缺乏有效性。因此，平均而言， 将一种药物推向市场需要10年以上的时间，花费超过10亿美元，往往导致药物价格高昂。在 在过去的十年中，患者衍生的类器官和芯片上器官等技术带来了一些希望。 然而，这些方法具有比传统细胞培养低得多的通量，并且通常不能用于细胞培养。 完全重建完整组织的TME。这些限制是个性化的基本障碍 通常需要根据患者的独特TME定制的治疗方法， 联合免疫疗法，目标是TME，数量呈指数增长。因此 为了帮助肿瘤学转型，迫切需要一种保留人类TME的药物测试模式 进入一个更便宜和快速发展的治疗阶段。 Folch和Gujral实验室开发了一种完整组织微流体药物测试平台，该平台基于定期- 用组织机械切割的大小、长方体形状的显微解剖组织（称为“长方体 直升机在不到一个小时的时间里，可以从约1 cm 3的固体中产生超过10，000个长方体（约400 µ m宽） 肿瘤长方体从不解离并保留大部分天然TME（例如，免疫细胞和免疫球蛋白）。 微脉管系统）。该平台是一种用户友好的多孔装置，可以微流体捕获和选择性地 处理大量长方体。在这里，我们建议将电化学适体传感器整合到我们的长方体中 该平台能够自动、多时间点监测分泌的化合物（细胞因子或细胞死亡 指示器）以及从大长方体阵列直接实现电读出。 作为概念验证，我们将使用96孔格式的小鼠肿瘤和患者样本的长方体。我们将 使用结肠直肠癌（CRC）肝转移的小鼠模型和患者样品。使用机器学习 技术，我们将实现一个概念验证的药物屏幕上的小鼠立方体和概念验证 对患者长方体的联合免疫治疗药物评价。

ABSTRACT: The integration of electrochemical biosensors with microelectronics offers a unique avenue for miniaturized, low-cost systems that merge biomolecular sensing with digital computing, programming, and communication. Here we propose to integrate electrochemical aptamer-based sensors (“aptasensors”) into a microfluidic multi-well platform to enable multi-time-point, highly parallel readouts of cell death and cytokine secretion from intact tumor biopsies during and after drug treatment. Our platform will allow for gathering the large amounts of molecular data that are needed for machine learning approaches to drug testing. Cancer drug testing – a central process in cancer drug development and personalized oncology – is often inaccurate and inefficient because it typically relies on studies in cell cultures or animals that lack the human tumor microenvironment (TME). Only <4% of cancer drugs out of the ~1,000 drugs in clinical trials each year pass the safety and efficacy tests; more than half of the failures are due to lack of efficacy. Hence, on average, bringing a drug to market takes >10 years and costs >$1 billion, often resulting in high prices for the drugs. In the last decade, technologies such as patient-derived organoids and organs-on-chips have brought some hope. However, these approaches have much lower throughput than traditional cell cultures and are generally unable to fully recreate the TME of an intact tissue. These limitations are a fundamental hurdle for the personalization of therapies which often need to be customized to the unique TME of the patient, and also for the development of combination immunotherapies, which target the TME and are exponentially increasing in number. Thus, a different paradigm for drug testing that preserves the human TME is critically needed to help transition oncology into a stage of more affordable and rapidly evolving treatments. The Folch and Gujral labs have developed an intact-tissue microfluidic drug testing platform based on regularly- sized, cuboidal-shaped microdissected tissues (referred to as “cuboids”) that are mechanically cut with a tissue chopper. In under an hour, more than 10,000 cuboids (~400 µm-wide) can be produced from ~1 cm3 of solid tumor. The cuboids are never dissociated and retain much of the native TME (e.g. immune cells and microvasculature). The platform is a user-friendly multi-well device that can microfluidically trap and selectively treat a large array of cuboids. Here we propose to integrate electrochemical aptasensors into our cuboids platform to enable the automated, multi-time-point monitoring of secreted compounds (cytokines or cell death indicators) within the wells and the straightforward implementation of electrical readouts from large cuboid arrays. As a proof of concept, we will use cuboids from mouse tumors and patient samples in a 96-well format. We will use a mouse model and patient samples of colorectal cancer (CRC) liver metastases. Using machine learning techniques, we will implement a proof-of-concept drug screen on mouse cuboids and a proof-of-concept combination immunotherapy drug evaluation on patient cuboids.