Engineering chimeric antigen receptor T cells towards enhanced safety and efficacy in cancer immunotherapy

Adoptive transfer of chimeric antigen receptor (CAR) engineered T cells is an emerging cancer immunotherapy and has achieved many encouraging results in clinical trials for cancer treatment, especially in treating hematological malignancies. CAR-T cells can be redirected to specifically recognize tumor-associated antigens (TAAs) and induce high levels of antitumor activity. Despite its great potential, CAR-T cell therapy has some major limitations to overcome. CAR-T cells often have less satisfactory efficacy in solid tumor treatment and they may cause dangerous side effects sometime, such as on-target off-tumor toxicities resulting from low-level expression of TAAs in healthy tissues. These adverse effects have raised considerable safety concerns and limited the clinical application of this otherwise very promising therapeutic modality. Further optimization of CAR to enhance its safety and efficacy could help to expand this therapeutic success to more patients and more cancer types. ❧ The first two studies included in this dissertation use epithelial growth factor receptor (EGFR) specific CARs as model. EGFR is overexpressed in many types of tumor and is an attractive target antigen. However, since it is expressed in low levels in healthy tissues, most of targeted therapeutics for EGFR have considerable side effects. ❧ To minimize the on-target off-tumor toxicities of anti-EGFR CAR T cells, we have designed a masked CAR (mCAR) in the first study. It consists of a masking peptide that blocks the antigen-binding site and a protease-sensitive linker that can be cleaved in the tumor microenvironment. mCAR T cells can only recognize target antigens at the tumor site. They showed dramatically reduced antigen binding and antigen-specific activation in the absence of proteases, but normal levels of binding and activity upon treatment with certain proteases in vitro. Masked CAR-T cells also showed antitumor efficacy in vivo comparable to that of unmasked CAR. Our study demonstrates the feasibility of improving the safety profile of conventional CARs and may also inspire future design of CAR molecules targeting broadly expressed TAAs. ❧ In the second study, we designed novel adnectin-based CARs. Most CAR constructs used in clinical trials incorporate single-chain variable fragment (scFv) as the extracellular antigen recognition domain. However, the immunogenicity of nonhuman scFv could cause host rejection against CAR T cells and compromise their persistence and efficacy. The limited availability of scFvs and slow discovery of new monoclonal antibodies also limit the development of novel CAR constructs. Adnectin, a class of affinity molecules derived from the 10th type III domain of human fibronectin, can be efficiently adapted to target proteins of interest with high specificity and affinity. Adnectin could be an alternative to scFv as an antigen binding moiety in the design of CAR molecules. We constructed adnectin-based CARs targeting EGFR and compared the results to scFv-based CAR. Adnectin-based CAR T cells exhibited equivalent cell-killing activity against target H292 lung cancer cells in vitro and had comparable antitumor efficacy in xenograft tumor-bearing mice in vivo. With optimal affinity-tuning, adnectin-based CAR showed higher selectivity to target cells with high EGFR expression compared to those with low expression. This new design of adnectin CARs can potentially facilitate the development of T cell immunotherapy for cancer and other diseases. ❧ Hematopoietic stem cells (HSCs) engineered with CARs could have long-term engraftment and hence continuously replenish anti-cancer effector cells, such as myeloid, NK and T cells with CAR expression. In the third study, we mainly focused on enhancing ex vivo expansion of cord blood HSCs. Umbilical cord blood (UCB) has been established as an important source of hematopoietic stem cells (HSC) for transplantation during the past two decades. However, cord blood transplantation in adult recipients is severely limited due to the insufficient number of HSC in each UCB unit. Many efforts have been made to expand HSCs ex vivo to achieve sufficient cell dose. Yet the most common expansion conditions using serum free medium with cytokine cocktail often caused compromised re-constitutive potential of the cells after expansion. In this study, we use a culture system based on fibrin gel, a biologically compatible matrix, to culture cord blood HSCs. Compared to the conventional culture condition, fibrin gel based culture system can promote ex vivo expansion of UCB HSC with less differentiation and loss of repopulating ability. Our results suggest that fibrin gel matrix is suitable for ex vivo expansion of HSCs and it may extend the clinical application of cord blood HSCs to more patients.

嵌合抗原受体（chimeric antigen receptor, CAR）工程化T细胞的过继性转移是新兴的癌症免疫疗法，在癌症治疗的临床试验中已取得诸多喜人成果，尤其在血液系统恶性肿瘤的治疗中表现突出。CAR-T细胞可被重定向以特异性识别肿瘤相关抗原（tumor-associated antigens, TAAs），并介导高水平的抗肿瘤活性。尽管其潜力巨大，但CAR-T细胞疗法仍存在多项亟待攻克的关键局限：在实体瘤治疗中，其疗效往往不尽如人意，且有时会引发严重不良反应，例如因健康组织低水平表达TAA而导致的靶向脱瘤毒性（on-target off-tumor toxicities）。这类不良事件引发了广泛的安全顾虑，限制了这一极具前景的治疗手段的临床应用。进一步优化CAR结构以提升其安全性与疗效，有望将该疗法的成功惠及更多患者与更多癌种。
本论文纳入的前两项研究以表皮生长因子受体（epithelial growth factor receptor, EGFR）特异性CAR为模型。EGFR在多种肿瘤中过表达，是极具吸引力的靶标抗原。但由于其在健康组织中也呈低水平表达，多数针对EGFR的靶向治疗药物均存在较为显著的不良反应。
为降低抗EGFR CAR-T细胞的靶向脱瘤毒性（on-target off-tumor toxicities），我们在第一项研究中设计了伪装型CAR（masked CAR, mCAR）。该结构包含一个可阻断抗原结合位点的掩蔽肽，以及一个可在肿瘤微环境中被切割的蛋白酶敏感接头。mCAR-T细胞仅能在肿瘤部位识别靶标抗原：在无蛋白酶的环境中，其抗原结合与抗原特异性激活能力显著受限；而在体外经特定蛋白酶处理后，则可恢复正常的结合与活性水平。体内实验显示，伪装型CAR-T细胞的抗肿瘤效果与未修饰的CAR-T细胞相当。本研究证实了优化传统CAR安全性的可行性，也可为靶向广泛表达TAA的CAR分子后续设计提供思路。
在第二项研究中，我们设计了新型纤连蛋白体（adnectin）基CAR。当前临床试验中多数CAR构建体均采用单链可变片段（single-chain variable fragment, scFv）作为细胞外抗原识别结构域，但非人源scFv的免疫原性可能引发宿主对CAR-T细胞的排斥反应，损害其存续性与疗效；此外，scFv的获取难度有限、新型单克隆抗体开发周期漫长，也制约了新型CAR构建体的研发。纤连蛋白体是一类源自人纤连蛋白第Ⅲ型10结构域的亲和分子，可被高效改造以高特异性、高亲和力靶向目标蛋白，有望作为scFv的替代物作为CAR设计中的抗原结合元件。我们构建了靶向EGFR的纤连蛋白体基CAR，并与scFv基CAR进行对比。体外实验显示，纤连蛋白体基CAR-T细胞对靶标H292肺癌细胞的杀伤活性与scFv基CAR-T细胞相当；在异种移植瘤荷瘤小鼠模型中，二者的抗肿瘤效果亦无显著差异。经过最优亲和力优化后，纤连蛋白体基CAR对高EGFR表达靶细胞的选择性优于低表达细胞。这一新型纤连蛋白体CAR设计有望推动癌症及其他疾病的T细胞免疫疗法开发。
第三项研究则聚焦于增强脐带血造血干细胞（hematopoietic stem cells, HSCs）的体外扩增。经CAR修饰的造血干细胞可实现长期植入，从而持续补充抗癌效应细胞（如表达CAR的髓系细胞、自然杀伤细胞（natural killer, NK）及T细胞）。在过去二十年中，脐带血（umbilical cord blood, UCB）已被确立为造血干细胞移植的重要造血干细胞来源。但由于每份脐带血单位中含有的造血干细胞数量不足，成人接受脐带血移植的应用受到严重限制。诸多研究致力于通过体外扩增获取足够剂量的造血干细胞，但当前最常用的无血清培养基联合细胞因子组合的扩增方案，往往会导致扩增后细胞的重建造血潜能受损。本研究采用基于纤维蛋白胶（一种生物相容性基质）的培养体系培养脐带血造血干细胞。与传统培养条件相比，基于纤维蛋白胶的培养体系可促进脐带血造血干细胞的体外扩增，同时减少细胞分化与重建造血能力的丢失。研究结果表明，纤维蛋白胶基质适用于造血干细胞的体外扩增，有望将脐带血造血干细胞的临床应用拓展至更多患者群体。