Watching documentaries as a child, I became fascinated by how genomes encode the instructions to make all the cells of an organism. I studied Biochemistry at Oxford University as the subject seemed to provide a mechanistic understanding of living systems. During my time at Oxford, I completed my part II thesis (similar to a master's project) in Prof. Doug Higgs' lab. I learned about the regulation of gene expression during development of blood cells and the disease ATRX which causes alpha thalassemia and neurological defects in patients via misregulation of gene expression. While we were studying the effects of this disease on gene expression within the blood system, I wondered if studying both the blood and the brain may help find generalizable principles and mechanisms driving the disease. For this reason, I wanted to do my Ph.D. in a system where I could study many different types of cells. I decided to work with stem cells and transcription factors involved in the specification of cell fate.

I did my Ph.D. at Imperial College London at the MRC London Medical Sciences Center with Dr. Meng Li. I studied how midbrain dopaminergic neurons are made in developing mouse and chick brains and applied this knowledge to stem cells to create dopaminergic neurons in a dish. The hope was that these stem cell-derived dopaminergic neurons would serve as a platform for drug screening and therapeutic approaches for patients with Parkinson's disease. While I value the stem cell system, at the time, it was not the ideal system to explore how genomes encode gene expression in time and space. My stem cell cultures were often heterogenous; a mixture of neural-like cells and other cells, most commonly cardiac cells beating in the dish. And one could never truly know if the cells in the dish recapitulated the endogenous dopaminergic neurons. Through my research experiences, I thought that experimental approaches in whole developing embryos would be better suited for understanding how our genomes encode the instructions for making an organism. I set about looking for a system in which I could study enhancers in high-throughput within whole developing organisms.

Prof. Mike Levine spoke about Ciona at a British Society of Developmental Biology meeting, and I was hooked. I realized that Ciona, with its close relation to vertebrates and the power of electroporation to incorporate plasmids into millions of embryos, would be an ideal organism for whole embryo high-throughput reporter assays to study enhancers. Thus, Ciona is an ideal system to decipher how the instructions for development are encoded in our genomes.

I started my postdoc with Prof. Mike Levine in 2012. I developed a synthetic enhancer library screen (SEL-seq) to test many millions of enhancers for activity in developing Ciona (Figure 1a). I used SEL-Seq to test 2.5 million variants of a neural Otx-a enhancer to examine how this enhancer activated by two pleiotropic factors (ETS and GATA) encodes neural-specific expression within the anterior sensory vesicle and dorsal nerve cord (Farley et al., 2015) (Figure 1b). From these screens, we discovered that enhancers need low or suboptimal affinity transcription factor binding sites to correctly encode tissue-specific expression. If these low-affinity sites are replaced with high-affinity sites, then the enhancer is no longer restricted to the neural lineages but is also active in many other tissues where FGF signaling or GATA are present. These studies illustrate that the use of suboptimal affinity sites is critical to ensure that the enhancer remains under combinatorial control of ETS and GATA and is only active where the concentration of these two factors is just right. Similar results showing that low-affinity Hox sites were important for specificity in flies were also reported (Crocker et al., 2015).

We also found that the organization of sites (order, spacing, and orientation) in the endogenous Otx-a sequence is not optimal for the highest level of transcription (Farley et al., 2015). Changing the spacing between the sites within the enhancer could increase the levels of transcription, giving stronger neural expression. Indeed, optimizing the affinity and spacing within a version of the Otx-a enhancer results in a complete loss of tissue specificity; expression is no longer restricted to the neural tissue (a6.5 and b6.5 lineages) but is also seen in the notochord, endoderm, and posterior sensory vesicle (Figure 1b,c).

The use of low-affinity sites that are non-optimally organized prevents aberrant activation of the enhancers by a single factor and means that combinatorial control is required to activate transcription (Farley et al., 2015). The realization that enhancers contain incredibly degenerate binding sites was alarming as it meant enhancers were even more complex than we originally thought. Luckily, we noticed a relationship between the affinity and organization of the binding sites. Low-affinity sites within native enhancers had an organization that led to higher levels of transcriptional output, while higher-affinity sites had less optimal spacing for transcriptional output. Thus, there appeared to be an interplay between affinity and organization of binding sites (enhancer grammar) (Farley et al., 2015; Farley et al., 2016; Jindal & Farley, 2021). As a postdoc and in my own lab, we have used these grammatical rules to find tissue-specific enhancers within the genome (Farley et al., 2016; Song et al., 2023).

In my own lab at UC San Diego, I continue to harness Ciona for high-throughput enhancer screens to find rules governing enhancers. We have found signatures of enhancer grammar conserved across chordates—in Ciona, mice, and humans (Song et al., 2023). We've also expanded to look at how violations in regulatory principles governing enhancers can drive organismal-level phenotypes in Ciona and other species, such as mice and humans (Jindal et al., 2023; Lim et al., 2022). We've found that single nucleotide variants (SNVs) can increase binding site affinity driving ectopic expression and organismal phenotypes as severe as a second heart in Ciona and extra fingers in mice and humans (Jindal et al., 2023; Lim et al., 2022). Thus, the principle of suboptimization of developmental enhancers to encode tissue-specific expression, which we initially discovered in Ciona, applies to other organisms too. Furthermore, violating this principle leads to major phenotypes in tunicates and vertebrates.

I am incredibly grateful to Prof. Mike Levine for supporting me while I pursued the high-throughput enhancer screens in Ciona. If it were not for him, I would not have been able to realize these experiments, which formed the foundation of the research conducted within my own lab. While occasionally challenging, the Levine lab was also fun. I'd never been around so many people who loved enhancers as much as I did. My time in the Levine lab was full of exciting conversations and brainstorming, which helped me develop into the scientist I am today. Now in my own lab, I enjoy being surrounded by enhancerophiles all the time. So far, two graduate students and one postdoc have graduated from my lab and joined industry. My lab now consists of two postdocs, three graduate students, and three undergraduates (Figure 2). I hope many of them remain in the Ciona community. To support my research, I've been fortunate to receive the NIH New Innovator Award, NSF CAREER Award, and NHGRI R01.

小时候看纪录片，我对基因组如何编码指令来制造有机体的所有细胞着迷。我在牛津大学学习生物化学，因为这门学科似乎提供了对生命系统的机械理解。在牛津大学期间，我在 Doug Higgs 教授的实验室完成了第二部分论文（类似于硕士项目）。我了解了血细胞发育过程中基因表达的调节以及 ATRX 疾病，该疾病通过基因表达的错误调节导致患者出现α地中海贫血和神经系统缺陷。当我们研究这种疾病对血液系统内基因表达的影响时，我想知道研究血液和大脑是否有助于找到驱动该疾病的普遍原理和机制。因此，我想攻读博士学位。在一个我可以研究许多不同类型细胞的系统中。我决定研究参与细胞命运规范的干细胞和转录因子。

我完成了博士学位。在伦敦帝国学院 MRC 伦敦医学科学中心与李萌博士一起。我研究了小鼠和小鸡大脑发育过程中中脑多巴胺能神经元的形成过程，并将这些知识应用于干细胞，在培养皿中产生多巴胺能神经元。人们希望这些干细胞衍生的多巴胺能神经元能够作为帕金森病患者药物筛选和治疗方法的平台。虽然我很看重干细胞系统，但当时它并不是探索基因组如何在时间和空间上编码基因表达的理想系统。我的干细胞培养物通常是异质的；神经样细胞和其他细胞的混合物，最常见的是在培养皿中跳动的心肌细胞。人们永远无法真正知道培养皿中的细胞是否再现了内源性多巴胺能神经元。通过我的研究经验，我认为整个发育胚胎的实验方法更适合了解我们的基因组如何编码制造有机体的指令。我开始寻找一个可以在整个发育生物体中高通量研究增强子的系统。

迈克·莱文教授在英国发育生物学会会议上谈到了Ciona ，我被迷住了。我意识到Ciona与脊椎动物关系密切，并且能够通过电穿孔将质粒整合到数百万个胚胎中，因此将成为用于全胚胎高通量报告分析以研究增强子的理想生物体。因此，Ciona是一个理想的系统，可以破译我们基因组中发育指令的编码方式。

我于 2012 年开始跟随 Mike Levine 教授进行博士后研究。我开发了一个合成增强子库筛选 (SEL-seq)，以测试数百万个增强子在开发Ciona过程中的活性（图 1a）。我使用 SEL-Seq 测试了神经 Otx-a 增强子的 250 万个变体，以检查这种由两种多效性因子（ETS 和 GATA）激活的增强子如何编码前感觉囊泡和背神经索内的神经特异性表达（Farley 等人） .，  2015）（图1b）。从这些筛选中，我们发现增强子需要低或次优亲和力转录因子结合位点才能正确编码组织特异性表达。如果这些低亲和力位点被高亲和力位点取代，那么增强子就不再局限于神经谱系，而且在存在 FGF 信号传导或 GATA 的许多其他组织中也具有活性。这些研究表明，使用次优亲和位点对于确保增强子保持在 ETS 和 GATA 的组合控制之下至关重要，并且仅在这两个因子的浓度恰到好处时才具有活性。类似的结果表明低亲和力 Hox 位点对于果蝇的特异性很重要（Crocker 等人，  2015）。

我们还发现内源 Otx-a 序列中的位点组织（顺序、间距和方向）对于最高水平的转录而言并不是最佳的（Farley 等人，  2015）。改变增强子内位点之间的间距可以提高转录水平，从而产生更强的神经表达。事实上，优化 Otx-a 增强剂版本内的亲和力和间距会导致组织特异性完全丧失；表达不再局限于神经组织（a6.5 和 b6.5 谱系），还可见于脊索、内胚层和后感觉囊泡（图 1b、c）。

使用非最佳组织的低亲和力位点可以防止单个因素对增强子的异常激活，并且意味着需要组合控制来激活转录（Farley et al.,  2015）。增强子含有令人难以置信的简并结合位点的认识令人震惊，因为这意味着增强子比我们最初想象的还要复杂。幸运的是，我们注意到结合位点的亲和力和组织之间的关系。天然增强子内的低亲和力位点的组织导致更高水平的转录输出，而高亲和力位点的转录输出的最佳间距较差。因此，结合位点的亲和力和组织（增强子语法）之间似乎存在相互作用（Farley 等人，  2015；Farley 等人，  2016；Jindal 和 Farley，  2021）。作为一名博士后，在我自己的实验室中，我们使用这些语法规则来寻找基因组内的组织特异性增强子（Farley 等人，  2016；Song 等人，  2023）。

在我自己在加州大学圣地亚哥分校的实验室中，我继续利用Ciona进行高通量增强子筛选，以找到管理增强子的规则。我们发现增强子语法的特征在脊索动物中保守——海鞘、小鼠和人类（Song et al.,  2023）。我们还进一步研究了违反增强子监管原则如何驱动Ciona和其他物种（例如小鼠和人类）的有机体水平表型（Jindal 等人，  2023 年；Lim 等人，  2022 年）。我们发现，单核苷酸变异 (SNV) 可以增加结合位点亲和力，从而驱动异位表达和生物体表型，其严重程度与海鞘中的第二颗心脏以及小鼠和人类中的额外手指一样严重（Jindal 等人，  2023；Lim 等人，  2022）。因此，我们最初在玻璃海鞘中发现的发育增强子编码组织特异性表达的次优化原理也适用于其他生物体。此外，违反这一原则会导致被囊动物和脊椎动物出现主要表型。

我非常感谢 Mike Levine 教授在我追求Ciona 的高通量增强子筛选过程中对我的支持。如果没有他，我就不可能实现这些实验，这些实验构成了我自己实验室中进行的研究的基础。莱文实验室虽然偶尔充满挑战，但也很有趣。我从来没有接触过这么多像我一样热爱增强剂的人。我在莱文实验室的时光充满了激动人心的对话和头脑风暴，这帮助我成长为今天的科学家。现在在我自己的实验室里，我喜欢一直被增强子包围。到目前为止，已有两名研究生和一名博士后从我的实验室毕业并进入行业。我的实验室现在由两名博士后、三名研究生和三名本科生组成（图 2）。我希望他们中的许多人留在Ciona社区。为了支持我的研究，我很幸运地获得了 NIH 新创新者奖、NSF 职业奖和 NHGRI R01。