Evolving Libraries of Bivalent Compounds (RMI)

描述(由申请人提供)：将准备六个库，每个库至少包含15个化合物，使库中的每个化合物都以1-2克的量存在。文库中的所有化合物都被设计成转弯模拟物，它们的侧链药效团将被设计成与蛋白质相互作用热点中最常见的天然氨基酸相对应。 我们已经开发出一种方法，允许一价化合物连接起来得到二价分子。该方法不需要偶联剂，适用于具有无保护侧链官能团的化合物。只需将两种分子的溶液(微管)与碳酸钾混合，就可以由单价分子形成二价分子的库。因此，一个由15个单价化合物组成的文库可以组装成105个二价化合物(总文库大小=105+15=120)。然而，随着单价化合物的数量(N)的增加{二价化合物的数量=n(n-L)/2}，可以得到的二价化合物的数量急剧增加。一个由90个(对应于所提出的15个化合物的6组)单价构筑块组成的文库可以给出4005个二价化合物。 从逻辑上讲，很难制造出组装4000多种二价化合物所需的单价化合物的数量；表征问题变得耗时得令人望而却步，而且这么多样品的库存也不容易保存。因此，我们提出了“进化图书馆”的概念，以保持数字可管理(1,000-3,000)，同时仍在探索多样性空间的有利区域。这是基于形成每组约100-200个二价化合物，选择导致二价命中的单价化合物，添加这些化合物的类似物以探索结构活性关系，以及从铅单价化合物、它们的类似物和一组新的单价化合物形成第二二价化合物文库。 所有产生的二价化合物都将被荧光标记。这将使得通过直接结合分析来筛选文库成为可能；与通常用于蛋白质-蛋白质相互作用靶标的竞争性结合分析相比，这些方法具有相当大的优势。首先，标记的蛋白质不是必需的。这一点很重要，因为许多标记的蛋白质太贵了，以至于不可能进行高通量筛选。其次，可以检测到弱结合，因为它们不与天然配体竞争。这种方法的另一个主要优点是标记的化合物选自可用作药理探针的检测方法。这种方法可以很容易地进行调整，以包括其他类型的标记，甚至第三种小分子。 二价分子很有价值，因为它们可以将蛋白质-蛋白质界面上的两个热点结合起来。为了实现这一点，提出了优化连接两个小分子所需的连接基尺寸的策略。随着二价化合物文库的形成，这些信息将不费吹灰之力就会出现。

DESCRIPTION (provided by applicant): Six libraries of at least 15 compounds each will be prepared such that every compound in the library is present in 1 -2 g amounts. The compounds in the library are all designed to be ¿-turn mimics, and their side chain pharmacophores will be designed to correspond to the natural amino acids that are found most frequently at hot spots for protein-protein interactions. We have developed a method that allows monovalent compounds to be joined to give bivalent molecules. This method does not require coupling agents and is applicable to compounds with unprotected side chain functional groups. Libraries of bivalent molecules can be formed from monovalent ones simply by mixing (micropipette) solutions of the two with potassium carbonate. Thus one library of 15 monovalent compounds could be assembled into 105 bivalent ones (total library size = 105 + 15 = 120). However, the number of bivalent compounds that can be made increases steeply as the number of monovalent compounds (n) increases {number of bivalent compounds = n(n - l)/2}. A library of 90 (corresponding to the 6 sets of 15 compounds proposed) monovalent building blocks could give 4005 bivalent compounds. Logistically, it is difficult to make the amounts of monovalent compounds required to assemble much more than 4,000 bivalent compounds; characterization issues become prohibitively time consuming and inventories of this many samples are not easy to keep. Consequently, we propose the concept of "evolving libraries" to keep the numbers manageable (1,000 - 3,000) while still exploring favorable regions of diversity space. This is based on formation of the bivalent compounds in sets of about 100 - 200 each, selection of the monovalent compounds that led to bivalent hits, addition of analogs of these to explore structure activity relationships, and formation of a second library of bivalent compounds from the lead monovalent compounds, their analogs, and a new set of monovalent compounds. All the bivalent compounds produced will be fluorescently labeled. This will make it possible to screen the libraries via direct binding assays; these have considerable advantages over the competitive binding assays that are often used for protein-protein interaction targets. First, labeled proteins are not required. This is important because many labeled proteins are so expensive that high throughput screens become impossible. Second, weak binders can be detected because they are not in competition with the native ligand. Another major advantage of this approach is that labeled compounds are selected from the assays that might be used as pharmacological probes. The approach could easily be adapted to include other types of label, or even a third small molecule. Bivalent molecules are valuable because they can bind two hot spots at a protein-protein interface. Strategies are proposed to optimize the linker dimensions required to join two small molecules together to achieve this. This information will emerge with little extra effort as the libraries of bivalent compounds are formed.