In the ongoing effort to develop new drugs for difficult therapeutic targets, the generally accepted guidelines of drug discovery are being challenged by advances in molecular scaffold design. Macrocyclization of molecules is an increasingly common strategy that pushes the boundaries of what are considered “drug-like properties.” Macrocycles (ring systems containing 12 or more heavy atoms) have been shown to provide opportunities for enhanced sp³ and stereochemical diversity, precise spatial distribution of binding interactions, and structural preorganization of key pharmacophores that elicit unique protein–ligand interactions (coverage often not obtainable by analogous acyclic compounds). This strategy has been shown to be particularly successful for scaffolds that are predisposed to a “U-shape”; macrocyclization can conformationally lock such compounds into a productive binding mode, often with a concomitant increase in desirable ADME properties (e.g., circumvention of efflux-mediated permeability limitations). Balancing and maintaining the desired drug-like properties of macrocycles to achieve therapeutic efficacy, however, remains a significant challenge. Although there are macrocyclic drugs that follow Lipinski’s rule of five and Veber’s rule (including lorlatinib and pacritinib), the majority of macrocycles fall into the “beyond the rule of 5 (bRo5)” space. Macrocycle design often relies on properties that are difficult to predict for bRo5 scaffolds, and is better understood by considering higher-order 3-dimensional metrics such as effective polar surface area (EPSA) and chameleonicity. As the data sets of known macrocycles grow, along with more extensive analyses of such sets, medicinal chemists will perhaps be enabled to intentionally design molecules with these types of properties at the outset of a drug discovery campaign (and move away from serendipity). The discovery of macrocycles with central nervous system (CNS) exposure presents an even greater challenge, and accordingly, there are very few examples of macrocycles developed to treat diseases of the CNS. To date, of the approximately 70 FDA-approved macrocyclic drugs, the majority are prescribed for 3 major classes of therapeutic indication: infectious disease (44%), oncology (21%), and immunology (11%). Of the only 24% of approved macrocycles miscellaneously designated as “other”, an even smaller percentage specifically target CNS indications (the melanocortin receptor agonist bremelanotide is one notable example). Importantly, only 26 of these drugs (∼39%) are dosed orally to achieve systemic exposure and target engagement. A similar breakdown of indications is observed for ongoing clinical trials; as of 2023, antibacterial agents fall just behind oncology as the combined major areas of current focus. Although an analysis of recent literature suggests that macrocycles are now being explored for a more diverse set of therapeutic indications, neuroscience continues to account for a disproportionately small percentage. Specifically, neurodegenerative diseases cumulatively account for less than 5% of ongoing macrocycle research, and no compound in this class has been approved. (1) Perhaps the two most widely studied classes of CNS macrocycles are the BACE1 (β-secretase) inhibitors for the treatment of Alzheimer’s disease (AD), and kinase inhibitors for a variety of neurooncology indications. Merck’s sulfonamide series of BACE1 inhibitors are a textbook example of the utility of macrocyclization toward improved target potency and membrane permeability. Macrocyclization of a weakly potent acyclic BACE1 inhibitor resulted in early compounds with increased potency, and subsequent SAR revealed a BACE1 macrocycle with high potency in cellular assays, excellent membrane permeability, and low P-gp efflux. Additionally, this advanced compound robustly decreased Aβ(40) levels in APP-YAC mice, and was therefore reported as being the first BACE1 inhibitor (macrocyclic or otherwise) to demonstrate efficacy after systemic administration in transgenic animals. (2) Although next-generation BACE inhibitors (with excellent properties and preclinical efficacy) are known, to date this class of compounds has not found success in the clinic. This apparent lack of translatability to human subjects is certainly not unique to BACE1 programs among neuroscience clinical trials. As a means by which to ferry challenging chemotypes into the CNS, however, the development of these pioneering BACE1 macrocycles should be applauded as an important proof of concept. Considering neurooncology indications, Pfizer’s development of lorlatinib is another excellent example of the successful implementation of a macrocyclization strategy. An earlier-generation compound, crizotinib, is an approved treatment for patients with locally advanced or metastatic ALK-positive non-small cell lung cancer (NSCLC). For many such patients, crizotinib initially demonstrated robust efficacy, but resistance eventually developed. A follow up drug discovery effort sought to develop a compound that potently inhibited drug-resistant ALK mutations, along with the necessary CNS exposure to treat brain metastases. As with the acyclic BACE1 inhibitors, P-gp efflux proved a significant challenge. Careful analysis of a common “U-shaped” motif within the available chemical matter revealed potential for macrocyclization, ultimately leading to the discovery of lorlatinib. Lorlatinib is approximately 80-fold more potent against cellular wild-type ALK compared to crizotinib, has low nanomolar potency against the primary drug-resistant mutant, and, importantly, is brain penetrant (MDR BA/AB = 1.5). (3) Additionally, lorlatinib is characterized by high oral bioavailability in humans (81%), and has a prolonged half-life of ∼25 h. Lorlatinib is currently approved for the treatment of ALK-positive metastatic NSCLC patients. (4) Despite this significant advance, to date there is no macrocyclic kinase inhibitor approved for the treatment of primary CNS tumors (e.g., glioblastomas). An understanding of the principles underlying brain penetration is critical to the success of CNS drug discovery programs, and is the topic of an excellent perspective (Figure 1). (5) But how well do these principles translate to macrocycles? As with any chemotype, quantification of passive permeability and any efflux/transport mechanisms will be crucial for an understanding of PK–PD, but the design of a successful CNS macrocycle will ultimately depend on a holistic understanding of the molecule’s entire topology. To what extent is its polarity “shielded” (EPSA)? Are intramolecular hydrogen bonds effective for controlling EPSA, or is a capping strategy (e.g., N-methylation) necessary? To what extent does the compound exhibit chameleonic character (an ability to expose polar groups in aqueous environments to maximize solubility, but shield them to cross lipid membranes)? In the absence of predictive data sets for a novel chemotype, such design considerations may best be left to the eye of a creative medicinal chemist. Figure 1. Representation of unbound macrocycle equilibrium distribution between brain, CSF, and plasma, and summarized design considerations. C_b,u = unbound macrocycle concentration in brain (total concentration between ICF (intracellular fluid) and ISF (interstitial fluid)), C_CSF = macrocycle concentration in cerebral spinal fluid, C_p,u = unbound macrocycle concentration in plasma, P-gp = P-glycoprotein, BCRP = breast cancer resistance protein, BBB = blood brain barrier, BCSFB = blood cerebral spinal fluid barrier. Reproduced with permission from ref (5). Copyright 2012 American Chemical Society. Historically, the macrocycle approach has found somewhat higher success in targeting proteins with relatively flat/extended ligand binding regions, or “groove-shaped” regions (although differentiating examples are known). (1,6) It remains to be seen how the approach will translate to CNS targets that do not necessarily share this type of topology. The class A G protein-coupled receptors, for example, differ from the peptidic/shallow binding site of some class B GPCRs (e.g., the calcitonin gene-related peptide receptor (CGRPR)) (6) in that the canonical orthosteric ligand binding region is buried within the 7-transmembrane domain, occupying a relatively 3-dimensional space among the helical contacts. Would a ≥ 12-membered macrocyclic ligand, which necessarily occupies a larger, more rigid space compared to a more flexible acyclic counterpart, be compatible with this type of pocket? For this class of receptors and others (class C GPCRs, CNS-relevant ion channels, etc.), the discovery of macrocyclic ligands (or lack thereof) will help guide the design of next generation compounds for important CNS indications, and further inform what limitations exist on macrocycle design for unconventional targets. Moreover, as advances in X-ray crystallography and cryogenic electron microscopy increasingly provide snapshots of these proteins’ various dynamic ensembles, it is tempting to speculate that alternative allosteric sites, perhaps more amenable to macrocycle binding, will be revealed (or will manifest via induced fit binding). Additionally, a high percentage of endogenous and synthetic orthosteric ligands for CNS targets contain basic amine functionalities. Despite their high pK_a, basic nitrogen-containing macrocycles have been shown to be quite membrane permeable in certain contexts, particularly when the charge is counterbalanced by appropriately placed hydrophobic regions within the macrocycle. (7) Could these features enhance the likelihood of success in the design of macrocyclic derivatives of classical orthosteric CNS ligands (e.g., serotonergic tryptamines)? The Journal of Medicinal Chemistry warmly welcomes the submission of manuscripts that address any of these topics, including studies in which the concept may prove limited or ineffective within a suite of superior/differentiating ligands. If a lead compound or series is predisposed to a “U-shape”, and has amenable chemistry, a singleton macrocyclic example is at least worth the attempt. For basic and translational science alike, the future for macrocycles is exciting! Journal of Medicinal Chemistry ─ most trusted, most cited, most read. This article references 7 other publications. This article has not yet been cited by other publications. Figure 1. Representation of unbound macrocycle equilibrium distribution between brain, CSF, and plasma, and summarized design considerations. C_b,u = unbound macrocycle concentration in brain (total concentration between ICF (intracellular fluid) and ISF (interstitial fluid)), C_CSF = macrocycle concentration in cerebral spinal fluid, C_p,u = unbound macrocycle concentration in plasma, P-gp = P-glycoprotein, BCRP = breast cancer resistance protein, BBB = blood brain barrier, BCSFB = blood cerebral spinal fluid barrier. Reproduced with permission from ref (5). Copyright 2012 American Chemical Society. This article references 7 other publications.

中文翻译：

---

脑渗透大环化合物：设计考虑、未来前景和论文征集

在针对困难治疗靶点开发新药的持续努力中，普遍接受的药物发现指南正受到分子支架设计进步的挑战。分子大环化是一种越来越常见的策略，它突破了“类药物特性”的界限。大环（含有 12 个或更多重原子的环系统）已被证明可以提供增强 sp ^3的机会立体化学多样性、结合相互作用的精确空间分布以及引发独特蛋白质-配体相互作用的关键药效​​团的结构预组织（覆盖范围通常无法通过类似的无环化合物获得）。该策略已被证明对于倾向于“U 形”的支架特别成功；大环化可以在构象上将此类化合物锁定为有效的结合模式，通常伴随着所需的 ADME 特性的增加（例如，规避外排介导的渗透性限制）。然而，平衡和维持大环化合物所需的药物样特性以实现治疗功效仍然是一个重大挑战。尽管有一些大环药物遵循 Lipinski 的 5 规则和 Veber 的规则（包括 lorlatinib 和 pacritinib），但大多数大环药物属于“超越 5 规则（bRo5）”空间。大环设计通常依赖于 bRo5 支架难以预测的特性，并且通过考虑高阶 3 维指标（如有效极性表面积 (EPSA) 和变色性）可以更好地理解。随着已知大环化合物数据集的增长，以及对此类数据集的更广泛分析，药物化学家也许能够在药物发现活动一开始就有意设计具有这些类型特性的分子（并摆脱偶然性）。中枢神经系统（CNS）暴露的大环化合物的发现提出了更大的挑战，因此，开发用于治疗中枢神经系统疾病的大环化合物的例子非常少。迄今为止，在 FDA 批准的大约 70 种大环药物中，大多数用于 3 类主要的治疗适应症：传染病 (44%)、肿瘤 (21%) 和免疫学 (11%)。在仅有 24% 的已批准大环化合物中，有多种被指定为“其他”，其中专门针对 CNS 适应症的比例甚至更小（黑皮质素受体激动剂布雷默诺肽就是一个值得注意的例子）。重要的是，这些药物中只有 26 种 (∼39%) 通过口服给药来实现全身暴露和目标参与。在正在进行的临床试验中观察到了类似的适应症细分；截至 2023 年，抗菌药物的综合主要关注领域仅次于肿瘤学。尽管对最近文献的分析表明，目前正在探索大环化合物用于更多样化的治疗适应症，但神经科学仍然占不成比例的小比例。具体而言，神经退行性疾病累计占正在进行的大环化合物研究的不到 5%，且此类化合物尚未获得批准。 (1) 也许研究最广泛的两类中枢神经系统大环化合物是用于治疗阿尔茨海默病 (AD) 的 BACE1（β-分泌酶）抑制剂和用于各种神经肿瘤学适应症的激酶抑制剂。Merck 的磺酰胺系列 BACE1 抑制剂是利用大环化来提高靶标效力和膜通透性的教科书示例。弱效无环 BACE1 抑制剂的大环化导致早期化合物的效力增强，随后的 SAR 揭示了 BACE1 大环化合物在细胞测定中具有高效力、优异的膜渗透性和低 P-gp 外流。此外，这种先进的化合物可显着降低 APP-YAC 小鼠的 Aβ(40) 水平，因此被报道为第一个在转基因动物全身给药后证明有效的 BACE1 抑制剂（大环或其他）。 (2) 尽管下一代BACE抑制剂（具有优异的性能和临床前疗效）已为人所知，但迄今为止此类化合物尚未在临床上取得成功。这种明显缺乏对人类受试者的可转化性当然不是神经科学临床试验中 BACE1 项目所独有的。然而，作为将具有挑战性的化学型运送到中枢神经系统的一种手段，这些开创性的 BACE1 大环化合物的开发应该被视为重要的概念证明。考虑到神经肿瘤学适应症，辉瑞开发的lorlatinib是大环化策略成功实施的又一个绝佳例子。克唑替尼是一种较早一代的化合物，已被批准用于治疗局部晚期或转移性 ALK 阳性非小细胞肺癌 (NSCLC) 患者。对于许多此类患者，克唑替尼最初表现出强大的疗效，但最终出现了耐药性。后续药物发现工作试图开发一种能够有效抑制耐药 ALK 突变的化合物，以及治疗脑转移所需的中枢神经系统暴露。与无环 BACE1 抑制剂一样，P-gp 外流被证明是一个重大挑战。对现有化学物质中常见的“U 形”基序的仔细分析揭示了大环化的潜力，最终导致了洛拉替尼的发现。与克唑替尼相比，Lorlatinib 对抗细胞野生型 ALK 的效力大约高 80 倍，对抗主要耐药突变体的纳摩尔效力较低，而且重要的是，它具有脑渗透性（MDR BA/AB = 1.5）。 (3) 此外，劳拉替尼在人体中具有较高的口服生物利用度（81%），并且半衰期较长，约为 25 小时。 Lorlatinib 目前被批准用于治疗 ALK 阳性转移性 NSCLC 患者。 (4) 尽管取得了这一重大进展，但迄今为止，还没有大环激酶抑制剂被批准用于治疗原发性中枢神经系统肿瘤（例如胶质母细胞瘤）。了解大脑渗透的原理对于中枢神经系统药物发现项目的成功至关重要，也是一个很好的视角主题（图 1）。 (5) 但是这些原理如何转化为大环呢？与任何化学型一样，被动渗透性和任何外排/运输机制的量化对于理解 PK-PD 至关重要，但成功的 CNS 大环的设计最终将取决于对分子整个拓扑结构的整体理解。其极性“屏蔽”程度如何（EPSA）？分子内氢键对于控制 EPSA 是否有效，或者是一种封端策略（例如，N-甲基化）有必要吗？该化合物在多大程度上表现出变色龙特征（能够在水性环境中暴露极性基团以最大程度地提高溶解度，但保护它们穿过脂质膜）？在缺乏新化学型的预测数据集的情况下，这种设计考虑最好留给有创造力的药物化学家来考虑。图 1. 大脑、脑脊液和血浆之间未结合大环化合物平衡分布的图示，并总结了设计注意事项。 C _b,u = 脑中未结合的大环化合物浓度（ICF（细胞内液）和 ISF（间质液）之间的总浓度），C _CSF = 脑脊液中的大环化合物浓度，C _p,u = 血浆中未结合的大环化合物浓度，P- gp = P-糖蛋白，BCRP = 乳腺癌抗性蛋白，BBB = 血脑屏障，BCSFB = 血脑脊髓液屏障。经参考文献 (5) 许可转载。版权所有 2012 美国化学会。从历史上看，大环方法在靶向具有相对平坦/延伸的配体结合区域或“凹槽形”区域的蛋白质方面取得了较高的成功（尽管不同的例子是已知的）。 (1,6) 该方法如何转化为不一定共享此类拓扑的 CNS 目标仍有待观察。例如，AG 类蛋白偶联受体与某些 B 类 GPCR（例如降钙素基因相关肽受体 (CGRPR)）的肽/浅结合位点不同 (6)，因为典型的正构配体结合区域是埋藏在7次跨膜域内，在螺旋接触之间占据相对3维的空间。与更灵活的无环配体相比，≥ 12 元大环配体必然占据更大、更刚性的空间，是否与这种类型的口袋兼容？对于此类受体和其他受体（C 类 GPCR、CNS 相关离子通道等），大环配体（或缺乏）的发现将有助于指导针对重要 CNS 适应症的下一代化合物的设计，并进一步告知什么非常规目标的宏观周期设计存在局限性。此外，随着 X 射线晶体学和低温电子显微镜的进步越来越多地提供这些蛋白质的各种动态整体的快照，人们很容易推测可能更适合大环结合的替代变构位点将被揭示（或将通过诱导适合装订）。此外，中枢神经系统靶标的内源性和合成正构配体的高比例含有碱性胺官​​能团。尽管它们_的p Ka，碱性含氮大环已被证明在某些情况下具有相当的膜渗透性，特别是当电荷通过大环内适当放置的疏水区域平衡时。 (7) 这些特征能否提高经典正构中枢神经系统配体（例如血清素色胺）大环衍生物设计成功的可能性？《药物化学杂志》热烈欢迎提交涉及任何这些主题的手稿，包括在一系列高级/差异化配体中该概念可能被证明有限或无效的研究。如果先导化合物或系列易于形成“U 形”，并且具有良好的化学性质，那么单个大环化合物的例子至少值得尝试。对于基础科学和转化科学来说，大环化合物的未来是令人兴奋的！《药物化学杂志》─ 最值得信赖、引用最多、阅读最多的杂志。本文参考了其他 7 篇出版物。这篇文章尚未被其他出版物引用。图 1. 大脑、脑脊液和血浆之间未结合大环化合物平衡分布的图示，并总结了设计注意事项。 C _b,u = 脑中未结合的大环化合物浓度（ICF（细胞内液）和 ISF（间质液）之间的总浓度），C _CSF = 脑脊液中的大环化合物浓度，C _p,u = 血浆中未结合的大环化合物浓度，P- gp = P-糖蛋白，BCRP = 乳腺癌抗性蛋白，BBB = 血脑屏障，BCSFB = 血脑脊髓液屏障。经参考文献 (5) 许可转载。版权所有 2012 美国化学会。本文参考了其他 7 篇出版物。