先证明它必需，再去找分子：脓肿分枝杆菌的 MshC 成药之路
下划线 = 中文容易“变味”或丢信息的词句：红线带编号 = 15 条注释卡（悬停看、点击跳转）。这篇要回答的问题很直白：一个“分枝杆菌版谷胱甘肽”通路里的酶，能不能做成药靶？作者用四步把它答到“有分子、有口袋、有配伍方案”。英文原文附于文末
作者：Thanh Quang Nguyen、Claire Hamela、Luping Pang、Yi Liu、Greta Carola Giannini、Lucas Boeck、Gerald Larrouy-Maumus、Laurent Kremer（共 8 人；通讯作者为末位 Laurent Kremer，法国蒙彼利埃）
出处：Journal of Biological Chemistry 2026;302(6):111433（开放获取）
DOI：10.1016/j.jbc.2026.111433 PMID：41962868
数据：LC-MS 代谢组原始数据存于 MetaboLights，编号 MTBLS13995
摘要（全译）
由脓肿分枝杆菌（Mycobacterium abscessus，Mab）引起的肺部疾病，患病率正在上升，在许多国家已经超过结核病。然而，这种病原体对多种抗生素的固有耐药性，严重限制了我们的治疗武器库，新的化学治疗药物与新的药物靶点都亟需被找到。
分枝杆菌中含量最高的硫醇是分枝杆菌硫醇（mycothiol，MSH），它在抵抗氧化应激中起重要作用。此前在耻垢分枝杆菌（Mycobacterium smegmatis）与结核分枝杆菌（Mycobacterium tuberculosis）中的工作表明，依赖 ATP 的半胱氨酸连接酶 MshC 催化 MSH 生物合成的倒数第二步。在本研究中，我们检验了 MshC 能否成为脓肿分枝杆菌中一个有价值的药物靶点。
我们用 CRISPRi 证明 mshC 在体外、以及在斑马鱼幼鱼的感染过程中都是必需的。随后，计算机药物筛选找到三个可能与 MshC 结合的分子，即 rebastinib、olverembatinib 与 SB-224289，它们显著抑制脓肿分枝杆菌的生长与 MSH 的产生。MshC 中自发产生的错义突变，使菌株对这三个化合物交叉耐药。把这些导致耐药的突变投射到 MshC 的三维同源模型上，把酶的催化位点定为可能的药物结合腔。此外，在敏感菌株中过表达 mshC，会提高对 rebastinib、olverembatinib 与 SB-224289 的耐药水平，但对利福布汀（rifabutin）没有影响。反过来，在 CRISPRi 突变体中降低 mshC 的表达水平，会提高细菌对这些抑制剂的敏感性。系统评估 olverembatinib 与 SB-224289 同另外 18 种药物的组合，揭示出明确的协同与拮抗相互作用。这些结果突出了一个此前未被开发的化学结构类型对脓肿分枝杆菌的潜力，也说明了靶向 MSH 通路在脓肿分枝杆菌肺病未来转化研究中的可能性。
引言：脓肿分枝杆菌为什么难治
〔第 1 段·精译〕非结核分枝杆菌（NTM）感染是一个正在变大的健康问题。NTM 是环境分枝杆菌，毒力高低不一，临床谱很宽。与鸟分枝杆菌复合群并列，脓肿分枝杆菌复合群是最常被检出的 NTM 肺病病原之一。Mab 复合群感染中约 50% 累及肺，主要发生在免疫功能低下的人、或已有肺部基础病（支气管扩张、囊性纤维化）的人身上。在这类患者里，Mab 会加速炎症性肺损伤，推高发病率与死亡率。
治疗上的数字更难看：目前至少要12 个月的多药方案——大环内酯（克拉霉素或阿奇霉素）＋氨基糖苷（阿米卡星）＋静脉 β-内酰胺（头孢西丁或亚胺培南）；而大环内酯耐药的病例，治愈率只有 25–40%，这类耐药分离株占全部分离株的 40–60%。再加上一部分人内科治疗或手术切除后仍会复发，所以“新靶点、新药效团”在这个病原上不是锦上添花，是唯一出路。
〔第 2 段·精译〕分枝杆菌与多数放线菌都合成 MSH（1-D-myo-inosityl-2-(N-acetyl-L-cysteinyl)amido-2-deoxy-α-D-glucopyranoside），它是真核细胞里谷胱甘肽的功能等价物：分枝杆菌最主要的低分子量硫醇，用来抵挡氧化应激、亲电毒素和抗生素。MSH 的生物合成分五步，依次是糖基转移酶 MshA、磷酸酶 MshA2、脱酰酶 MshB、半胱氨酸连接酶 MshC、乙酰转移酶 MshD；其中 MshC 催化半胱氨酸与 GlcN-Ins 中间体在 ATP 依赖下的缩合。
但“必需”这件事在不同菌里完全不是一回事。结核分枝杆菌里 mshA 与 mshC 必需，mshB 与 mshD 可以缺；耻垢分枝杆菌里四个基因全都能敲掉，突变株照样活；脓肿分枝杆菌此前没做过系统验证，只有一项饱和转座子诱变的全基因组分析提示 mshC 可能必需。
〔第 3 段·精译〕所以本研究就三件事：查清 mshC 在 Mab 里到底必需不必需；用 CRISPRi 在体外与体内两边验证；再用计算机筛选找能结合 MshC、并抑制 Mab 生长的分子，把它做成一个“从靶点到苗头化合物”的起点。
读到这儿你已经能看出这篇的骨架了：先证明靶点必需，再去找分子，最后证明分子打的就是这个靶点。这个顺序不是形式主义。反过来做（先筛到能抑菌的分子，再回头找靶点）也能发文章，但那样你永远说不清抑菌是靶点效应还是脱靶毒性；而先把“必需性”钉住，后面每一个抑菌数字才有一个可以解释的落点。
另外注意一个方法学上的便宜：CRISPRi 给的是可调的必需性。同一套菌，加多少诱导剂（ATc）就压低多少 mshC 表达。这不但能证明“必需”，还顺手为后面“表达降一半 → 对抑制剂更敏感”的实验铺好了路。整篇文章里，这个设计是最省事也最漂亮的一笔。
结果 1：CRISPRi 证明 mshC 是必需基因
Mab 的 msh 基因散落在基因组各处：mshA（MAB_4057c）、mshB（MAB_1316）、mshC（MAB_2116）、mshD（MAB_0745）。作者用的是 pLJR962 这套 CRISPRi 载体：一个受脱水四环素（ATc）诱导的、来自嗜热链球菌的失活 Cas9（dCas9Sth1），加一个同样受 ATc 诱导的 sgRNA。挑好 PAM 位点后，把四条靶向 mshC 的 sgRNA 分别装进载体，得到四个 Mab 突变株。
关键的一步是“筛 guide”。四条 sgRNA 里，sgRNA3 与 sgRNA4 只差一个核苷酸，但压制效果差得很远：sgRNA1/2/3 的菌在 ATc >250 ng/ml 时生长受损，而带 sgRNA4 的菌在 250 ng/ml 就基本停住，ATc 对它的 IC50 是 220 ng/ml。后续全部实验都用这个最强的一株，命名为 Mab_pJLR962:mshC。
压制效果是定量过的：ATc 浓度与 mshC 的转录抑制水平强相关（R²=0.76），110 ng/ml 与 220 ng/ml ATc 分别让 mshC 转录下降 22.4% 与 47.6%（相对空载体菌）。注意，这只是“降了一半”，不是“敲没了”——这正是 CRISPRi 与基因敲除的区别，也是后面做剂量效应实验的前提。
生长上的后果很硬：空载体菌在 1000 ng/ml ATc 下照长；而 mshC_sgRNA4 菌在 250 ng/ml ATc 处理 3 天后，CFU 下降 3 个数量级。平板点样（7H10 琼脂、阳离子调节 Mueller-Hinton 琼脂）复现了同样趋势；把 ATc 换成多西环素也能诱导，只是低浓度下弱一些。
体内那一步用斑马鱼幼鱼：受精后 30 小时（30 hpf）的胚胎，尾静脉注射约 200 CFU（菌株带 pTEC15/TdTomato，方便活体成像）；感染后 1 天（1 dpi）注射 10 ng ATc 压住 mshC。结果是：不压制的感染组到 11 dpi 约 80% 死亡；压制 mshC 并给 ATc 的组只有约 30% 死亡，整鱼成像里的细菌灶也明显更少。也就是说，在活体感染过程中把 mshC 关小，细菌就撑不住——这一条把“体外必需”升级成了“感染中必需”。
这里有个容易被读者跳过的细节：斑马鱼是28.5 °C 培养的，比人的 37 °C 低；而 Mab 是能在 37 °C 生长的少数 NTM 之一。所以这个模型证明的是“宿主体内环境 + 免疫压力”下 mshC 必需，不等于“人体肺里也如此”。做抗菌研究的人都懂这条边界，但写讨论时值得自己先说出来。
结果 2：虚拟筛选得到三个分子
脓肿分枝杆菌的 MshC 没有晶体结构，结核的也没有。作者的做法很标准：拿同属的耻垢分枝杆菌 MshC（PDB 8HFN）当模板，序列同一性 75.7%，建出同源模型，质量评估 GMQE 0.94、QMEAN 0.9——这个分数在同源模建里算相当体面，意思是“主链大体可信，侧链别太当真”。
然后是虚拟筛选。DrugBank 里 15,236 个已批准或研究中的化合物，逐一对接进模型，结合能从 −0.6 到 −12.6 kcal/mol 铺开。作者先做了两件常被省略的对照：把已知的结核 MshC 抑制剂 dequalinium（−6.9 kcal/mol）和反应中间体类似物 CysSA（−8.5 kcal/mol）也丢进去，确认这个模型能把“真抑制剂”排到前列——没有这一步，后面所有打分都没有意义。筛出的少数高分分子要求结合能 < −10 kcal/mol 且可买到，最后只有三个进入湿实验：
rebastinib（−10.8 kcal/mol），一个尿素类的多靶点酪氨酸激酶抑制剂，原本在慢性髓性白血病（CML）的临床试验里；对接显示它的 Pi-Anion 作用朝向 D87/E91，Pi-Alkyl 朝向 P88。
olverembatinib（−10.9 kcal/mol），第三代 BCR-ABL1 抑制剂，在中国已获批用于部分 CML 患者；它的三氟甲基与 W227 主链形成 2.3 Å 氢键，另有一批碳氢键与 G226、G189 以及 D86/G189/W227/L204 的疏水作用。
SB-224289（−12.1 kcal/mol，三药里最高），原本是人 5-羟色胺受体的拮抗剂，还被报道过能拮抗 Papuamide A 的抗真菌机制；它与 D87 形成强 Pi-Anion，与 W227 有 Pi-Alkyl/π-π 堆积，另在 R188/G189 主链和 P88 上有接触。注意这三个分子在化学上彼此毫无关系，唯一的共同点是它们都被同一个模型指到了同一个腔——这一点后面会被抗性突变证明不是巧合。
| 化合物 | MIC50（μM） | MBC（μM） | MBC/MIC | 判读 |
|---|---|---|---|---|
| rebastinib | 20.79 ± 0.74 | 100 | 4.81 | 抑菌 |
| olverembatinib | 36.90 ± 0.59 | 200 | 5.42 | 抑菌 |
| SB-224289 | 2.39 ± 0.31 | 6.25 | 2.61 | 杀菌 |
| rifabutin（对照） | 2.68 ± 0.08 | 6.25 | 2.33 | 杀菌 |
表 1 三个命中分子对脓肿分枝杆菌标准株（CIP104536）的活性（数据自原文表 1；MIC 为三次独立实验均值 ± SD，MBC 为三次实验中的最大一致值；MBC/MIC ≤ 4 判为杀菌，> 4–32 判为抑菌）
抑菌只是第一步，还要证明它们真的打到了通路。作者用 LC-MS 直接测 MSH：正离子模式下 MSH 的 [M+H]+ 在 487.1592 m/z、保留时间 4.2 min。用 0.1× MIC50 处理，MSH 的离子计数大幅下降；用 0.5× MIC50 处理，峰干脆消失。同时 MSH 的上游底物 GlcN-Ins（[M−H]− 340.1249 m/z）在处理与未处理之间没有显著差别——也就是说，堵点在 MshC 这一步，前一步的产物堆不起来，后一步的产物掉下去了。这是全篇最“分子生物学”的一段证据，比任何抑菌曲线都硬。
体外活性谱也测了：14 株脓肿分枝杆菌复合群临床分离株，SB-224289 的MIC50 在 1.15–6.72 μM，三药里最好；rebastinib 6.74–55.68 μM，olverembatinib 9.93–61.95 μM；对照的克拉霉素是 0.11–0.74 μM。同复合群的 M. massiliense 也对三药敏感（敏感与耐药变体都在），M. tuberculosis mc²6230 是 10.1–25.5 μM——说明这套化学不挑菌种，挑的是通路。有意思的反向结果：M. smegmatis 对三药全部不敏感。作者的解释干净利落：mshC 在耻垢分枝杆菌里本来就不是必需基因，靶点不必需，抑制剂当然压不住生长。
最后补了一组结构类似物做“靶向性指纹”：SB-224289 的类似物 GR-127935（MIC50 24 μM）、SB-216641（108 μM）、SB-616234-A（9.5 μM）都明显弱于 SB-224289（2.39 μM）；而三个化合物对过表达 mshC 的菌株，MIC 都掉到 >200 μM。同一类化学结构，靶点多了就失效——这比“我筛到了它就相信它”要可靠得多。
结果 3：抗性突变把结合位点定在催化口袋
判断一个化合物是否打在酶上，最便宜也最硬的方法不是做结合实验，而是让细菌自己招供：在亚致死浓度下把菌铺开，挑出自发抗性克隆，测突变在哪。作者的频率是这样的：rebastinib 与 olverembatinib 约 10−6–4×10−6，SB-224289 在 1× MIC50 时 5.5×10−6，到 10× MIC50 降到 5×10−8——剂量越高、越难挑到，说明这些突变不是“随便一敲就抗”，而是有代价的。
挑出来的突变株很集中，只有两个位点：D87（D87A、D87G）和 W227（W227R、W227G）。而且这三组抗性株互相交叉：rebastinib 抗性株对 olverembatinib 的 MIC 升 4–5 倍，olverembatinib 抗性株对 rebastinib 升 5 倍，SB-224289 抗性株对另两个直接 >200 μM；但它们对利福布汀仍然敏感（3.00 ± 0.63 μM 一类）。交叉耐药出现在两个化学上完全不同的分子之间，通常意味着它们争的是同一个位置；而利福布汀不跟着变，说明这不是外排泵被打开这种“广谱抗性”。
| 抗性株来源 | MIC 相对野生株 | 突变（核苷酸） | 突变（氨基酸） |
|---|---|---|---|
| rebastinib（R1–R3） | >10× | A260C | D87A（R1、R3）/ D87G（R2） |
| olverembatinib（O1–O3） | >5× | T746A | W227R（O1、O2）/ W227G（O3） |
| SB-224289（SB1–SB4） | >17× | — | D87A 和/或 W227R |
表 2 自发抗性突变株的突变位点（数据自原文表 3 与结果文字。注意两个药物挑出的位点不重合，而 SB-224289 同时能碰到两个位点——这与它打分最高、作用基团最多是一致的）
但是“抗性株带突变”和“突变导致抗性”是两件事。作者做了一步很多人偷懒不做的验证：用 recombineering（pJV53 系统，70 nt 滞后链寡核苷酸，乙酰胺/琥珀酸诱导）把 O1 的 227 位、R1 的 87 位改回野生型。回复后的菌对三个化合物的敏感性完全恢复——因果闭环。这一步花的时间比筛选多得多，但它把“相关”变成了“导致”。
第二组证据是靶点剂量效应，双向都做了：过表达 mshC（pMV261 上的 mshC-HA，hsp60 启动子）让菌对三个化合物的耐药显著上升、对利福布汀不变；反过来，CRISPRi 菌在 110 ng/ml ATc 这种亚抑菌剂量下，对 rebastinib/olverembatinib 的敏感性翻倍、对 SB-224289 翻四倍，对利福布汀依然不动。同一支枪，靶子多一点就打不穿，靶子少一点就打得穿——这是靶点归属最朴素也最可信的证据。
把这些突变投到同源模型上，D87 和 W227 都落在催化口袋内壁。于是整篇的逻辑链闭合了：必需性（基因）→ 命中（分子）→ 产物下降（通路）→ 抗性突变定位（口袋）→ 剂量效应（靶点归属）。六条证据叠在一起，比任何单一实验都难反驳。你要是只记一件事，就记这条链的形状。
结果 4：与 18 种药联用的协同与拮抗
抗结核/抗 NTM 的药从来不是单打，所以任何新先导化合物都要回答“它跟我手上这些药能不能一起用”。作者的方案叫 DiaMOND：在 96 孔板的一整条对角线上做 9×9 的药物浓度矩阵，用 Loewe 加和模型算 FIC，再把 log2 FIC 映射成颜色——蓝的是协同，红的是拮抗，点的大小表示强度。这套图的好处是一眼能看出整片区域，而不是只报“某两药协同”的孤立数字。
结果不太好看，但很有信息量：多数组合是拮抗，尤其是与氨基糖苷类、β-内酰胺类和喹诺酮类一起用时。真正协同的少数几个很有规律：SB-224289 与贝达喹啉（bedaquiline）、氯法齐明（clofazimine）强协同；olverembatinib 与贝达喹啉、大环内酯强协同；利奈唑胺与两者都是加和。更妙的是，两种 MshC 抑制剂的整张相互作用谱高度相关（R²=0.82）——两个化学结构毫不相干的分子，跟同一批药的配伍行为却几乎一样，这又是一条“它们打同一个靶”的旁证。
这里必须指出一个数字上的不一致，因为你自己读原文时会卡住：摘要写“与另外 18 种药物联用”，结果正文写“19 种对 Mab 有活性的药”，图注写“19 种（不含 SB-224289 本身）”，方法里又写“共 20 种单药（含 olverembatinib 与 SB-224289）”。三种说法里，18 才是自洽的那个（20 减去自身两个）。这类小错在长篇实验文章里很常见，但它提醒你：数字要交叉核对，别默认作者写的就是对的。
至于为什么和氨基糖苷、β-内酰胺拮抗，原文没有给出机制解释。一个合理的猜测（这是我的推测，不是原文结论）是：MSH 被压下去之后，细菌整体的解毒与氧化还原缓冲能力下降，反而更早进入低代谢的“被按住”状态；而氨基糖苷和 β-内酰胺的杀菌都依赖细菌还在活跃生长、还在合成细胞壁。你抑制了它生长，也就顺手削弱了这些药的杀灭条件。这个假说要成立，得测处理后的呼吸速率、ATP 水平或者持留菌比例——如果哪天你要接一个组合用药的课题，这就是一个现成的入口。
临床上的含义也很直接：如果 MshC 抑制剂真的往下走，它最可能的搭档是口服二线药（大环内酯、氯法齐明、贝达喹啉、利奈唑胺），而不是静脉的氨基糖苷和 β-内酰胺。而 Mab 治疗恰恰要靠长期口服方案维持——这条配伍规律跟临床需求是对得上的。
讨论：从一个基因到一类化学结构
讨论第一段先把成绩摆出来：Mab 内在耐药、新药管线空，老药新用（repurposing）能缩短研发周期、降低成本，而这篇给出了“一个此前未被开发的化学结构类型”对 Mab 的活性。然后是那六条证据的汇总——我在上节已经列过，这里不重复；值得抄下来的是它的结构：基因必需性、通路产物、抗性突变、表达量双向调控、对接同腔、以及不同化学结构的交叉耐药。任何一篇“我找到了靶点抑制剂”的文章，都该能凑出这六条里的至少四条。
接着是机制上的新东西。把三个分子叠到耻垢分枝杆菌 MshC 的两个晶体结构上（PDB 3C8Z，带 CysSA；PDB 8HFN，带化合物 7b），三个分子都落在催化口袋。SB-224289 最特别：它的吲哚-噁二唑基团占据的正是半胱氨酸-AMP 反应中间体该待的位置，而它其余的杂环基团伸向一条由 I178–R188 螺旋 C 端、D86–G97 环、以及 W227–L239 螺旋 N 端围出来的通道。原文把这条通道称为“此前未被描述的辅助口袋”，并推测它可能参与底物进入或产物释放。这个说法要小心：模型 + 对接 + 突变能支持“这里重要”，但支持不了“这里是通道”。要坐实，得等 Mab-MshC 与抑制剂的共晶。
作者也老实交代了局限，而且都是硬伤级别的：SB-224289 水溶性太差，加不进鱼缸的水里，所以斑马鱼实验只做了靶点必需性（CRISPRi），没做抑制剂疗效；同样因为溶解度，三个化合物与纯化 MshC 的结合常数测不出来，靶点结合只有计算证据；三个化合物 MIC 比阿米卡星、克拉霉素高一个量级（SB-224289 与利福布汀相当）；SAR 没有共晶指导，只能靠类似物摸索。作者给出的下一步也很清楚：拿 Mab-MshC·抑制剂共晶来指导结构优化。
还有一个对照值得记住：此前报道的结核 MshC 抑制剂是半胱氨酰-磺酰胺类、NTF1836、dequalinium 这些，与本文三个分子在化学上毫无关系。同一个靶点、完全不同的化学空间，都能打出活性——这既说明 MshC 这个口袋“好打”，也说明虚拟筛选在这种场景下确实能省钱。
这篇在轨迹里的位置
把六篇按时间排：2021 年 IJMS 做人源 ProRS 抑制剂 → 2022 年 Communications Biology 做 LeuRS 的机制 → 2022 年 IJMS 做 MshC 的结构 → 2023 年 STTT 写 CLK 家族综述 → 2026 年这篇 JBC → 同年 ACS Omega 那篇吡嗪酰胺衍生物（本系列 06）。
这条线里，2022 年那篇 MshC 结构是这篇的直接前身：先把这个酶的结构弄清楚，四年后才谈“它能不能成药”。而更早的 aaRS 训练（机制 + 抑制剂两条腿走路）其实已经把这套方法论练过一遍了——先机制、再结构、后抑制剂，这个顺序在 MshC 上被完整复刻了一次。
然后是这一页跟下一页的对照，这是这套精读最想让你看见的东西。05 是从靶点出发找分子：先证明基因必需，再筛化合物，再用抗性突变把结合位点钉在口袋上，最后测组合。06 是从化合物出发优化结构：手上已经有一个用了七十年的老药骨架，改它的烷基链、改它的苯环，看 MIC 怎么动，再看它对耐药株还有没有效、机制跟老药是不是一样。两条路方向相反，但最后都撞在同一堵墙上——分子能不能进细菌、能不能穿过那层富含分枝菌酸的细胞壁。05 里三个分子都是亲脂性的（这也是它们能被筛出来的原因之一，DrugBank 里的激酶抑制剂亲脂性普遍偏高），06 里整篇文章就是在调亲脂性。做抗菌药物化学，绕不开这一个物理化学参数。
还有一件小事可以说明她在这篇里不是挂名：作者贡献（CRediT）里，L.P. 出现在 conceptualization、methodology、investigation 和 writing–original draft 四项里，但不在 validation、formal analysis 里——换句话说，她参与了设计、方法、实验和初稿，把数据验证留给了别人。读作者列表时顺手看一眼 CRediT，比数作者位次有用得多。
给刚进组的你
一、见到 CRISPRi，先问三个数。诱导剂剂量、转录被压低了多少百分比、表型是否随剂量单调变化。这篇里 220 ng/ml ATc 只把 mshC 转录压了 47.6%，就已经能让 CFU 掉三个数量级——这说明 MshC 的蛋白水平对细菌很敏感，也说明“必需”是个定量结论，不是一个开关。你以后自己用 CRISPRi，第一件事就是把这条曲线做出来，否则审稿人一问“你确定敲下去了吗”你就没话说。
二、抗性突变是你能做的最便宜的靶点验证。不需要纯化蛋白、不需要标记化合物、不需要买抗体：铺板、挑克隆、提基因组、Sanger 测序，一周能出结果。但有两件事要提前想清楚：自发抗性频率有多高（10−8 是低风险，10−6 就要小心临床上很快出现耐药），以及你的突变是不是落在预期位点。如果突变落在别处（比如外排泵调控基因），那说明你的化合物可能根本不是打在你想的靶点上——这比做出一个漂亮 MIC 更有价值。
三、靶点归属要做双向剂量实验。单侧证据（过表达→耐药上升）可以被“蛋白负担”“折叠应激”这类解释糊弄过去；加上反向（敲低→敏感性上升，且对无关药物无影响），说服力立刻翻倍。这篇里利福布汀作为阴性对照贯穿始终，这个设计值得抄。
四、组合用药矩阵要在动物实验之前跑。如果体外就是拮抗，你进了小鼠只会得到“无效”这个结论，浪费半年。DiaMOND 这类对角矩阵跑一遍只要一周，能提前告诉你该跟谁搭、不该跟谁搭。
五、读文章时把“作者说的”和“数据支持的”分开记。这篇里“辅助口袋可能参与底物进入”是推测，“D87/W227 突变导致抗性”是数据；前者可以讨论，后者不必再验。你写自己的讨论时，也要有意识地把这两类句子分开——审稿人最爱挑的就是把推测写成结论。
▲ 本页翻译难点（对照学习用）
附录 · 英文原文
Pharmacological targeting of the mycothiol cysteine ligase MshC in Mycobacteriumabscessus
Abstract
The prevalence of pulmonary diseases caused by Mycobacterium abscessus is on the rise and surpasses tuberculosis in many countries. However, intrinsic resistance of this pathogen to antibiotics profoundly limits our therapeutic arsenal, and new chemotherapeutic agents and targets are urgently needed. Mycothiol (MSH), the major thiol in mycobacteria, plays an important role in resistance to oxidative stress. Previous work in Mycobacterium smegmatis and Mycobacterium tuberculosis indicates that the ATP-dependent cysteine ligase MshC catalyzes the penultimate step in MSH biogenesis. Here, we investigated whether MshC represents a potent drug target in M. abscessus. CRISPRi was used to demonstrate that mshC is essential in vitro and during infection in zebrafish larvae. Subsequently, in silico drug screening identified three potential MshC-binding molecules, namely rebastinib, olverembatinib and SB-224289, significantly inhibiting M. abscessus growth and MSH production. Spontaneous missense mutations in MshC generated cross-resistance to all three compounds. Mapping the mutations conferring resistance on a MshC three-dimensional homology model defined the enzyme catalytic site as the potential drug binding cavity. Additionally, overexpression of mshC in a susceptible M. abscessus strain increased resistance levels to rebastinib, olverembatinib and SB-224289, but not to rifabutin. Conversely, reduced mshC expression levels in the CRISPRi mutant increased susceptibility to the inhibitors. Systematic assessment of olverembatinib and SB-224289 in combination with 18 additional drugs revealed distinct synergistic and antagonistic interactions. These results emphasize a yet unexploited chemical structure class against M. abscessus and highlight the potential of targeting the MSH pathway for future translational developments against M. abscessus lung diseases.
Keywords: antibiotic resistance, CRISPRi, drug screening, inhibition, MshC, Mycobacterium abscessus, mycothiol
Infections with non-tuberculous mycobacteria (NTM) represent an increasing health problem ([1]). NTM are environmental mycobacteria, displaying various degrees of virulence and associated with a wide spectrum of clinical manifestations. Alongside the Mycobacterium avium complex, the Mycobacterium abscessus (Mab) complex represents one of the most commonly identified pathogens responsible for NTM pulmonary diseases ([2], [3]). Approximately 50% of infections caused by the Mab complex involve the lungs and occur predominantly in immunocompromised individuals or in patients with pre-existing pulmonary conditions, such as bronchiectasis or cystic fibrosis (CF) ([1], [4], [5]). In these patients, Mab accelerates inflammatory lung damage, leading to increased morbidity and mortality ([6], [7]). Infections caused by Mab are known to be extremely difficult to treat, due to the intrinsic or induced drug resistance mechanisms developed by these microorganisms, making them highly resistant to most available antimicrobials ([8], [9], [10], [11], [12]). In addition, various patients experience relapse after medical treatment or surgical removal of the pathogen. Currently, the treatment requires long-term multidrug therapy including a macrolide (clarithromycin or azithromycin), an aminoglycoside (amikacin) and an intravenous β-lactam (cefoxitin or imipenem) for at least 12 months ([13]), resulting in a cure rate of only 25 to 40% in the case of macrolide resistance, which occurs in 40 to 60% of the isolates ([14]). Thus, novel mycobacterial targets and pharmacophores need to be explored to relieve the burden of drug resistance and effectively reduce the bacterial loads in the infected patient. Therefore, inhibiting enzymes involved in essential cellular processes of Mab, such as detoxification processes, may be of interest for future therapeutic developments.
Mycobacteria and most Actinobacteria produce mycothiol (MSH; 1-D-myo-inosityl-2-(N-acetyl-L-cysteinyl)amido-2-deoxy-α-D-glucopyranoside), the functional equivalent of glutathione in eukaryotes as the major low-molecular-weight thiol to protect against oxidative stress, electrophilic toxins and antibiotics ([15], [16]). MSH biosynthesis is carried out in five steps catalyzed by a glycosyltransferase (MshA), a phosphatase (MshA2), a deacylase (MshB), a cysteine ligase (MshC), and an acetyltransferase (MshD). MshC is a cysteine ligase catalyzing the ATP-dependent condensation of cysteine and the GlcN-Ins intermediate (A) ([17]). MshC is essential for the maintenance of the redox balance and metabolic homeostasis of Mycobacterium tuberculosis ([18], [19]). Targeted genetic disruption studies have shown that mshA and mshC are essential while mshB and mshD are dispensable for in vitro growth of M. tuberculosis ([20], [21], [22]), suggesting that the MSH pathway, particularly MshC, may represent an attractive drug target for the development of new anti-tuberculosis drugs ([23]). In contrast, MSH is not essential for the growth of Mycobacterium smegmatis, since disruption of mshA, mshB, mshC, or mshD resulted in viable mutant strains ([24]). In Mab, however, the importance of MSH has not been addressed yet, although a recent genome-wide analysis by saturated transposon mutagenesis suggested that mshC is essential ([25]).
In this study, we investigated the essential character of mshC and the potential of MshC as an attractive drug target in Mab by identifying new inhibitors. CRISPR interference assay was employed to demonstrate that mshC is indispensable in vitro and in vivo, while in silico drug screening identified three molecules targeting MshC and inhibiting Mab growth. Collectively, our results validate MshC as an attractive drug target in Mab and provide a starting point for further drug design and optimization developments.
Results
In vitro essentiality of mshC in M. abscessus
The msh genes are scattered throughout the genome in Mab: mshA (MAB_4057c), mshB (MAB_1316), mshC (MAB_2116) and mshD (MAB_0745). To evaluate the importance of mshC in Mab growth, the CRISPR interference (CRISPRi) assay was applied. The pLJR962 (), containing an anhydrotetracycline (ATc)-inducible nuclease-deactivated Cas9 (from Streptococcus thermophilus) as well as an ATc-inducible single guide RNA (sgRNA) was employed as the vector for CRISPRi in Mab, as described earlier ([26], [27]). Protospacer adjacent motif (PAM) sequences were identified in mshC, and several sequences complementary to the non-template target sequence were cloned into pLJR962, upstream of dCas9 (B) ([28]). Four pLJR962_mshC_sgRNA constructs were generated, with mshC_sgRNA3 and mshC_sgRNA4 differing by only one nucleotide (). After transformation, the different Mab CRISPRi mutants were cultivated in 7H9 broth in the presence of increasing concentrations of ATc and tested for growth. C shows that not all mutants were equally responsive to ATc. While the strains harboring mshC_sgRNA1, mshC_sgRNA2, or mshC_sgRNA3 displayed impaired growth (OD600 measurements) at ATc concentrations >250 ng/ml, growth of the strain carrying mshC_sgRNA4 was stopped at 250 ng/ml ATc. The IC50 of ATc against MAB_mshC-sgRNA4 was 220 ng/ml (). Given the highest susceptibility of the mshC_sgRNA4 strain to ATc, this CRISPRi mutant was further studied. Increasing ATc concentrations were added to mshC_sgRNA4 cultures, RNA was extracted, and quantitative RT-PCR (RT-qPCR) was performed. There was a strong correlation (R2= 0.76) between the concentration of ATc and the level of mshC transcriptional repression, with a 22.4% and 47.6% decrease in the presence of 110 and 220 ng/ml of ATc, respectively, in MAB_mshC-sgRNA4 relative to the strain carrying the empty pLJR962 ().
While ATc (up to 1000 ng/ml) did not affect the growth of the strain carrying the empty vector, CFU determination of the strain containing mshC_sgRNA4 showed that 250 ng/ml of ATc was associated with a strong decrease in the CFU (3-log reduction after 3 days of incubation as compared to the condition without ATc), validating the OD measurements. This effect was also confirmed by visual inspection of the culture in the presence or absence of ATc (D). Next, diluted cultures were spotted onto 7H10 agar supplemented with ATc concentrations ranging from 0 to 1000 ng/ml. Growth of the CRISPRi strains was inhibited with increasing ATc concentrations, showing a notable effect with the mshC_sgRNA4 strain (E, upper panels). Similar results were obtained when spotting the cultures on Cation-Adjusted Mueller-Hinton (MH) agar medium (E, middle panels) or when replacing ATc by doxycycline, although with a less pronounced effect at low concentrations (E, lower panels).
Together, these results demonstrate that the CRISPRi produces a robust impairment of mshC gene expression associated with a striking mycobacterial growth defect, strongly suggesting that mshC is indispensable for Mab in vitro growth. All subsequent studies were based on the Mab mshC_sgRNA4 strain, hereafter referred to as Mab_pJLR962:mshC.
CRISPRi-mediated repression of mshC impairs growth and pathogenicity of M. abscessus in zebrafish
We used the zebrafish model to address the effect of mshC silencing on the infection outcome by treating intravenously infected embryos with ATc, following a previously established protocol ([29]). We injected 30 h post-fertilization (hpf) embryos in the caudal vein with ≈200 CFU of Mab R (parental strain), Mab_pLJR962 or Mab_pLJR962:mshC, transformed with pTEC15 expressing TdTomato. At 1 day post-infection (dpi), larvae were treated with 10 ng of ATc to repress the expression of mshC, a concentration that failed to show toxicity signs in ATc-treated larvae ([29]). Daily monitoring of larval survival demonstrated increased mortality over time in the non-treated groups (≈80% mortality at 11 dpi) as expected, while only ≈30% of embryos infected with Mab_pLJR962:mshC and treated with ATc succumbed to the infection (F, left). Imaging whole infected larvae showed lower bacterial foci in Mab_pLJR962:mshC treated with ATc as compared to the ATc-untreated larvae (F, right panel), in agreement with the survival results. Overall, this underscores that temporal control of mshC repression is working in vivo.
These results confirm the CRISPRi system to reduce mshC expression in Mab in the infected host and further strengthen the essentiality of MshC as a valuable drug target for the treatment of Mab diseases.
Homology modeling and in silico molecular docking of MshC inhibitors
Since the crystal structure of MshC from Mab is not available yet, a structural homology model was built using MshC from M. smegmatis (PDB ID: 8HFN), sharing 75.7% sequence identity with MshC from Mab with a GMQE score of 0.94 and a QMEAN score of 0.9, indicating good model quality and reliability (A). SWISS-MODEL evaluates the quality of the predicted structure using the QMEAN scoring function both locally for each residue and globally for the entire model. To assess the local quality of the predicted protein model, a graph was generated showing the similarity scores of individual residues plotted against their position in the sequence. The majority of residues displayed scores near 1, indicating strong local reliability (not shown).
[Figure 2] Three-dimensional structural predictions of MshC in complex with ligands and effect on MSH synthesis.A, crystal structure of MshC from M. smegmatis (PDB: 8HFN) with α-helices in red, β-sheets in yellow, loops in green (left panel) and predicted 3D structure of MshC from Mab (green) versus MshC from M. smegmatis (pink) (right panel). B, chemical structures of rebastinib, olverembatinib and SB-224289. C, 3D docking interaction of rebastinib, olverembatinib and SB-224289 (yellow stick) with key residues of Mab-MshC, including D87 and W227. Hydrogen bonds and other interactions are shown as dashed lines. D, effect of MshC inhibitors on MSH production. Mab cultures were treated with 0.1× MIC50 of rebastinib, olverembatinib, or SB-224289 prior to metabolite extraction and LC-MS analyses. Extracted ion chromatograms in the positive ion mode for MSH (m/z 487.1592 [M + H]+) by LC-MS is shown. The grey shaded region indicates the retention time corresponding to the MSH peak.
Next, in silico molecular docking was performed to identify the most favorable binding modes of ligands interacting with Mab-MshC. The quality of docking results was evaluated by binding affinity. More negative binding energy indicates stronger and more stable ligand-protein interactions. The homology modeled MshC structure was used as the receptor for docking 15,236 compounds from the DrugBank database with binding affinities ranging from −0.6 to −12.6 kcal/mol. Of note, dequalinium and 5′-O-[N-(L-cysteinyl)-sulfamonyl] adenosine (CysSA) are 2 M. tuberculosis MshC inhibitors discovered previously ([30], [31]) displaying binding affinities of −6.9 and −8.5, respectively, thus validating our screen (not shown). Potential inhibitors with a binding affinity above −10 kcal/mol and market availability were prioritized, leading to a subset of a few potential ligands. Among these, rebastinib (B) showed a binding affinity of −10.8 kcal/mol and exhibits Pi-Anion interactions with the side chains of the D87 and E91 and Pi-Alkyl binding with P88 (C and A). Rebastinib is a member of the class of ureas and used in trials for the treatment of chronic myeloid leukemia. It is an orally bioavailable inhibitor of multiple tyrosine kinases with potential antineoplastic activity, offering new therapeutic possibilities for hematologic malignancies ([32], [33]). Additionally, anti-cancer effects of rebastinib have been observed in multiple types of tumors, including breast cancer and non-small cell lung cancer, underscoring its potential value in anti-cancer drug development ([34]). The second potential inhibitor, olverembatinib (B) showed a binding affinity of −10.9 kcal/mol with the modeled MshC structure (C). While the trifluoromethyl group of olverembatinib forms a hydrogen bond with the main chain of W227 (2.3 Å), the compound also establishes two carbon hydrogen bonds with G226 and G189 and forms additional interactions with surrounding residues including D86, G189, W227 and L204 (B). Similar to rebastinib, olverembatinib is also a BCR-ABL1 tyrosine kinase inhibitor having been approved in China for the treatment of some Chronic Myeloid Leukemia patients ([35], [36]). One other top-list compound, SB-224289 (B), with a binding affinity of −12.1 kcal/mol, is a small molecule initially discovered as a human serotonin receptor antagonist ([37]) and found to antagonize the antifungal mechanisms of Papuamide A ([38]). In MshC, SB-224289 (C) is anchored mainly through a strong π-anion interaction with D87 and a π–alkyl/π–π stacking interaction with W227. Additional stabilization is provided by stacking hydrophobic interactions with the backbone of R188 and G189 on one side and P88 on the other side (C).
MshC inhibitors disrupt mycothiol biosynthesis in M. abscessus
To investigate the effect of rebastinib, olverembatinib, and SB-224289 on MSH production, Mab was treated with each inhibitor and the MSH concentrations were measured using LC-MS. While a clear peak representing MSH could be visualized at retention time of 4.2 min on the extracted ion chromatogram at [M + H]+= 487.1592 m/z in the untreated samples, the ion counts were strongly reduced in bacteria treated with 0.1× MIC50 of any of the inhibitor (D), and the MSH peak disappeared completely when cultures were treated with 0.5× MIC50 of each inhibitor (A), suggesting a dose-dependent effect of the inhibitors on MSH production. To exclude the effects of altered overall metabolism, glucosamine inositol (1-D-myo-inosityl-2-amino-2-deoxy-α-D-glucopyranoside)—the substrate of MshC characterized by [M-H]−= 340.1249 m/z—was extracted and normalized to protein concentrations. No significant difference was found between untreated or treated cultures (B).
Activity of rebastinib, olverembatinb, and SB-224289 against mycobacteria
Exposure of exponentially-growing Mab to increasing concentrations of rebastinib, olverembatinib, and SB-224289 shows that the three compounds inhibit growth, with the latter one being the most active compound sharing a comparable effect to rifabutin, included as a control drug ([39]) (, A–C). The MIC50 values of rebastinib, olverembatinib, and SB-224298 were 20.8, 36.9, and 2.4 μM, respectively. While both rebastinib and olverembatinib showed a bacteriostatic effect, SB-224298 exhibited a bactericidal effect, like rifabutin (). The activity of all three compounds was further confirmed following determination of the CFUs when treating the cultures with 1×, 5×, and 10× MIC50 of rebastinib (D), olverembatinib (E) and SB-244289 (F). The CFU numbers at 72 h of treatment were comparable to those of the inoculum at all concentrations tested for rebastinib and olverembatinib. In contrast, exposure to SB-244289 was associated with a more pronounced killing effect, similar to rifabutin ([39]).
[Figure 3] Rebastinib, olverembatinib and SB-224289inhibit M. abscessus growth of in vitro.A–C, susceptibility of Mab S to rebastinib, olverembatinib, SB-224289 and rifabutin. Overnight bacterial cultures (OD600∼0.8) were collected and adjusted to an OD of 0.005 in MH medium in 96-well plates. Compounds were added to the wells in 2-fold serial dilutions and plates were incubated at 37 °C for 3 days prior to the addition of 40 μl of 0.025% resazurin/well, incubation at 37 °C overnight and fluorescence measurements. MIC50 values were determined by fitting the curve with a sigmoidal dose-response using the GraphPad Prism software. D–F, killing curves of Mab S exposed to 1×, 5×, 10× MIC50 of rebastinib, olverembatinib or SB-224289 in CaMHB at 37 °C. Rifabutin at 100 μM was added as a control drug. At various time points, bacteria were plated on 7H10 agar and further incubated at 37 °C for 3 days prior to CFU counting. Results are expressed as the mean of three independent experiments done in triplicates ±SD. Data were analyzed using a one-way ANOVA Dunnett’s test. RFU, Relative Fluorescence Unit. ∗∗∗, p≤ 0.001.
[Table 1] MIC and MBC values of rebastinib, olverembatinib, SB-224289 and rifabutin against M. abscessus CIP104536 (S) grown in MH medium
| Parameter | Rebastinib | Olverembatinib | SB-224289 | Rifabutin |
|---|---|---|---|---|
| MIC (μM) | 20.79 ± 0.74 | 36.90 ± 0.59 | 2.39 ± 0.31 | 2.68 ± 0.08 |
| MBC (μM) | 100 | 200 | 6.25 | 6.25 |
| MBC/MIC | 4.81 | 5.42 | 2.61 | 2.33 |
| Effect | Bacteriostatic | Bacteriostatic | Bactericidal | Bactericidal |
The activity of all three compounds was next tested using a wide panel of clinical isolates belonging to the Mab complex, as well as against other mycobacterial species. All showed activity against the different clinical isolates, with rebastinib outperforming the efficacy of olverembatinib (). Among the three inhibitors, SB-244289showed the highest inhibitory activity against most strains, but the MIC50 values of SB-244289 were higher than those of clarithromycin. Interestingly, the three inhibitors were active against M. tuberculosis (10.1–25.5 μM range) but inactive against M. smegmatis (). The potent activity of SB-224289 prompted to assess the ability of three closely related structural analogs, designated GR-127935, SB-216641 and SB-616234-A (A), to inhibit Mab growth. With MIC50 values of 24, 108 and 9.5 μM, respectively, these compounds appear less active than SB-224289, albeit targeting MshC as can be inferred from their very low activity (MIC >200 μM) against the Mab strain overexpressing mshC, B and C).
[Table 2] Activity of rebastinib, olverembatinib and SB-224289 against a panel of M. abscessus clinical isolates and other mycobacterial species
| MIC50 (μM) | Rebastinib | Olverembatinib | SB-224289 | Clarithromycin |
|---|---|---|---|---|
| M. abscessus complex | ||||
| GD17 | 38.26 ± 3.21 | 52.82 ± 7.54 | 1.64 ± 0.20 | 0.22 ± 0.04 |
| GD22 | 55.68 ± 1.88 | 61.95 ± 2.11 | 2.35 ± 0.60 | 0.28 ± 0.05 |
| GD38 | 17.25 ± 2.32 | 52.86 ± 1.5 | 1.66 ± 0.71 | 0.39 ± 0.03 |
| GD59 | 53.43 ± 0.82 | 58.68 ± 0.78 | 6.72 ± 1.12 | 0.3 ± 0.05 |
| GD163 | 23.22 ± 1.07 | 32.81 ± 1.48 | 5.75 ± 2.15 | 0.21 ± 0.06 |
| GD180 | 33.96 ± 4.57 | 46.11 ± 0.67 | 3.41 ± 0.82 | 0.11 ± 0.02 |
| 330 | 22.5 ± 1.67 | 29.45 ± 1.26 | 1.15 ± 0.52 | 0.16 ± 0.03 |
| 383 | 25.56 ± 3.6 | 21.79 ± 1.03 | 1.63 ± 0.20 | 0.35 ± 0.01 |
| 386 | 16.22 ± 2.06 | 32.64 ± 2.27 | 1.68 ± 0.19 | 0.25 ± 0.09 |
| 420 | 13.61 ± 2.88 | 46.15 ± 0.79 | 1.75 ± 0.45 | 0.35 ± 0.03 |
| 442 | 6.74 ± 0.08 | 9.93 ± 1.84 | 1.70 ± 0.90 | 0.37 ± 0.38 |
| 759 | 7.02 ± 0.7 | 22.06 ± 1.02 | 3.34 ± 0.22 | 0.74 ± 0 |
| 783 | 11.54 ± 0.96 | 43.19 ± 1.89 | 2.45 ± 1.12 | 0.41 ± 0.04 |
| 793 | 14.36 ± 1.61 | 22.07 ± 0.64 | 3.45 ± 0.60 | 0.12 ± 0 |
| M. massiliense CIP108297 (S) | 12.35 ± 2 | 16.82 ± 2.63 | 3.15 ± 1.05 | 0.004 ± 0.002 |
| M. massiliense CIP108297 (R) | 14.03 ± 4.69 | 46.52 ± 0.81 | 4.21 ± 0.91 | 0.35 ± 0.02 |
| Other mycobacterial species | ||||
| M. goodii | 8.3 ± 1.5 | 14.91 ± 4.41 | 2.85 ± 1.80 | Inactive |
| M. mageritense | 20.75 ± 3.5 | 45.72 ± 1.9 | 5.89 ± 1.15 | 1.98 ± 1.52 |
| M. fortuitum | 6.09 ± 1.13 | 20.04 ± 0.29 | 1.40 ± 2.20 | 0.19 ± 0.04 |
| M. smegmatis | Inactive | Inactive | Inactive | ND |
| M. tuberculosis mc26230 | 12.31 ± 0.82 | 25.53 ± 1.24 | 10.15 ± 0.33 | ND |
Overall, these results demonstrate that rebastinib, olverembatinib, and SB-244289 exert potent activity against Mab and other mycobacteria.
Mutations in MshC confer resistance to rebastinib, olverembatinib, and SB-224289
To address whether mutation in MshC may be responsible for increased levels of resistance to rebastinib, olverembatinib, and SB-224289, spontaneous resistant mutants were selected by plating Mab CIP104536 (S) in the presence of 1×, 5× or 10× MIC50 of each inhibitor. The frequency of resistance for each compound is presented in. While this frequency was relatively high for rebastinib and olverembatinib (varying between 1.1 × 10−6 and 4 × 10−6, depending on the drug concentration used for selection), it was much lower in the case of SB-224289 (varying between 5.5 × 10−6 at 1× MIC50 to 5 × 10−8 at 10× MIC50). Three rebastinib resistors (designated R1 to R3) and three olverembatinib resistors (designated O1 to O3) were isolated, exhibiting increased resistance levels to rebastinib and olverembatinib, respectively, as compared to the parental strain, while all mutants remained susceptible to rifabutin (A). MIC determination pointed to a >10-fold increased resistance level for R1-R3 (>200 μM) and a >5-fold increased resistance level for O1-O3 (>200 μM). Four selected SB-224289 resistors (designated SB1-SB4) displayed >17-fold increased resistance levels (>50 μM) as compared the parental strain (). Interestingly, all three mutants selected for resistance to olverembatinib (O1 to O3) showed cross-resistance to rebastinib (5-fold increased resistance as compared to the parental strain) and conversely, all three mutants selected for resistance to rebastinib (R1 to R3) showed cross-resistance to olverembatinib (4 to 5-fold increased resistance as compared to the parental strain) (B). Similarly, all four SB-224289 resistors (SB1-SB4) were highly resistant to the two other compounds (>200 μM) (). Together, these results suggest that the three inhibitors are very likely sharing the same target and mode of action in Mab.
[Figure 4] Spontaneous mshC mutants are cross-resistant to rebastinib, olverembatinib and SB-224289.A, susceptibility profile Mab spontaneous resistance mutants to rebastinib, olverembatinib, and rifabutin. Overnight bacterial cultures were collected and adjusted to an OD of 0.005 in CaMHB and resuspended into 96-well plates. Rebastinib, olverembatinib or rifabutin were added to the wells in 2 fold-serial dilutions. Plates were incubated at 37 °C for 3 days prior to the addition of 40 μl of 0.025% resazurin per well, further incubated at 37 °C overnight before fluorescence measurements. The MIC50 was determined by fitting the curve with a sigmoidal dose response using the GraphPad Prism software. R1, R2 and R3 correspond to three independent strains selected on agar in the presence of rebastinib. O1, O2 and O3 correspond to three independent strains selected on agar in the presence of olverembatinib. B, cross-resistance of O1-O3 strains to rebastinib (left panel), R1-R3 strains to olverembatinib (middle panel) and R1-R3, O1-O3 to SB-224289. RFU, Relative Fluorescence Unit.
[Table 3] Mutations in mshC confer cross-resistance to rebastinib, olverembatinib and SB-224289
| Strain | MIC50 rebastinib (μM) | MIC50 olverembatinib (μM) | MIC50 SB-224289 (μM) | MIC50 rifabutin (μM) | SNP | AA change |
|---|---|---|---|---|---|---|
| WT | 19.49 ± 0.49 | 39.19 ± 1.2 | 2.89 ± 0.15 | 3.05 ± 0.33 | - | - |
| R1 | >200 | 137.5 ± 0.22 | >50 | 3.00 ± 0.63 | A260C | D87A |
| R2 | >200 | >200 | >50 | 3.05 ± 0.42 | A260C | D87G |
| R3 | >200 | 155.3 ± 0.25 | >50 | 2.95 ± 0.36 | A260G | D87A |
| O1 | 114.7 ± 0.68 | >200 | >50 | 2.94 ± 0.40 | T746A | W227R |
| O2 | 105.2 ± 0.75 | >200 | >50 | 3.02 ± 0.40 | T746A | W227R |
| O3 | 108.2 ± 0.62 | >200 | >50 | 3.04 ± 0.31 | T746G | W227G |
| SB1 | >200 | >200 | >50 | 3.15 ± 0.68 | A260C T746A | D87A W227R |
| SB2 | >200 | >200 | >50 | 3.14 ± 0.65 | T746A | W227R |
| SB3 | >200 | >200 | >50 | 3.20 ± 0.72 | A260G | D87A |
| SB4 | >200 | >200 | >50 | 3.35 ± 0.52 | A260G | D87A |
Comparative analysis of the mshC sequences from R1-R3 with the parental strain identified a single nucleotide polymorphism (SNP) in mshC in all three resistors at position D87 (D87A replacements in R1 and R3 and D87G replacement in R2) (). In O1 to O3, the SNP in mshC was found at position W227 (W227R replacement in O1 and O2 and W227G replacement in O3). Mutants selected on SB-224289 harbored either a D87A replacement and/or a W227R mutation (). To further confirm that these sole mutations are responsible for resistance to the inhibitors, recombineering was used to restore a wild-type mshC gene into the O1 and R1 resistant mutants (A). Bacteria in which the R227 and A87 were replaced by W227 and D87 residues, respectively, became fully susceptible to all three inhibitors (B).
Overall, these results highlight the presence of common missense mutations in MshC conferring cross-resistance to rebastinib, olverembatinib and SB-224289 in Mab.
Modulating mshC transcriptional levels affects the activity of rebastinib and olverembatinib
We next investigated whether low mshC transcriptional levels sensitize Mab to the three MshC inhibitors. This was achieved by treating the Mab_pJLR962:mshC with a sub-MIC concentration of ATc (110 ng/ml) (). ATc did not influence the activity of rebastinib, olverembatinib, SB-224289 or rifabutin in Mab_pLJR962 (A, upper panels). In contrast, Mab_pJLR962:mshC showed a 2-fold increased susceptibility to rebastinib, olverembatinib, and a 4-fold increased susceptibility to SB-224289, but not to rifabutin, when increasing the ATc concentration from 0 to 110 ng/ml (A, lower panels and). These effects could be reproduced when spotting the cultures onto CaMH agar and were particularly pronounced with SB-224289 (B).
[Figure 5] Modulation of mshC expression alters the susceptibility to rebastinib, olverembatinib and SB-224289.A, susceptibility of Mab R strains carrying either the empty pLJR962 (upper panels) or the pLJR962-mshC_sgRNA4 (lower panels) to rebastinib, olverembatinib, SB-224289 and rifabutin was determined in the absence or presence of 110 ng/ml of ATc. Overnight cultures were collected, adjusted to an OD of 0.005 in CaMHB supplemented with 250 μg/ml kanamycin. Drugs were added to the well in 2-fold serial dilutions and MIC50 values were determined using the REMA assay and by fitting the curve with a sigmoidal dose response using the GraphPad Prism software. Results are expressed as the mean of three independent experiment done in triplicates ± SD. B, serial dilutions of the pLJR962-and pLJR962-mshC-containing strains were spotted onto CaMHA plates supplemented with 250 μg/ml kanamycin and the indicated concentrations of rebastinib, olverembatinib, SB-224289, and rifabutin (RFB), in the absence and presence of ATc (110 ng/ml). C, crude lysates of Mab S WT or Mab S carrying the pMV261:mshC (fused with an HA-tag) were probed with anti-HA antibodies (upper panel) or with anti-Hsp65 antibodies (loading control; lower panel). D, susceptibility profile of Mab S WT or Mab S carrying the pMV261:mshC to rebastinib, olverembatinib, SB-224289 and RFB. Growth conditions were as described in (A). Results are expressed as the mean of three independent experiments done in triplicates ±SD.
In contrast, overexpression of a wild-type copy of mshC under the control of the strong constitutive hsp60 promoter in wild-type Mab (validated by Western blotting using anti-HA antibodies) (C) resulted in high levels of resistance to all three MshC inhibitors, but not to rifabutin (D). This indicates that the mshC expression levels influence the susceptibility profile of Mab to rebastinib, olverembatinib and SB-224289. Overall, these data further support the hypothesis that rebastinib, olverembatinib, and SB-224289 inhibit specifically the MSH pathway.
Systematic assessment of olverembatinib and SB-224289 drug combinations
To evaluate how MshC inhibitors could be integrated into future multidrug regimens, we systematically assessed pairwise combinations of olverembatinib and SB-224289 with 19 additional Mab-active drugs using a diagonal checkerboard approach (DiaMOND; ([40])). Most interactions were antagonistic, particularly when MshC inhibitors were combined with aminoglycosides, β-lactams, or quinolones (, A–C and, A and B). Despite this overall trend, several potent drug-drug synergies emerged, including multiple combinations with oral agents. SB-224289 synergized strongly with bedaquiline and clofazimine, while olverembatinib showed marked synergy with bedaquiline and macrolides. Linezolid, another commonly used oral drug, was additive with both MshC inhibitors. Overall, drug interaction profiles of olverembatinib and SB-224289 were highly correlated (R2= 0.82).
[Figure 6] Pairwise drug interactions of SB-224289.A, interaction of pairwise combinations of SB-224289 with 19 other drugs active against Mab (+SB-224289 itself to show no interaction). The interaction is determined by the log2 of the fractional inhibitory concentration (FIC), with dot colour showing the interaction type (blue for synergistic and red for antagonistic interactions), and dot size showing the interaction strength (absolute value). B, DiaMOND heatmaps ([40]) of Mab growth for all pairwise combinations with SB-224289 across increasing drug concentrations (same combinations as in panel A). Drug concentrations were centred on the single-drug MIC50 and increased from left to right in log√2-fold dilution steps. Diamonds indicate the expected drug combination MIC50 based on single drug activity and dots show the experimentally observed MIC50 of the combination. Dot colours indicate the interaction type (yellow: additive; blue: synergistic; and red: antagonistic). C, representative dose response curves for a synergistic (SB-224289 + bedaquiline), antagonistic (SB-224289 + imipenem) and additive (SB-224289 + olverembatinib) drug combinations.
Discussion
Mab pulmonary diseases pose a significant clinical challenge mainly because of their intrinsic resistance level to many antibiotic classes ([11], [41], [42]). This resistance highlights the urgent need for new therapeutic strategies. However, the current Mab drug pipeline remains largely underpopulated, emphasizing the need for novel chemical scaffolds and drug targets to combat this pathogen ([43], [44]). Drug repurposing of compounds for new therapeutic applications is particularly attractive in antibacterials, considering that the repurposing of previously approved drugs can shorten development time and lower costs. While the opportunities and challenges associated with drug repurposing for the treatment of Mab pulmonary infections have been proposed ([11], [43], [45], [46]), only a few reports discuss novel compounds that are being developed against these emerging pathogens ([47]).
In this study, CRISPRi-based gene silencing of mshC in Mab indicates that mshC is essential for Mab growth in vitro and during infection in zebrafish larvae, thus validating MshC as an attractive vulnerable site of inhibition, as suggested previously in M. tuberculosis ([22], [23]). This differs, however, from the situation in M. smegmatis, where mshC is dispensable ([24]). Our results, thus, support the importance of mshC and MSH production in Mab, as proposed in recent saturated transposon mutagenesis studies ([25], [48]).
To leverage the essentiality of MshC for drug design, in silico drug screening identified the tyrosine kinase inhibitors, rebastinib and olverembatinib, considered for the treatment of Chronic Myeloid Leukemia, and the human serotonine receptor antagonist SB-224289. While these molecules have not previously been reported as antibacterials, our results emphasize their potent antimycobacterial activity and their mode of action by targeting the MSH pathway in Mab. This is supported by various complementary findings: 1) spontaneous rebastinib-resistant mutants are cross-resistant to olverembatinib and SB-224289 and vice versa, suggesting that the three compounds, which are structurally unrelated, share the same target; 2) missense mutations in MshC at residues 87 and/or 227 are found in all spontaneous resistant mutants; 3) increased resistance to all three compounds in a Mab strain overproducing MshC; 4) reduced MIC values of rebastinib, olverembatinib and SB-224289 in the mshC CRISPRi mutant treated at low ATc concentrations; 5) docking studies revealing binding of all three inhibitors into the same cavity in MshC; 6) a dose-dependent decrease of MSH production in Mab cultures exposed to sub-MIC levels of each inhibitor.
While MIC values of rebastinib and olverembatinib are higher than those of standard drugs, such as amikacin or clarithromycin, the MIC of SB-224289 is comparable to the one of rifabutin and found to be active in all Mab isolates tested. However, all three compounds were effective against a panel of M. abscessus complex clinical strains. Interestingly, these inhibitors also showed potent in vitro activity against other opportunistic NTM species, including Mycobacterium fortuitum as well as against M. tuberculosis, but were inactive against M. smegmatis, presumably because mshC is not essential in this species ([24]). Thus, the MSH pathway represents a therapeutic target to be further explored in Mab, which aligns with the recent demonstration that the NADPH-dependent MSH reductase (Mtr), catalyzing the reduction of oxidized mycothione (MSSM) into MSH, while not essential in vitro, is important for intramacrophage replication and survival of Mab in Galleria mellonella ([49]).
Rebastinib, olverembatinib and SB-224289 are structurally unrelated and do not share structural relatedness with previously characterized M. tuberculosis MshC inhibitors, such as cysteinyl-sulfonamides ([31]), NTF1836 ([50]) or dequalinium ([30]). Our work shows that these three inhibitors represent promising molecular scaffolds targeting MshC in Mab. Structural superposition of modeled Mab-MshC⋅SB-224289 complex with the crystallographic structures of Ms-MshC⋅CysSA (PDB: 3C8Z) and Ms-MshC⋅Compound 7b (PDB: 8HFN) showed that all three ligands bound in the catalytic pocket (, A and B). Similar to the CysSA and 7b, the indole-oxadiazole moiety of SB-224289 likely occupies the same canonical cysteinyl-adenylate (Cys-AMP) reaction intermediate binding site. However, the multi-heterocycle group of SB-224289 extends toward the new binding channel formed the C-terminal residues of I178-R188 helix and N-terminal ends of the other two α-helixes (D86-G97 and W227-L239) (, C and D). As the three identified Mab-MshC inhibitors are affected by residues D87 and W227, both located in the same channel region, this suggests that Mab-MshC possesses an unexpected potential channel, as an auxiliary pocket, that can be targeted.
SB-224289, the most potent compound in vitro, could not be assessed in the zebrafish model due to its low solubility in water, which requires adding the compound directly to fish water ([51]). However, our data on a small set of structural analogs of SB-224289 () clearly shows the possibility to manipulate these compounds, implying that future chemical improvements may help to translate into new analogs with optimized biophysical and pharmacological properties and to evaluate their safety profile and efficacy in pre-clinical models. The elucidation of the crystal structure of the Mab-MshC.inhibitor complexes should greatly assist in guiding structure-activity relationship studies to generate a new generation of analogs with enhanced effect. Structural characterization of these complexes is particularly important, as the poor aqueous solubility of rebastinib, olverembatinib, and SB-224289 prevented reliable measurement of binding constants with purified MshC (data not shown).
Given the potent anti-M. abscessus activity of MshC inhibitors, we also examined whether they could be effectively integrated into multidrug treatment regimens by assessing pairwise drug combinations. Consistent with prior reports in mycobacteria and other bacterial species, most drug interactions were not synergistic ([52], [53]). Nevertheless, MshC inhibitors showed strong synergies with several of the most effective oral agents currently used to treat Mab disease, including macrolides, clofazimine and bedaquiline. Linezolid, another important component of current oral regimens, showed additive interactions with olverembatinib and SB-224289. These findings suggest that MshC inhibitors have potential not only as potent standalone agents but also as promising partners for oral, long-term combination therapy for Mab.
In conclusion, our work unraveled new MshC-specific inhibitors as promising antimicrobial agents with potential for clinical application in Mab infections, and identified a new binding channel for drug discovery besides the canonical intermediate binding site. Their unique mode of action, synergistic effects with standard and non-standard antibiotics position them as candidates for further developments against Mab pulmonary diseases.
Experimental procedures
Plasmids and primers
Plasmids and primers are listed in and, respectively.
Reagents
Rebastinib, olverembatinib and SB-224289 were purchased from MedChemExpress. Rifabutin was obtained from Selleckchem. All reagents were dissolved in dimethyl sulfoxide (DMSO).
Bacterial growth conditions
Smooth (S) and rough (R) variants of Mab CIP104536T ([54]) were used to inhibit mshC expression. All mycobacterial strains, including clinical isolates, were routinely cultured at 37 °C in Middlebrook 7H9 broth (BD Difco) supplemented with 0.025% Tyloxapol (Sigma-Aldrich) and 10% oleic acid, albumin, dextrose, and catalase (OADC enrichment; BD Difco) (7H9T/OADC) or on Middlebrook 7H10 agar (BD Difco) containing 10% OADC (7H10OADC) with antibiotics added when necessary. For drug susceptibility testing, bacteria were grown in cation-adjusted Mueller-Hinton broth (CaMHB; Sigma-Aldrich). To evaluate the growth of the mshC_CRISPRi mutants, cultures were spotted onto 7H10OADC and CaMH agar (CaMHA), prepared by mixing 8.8g of CaMHB with 6g of agar (Sigma-Aldrich) in 400 ml of water.
Drug susceptibility testing
The minimum inhibitory concentration (MIC) was determined by using the resazurin microtiter assay (REMA) in which bacterial cultures from the exponential growth phase were diluted to an optical density 600 nm (OD600) of 0.01. 50 μl of bacterial culture was added to each well in a sterile 96-well plate along with 50 μl of serial two-fold dilutions of the test compound. Each plate also contained a drug-free growth control. 200 μl of sterile water was added to the outer perimeter wells to avoid evaporation during incubation. The plates were covered with self-adhesive membranes and incubated at 37 °C for one to five days, depending on the strain. Then, 40 μl of 0.025% (w/v) resazurin solution was added to each well before overnight incubation. Fluorescence was measured by using a spectrophotometer multimode microplate reader (Tecan Spark 10M; Tecan Group Ltd). Results are expressed as RFU (%), corresponding to treated RFU / untreated control RFU × 100, with the untreated control set to 100%, reflecting bacterial growth measured by the resazurin assay. The concentrations required to inhibit bacterial growth by 50% (MIC50) were determined by fitting the curve with a sigmoidal dose-response using the GraphPad Prism software (version 6.05).
To evaluate whether the compound had a bactericidal or bacteriostatic effect, plates were incubated at 37 °C for 3 days. Then, 10 μl of bacterial suspension was transferred to each well in new plates containing 7H10OADC and further incubated at 37 °C for three more days before taking a picture. The MBC (Minimum Bactericidal Concentration) is the lowest drug concentration that results in no visible bacterial growth. The drug effect was classified as bactericidal if the ratio between MBC and MIC was ≤4, bacteriostatic if the ratio >4 but ≤32, and resistant if the ratio was >32 ([55]).
In vitro growth inhibition assay
Mab CIP104536T S cultures from the exponential growth phase were diluted to an OD600 of 0.01 and exposed to 1×, 5×, or 10× the MIC50 of rebastinib, olverembatinib and SB-224289 in CaMHB at 37 °C. Rifabutin at 100 μM was used as the control drug. Aliquots were taken at various time points, plated onto 7H10OADC, and incubated at 37 °C for 3 days prior to CFU counting.
Generation of resistant mutants and target identification
250 μl of Mab CIP104536T S cultures from the exponential growth phase were spread onto 7H10OADC supplemented with either rebastinib, olverembatinib or SB-224289 (at up to 10× MIC50), followed by incubation at 37 °C for 3 days. Individual colonies picked up and transferred into 7H9T/OADC supplemented with either rebastinib or olverembatinib (at up to 10× the MIC50) until OD600 reaches 0.8. The MIC of laboratory-generated resistant mutants was determined by using the REMA. Identification of SNPs in the resistant strains was done by PCR amplification using MAB_2116 val F and MAB_2116 val R () for full-coverage sequencing of mshC (MAB_2116).
Reversion of mutations in mshC into a wild-type copy
Restoring the wild-type mshC sequence into the spontaneous resistant strains O1 and R1 (carrying the W227R or D87A mutations, respectively) was done using the recombineering method ([56]). The wild-type mshC sequence was centered in lagging-strand oligonucleotides of 70 nucleotides. O1 and R1 strains were transformed with pJV53, grown to an OD600 of 0.8 in 7H9 broth containing 250 μg/ml kanamycin and exposed to 0.2% acetamide and 0.2% succinate for 24 h in the presence of 0.2 M glycine for 16 h. Competent cells were prepared and transformed with 100 ng of lagging-strand oligonucleotides. After overnight rescue at 37 °C, bacteria were plated on 7H10 agar plates containing 250 μg/ml kanamycin and transformants were screened for restoration of susceptibility to rebastinib, olverembatinib or SB-224289. Reversion of the wild-type mshC gene was subsequently confirmed by PCR and Sanger sequencing.
DNA constructs
To confirm whether mshC is the target gene of the three inhibitors, mshC was PCR-amplified using Mab CIP104536T S chromosomal DNA and the primers MAB_2116 InFus F and MAB_2116 InFus R (). The mshC gene in fusion with an HA-tag was cloned into the pMV261 ([57]) to generate pMV261:mshC-HA, and introduced into Mab CIP104536T S. Overexpression of MshC-HA was confirmed by Western blotting using anti-HA antibodies.
To generate the mshC conditional knockdown strains, the pLJR962 was used and sgRNAs targeting mshC were designed using the CHOPCHOP web tool. The forward and reverse 29-nucleotide sgRNA oligonucleotides were synthesized by Sigma-Aldrich. The sgRNA oligonucleotides were annealed and ligated into the BsmBI-linearized pLJR962 using T4 DNA ligase (New England BioLabs). The resulting constructs were transformed into Escherichia coli Stellar competent cells and selected on LB agar supplemented with 50 μg/ml kanamycin (Euromedex). Plasmids were extracted and verified by sequencing using primer 1834 () and electroporated into Mab CIP104536T (R). The transformants were selected in the presence of 500 μg/ml kanamycin. Mab CIP104536T (R) mshC_CRISPRi mutants were grown in 7H9T/OADC (OD600 of 1), and 10-fold diluted cultures were spotted onto 7H10OADC or CaMHA supplemented with increasing concentrations of anhydrotetracycline (ATc; Takara-Bio) or doxycycline (Sigma-Aldrich) and 500 μg/ml kanamycin, and incubated at 37 °C for 3 days.
To generate Mab CIP104536T (R) mshC_CRISPRi mutants expressing tdTomato, the pTEC27 was electroporated into Mab CIP104536T (R) carrying pLJR962:mshC_sgRNA4 and selected in the presence of 1000 μg/ml hygromycin (Sigma-Aldrich) and 500 μg/ml kanamycin.
Western blotting
Mab CIP104536T (S) carrying pMV261:mshC was grown in Sauton’s broth containing 0.025% tyloxapol and 500 μg/ml kanamycin. Bacteria were harvested by centrifugation at 3000g for 10 min at 4 °C and resuspended in cold PBS containing a protease inhibitor cocktail (Sigma-Aldrich). Cell lysis was performed using 1-mm glass beads and four times 60 s high-speed cycles in a bead beater (Mixer Mill MM 301, Retsch). The lysates were collected, and total protein concentration was determined using the BCA Protein Assay Reagent Kit (ThermoFisher Scientific), following the manufacturer's instructions. Equal amounts of protein (10 μg) were separated on a 12% SDS-PAGE gel and transferred onto a PVDF membrane (Immobilon-P, Merck Millipore). Membranes were probed using either rat anti-HA antibodies (Roche) at a 1:5000 dilution or mouse monoclonal antibodies anti-Hsp65 (Santa Cruz Biotechnology) at a 1:3000 dilution. Membranes were then incubated with either anti-rat or anti-mouse secondary antibodies conjugated to HRP (horseradish peroxidase) at a 1:5000 dilution, and the signal visualized using the SuperSignal West Femto substrate (ThermoFisher Scientific) on a ChemiDoc MP imaging system (Bio-Rad).
Transcriptional analyses
Total RNA was isolated, followed by reverse transcription and quantitative real-time PCR (RT-qPCR) as previously described ([58]), using the primers listed in (). The housekeeping gene rpoB served as the reference for relative gene expression analysis.
Metabolite extraction
Mab ATCC19977 strain was cultured in 7H9OADC in a shaking incubator set at 180 rpm and 37 °C. MSH inhibitors (rebastinib, olverembatinib, SB-224289) were spiked into media at 0.5× MIC50 at the same time of bacterial inoculation, together with a group of untreated controls. Bacteria were grown until the OD600 reached 0.5 to 0.8. 10 ml of bacterial cultures were centrifuged at 3000 rpm, 4 °C, to remove the supernatant and washed once with cold sterile PBS. Bacterial pellets were then metabolically quenched by resuspending in 1 ml of cold extraction solution composed of acetonitrile/methanol/H2O (2:2:1; v/v/v). Metabolites were extracted by mechanical lysis of the bacteria-containing solution with 0.1 mm acid-washed zirconia beads twice, each for 30 s using a FastPrep (MP Bio) set to 6.0 m/s, with a 5-min interval on ice. Cell debris was precipitated by centrifugation at 15, 000× g for 5 min, and the cell lysates were transferred to 0.22 μm Spin-X column filters (CoStar) and filtered by centrifugation at 15, 000g for 30 min at room temperature. A 100 μl aliquot of the metabolite solution was then mixed with 100 μl of cold acetonitrile and centrifuged for 10 min at 17, 000× g. The final concentration of 70% acetonitrile was compatible with the starting conditions of chromatography. 100 μl of the supernatant was then transferred into LC-MS V-shaped vials (Agilent 5188–2788), and 4 μl was injected into the LC-MS instrument (Agilent 6545 QToF). The bacterial biomass of the individual samples was determined by measuring the residual protein content of the metabolite extracts using the BCA assay kit (Thermo).
Liquid chromatography-mass spectrometry
Aqueous normal-phase liquid chromatography was performed using an Agilent 1290 Infinity II LC system equipped with a binary pump, temperature-controlled autosampler (set at 4 °C) and temperature-controlled column compartment (set at 25 °C) containing a Cogent Diamond Hydride Type C silica column (150 mm × 2.1 mm; dead volume of 315 μl). A flow rate of 0.4 ml/min was used. The elution of polar metabolites was performed using solvent A, consisting of deionized water (resistivity ∼18 MW cm) and 0.2% acetic acid, and solvent B, consisting of 0.2% acetic acid in acetonitrile. The following gradient was used at a flow rate of 0.4 ml/min: 0 min, 85% B; 0 to 2 min, 85% B; 3 to 5 min, 80% B; 6 to 7 min, 75% B; 8 to 9 min, 70% B; 10 to 11 min, 50% B; 11.1 to 14 min, 20% B; 14.1 to 25 min, 20% B; and 5-min re-equilibration period at 85% B. For quantification of glucosamine inositol, chromatographic separation was performed with an Agilent InfinityLab Poroshell 120 HILIC-Z (2.1 × 100 mm, 2.7 μm (p/n 675,775–924)) column. The HILIC-Z methodology was optimized for polar acidic metabolites. Column compartment was set at 50 °C. For easy and consistent mobile-phase preparation, a concentrated 10x solution consisting of 100 mM ammonium acetate (pH 9.0) in water was prepared to produce mobile phases A and B. Mobile phase A consisted of 10 mM ammonium acetate in water (pH 9) with a 5 μM Agilent InfinityLab deactivator additive (p/n 5191–4506), and mobile phase B consisted of 10 mM ammonium acetate (pH 9) in 10:90 (v:v) water/acetonitrile with a 5 μM Agilent InfinityLab deactivator additive (p/n 5191–4506). The following gradient was applied at a flow rate of 0.25 ml/min: 0 min, 96% B; 2 min, 96% B; 5.5 min, 88% B; 8.5 min, 88% B; 9 min, 86% B; 14 min, 86% B; 17 min, 82% B; 23 min, 65% B; 24 min, 65% B; 24.5 min, 96% B; 26 min, 96% B and 3-min of re-equilibration at 96% B.
Accurate mass spectrometry was performed using an Agilent Accurate Mass 6545 QTOF apparatus. Dynamic mass axis calibration was achieved by continuous infusion after chromatography of a reference mass solution using an isocratic pump connected to an ESI ionization source operated in the positive-ion mode. The nozzle and fragmentor voltages were set to 2000 V and 100 V, respectively. The nebulizer pressure was set to 50 psig, and the nitrogen drying gas flow rate was set to 5 L/min. The drying gas temperature was maintained at 300 °C. The MS acquisition rate was 1.5 spectra/sec, and m/z data ranging from 50 to 1200 were stored. This instrument enabled accurate mass spectral measurements with an error of less than 5 parts per million (ppm), a mass resolution ranging from 10, 000 to 45, 000 over the m/z range of 121 to 955 atomic mass units, and a 100, 000-fold dynamic range with picomolar sensitivity. The data were collected in the centroid 4-GHz (extended dynamic range) mode. The detected m/z data were deemed to be metabolites identified based on unique accurate mass-retention time and MS/MS fragmentation identifiers for masses exhibiting the expected distribution of accompanying isotopomers. Typical variation in abundance for most of the metabolites remained between 5 and 10% under these experimental conditions.
Metabolomics data analysis
The extracted ion spectra (EIC) of mycothiol (C17H30N2O12S, [M + H]+= 487.1592 m/z) were extracted from all samples in positive ion mode using the Agilent Qualitative Navigator B.08.00 software. The retention times of the peaks in each sample were aligned to ensure accurate annotation of the molecule, and the abundances over time were extracted from the software and processed using GraphPad Prism.
Zebrafish maintenance and ethics
Zebrafish (Danio rerio) were kept and handled in compliance with the guidelines of the European Union for handling laboratory animals and approved by the Direction Sanitaire et Vétérinaire de l’Hérault for the ZEFIX-CRBM zebrafish facility (Montpellier) (registration number C-34–172–39). Experiments were approved by the Ministère de l’Enseignement Supérieur, de la Recherche et de l’Innovation under the reference APAFIS#24406 to 2020022815234677 V3. Experiments were conducted using wild-type AB zebrafish, raised in the ZEFIX-CRBM zebrafish facility and kept on a 12/12 h light/dark cycle. Eggs were obtained by natural spawning, quickly bleached and incubated at 28.5 °C in Petri dishes containing E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4).
Zebrafish infection and ATc treatment
At 24 h post-fertilization (hpf), embryos were dechorionated using Pronase (10 mg/ml; Millipore Sigma) for up to 5 min at room temperature, followed by extensive washing in E3 medium and placed in 60 mm Petri dishes containing E3 at 28.5 °C. At 30 hpf, embryos were anesthetized with 0.02% buffered MS222 (Tricaine; ethyl-3-aminobenzoate methanesulfonate salt). Borosilicate glass capillaries (O.D. = 1 mm; I.D. = 0.78 mm; 10 cm length) were pulled using a micropipette puller (Sutter instruments P-97). Needles were opened with fine forceps and loaded with 8 μl of the desired inoculum resuspended in Polyvinylpyrolidone 2% (PVP, Cas#9003–39–8) supplemented with 0.05% phenol red for microinjection visualization. Larvae were injected in the caudal vein with 3 nl (containing 212 ± 14 CFU) of either the Mab R_pLJR962 (control) or the Mab R_pLJR962:mshC, previously transformed with pTEC27 expressing TdTomato ([59]) and selected in the presence of 1000 μg/ml hygromycin or injected with the PVP control. Injected larvae were then rinsed twice and transferred in 60 mm small Petri dish containing embryo media at 28.5 °C. Inoculum was checked a posteriori by microinjecting a PBS drop and plating it on 7H10OADC.
To repress expression of mshC upon infection, half of the larvae from both groups (control and infected) were injected at 1 day post-infection (dpi) intravenously with 10 ng of ATc (5 nl of a 2.0 mg/ml solution in sterile PBS) and the other half with the PBS control ([29]). Larvae were randomized over survival assays and monitored daily for 11 days (n ≈24 per group per experiment).
Homology modeling and docking studies
The MshC protein sequence from Mab (ATCC19977) was retrieved from the UniProtKB database (accession code: B1MPE6) and submitted to the SWISS-MODEL online workspace (https://www.swissmodel.expasy.org) for structural validation. Suitable structural templates were identified using Basic Local Alignment Search Tool (BLAST) and a Hidden Markov model (HMM-HMM)-based lightning-fast iterative sequence search (HHblits) ([60], [61]). The top-ranked template was selected to generate an energy-minimised 3D model using the ProMod3 modeling engine ([62]). The model's quality was evaluated using QMEAN and GMQE scores where a QMEAN score below −4.0 indicates low model quality and GMQE values range from 0 to 1, with higher scores reflecting greater reliability ([62], [63]). Additional structural validation, including stereochemical quality and Ramachandran plot analysis, was also performed using the SWISS-MODEL platform ([64]). The homology modeled MshC target protein in PDB format was downloaded from the SWISS-MODEL workspace. The protein structure was then prepared by AutoDock Vina ([65]).
The docking simulation of a ligand library from the DrugBank database ([66]), consisting of 15,236 compounds was performed using PyRx virtual screening software with AutoDock Vina ([65], [67], [68]). The Vina algorithm applies a gradient based search to predict binding affinities and interaction modes of ligands within the homology modeled MshC protein. The top-scoring docking results were further visualized to analyze molecular interactions using Discovery Studio Visualizer v16.1.0.15350.
Drug–drug combination and interaction assessments
Mab ATCC19977 cultures were grown in 7H9 medium supplemented with OADC (10%), glycerol (0.4%) and Tween 80 (0.05%) to logarithmic phase, stored at −80 °C, and diluted to approximately 3 × 106 CFU/ml prior to each assay. Approximately 1.2 × 105 CFUs were dispensed into each well of 384-well plates. Compounds were prepared in 7H9 medium and added to the plates, which were then sealed and incubated at 37 °C with shaking (250 rpm) for 72 h. Optical density (OD600) was measured, and bacterial fitness was calculated as background-corrected OD600 normalized to the median OD600 of growth-control wells on each plate. Twenty single drugs, including olverembatinib and SB-224289, were tested across nine concentrations in log√2-fold dilution steps, with MIC50 set to be around the sixth concentration out of nine for a complete growth spectrum. Pairwise combinations were assessed using the DiaMOND diagonal sampling design (diagonal of a 9 × 9 checkerboard), in which each drug contributes half of the final combination concentration ([40]). Four technical replicates were assessed for each condition. Interaction scores were derived using DiaMOND analysis:
Additivity reference lines were determined according to the Loewe additivity model, which assumes additivity when the combined effect equals that of an equivalent total dose of either compound alone.
Statistical analyses
Statistical analyses were performed on GraphPad Prism software (version 6.05; San Diego, CA, USA) and detailed in each figure legend. The normality of data was evaluated using a Shapiro-Wilk test. Data were analyzed using a one-way ANOVA Dunnett’s test after validating the normality of the data. ns, no significant; ∗, p≤ 0.05; ∗∗, p≤ 0.01; ∗∗∗, p≤ 0.001; ∗∗∗∗, p≤ 0.0001.
Data availability
All data are contained within the manuscript and the. The raw data can be share upon request. All LC-MS metabolomics data has been uploaded into MetaboLights data repository with the accession code MTBLS13995 ([69]).
Supporting information
This article contains ([57]).
Conflict of interest
The authors declare that they do not have any conflicts of interest with the content of this article.
Acknowledgments
We are grateful to M. Plays, S. Decamps, A. Morel and A. Ollier for zebrafish husbandry.
Author contributions
L. P., C. H., T. Q. N., L. K., G. L-M., L. B., Y. L., and G. C. G. writing – original draft; L. P., C. H., T. Q. N., G. L-M., L. B., Y. L., and G. C. G. methodology; L. P., C. H., G. L-M., L. B., Y. L., and G. C. G. investigation; L. P., T. Q. N., L. K., G. L-M., and Y. L. conceptualization; C. H., T. Q. N., L. B., Y. L., and G. C. G. validation; C. H., T. Q. N., and G. C. G. formal analysis; T. Q. N., L. K., and L. B. writing – review & editing; L. K. supervision; L. K. project administration; L. K., G. L-M., and L. B. funding acquisition.
Funding and additional information
This work was supported by the Fondation pour la Recherche Médicale (FRM) [grant number DEQ20150331719] to L. K and the National Research Agency grant ANR-24-CE18-7176 (TriSLa-4-NTM) to L. K. GLM received funding from the Medical Research Council MR/W018756/1. L. B received funding from the Swiss National Science Foundation (grant no. 215557).
Reviewed by members of the JBC Editorial Board. Edited by Chris Whitfield
Supporting information
References
Associated Data
Supplementary Materials
Data Availability Statement
All data are contained within the manuscript and the. The raw data can be share upon request. All LC-MS metabolomics data has been uploaded into MetaboLights data repository with the accession code MTBLS13995 ([69]).