一句话亮点

这篇《Drug Resistance Updates》的研究通过体内CRISPR/Cas9筛选锁定SUMO化修饰通路，并揭示SUMO1通过TRIM28介导的PKM2 SUMO化增强糖酵解，从而驱动肝细胞癌索拉非尼耐药——为克服耐药提供了"代谢检查点+靶向SUMO化"的双重干预策略。

背景/痛点

索拉非尼一直是晚期肝癌系统治疗的基石药物，但几乎不可避免地出现获得性耐药。过去人们已经知道，耐药背后有旁路信号激活、自噬、上皮间质转化等多重机制，而代谢重编程——尤其是瓦伯格效应（有氧糖酵解增强）——被认为是肿瘤细胞在治疗压力下"自救"的核心手段。

但问题在于：到底是什么上游事件"踩下了糖酵解的油门"？过往研究多集中在转录水平或磷酸化等常见翻译后修饰，而SUMO化这种"类泛素化"修饰在肝癌耐药中是否扮演角色，仍不清晰。

作者正是从这个缺口切入——他们推测：耐药状态下，肿瘤细胞可能通过某种翻译后修饰快速"劫持"代谢通路，而不需要等待转录重编程。这个假设的有趣之处在于，它解释了耐药为何可以在药物压力下迅速建立。

推理链分步拆解

第一步：CRISPR/Cas9筛选——让"SUMO化"浮出水面

作者没有先入为主地盯住某个候选基因，而是做了个非常"干净"的设计：他们在原位肝癌小鼠模型中转导全基因组sgRNA文库，分别在肿瘤"敏感期（缩小）“和"耐药期（反弹）“两个时间点取样测序。

关键逻辑是：如果在耐药期某些sgRNA显著减少，说明这些基因被敲除后肿瘤活不下去——它们是耐药所必需的。结果令人意外：富集最显著的居然是SUMO化相关基因，其中SUMO1的临床相关性最强（高表达预示索拉非尼治疗患者生存更差）。

接着他们回到细胞模型，用低剂量索拉非尼"驯化"14天，果然看到SUMO1上调、整体SUMO化水平升高。

@方法论点评：这就是"体内正向遗传筛选+体外因果验证"的双保险设计。先用无偏筛选找线索，再用靶向操作验证，排除了"只是伴随现象"的可能。体内筛选用sgRNA缺失作为读出的设计很巧妙，相当于把"必需基因"这个概念直接放在药物压力下检验。

Fig. 1：CRISPR/Cas9 screening identifies protein SUMOylation in sorafenib resistance. (a) Workflow of in vivo CRISPR/Cas9 screen in orthotopic HCC model. (b) Tumor growth curves: sensitive (regression) vs. resistant (rebound) phases under sorafenib. (c) Top enriched sgRNAs in resistant tumors. (d) Plot of fold changes with p-value of selected SUMOylation-related genes in CRISPR/Cas9 screen. (e) Kaplan-Meier overall survival in sorafenib-treated HCC patients stratified by the expression of top 4 genes. (f) mRNA and (g) protein expression of SUMO1 in sorafenib-sensitive and -resistant HCC tumors. (h) cellular protein SUMOylation in sorafenib-sensitive and resistant tumors. (i) mRNA and (j) protein expression of SUMO1 in sorafenib-sensitive and -resistant HCC tumors. (k) cellular protein SUMOylation in sorafenib-sensitive and resistant tumors. in HCC cells after 14-day 5 μM sorafenib challenge.

Fig. 1. CRISPR/Cas9 screening identifies protein SUMOylation in sorafenib resistance. (a) Workflow of in vivo CRISPR/Cas9 screen in orthotopic HCC model. (b) Tumor growth curves: sensitive (regression) vs. resistant (rebound) phases under sorafenib. (c) Top enriched sgRNAs in resistant tumors. (d) Plot of fold changes with p-value of selected SUMOylation-related genes in CRISPR/Cas9 screen. (e) Kaplan-Meier overall survival in sorafenib-treated HCC patients stratified by the expression of top 4 genes. (f) mRNA and (g) protein expression of SUMO1 in sorafenib-sensitive and -resistant HCC tumors. (h) cellular protein SUMOylation in sorafenib-sensitive and resistant tumors. (i) mRNA and (j) protein expression of SUMO1 in sorafenib-sensitive and -resistant HCC tumors. (k) cellular protein SUMOylation in sorafenib-sensitive and resistant tumors. in HCC cells after 14-day 5 μM sorafenib challenge.（图注取自PDF文本层，来源：Drug Resistance Updates, 2026）

第二步：SUMO1——是"看客"还是"操盘手”？

发现相关性之后，作者问了一个更尖锐的问题：改变SUMO1的水平，耐药表型会不会跟着变？

他们分别在两种肝癌细胞中敲低或过表达SUMO1，然后用索拉非尼处理。结果很干净：敲低SUMO1 → 细胞对索拉非尼更敏感（IC₅₀下降、凋亡增多）；过表达SUMO1 → 耐药增强。原位肝癌模型也印证了这个趋势。

这说明SUMO1不是旁观者，而是耐药过程的功能性驱动因子。但机制呢？SUMO化修饰成千上万的蛋白，它具体改变了什么"下游事件”？

@方法论点评：这一步用了"双向操作"（gain- and loss-of-function）——敲低和过表达都做，结论的说服力远超只做单向。体内实验又加了一层正交验证，排除了细胞模型的人为干扰。

Fig. 2：SUMO1 regulates sorafenib sensitivity in vitro and in vivo. (a) Immunoblot of SUMO1 in knockdown (shSUMO1) or overexpressing (OE-SUMO1) cells. (b) cellular protein SUMOylation in shSUMO1 and OE-SUMO1 cells in the presence or absence of sorafenib. The (c) cytotoxicity analysis, (d) colony formation, (e) apoptosis analysis in shSUMO1 and OE-SUMO1 cells in the presence or absence of sorafenib. (f) In vivo bioluminescence of orthotopic tumors. (g) the size and (h) weight of orthotopic tumors with SUMO1 knockdown or overexpression from mice with sorafenib treatment.

Fig. 2. SUMO1 regulates sorafenib sensitivity in vitro and in vivo. (a) Immunoblot of SUMO1 in knockdown (shSUMO1) or overexpressing (OE-SUMO1) cells. (b) cellular protein SUMOylation in shSUMO1 and OE-SUMO1 cells in the presence or absence of sorafenib. The (c) cytotoxicity analysis, (d) colony formation, (e) apoptosis analysis in shSUMO1 and OE-SUMO1 cells in the presence or absence of sorafenib. (f) In vivo bioluminescence of orthotopic tumors. (g) the size and (h) weight of orthotopic tumors with SUMO1 knockdown or overexpression from mice with sorafenib treatment.（图注取自PDF文本层，来源：Drug Resistance Updates, 2026）

第三步：蛋白组学+代谢流——锁定"糖酵解"这个下游事件

为了看清SUMO1过表达后到底改变了什么，作者做了无偏蛋白组学。GO分析和KEGG富集都指向同一个方向：糖酵解相关蛋白显著上调。而且不是单个酶的变化，是整个通路的系统性增强。

接下来用Seahorse测代谢流：SUMO1过表达细胞ECAR（细胞外酸化率，代表糖酵解）飙升，OCR（耗氧率，代表线粒体呼吸）相对下降——典型的瓦伯格效应。用2-DG（糖酵解抑制剂）回补实验进一步确认：抑制糖酵解后，SUMO1带来的耐药优势被抹平。

这个推理链条非常清晰：SUMO1 → 糖酵解增强 → 耐药。但还缺最后一环：SUMO化到底修饰了哪个关键蛋白来"点燃"糖酵解？

@方法论点评：注意作者用的是"蛋白组学+功能代谢验证"的组合——蛋白组学看"谁变了"，Seahorse看"功能输出变了没"，2-DG做"因果回补"。三步走把代谢重编程从"相关性"推到"必要性"。

Fig. 3：SUMO1 induces glycolysis to promote sorafenib resistance. (a) Proteomic analysis of protein expression in sorafenib-treated vehicle-expressing or SUMO1- overexpressing MHCC97L. GO analysis of upregulated (b) and downregulated (c) proteins. (d) KEGG enrichment of metabolic pathways. (e) the ECAR (f) the basal glycolytic flux, (g) the maximal glycolytic flux, (h) the OCR, (i) the basal oxygen consumption, and (j) the maximal oxygen consumption of vehicle-expressing or SUMO1-overexpressing HCC cells in the presence or absence of sorafenib by Seahorse assay. (k) the glucose uptake, (l) lactate secretion and (m) ATP production of sorafenib-treated SUMO1-overexpressing HCC cells in the presence of glycolysis inhibitor 2-DG. The (n) cytotoxicity analysis, (o) colony formation, (p) apoptosis analysis in sorafenib-treated SUMO1-overexpressing HCC cells in the presence of glycolysis inhibitor 2-DG.

Fig. 3. SUMO1 induces glycolysis to promote sorafenib resistance. (a) Proteomic analysis of protein expression in sorafenib-treated vehicle-expressing or SUMO1- overexpressing MHCC97L. GO analysis of upregulated (b) and downregulated (c) proteins. (d) KEGG enrichment of metabolic pathways. (e) the ECAR (f) the basal glycolytic flux, (g) the maximal glycolytic flux, (h) the OCR, (i) the basal oxygen consumption, and (j) the maximal oxygen consumption of vehicle-expressing or SUMO1-overexpressing HCC cells in the presence or absence of sorafenib by Seahorse assay. (k) the glucose uptake, (l) lactate secretion and (m) ATP production of sorafenib-treated SUMO1-overexpressing HCC cells in the presence of glycolysis inhibitor 2-DG. The (n) cytotoxicity analysis, (o) colony formation, (p) apoptosis analysis in sorafenib-treated SUMO1-overexpressing HCC cells in the presence of glycolysis inhibitor 2-DG.（图注取自PDF文本层，来源：Drug Resistance Updates, 2026）

第四步：拉下SUMO化底物——PKM2浮出水面

作者用SUMO抗体做免疫沉淀，把SUMO化修饰的蛋白"一网打尽"，然后送质谱鉴定。在糖酵解相关的候选蛋白中，PKM2的打分最高——这很合理，因为PKM2是糖酵解最后一步的限速酶。

接下来一系列生化和细胞实验回答了四个关键问题：

谁修饰谁？SUMO1对PKM2的修饰效率高于SUMO2/3。 谁来"拧螺丝"（E3连接酶）？敲除筛选发现TRIM28是主导，PIAS1影响不大。 修饰在哪个位点？质谱锁定K270和K336两个赖氨酸位点，突变这两个位点后SUMO化大幅降低。 修饰后有什么功能后果？SUMO1过表达后PKM2酶活性显著升高。

@方法论点评：E3连接酶的鉴定用了个非常标准的"组合拳"——先过表达候选E3看谁结合，再逐个敲除看谁消失。TRIM28和PIAS1的对比实验尤其关键：如果只做TRIM28敲除不做PIAS1对照，会被质疑"随便敲个E3都会有效果"。

Fig. 4：PKM2 is hyper-sumoylated in sorafenib-resistant HCC. (a) the SDS-PAGE gel electrophoresis of sumolyated proteins in sorafenib-treated MHCC97L cells pull- downed by SUMO1 antibody (b) Proteomic analysis of pull-downed sumoylated proteins. (c) immunoblotting of sumolyated PKM2 pull-downed by His antibody in sorafenib-treated HCC cells expressing either His-SUMO1 or His-SUMO2/3. (d) Immunoblotting of endogenous PKM2 pulldowned by SUMO1 antibody in sorafenib- treated HCC cells. (e) Immunoblotting of E3 ligase TRIM28 and PIAS1 pulldowned by anti-PKM2 antibody in sorafenib-treated SUMO1-overexpressing HCC cells expressing GFP-TRIM28 or HA-PIAS1. (f) Immunoblotting of PKM2 SUMOylation after knockdown of TRIM28 and/or PIAS1 in HCC cells. The (g) Immunoblotting and (h) cytotoxicity analysis in sorafenib-treated HCC cells in the presence of si-PIAS1 and si-TRIM28. (i) PKM2 enzyme activity in vector-expressing or SUMO1- overexpressing HCC cells. (j)MS/MS spectra of PKM2 SUMOylation at lysine 270 and lysine 336. (k) Immunoblotting of PKM2 SUMOylation in PKM2- knockdown HCC cells reconstituted with wild-type, K270R, or K336R PKM2.

Fig. 4. PKM2 is hyper-sumoylated in sorafenib-resistant HCC. (a) the SDS-PAGE gel electrophoresis of sumolyated proteins in sorafenib-treated MHCC97L cells pull- downed by SUMO1 antibody (b) Proteomic analysis of pull-downed sumoylated proteins. (c) immunoblotting of sumolyated PKM2 pull-downed by His antibody in sorafenib-treated HCC cells expressing either His-SUMO1 or His-SUMO2/3. (d) Immunoblotting of endogenous PKM2 pulldowned by SUMO1 antibody in sorafenib- treated HCC cells. (e) Immunoblotting of E3 ligase TRIM28 and PIAS1 pulldowned by anti-PKM2 antibody in sorafenib-treated SUMO1-overexpressing HCC cells expressing GFP-TRIM28 or HA-PIAS1. (f) Immunoblotting of PKM2 SUMOylation after knockdown of TRIM28 and/or PIAS1 in HCC cells. The (g) Immunoblotting and (h) cytotoxicity analysis in sorafenib-treated HCC cells in the presence of si-PIAS1 and si-TRIM28. (i) PKM2 enzyme activity in vector-expressing or SUMO1- overexpressing HCC cells. (j)MS/MS spectra of PKM2 SUMOylation at lysine 270 and lysine 336. (k) Immunoblotting of PKM2 SUMOylation in PKM2- knockdown HCC cells reconstituted with wild-type, K270R, or K336R PKM2.（图注取自PDF文本层，来源：Drug Resistance Updates, 2026）

第五步：回补实验和临床数据——闭环

最后作者做了"反向验证"：用PKM2特异性抑制剂Compound 3k处理SUMO1过表达细胞，结果糖酵解被抑制、索拉非尼敏感性恢复，体内原位瘤模型也证实联合用药显著增强疗效。

临床组织芯片数据显示：SUMO化PKM2在肝癌组织中高表达，与复发正相关，与p-ERK（索拉非尼响应标志物）负相关——提示SUMO化PKM2可能是耐药患者的"逃生通道"。

@方法论点评：“抑制剂回补"是整条因果链的最后一颗钉子——如果SUMO1通过PKM2起作用，那么直接抑制PKM2应该能"模拟"敲低SUMO1的表型。结果确实如此，推理链闭环。

Fig. 5：PKM2 SUMOylated drives glycolysis and sorafenib resistance. (a) the glucose uptake, (b) lactate secretion and (c) ATP production of sorafenib-treated SUMO1-overexpressing HCC cells in the presence of PKM2 inhibitor compound 3k. (d) the ECAR (e) the basal glycolytic flux, (f) the maximal glycolytic flux of sorafenib-treated SUMO1-overexpressing HCC cells in the presence of PKM2 inhibitor compound 3k. The (g) cytotoxicity analysis, (h) colony formation, (i) apoptosis analysis in sorafenib-treated SUMO1-overexpressing HCC cells in the presence of PKM2 inhibitor compound 3k. (j) In vivo bioluminescence of orthotopic tumors. (k) the size and (l) weight of orthotopic tumors with SUMO1 overexpression from mice with sorafenib treatment in the presence or absence of compound 3k.

Fig. 5. PKM2 SUMOylated drives glycolysis and sorafenib resistance. (a) the glucose uptake, (b) lactate secretion and (c) ATP production of sorafenib-treated SUMO1-overexpressing HCC cells in the presence of PKM2 inhibitor compound 3k. (d) the ECAR (e) the basal glycolytic flux, (f) the maximal glycolytic flux of sorafenib-treated SUMO1-overexpressing HCC cells in the presence of PKM2 inhibitor compound 3k. The (g) cytotoxicity analysis, (h) colony formation, (i) apoptosis analysis in sorafenib-treated SUMO1-overexpressing HCC cells in the presence of PKM2 inhibitor compound 3k. (j) In vivo bioluminescence of orthotopic tumors. (k) the size and (l) weight of orthotopic tumors with SUMO1 overexpression from mice with sorafenib treatment in the presence or absence of compound 3k.（图注取自PDF文本层，来源：Drug Resistance Updates, 2026）

Fig. 6：Clinical relevance of PKM2 SUMOylation in HCC. (a) Staining of tissue microarray (TMA) of 90 HCC samples with its adjacent livers. (b) Representative multiplex immunofluorescence of TMA tissue sections. (c) SUMO1 expression in tumors vs. adjacent liver. (d, e) SUMO1 expression in tumors vs. adjacent liver from two patients cohorts GDS4887 and TCGA database. (f) SUMOylated and total PKM2 expression in tumors vs. adjacent tissue of HCC patients. Expression of SUMOylated PKM2 in cirrhotic v.s. non-cirrhotic patients with HCC (g), and in patients with viral v.s. non-viral background (h). Correlation of SUMOylated PKM2 expression with (i) tumor size or (j) serum AFP level. (k) SUMOylated and total PKM2 expression in recurrent vs. non-recurrent HCC. Correlation of SUMOylated PKM2 with TRIM28 (l), LDHA (m), CD31 (n), and phosphor-Erk (o).

Fig. 6. Clinical relevance of PKM2 SUMOylation in HCC. (a) Staining of tissue microarray (TMA) of 90 HCC samples with its adjacent livers. (b) Representative multiplex immunofluorescence of TMA tissue sections. (c) SUMO1 expression in tumors vs. adjacent liver. (d, e) SUMO1 expression in tumors vs. adjacent liver from two patients cohorts GDS4887 and TCGA database. (f) SUMOylated and total PKM2 expression in tumors vs. adjacent tissue of HCC patients. Expression of SUMOylated PKM2 in cirrhotic v.s. non-cirrhotic patients with HCC (g), and in patients with viral v.s. non-viral background (h). Correlation of SUMOylated PKM2 expression with (i) tumor size or (j) serum AFP level. (k) SUMOylated and total PKM2 expression in recurrent vs. non-recurrent HCC. Correlation of SUMOylated PKM2 with TRIM28 (l), LDHA (m), CD31 (n), and phosphor-Erk (o).（图注取自PDF文本层，来源：Drug Resistance Updates, 2026）

核心结论

本文发现了一条清晰的耐药驱动链条：索拉非尼压力 → SUMO1上调 → TRIM28介导PKM2在K270/K336位点SUMO化 → PKM2酶活性增强 → 糖酵解通量升高 → 肿瘤细胞获得生存优势 → 索拉非尼耐药。

这个机制的"精妙"之处在于：SUMO化作为一种快速、可逆的翻译后修饰，让肿瘤细胞无需通过转录重编程就能在药物压力下迅速"切换"代谢模式。而PKM2作为糖酵解"闸门”，被SUMO化直接激活——比上调表达更直接、更快速。

从治疗角度看，该研究提出了两个潜在干预策略：靶向SUMO化通路（如抑制TRIM28）或直接抑制PKM2（Compound 3k），均有可能逆转耐药。

对耐药/DTP/PGCC 的启示

DTP细胞可能依赖SUMO化维持"代谢可塑性"：药物耐受持久细胞（DTP）的核心特征之一是代谢灵活性。本文提示SUMO化可能是DTP细胞快速切换代谢模式的"开关"——值得在DTP模型中检测SUMO1/PKM2轴是否被优先激活。 PGCC的代谢优势或与SUMO化相关：多倍体巨细胞（PGCC）在应激下存活能力强，往往伴随代谢重编程。SUMO化作为应激响应修饰，是否在PGCC形成和维持中参与调控PKM2活性，是值得探索的方向。 联合治疗策略的启示：单用索拉非尼时，SUMO1-PKM2轴成为"代偿逃生路线"；联合SUMO化抑制剂或PKM2抑制剂可能堵死这条路。文中Compound 3k的数据为这类联合策略提供了初步临床前依据。

局限

作者自己也承认几个局限：①用的是免疫缺陷小鼠，无法评估肿瘤免疫微环境在SUMO化介导耐药中的作用；②虽然锁定PKM2为关键靶点，但SUMO化可能还影响其他蛋白，形成"网络效应"；③K270/K336位点突变体的功能回补实验尚未完成，特定SUMO化位点的因果贡献需进一步确认；④临床队列为单中心回顾性分析，需更大样本前瞻性验证。

来源

期刊：Drug Resistance Updates，2026。DOI: 10.1016/j.drup.2026.101451