《Science》文章被撤回源自学术不端？

　　

生物通报道：来自《The Scientist》的消息，杜克大学医学中心已经同意针对有关杜克大学蛋白生化学家Homme Hellinga学术不端行为的指控展开调查，上周（7月24日）《自然》公开了Hellinga这一相关内容的来信。

Hellinga在计算机设计蛋白，蛋白质工程领域非常知名，他今年年初撤回了的两篇论文（《科学》和《分子生物学》）主要内容是重新设计了催化磷酸丙糖异构酶（triose phosphate isomerase ，TIM）活性的核糖体结合蛋白（ribose-binding protein，RBP），这个反应对于几乎所有类型细胞的糖酵解（glycolysis）都是一个关键的过程。

这个研究结果当时被认为是蛋白工程方面的一项突破性成果，因为之前设计并改造一个没有催化活性的蛋白，使之具有高度的酶活性从来没有实现过，这项成果在《Science》上发表了（到2008年止，这篇文章被引用了超过160次）。然而论文发表之后，纽约州立大学的生化学家John Richard尝试重复这一实验，结果发现在实验中纯化蛋白的方法出错了，蛋白的这个酶活性是由于野生型的蛋白污染导致的，因此整个文章的所有结果都不复存在了。因此Hellinga撤回了相关论文，并解释认为是最初的实验出现了问题，即“Innocent mistake”。然而Richard以及加州大学伯克利分校的John Kirsch并不满意其解释，认为对于Hellinga这样规模的实验室来说，这样错误的发生是”innocent”可能性几乎为零。

在上周Helinga给《Nature》的信中，他认为Nature对于撤回论文的报道以及之后的事件发生导致了一些误解，“由于我的行为引起了争论，所以我请求杜克大学医学中心针对这些论文撤销，以及之后的事件进行正式的，公平的调查。”

目前杜克大学医学院拒绝承认或否定这一调查是否在进行，但发表声名表示，“杜克大学致力于培养高素质的科学人才，我们审阅了所有针对相关研究的指控，我们相信大家能理解由于机密性和其它限制，杜克大学目前公开作出评论是不合适的。”

（生物通：张迪）

附：John Richard和John Kirsch的评论

Jack写到：“In my view, the Dwyer et al. Retraction (1 February 2008, p. 569) of their 2004 Report (1) does not explain the published results. They attribute the triosephosphate isomerase (TIM) activity reported in the earlier work solely to contamination by wild-type (TIM). The contamination is certainly real as originally documented by John Richard, but this fact does not explain the reported Km values, which are more than an order of magnitude reduced from that of the wild-type enzyme. While Vmax values can be in error due to poor estimates of enzyme concentration or the presence of inactive enzyme, and Km values may be artificially higher due to the presence of competitive inhibitors, it is hard to explain the much lower values reported for the designed enzyme. The Retraction does not address this issue, which is fundamental. Second, contamination by wild-type enzyme would be expected to compromise randomly (dependent on expression levels) the kinetic parameters of all designed TIM variants, yet Fig. 4C would make sense only if the design were successful. The active TIM design included three active site residues. The reported data show systematic reductions in activity from the complete designed TIM as each of the design entities is replaced by alanine, and the triple alanine replacement is the most compromised of all. Simple contamination with wild-type TIM cannot explain such a result. Additionally the statement, the in vivo experiments have not been reexamined.?does not address what to many is the most convincing part of the paper i.e., the in vivo rescue of a TIM knockout in Escherichia coli by the designed NovoTIM. There is no way that wild-type TIM contamination in the in vitro experiments could account for that in vivo result”


John写到:"I read with interest the Retraction of the Report "Computational design of a biologically active enzyme" (1) that was published on page 569 of your 1 February issue. In this Retraction, the authors mention a "reanalysis" carried out in my laboratory at the University at Buffalo. Reanalysis is an odd way to characterize our routine separation of host wild-type triosephosphate isomerase from an inactive engineered overexpressed protein. This purification was facilitated by the determination that the Michaelis constant Km = 2.3 mM for isomerization of dihydroxyacetone phosphate (DHAP) catalyzed by the active enzyme in our preparation of the designed protein is that expected for wild-type Escherichia coli triosephosphate isomerase. Dr. Hellinga's paper in Science reported a range of values of Km = 0.10 to 0.33 mM for isomerization of DHAP catalyzed by various iterations of his designed proteins (Table 2 of the Report in Science), which are considerably smaller than Km ≈ 2 mM expected for wild-type triosephosphate isomerase from E. coli. In a later report in Journal of Molecular Biology (2) the value of Km = 0.18 mM for the ecNovoTIM1.2 construct was amended, without adequate explanation, to a 40-fold larger value of 7.1 mM (Table 1 of that paper). The problems experienced by members of Dr. Hellinga's laboratory in determining accurate values of Km are closely tied to their failure to identify the isolated enzymatic activity as wild-type E. coli triosephosphate isomerase and to the unfortunate acceptance of the original papers for publication in Science and the Journal of Molecular Biology. In order to minimize the speculation that will arise when these problems become public knowledge, I think that they should have been addressed in the recent Retraction.”
 

原文检索：

Computational design of a biologically active enzyme