作者承认文章有两方面错误；同时认为其成果仍然有效

一篇发表于2004年《科学》杂志、有关一种新的脂肪因子Visfatin文章的作者在10月25日《科学》撤回了这一研究发现，但是他们仍然认为其原发现的成果有效。

文章作者为来自日本大阪大学前沿生命科学研究院、医学研究院的研究人员，第一作者为Atsunori Fukuhara，其与其他同事在撤回信中写道：“虽然我们仍然坚持我们的研究结论，但是根据《科学》编辑的意见，我们同意撤回这篇文章。”

这篇文章曾被引用200多次，研究人员在文章中提出Visfatin是一种在脂肪组织中新发现的蛋白，具有拟胰岛素（insulin-mimetic）特 征，这篇文章自2004年12月发表于《科学》杂志之后，其它的研究也相继发现 visfatin与一种已知的生长因子：前B细胞克隆刺激因子（pre-B cell colony-enhancing factor，PBEF，这是美国Amgen公司的Samal博士在1994年首次报告的一个从人外周血液淋巴细胞基因库克隆的细胞因子）遗传上相似，但 也对这种蛋白是否具有拟胰岛素特征提出了疑问。

大阪大学医学研究院的主任Masaya Tohyama在研究院对这一研究成果进行了为期一年的调查之后，去年6月向研究小组提出撤回这篇文章的要求，《科学》杂志表示，研究院并没有指出这属于 科学不端行为，但是对于这项研究提出了“许多疑问”，比如数据表中提出的雌性异质结合（heterozygous）敲除小鼠的排异反应。但是研究人员那时 拒绝撤回这篇文章，通讯作者 Iichiro Shimomura表示研究小组将“采取切实行动”对抗大学。

Shimomura说，“在收到大阪大学医学院的报告之后，《科学》杂志作出了评判，向我们提出在征得作者同意前提下撤回文章”，“我们讨论了许多次，最 后决定撤回这篇文章”。同时Shimomura也提出了两方面错误：1251标记处带有低特异放射性的visfatin的使用，以及41图上有关胰岛素与 visfation在HEK-293细胞上的绑定的一个描述性的错误。他表示，“我们承认我们利用放射标记蛋白的更高特异活性增加了数据的完整性”，而这 个描述性的错误“减少了文章内容的完整性”。

来自哈佛Joslin糖尿病中心（Joslin Diabetes Center）的Alessandro Doria认为，“Visfatin/PBEF被认为是一种炎症因子”。基于其自身在visfatin基因方面的研究，Doria推测visfatin的 增多与炎症增加有关，进而与II型糖尿病产生关联，“这好像与visfatin具有拟胰岛素行为这一看法相左。”

来自华盛顿大学医学院的Shin-ichiro Imai提出，“无论理由是什么，他们最终走出了正确的一步”，Imai有关visfatin/PBEF的研究报告反驳了Fukuhara研究小组的发 现，“在学术上和工厂中的研究不能复制他们的结果”。但是在Fukuhara他们的撤回信中提到，“最近有其它实验室报道visfatin在成骨培养细胞 具有拟胰岛素效应。”

在Fukuhara研究小组提交给Science的撤回信中，他们写道，“至今我们发现了四个不同的纯化重组visfatin蛋白，它们具有成脂（adipogenic）和拟胰岛素特征……我们愿意将这些蛋白寄送给其他研究人员，我们也将继续进行这个分子的研究。”

visfatin是新近发现的主要由人和小鼠内脏脂肪组织分泌的一 种脂肪细胞因子，其结构与pre-B细胞集落增强因子相似。它能够发挥类似胰岛素的作用，与Ⅱ型糖尿病相关联,降低血糖，促进糖摄取，可结合并活化胰岛素 受体，激活胰岛素信号通路。Visfatin与肥胖密切相关并能够促进脂肪细胞的分化，还能促进血管平滑肌细胞成熟。Visfatin的表达受炎症反应因 子和多种激素的调节。Visfatin可能是联系机体糖脂代谢的重要分子，它的发现可为揭示糖尿病与肥胖的发生发展机制提供新的研究思路，为代谢综合征的 治疗提供新方案。

http://www.sciencemag.org/cgi/content/full/318/5850/565b

Retraction of Fukuhara et al., Science 307 (5708) 426-430.

Science 26 October 2007: Vol. 318. no. 5850, p. 565 DOI: 10.1126/science.318.5850.565b

Letters

Retraction

In a paper titled "Visfatin: a protein secreted by visceral fat that mimics the effects of insulin" (1), we identified a protein that is secreted by visceral fat of humans and mice and named it "visfatin." The same protein had been identified previously by other laboratories as "pre-B cell--colony enhancing factor," a cytokine that is expressed by lymphocytes (2) and that displays nicotinamide phosphoribosyltransferase activity (3). In the Science Report, we investigated visfatin-s metabolic effects and the biochemical mechanism by which it might exert these effects. We showed that visfatin induces adipocyte differentiation in vitro and that plasma levels of visfatin correlate with visceral fat mass in humans. We also showed that male mice with only one functional copy of the visfatin gene have modest elevations in plasma glucose and that adenovirus-mediated delivery of the visfatin gene to c57BL/6J or KKAy mice resulted in a lowering of plasma glucose and insulin levels. Finally, we reported that visfatin binds to and activates the insulin receptor and we speculated that its activity as an insulin mimetic might explain its metabolic effects.

The visfatin work performed in our laboratory was recently investigated by the Committee for Research Integrity (CRI) of Osaka University Graduate School of Medicine. On the basis of the CRI report, which focused largely on our biochemical experiments examining visfatin-s interaction with the insulin receptor, the Faculty Council of Osaka University Medical School recommended that we retract the entire paper. At the suggestion of the Editor of Science, we have agreed to retract the paper, even though we continue to stand by our conclusions. We note that over a dozen subsequent publications have shown that plasma visfatin levels in humans correlate with various metabolic states, including obesity, visceral fat mass, and diabetes [for example, (4--6)]. We note also that another laboratory recently reported that visfatin has insulin mimetic effects in cultured osteoblasts (7). We acknowledge that, since publication of the Science Report, we have found that not all preparations of visfatin bind to and activate the insulin receptor. Thus far, we have found four different lots of purified recombinant visfatin protein that have both adipogenic and insulin mimetic activities. We still have the preparations of visfatin that show insulin mimetic activity, although the amount is limited, and we are willing to send them to other investigators for independent validation. We are continuing to investigate the significance of this molecule.

We regret any inconvenience caused by this retraction to researchers and readers. The corresponding author is responsible for the retraction.