To the Editor:

Although we recognise that paraformaldehyde may influence the detection of NPM1 and PU.1 in immunofluorescence, the main goal of our study (Pianigiani et al., 2020) was to test whether PU.1 localisation could be used to diagnose NPM1‐mutated AML through immunohistochemistry in B5‐fixed bone marrow biopsies. B5 fixation is routinely used in our laboratory to detect cytoplasmic localisation of NPM1 in acute myeloid leukaemia, based on Falini et al. (2005). It is unlikely that B5 fixation allows for correct detection of NPM1 but not that of PU.1.

The data provided by Gu et al. in this Letter and in previous work (2018) support the hypothesis that NPM1 directly binds PU.1, leading to PU.1 relocation from the nucleus to the cytoplasm. Given the heterozygous nature of NPM1 mutations, a significant proportion of PU.1 would be expected to still be localised in the nuclei of NPM1‐mutated cells (even accounting for the small amount of wild‐type NPM1 dragged to the cytoplasm by the mutant protein). However, no PU.1 is detected in the nuclei of NPM1‐mutated untreated cells in the vast majority of immunofluorescence and western blot experiments reported by Gu et al. , arguing for an overestimation of the actual amount of PU.1 in the cytoplasm. Also, all the experiments by Gu et al. have been performed using a polyclonal anti‐PU.1 antibody whose production has been discontinued years ago, making it difficult to reproduce the data.

Concerning the nuclear/cytoplasmic fractionation, it is possible that in our experiments a proportion of PU.1 was not separated from the nuclear fraction. Considering the correct localisation of control proteins in our blot, and assuming, as claimed by Gu et al. (2018), that almost all PU.1 should be in the cytoplasm of NPM1‐mutated cells, one would expect to find a visible proportion of PU.1 in the cytoplasm, even after an incomplete separation. However, we did not detect PU.1 at all in any of the cytoplasmic fractions, even after applying longer exposure times.

We would like to emphasize that we do not rule out that a proportion of PU.1 may be found in the cytoplasm of AML cells. However, our data indicate that PU.1 localisation studied by immunohistochemistry should not be used to diagnose NPM1‐mutated AML. More experiments are necessary to establish the exact proportion of PU.1 localised to the cytoplasm of leukemic cells, and to define the contribution of cytoplasmic PU.1 to the development and maintenance of NPM1‐mutated AML.

致编辑：

尽管我们认识到多聚甲醛可能会影响免疫荧光中NPM1和PU.1的检测，但我们研究的主要目标（Pianigiani等人，2020年）是测试PU.1的定位是否可用于通过以下方法诊断NPM1突变型AML： B5固定的骨髓活检中的免疫组织化学。根据Falini等人的研究，在我们的实验室中常规使用B5固定来检测NPM1在急性髓样白血病中的胞质定位。（2005）。B5固定不可能正确检测NPM1，而不能正确检测PU.1。

顾等人提供的数据。这封信和以前的工作（2018）中的文章支持NPM1直接结合PU.1从而导致PU.1从细胞核重新定位到细胞质的假设。鉴于杂合性质NPM1突变，PU.1的显著比例预计将在核仍然是局部的NPM1 -mutated细胞（即使在考虑了野生型的少量NPM1由突变蛋白拖动到细胞质）。但是，在Gu等人报告的绝大多数免疫荧光和Western印迹实验中，在NPM1突变的未处理细胞核中未检测到PU.1 。，认为细胞质中PU.1的实际含量被高估了。同样，Gu等人的所有实验。已使用多克隆抗PU.1抗体进行了克隆，该抗体的生产已于几年前停止，因此难以复制数据。

关于核/细胞质分级分离，有可能在我们的实验中没有从核级分中分离出一部分PU.1。考虑到我们的印迹中对照蛋白的正确定位，并假设，如Gu等人所述。（2018），几乎所有PU.1都应在NPM1突变细胞的细胞质中，即使分离不完全后，人们仍希望在细胞质中发现可见比例的PU.1。但是，即使应用了更长的暴露时间，我们也没有在任何细胞质部分中检测到PU.1。

我们要强调的是，我们不排除在AML细胞的细胞质中可以发现一定比例的PU.1。但是，我们的数据表明，不应使用免疫组织化学研究的PU.1定位来诊断NPM1突变的AML。必须进行更多的实验才能确定定位在白血病细胞质中的PU.1的确切比例，并确定细胞质PU.1对NPM1突变AML的发生和维持的作用。