To the Editor:

Nucleophosmin (NPM1) is the most recurrently mutated gene in de novo acute myeloid leukaemia (AML), producing mutant‐NPM1 that aberrantly accumulates in cytoplasm instead of nuclei. Why this should transform myeloid precursors was unknown. We discovered, using unbiased proteomic analyses, that NPM1 is a cofactor for the master transcription factor driver of granulo‐monocytic lineage‐fates PU.1, and that, crucially, mutant‐NPM1 dislocates PU.1 into cytoplasm with it (Gu et al., 2018). We showed that disruption of the granulo‐monocytic master transcription factor hub in this way decouples exponential proliferation of myeloid progenitors from forward‐differentiation to produce exponential self‐replication (a transforming action) (Gu et al., 2018). In a letter to the British Journal of Haematology, Pianigiani et al. (2020) contradicted our report by describing PU.1 location in nuclei, not cytoplasm, of NPM1‐mutated AML cells. Here, we show why Pianigiani et al. incorrectly found the bulk of NPM1 and PU.1 in nuclei instead of cytoplasm of NPM1‐mutated AML cells, and confirm again dislocation of both into cytoplasm.

Detection of intra‐cellular proteins by immune‐histochemistry or immune‐fluorescence requires cell fixation/permeabilization. Paraformaldehyde, a commonly used fixative, is known to affect apparent subcellular locations of proteins and to confound such determination (Frisch, 2004; Howat & Wilson, 2014). We thus evaluated different fixatives in a model of myelopoiesis in which PU.1 is known to be located in cytoplasm: PU.1‐knockout murine myeloid precursors transduced with PU.1 fused to the estrogen receptor (PU.1‐ER cells). These cells have PU.1 in cytoplasm instead of nuclei (DeKoter et al., 1998; Gu et al., 2018). Use of cold methanol as the fixative for immunofluorescence microscopy enabled correct localisation of PU.1 in cytoplasm, not nuclei, of PU.1‐ER cells (Fig 1A). Use of paraformaldehyde as the fixative, however, caused incorrect apparent localisation of PU.1 to their nuclei. If PU.1 is actually in the nuclei of these myeloid precursors (e.g., via application of an estrogen agonist), they rapidly terminally differentiate into monocyte/macrophages, then die (DeKoter et al., 1998; Gu et al., 2018) (Fig 1A). We then evaluated the impact of different fixatives on immunofluorescence microscopy studies of NPM1‐mutated (OCI‐AML3, IMS‐M2) and NPM1‐wildtype (THP1, MOLM13) AML cells. With paraformaldehyde as the fixative, NPM1 and PU.1 were apparently located in nuclei of NPM1‐mutated as well as wildtype AML cells (Fig 1B). However, with methanol (Gu et al., 2018) or ethanol fixation, NPM1 and PU.1 were predominantly in cytoplasm of NPM1‐mutated cells and nuclei of NPM1‐wildtype AML cells (Fig 1B). To show the biological significance of NPM1 and PU.1 location in cytoplasm of NPM1‐mutated AML cells, we treated these cells with the nuclear export inhibitor selinexor. Using immunofluorescence microscopy of methanol‐fixed NPM1‐mutated AML cells, selinexor treatment was seen to rapidly lock both NPM1 and PU.1 in nuclei (Fig 1C), the consequence of which was AML cytoreduction by terminal monocytic differentiation, both in vitro and in vivo (Gu et al., 2018).

Thus, methanol or ethanol but not paraformaldehyde fixation permits biologically accurate subcellular localisation of PU.1 and NPM1 in cells (including primary AML cells) (Gu et al., 2018). Accordingly, we used cold methanol as the method of cellular fixation for our previously published study and cautioned that paraformaldehyde fixation did not permit evaluation of PU.1 location by immunofluorescence (Gu et al., 2018). Pianigiani et al. (2020) used paraformaldehyde fixation for their analyses, and therefore incorrectly found predominant nuclear location in NPM1‐mutated AML cells of not just PU.1, but even NPM1 (see Fig 1A,B of the Pianigiani et al., 2020 report). Pianigiani et al. also performed a Western blot for mutant‐NPM1 in nuclear and cytoplasmic fractions of NPM1‐mutated (IMS‐M2 and OCI‐AML3) cells. They found weak cytoplasmic signals for mutant‐NPM1 and actin loading control. This weak mutant‐NPM1 cytoplasmic Western blot signal, together with a faint nuclear mutant‐NPM1 signal they also observed, is inconsistent with previously reported results, not just from us (Gu et al., 2018) but others also (Balusu et al., 2011). This is likely because Pianigiani et al. used a single extraction of cytoplasmic proteins using 15 mmol/l salt solution and 1% non‐ionic detergent. Such an approach limits detection of cytoplasmic proteins, especially if the proteins are still associated with membranes. This technical limitation explains why they did not detect PU.1, which is present in less abundance than NPM1. This technical issue can be addressed by a second and third wash using higher salt concentration buffer (e.g., 150 mmol/l), then combining these washes with the original cytoplasmic extraction to increase cytoplasmic protein yield while simultaneously decreasing contamination of the nuclear fraction by cytoplasmic components (Gu et al., 2018). This could have accounted for the faint nuclear‐fraction mutant‐NPM1 signal observed by Pianigiani et al.

NPM1 dislocation into cytoplasm of NPM1‐mutated AML cells was first observed by Falini et al. (2005) via immunohistochemical analysis of paraformaldehyde‐fixed AML cells. Thus, the cellular abundance of NPM1, and the extent of its actual dislocation into cytoplasm, permitted discovery of some NPM1 in cytoplasm despite paraformaldehyde fixation. That even this original report may have under‐estimated the extent of NPM1 dislocation into cytoplasm, and that, in fact, most NPM1 is dislocated into cytoplasm, taking PU.1 with it, is evident by careful partition of NPM1‐mutated AML cells into nuclear and cytoplasmic fractions, then rigorous proteomic analyses by immunoprecipitation, liquid‐chromatography/mass spectrometry and Western blot (Gu et al., 2018). Immno‐fluorescence microscopy, whether standard (Fig 1B) or confocal (Fig 1C), when using methanol or ethanol for fixation instead of paraformaldehyde, also clearly shows this. Moreover, use of these improved methods enables observation of rapid retention of both NPM1 and PU.1 in nuclei by nuclear‐export inhibition, which triggers terminal monocytic differentiation of chemorefractory NPM1‐mutated cells (including primary cells) in vitro or in vivo (Gu et al., 2018).

In conclusion, identifiable, correctable technical impediments underlie the results from Pianigiani et al. (2020), and we confirm again our extensive body of work showing that mutant‐NPM1 dislocates the myeloid lineage master transcription factor PU.1 into cytoplasm: a transforming action that can importantly be reversed using small molecule drugs (Gu et al., 2018).

致编辑：

核蛋白（NPM1）是从头急性髓细胞性白血病（AML）中最经常发生突变的基因，产生的突变型NPM1异常积累在细胞质而不是细胞核中。为什么这应该转化髓样前体尚不清楚。我们使用无偏蛋白质组分析发现，NPM1是粒单核细胞谱系命运PU.1的主转录因子驱动因子的辅因子，并且至关重要的是，突变NPM1将PU.1与其定位在细胞质中（Gu等。，2018）。我们表明，以这种方式破坏颗粒单核细胞主转录因子集线器可以使髓系祖细胞的指数增殖与正向分化解耦，从而产生指数自我复制（一种转化作用）（Gu等。，2018）。在给英国血液学杂志的一封信中，Pianigiani等人。（2020）通过描述PU.1在NPM1突变的AML细胞核而不是细胞质中的位置与我们的报告相矛盾。在这里，我们说明为什么Pianigiani等。错误地在细胞核中发现了NPM1和PU.1的大部分，而不是在NPM1突变的AML细胞的细胞质中，并再次确认两者均脱位到细胞质中。

通过免疫组织化学或免疫荧光检测细胞内蛋白质需要细胞固定/通透性。众所周知，低聚甲醛是一种常用的固定剂，会影响蛋白质的明显亚细胞位置并混淆这种测定（Frisch，2004； Howat＆Wilson，2014）。因此，我们在骨髓生成模型中评估了不同的固定剂，其中PU.1已知位于细胞质中：PU.1敲除的鼠髓样前体由PU.1转导至雌激素受体（PU.1-ER细胞）。这些细胞在细胞质而不是细胞核中具有PU.1（DeKoter等人，1998 ; Gu等人，2018）。使用冷甲醇作为免疫荧光显微镜的固定剂可以使PU.1在PU.1-ER细胞的细胞质而非细胞核中正确定位（图1A）。但是，使用多聚甲醛作为固定剂会导致PU.1的核表面出现错误的表观定位。如果PU.1实际上存在于这些髓样前体的核中（例如，通过应用雌激素激动剂），它们会迅速地最终分化为单核细胞/巨噬细胞，然后死亡（DeKoter等，1998 ; Gu等，2018） （图1A）。然后，我们评估了不同固定剂对NPM1突变（OCI-AML3，IMS-M2）和NPM1的免疫荧光显微镜研究的影响-野生型（THP1，MOLM13）AML细胞。以低聚甲醛为固定剂，NPM1和PU.1显然位于NPM1突变的核以及野生型AML细胞的核中（图1B）。然而，通过甲醇（Gu等人，2018）或乙醇固定，NPM1和PU.1主要存在于NPM1突变细胞的细胞质和NPM1野生型AML细胞的细胞核中（图1B）。为了显示NPM1的细胞质中的生物学意义和PU.1位置的NPM1 -mutated AML细胞，我们来处理这些细胞与核出口抑制剂selinexor。使用甲醇固定的NPM1的免疫荧光显微镜检查突变的AML细胞，已被SEX的治疗迅速将NPM1和PU.1都锁定在细胞核中（图1C），其结果是在体外和体内通过终末单核细胞分化而导致AML细胞减少（Gu et al。，2018）。

因此，甲醇或乙醇而不是多聚甲醛的固定可以使PU.1和NPM1在细胞（包括原代AML细胞）中进行生物学上准确的亚细胞定位（Gu等，2018）。因此，我们在先前发表的研究中使用冷甲醇作为细胞固定的方法，并告诫多聚甲醛固定不允许通过免疫荧光法评估PU.1的位置（Gu等，2018）。Pianigiani等。（2020年）使用多聚甲醛固定进行分析，因此错误地在NPM1突变的AML细胞中不仅在PU.1，甚至在NPM1中都发现了主要的核位置（参见Pianigiani的图1A，B）等。（2020年报告）。Pianigiani等。还对NPM1突变（IMS-M2和OCI-AML3）细胞的核和细胞质级分中的突变NPM1进行了蛋白质印迹分析。他们发现了弱的细胞质信号用于突变NPM1和肌动蛋白加载控制。他们还观察到了这种弱的突变体-NPM1细胞质Western印迹信号，以及他们也观察到的微弱的核突变体-NPM1信号，这与以前报道的结果不一致，不仅来自我们（Gu et al。，2018），而且也来自其他报道（Balusu  et al。，2011年）。这可能是因为Pianigiani等人。使用15 mmol / l盐溶液和1％非离子型去污剂对细胞质蛋白进行一次提取。这种方法限制了细胞质蛋白的检测，特别是如果蛋白仍与膜结合时。此技术局限性说明了为什么他们没有检测到PU.1的存在，而PU.1的存在量少于NPM1。此技术问题可以通过使用较高盐浓度的缓冲液（例如150 mmol / l）进行第二次和第三次洗涤来解决，然后将这些洗涤液与原始细胞质提取液结合使用，以增加细胞质蛋白产量，同时减少细胞质对核级分的污染组件（Gu et al。，2018）。这可能解释了皮亚尼贾尼（Pianigiani）观察到的微弱的核裂变突变体NPM1信号等。

Falini等人首先观察到NPM1脱位到NPM1突变的AML细胞的细胞质中。（2005）通过对多聚甲醛固定的AML细胞进行免疫组织化学分析。因此，尽管存在多聚甲醛固定，但NPM1的细胞丰度及其在细胞质中实际脱位的程度允许在细胞质中发现一些NPM1。通过仔细分配NPM1可以明显看出，即使是此原始报告也可能低估了NPM1脱位进入细胞质的程度，事实上，大多数NPM1随PU.1脱位进入细胞质。将AML细胞突变为核和细胞质部分，然后通过免疫沉淀，液相色谱/质谱和Western印迹进行严格的蛋白质组学分析（Gu et al。，2018）。当使用甲醇或乙醇代替多聚甲醛进行固定时，无论是标准（图1B）还是共聚焦（图1C）的免疫荧光显微镜也能清楚地表明这一点。此外，使用这些改进的方法可以通过核出口抑制观察NPM1和PU.1在核中的快速保留，从而触发化学难治性NPM1突变的细胞（包括原代细胞）在体外或体内的终末单核细胞分化（Gu等，2018）。

总之，Pianigiani等人的结果是可识别的，可纠正的技术障碍。（2020），我们再次证实我们广泛的研究工作表明，突变NPM1将髓系谱系主转录因子PU.1转移到细胞质中：使用小分子药物可以重要地逆转这种转化作用（Gu等人，2018年））。