The analysis of rare cells in flow cytometry data requires the acquisition of large data sets to visualize a sufficient number of events. Accurate identification of rare cells within a large number of events is hampered by the presence of artifacts and requires setting exclusionary gates to eliminate dying cells, cellular debris, autofluorescent cells, and cell doublets within analyzed populations (1). Doublet events are often observed in flow cytometry data, whether a sample is obtained from tissue or blood (2, 3). As cytometry data are further parsed to describe new rare populations of cells, the need for methods to properly discriminate contaminating doublets from true single cells has become an important aspect of data analysis.

A newly described rare population of lymphocytes coexpressing both a T‐cell receptor (TCR) and B‐cell receptor (BCR), termed dual expressing (DE) cells (4) has been identified in the blood of patients with Type 1 diabetes (T1D). In this issue, Burel and colleagues (in this issue, page XXX) determine that commonly used analysis methods of nonimaging flow cytometric data do not efficiently resolve DE cells from a contaminating doublet population consisting of T‐cell B‐cell conjugates. The authors develop a gating strategy to limit T‐and B‐cell conjugate contamination from the analysis of DE cell populations through the use of imaging flow cytometry, which provides additional key parameters, such as bright‐field area and bright‐field aspect ratio. Utilizing this new strategy for doublet discrimination, the authors argue that a majority of DE cells, identified by traditional singlet gating strategies, are actually T‐cell B‐cell conjugates.

Though they share a common stem cell progenitor, the development of the T‐ and B‐cell lineages occur in spatially distinct locations, thymus, and bone marrow, respectively, via highly regulated and selective processes centered on the expression and signaling events that emanate from the T‐ or B‐cell antigen receptors, TCR, and BCR (5, 6). Following development, naïve T cells circulate through the blood and secondary lymphoid organs surveilling an organism for the signs of foreign antigen (7). For major histocompatibility class II restricted CD4⁺ T cells, this surveillance occurs through interactions with antigen presenting cells (APCs) such as dendritic cells, monocytes, B cells, and macrophages (8). These interactions between T cells and APCs can result in the formation of cell: cell conjugates mediated through cell surface adhesion molecules (9). It is therefore not surprising that T‐cell conjugates have been observed in the peripheral blood of humans. The frequency of T‐cell conjugates and the phenotype of the cells that comprise them often change in response to an organisms' inflammatory state, such as following immunization or in cases of tuberculosis, dengue virus, and HIV infection (10, 11). These findings suggest that changes in T‐cell conjugates are likely hallmarks of an active immune response.

Two different types of gating strategies are commonly employed to exclude cell doublets from flow cytometry data. First, events can be excluded that deviate from the linear correlation between the forward scatter area (FSC‐A) and the FSC height (FSC‐H) parameters. Alternatively, events can be passed through two successive gates of FSC‐A by FSC width (FSC‐W) and side scatter area (SSC‐A) by SSC width (SSC‐W) utilizing the low pulse width signal indicative of single cells (Fig. 1) (3). While these methods are highly effective in eliminating the majority of contaminating doublets, not all T‐cell conjugates are removed from the data by these criteria alone (3).

To define cytometry analysis parameters that more efficiently discriminate T‐cell conjugates from DE cells, Burel and colleagues leverage their knowledge of T‐cell monocyte conjugates found in the peripheral blood of humans (10). Despite using a doublet discrimination gating strategy, they identify events expressing both T‐cell and monocyte‐specific markers CD3 and CD14. The authors find that mean fluorescent intensity (MFI) of both FCS‐A and SSC‐A increase between CD3⁺ CD14⁺ double‐positive and single‐positive populations. However, using high FSC‐A and SSC‐A intensity as exclusion criteria consequently removes single activated T cells as they are known to have increased size and granularity and thus have higher FCS‐A and SSC‐A values. When the levels of surface expressed lineage defining T cell and monocyte markers were investigated, it was determined that CD3⁺ CD14⁺ double positive population had comparable expression levels to single positive populations. One exception to this was the additive effect of the shared marker CD45, where the MFI substantially increased among the CD3⁺ CD14⁺ population relative to the single‐positive populations, suggesting the majority of this population contained T‐cell monocyte conjugates and not singlets.

The authors also investigated the levels of transcripts in CD3⁺ CD14⁺ double‐positive and single‐positive populations using single‐cell RNA sequencing (scRNAseq) analysis. They found a significant under representation of both monocyte and T‐cell‐enriched transcripts in the double‐positive population compared to the single‐positive populations. Furthermore, principle component analysis separates the double‐positive population as an intermediate grouping between the clusters of T‐cell and monocyte single‐positive populations. These findings suggest strategies can be developed to filter out contaminating doublets in scRNAseq data sets generated from conventionally sorted populations.

An effective strategy to identify CD3⁺ CD14⁺ single cells came through the use of imaging flow cytometry. By gating events with a high bright‐field aspect ratio and low bright‐field area, termed optimal (OPT) image gating (Fig. 1), data collected by imaging flow cytometry allowed efficient identification of single cells coexpressing the CD3 and CD14 markers and excluded T‐cell monocyte conjugates. The frequency of these CD3⁺ CD14⁺ single cells identified by OPT image gating and imaging flow cytometry, however, was 50‐fold reduced relative to the CD3⁺ CD14⁺ events observed using nonimaging flow cytometry and a traditional double exclusion gating strategy. The authors therefore argue that most CD3⁺ CD14⁺ events observed by nonimaging flow cytometry utilizing traditional double exclusion gating are often T‐cell monocyte conjugates and not coexpressing single cells.

The authors then applied their analysis protocol to the peripheral blood of healthy volunteers. Using nonimaging flow cytometric data and a traditional doublet discrimination gating strategy, they could identify the reported DE cells, defined as CD5⁺ CD19⁺ TCR⁺ cells (4). They observed that the DE population has a phenotype of higher MFIs for FSC‐A, SSC‐A, and CD45 than for CD5⁺ CD19⁻ T cells, CD19⁺ CD5⁻ B cells, or CD5⁺ CD19⁺ TCR⁻ cells, consistent with observations for the CD3⁺ CD14⁺ population that contained mostly T‐cell monocyte conjugates. The first report of TCR⁺ BCR⁺ DE cells came from cells isolated from the blood of people with T1D (4). B cells are thought to be critical APCs for the development of autoimmune diabetes in the nonobese diabetic animal models (12). Thus, in patients with T1D, islet‐specific autoreactive T cells may form cell: cell conjugates with B cells. When imaging flow cytometry data were collected from human peripheral blood, the majority of DE cells identified through traditional doublet exclusion gating contained T‐ and B‐cells conjugates, while with the OPT gating strategy most events were single cells that coexpressed CD19 and CD3.

The studies by Burel and colleagues (in this issue, page XXX) describe a robust and broadly applicable doublet discrimination analysis through the use of imaging flow cytometry. Utilization of this method to enhance doublet exclusion in future studies will lead to a more accurate enumeration of DE expressing cells relative to T‐ and B‐cell conjugates in people with Type 1 diabetes. Since T‐cell conjugates are often observed in people with ongoing inflammatory immune responses (10, 11), changes in frequencies of T‐cell conjugates and DE cells may provide a method to reveal the onset of autoimmune responses, as well as monitor the mobilization of immunity against infection in real time. Importantly, the accurate delineation of DE cells from T‐cell conjugates will allow specific exploration of DE cell function and contribution to immune responses.

在流式细胞仪数据中分析稀有细胞需要采集大量数据，以可视化足够多的事件。伪影的存在妨碍了在大量事件中准确识别稀有细胞，并且需要设置排除门以消除分析群体中的垂死细胞，细胞碎片，自发荧光细胞和细胞双峰（1）。无论是从组织还是从血液中获取样品，流式细胞仪数据中都经常观察到双重事件（2、3）。随着细胞计数数据被进一步解析以描述新的稀有细胞群，对从真正的单细胞中正确区分污染双峰的方法的需求已成为数据分析的重要方面。

新近描述的一种罕见的淋巴细胞群体，共表达T细胞受体（TCR）和B细胞受体（BCR），称为双重表达（DE）细胞（4）已在1型糖尿病（T1D）患者的血液中鉴定出来。在此问题中，Burel及其同事（在此问题中，第XXX页）确定，常用的非成像流式细胞仪数据分析方法不能有效地从包含T细胞B细胞共轭物的双重污染人群中解析DE细胞。作者开发了一种门控策略，通过使用成像流式细胞术从DE细胞群体分析中限制T细胞和B细胞共轭物的污染，该方法提供了其他关键参数，例如明视野面积和明视野纵横比。利用这种新的双态鉴别策略，作者认为，传统的单态门控策略识别出的大多数DE细胞实际上是T细胞B细胞结合物。

尽管它们共享一个共同的干细胞祖细胞，但T细胞和B细胞谱系的发育分别发生在空间不同的位置，胸腺和骨髓中，它们通过集中于表达和信号事件的高度调控和选择性过程而发生。 T细胞或B细胞抗原受体，TCR和BCR（5、6）。发育后，幼稚的T细胞在血液和继发性淋巴器官中循环，继而监视有机体以获取外源抗原的迹象（7）。对于主要的组织相容性II类受限制的CD4 ⁺ T细胞，这种监视通过与抗原呈递细胞（APC）如树突状细胞，单核细胞，B细胞和巨噬细胞的相互作用而发生（8）。T细胞和APC之间的这些相互作用可导致通过细胞表面粘附分子介导的细胞：细胞结合物的形成（9）。因此，在人类外周血中观察到T细胞偶联物并不奇怪。T细胞结合物的频率和构成它们的细胞表型通常会根据生物体的炎症状态而发生变化，例如在免疫后或在结核病，登革热病毒和HIV感染的情况下（10，11）。这些发现表明，T细胞结合物的变化可能是主动免疫反应的标志。

通常采用两种不同类型的门控策略从流式细胞仪数据中排除细胞双峰。首先，可以排除偏离前向散射区域（FSC-A）和FSC高度（FSC-H）参数之间线性相关性的事件。另外，事件可以通过表示单个单元的低脉冲宽度信号通过FSC宽度（FSC-W）穿过两个连续的FSC-A门，通过SSC宽度（SSC-W）通过侧面散射区（SSC-A）。图1）（3）。尽管这些方法在消除大多数污染双峰方面非常有效，但仅凭这些标准并不能从数据中去除所有T细胞共轭物（3）。

为了定义能更有效地区分来自DE细胞的T细胞偶联物的细胞计数分析参数，Burel及其同事利用了他们在人类外周血中发现的T细胞单核细胞偶联物的知识（10）。尽管使用了双重识别门控策略，但他们仍能识别同时表达T细胞和单核细胞特异性标志物CD3和CD14的事件。作者发现，FCS‐A和SSC‐A的平均荧光强度（MFI）在CD3 ⁺ CD14 ⁺之间增加。双阳性和单阳性人群。但是，使用高FSC-A和SSC-A强度作为排除标准会去除单个激活的T细胞，因为它们的大小和粒度增加，因此具有更高的FCS-A和SSC-A值。当研究限定T细胞和单核细胞标志物的表面表达谱系水平时，确定CD3 ⁺ CD14 ⁺双阳性群体具有与单阳性群体相当的表达水平。一个例外是共享标记CD45的累加效应，其中CD3 ⁺ CD14 ⁺中的MFI显着增加相对于单阳性人群，这表明该人群中的大多数包含T细胞单核细胞结合物，而不是单重态。

作者还使用单细胞RNA测序（scRNAseq）分析了CD3 ⁺ CD14 ⁺双阳性和单阳性人群的转录水平。他们发现，与单阳性人群相比，双阳性人群中单核细胞和T细胞富集的转录物均显着不足。此外，主成分分析将双阳性人群分隔为T细胞和单核细胞单阳性人群集群之间的中间群体。这些发现表明，可以制定策略来过滤掉从常规分选群体中产生的scRNAseq数据集中的污染双峰。

通过使用成像流式细胞术来鉴定CD3 ⁺ CD14 ⁺单细胞的有效策略。通过对具有高明场纵横比和低明场区域的事件进行选通（称为最佳（OPT）图像选通）（图1），通过流式细胞术成像收集的数据可有效鉴定共表达CD3和CD14标记的单细胞，排除T细胞单核细胞缀合物。通过OPT图像门控和成像流式细胞仪鉴定的这些CD3 ⁺ CD14 ⁺单细胞的频率相对于CD3 ⁺ CD14 ⁺降低了50倍使用非成像流式细胞仪和传统的双重排斥门控策略观察到的事件。因此，作者认为，使用传统的双排阻门控的非成像流式细胞术观察到的大多数CD3 ⁺ CD14 ⁺事件通常是T细胞单核细胞结合物，而不是共表达单细胞。

然后，作者将他们的分析方案应用于健康志愿者的外周血。使用非成像流式细胞仪数据和传统的双态判别门控策略，他们可以鉴定报告的DE细胞，定义为CD5 ⁺ CD19 ⁺ TCR ⁺细胞（4）。他们观察到，相对于CD5 ⁺ CD19 ^- T细胞，CD19 ⁺ CD5 ^- B细胞或CD5 ⁺ CD19 ⁺ TCR ⁻细胞，DE群体的FSC‐A，SSC‐A和CD45 MFI的表型更高。 CD3 ⁺ CD14 ^+的观察结果人群中主要包含T细胞单核细胞缀合物。TCR ⁺ BCR ⁺ DE细胞的第一个报道来自T1D患者血液中分离的细胞（4）。在非肥胖糖尿病动物模型中，B细胞被认为是自身免疫性糖尿病发展的关键APC（12）。因此，在患有T1D的患者中，胰岛特异性自身反应性T细胞可能与B细胞形成细胞：细胞结合物。当从人外周血收集成像流式细胞仪数据时，通过传统的双重排阻门控鉴定出的大多数DE细胞都含有T细胞和B细胞结合物，而采用OPT门控策略时，大多数事件是共表达CD19和CD3的单细胞。

Burel及其同事的研究（在本期XXX页）描述了通过使用成像流式细胞仪进行的稳健且广泛适用的双态判别分析。在未来的研究中，利用这种方法增强双态排斥，将使1型糖尿病患者中DE表达细胞相对于T细胞和B细胞缀合物的计数更为准确。由于T细胞结合物通常在进行性炎症免疫反应的人群中观察到（10，11），T细胞结合物和DE细胞的频率变化可能提供一种揭示自身免疫反应发生的方法，并实时监测针对感染的免疫动员。重要的是，从T细胞缀合物中准确描述DE细胞将允许对DE细胞功能的特定探索以及对免疫反应的贡献。