We read with interest the personal reflection of. Hans-Gustaf Ljunggren from the Karolinska Institute (KI) about the path towards natural killer (NK) cell-based cancer immunotherapy.¹ The manuscript includes a paragraph about mouse major histocompatibility complex (MHC) class I deficiency, and as we published several papers about the equivalent entity, Human Leukocyte Antigen (HLA) class I defects, we were inspired to briefly remind the NK cell status in these diseases.

Prof. Ljunggren was among the first to describe that NK cells preferentially kill targets with low or absent MHC class I molecules and postulated the existence of MHC class I-specific inhibitory receptors (IR), which would refrain NK cells from killing normal surrounding cells (missing self-hypothesis).² The existence of such receptors was demonstrated by several groups in human, rat and mouse.

The researchers from the KI started, among others, to look at mice genetically deficient in MHC class I molecules, such as beta-2-microglobulin (β2m) knockout (KO) animals. According to the missing self-concept, NK cells were expected to kill autologous cells because they lack expression of self MHC class I molecules, but this is not the case at baseline. The NK cells from such animals are tolerant towards autologous targets and unable to perform missing self-recognition in vitro and in vivo.³ These observations were confirmed in transporter associated with antigen processing (TAP) KO mice.⁴

An equivalent to TAP-deficient mice was described in 1994 by de la Salle et al⁵ in two siblings from a consanguineous marriage, who presented with chronic bacterial infections of the upper and lower respiratory tract. Their serologic HLA class I typing was negative, and the expression of HLA class I molecules assessed by flow cytometry appeared strongly reduced. An autosomal recessive mutation in the TAP-2 gene was identified. Interestingly, ex vivo NK cells from the patients displayed no cytotoxic activity towards K562 (the classical human HLA class I negative NK cell target) nor towards autologous cells.^{5, 6} Thus, these patients’ NK cells were (a) unable to perform missing self-recognition and (b) tolerant to the autologous MHC class I-deficient environment. Upon cytokine-mediated activation, however, they killed several cancer cell lines (including K562), and the autologous B lymphoblastoid cell lines (B-LCL) and skin fibroblasts.⁶

Moins-Teissserenc et al (5) and Furukawa et al. (1) described additional TAP-1 and TAP-2 deficient patients,^{7, 8} of whom some suffered not only from respiratory infections but had also debilitating granulomatous skin lesions or even destruction of the nasal cartilage.⁷ Regarding their NK cells, Furukawa et al confirmed our observations that fresh peripheral blood mononuclear cells were not cytolytic towards K562, Daudi, Molt4 and the HLA class I-negative B-LCL 721.221, in contrast to normal effectors.⁸ After activation with interleukin (IL)-2, IL-12, or IL-15, K562 and Molt4 were lysed by the patient's NK cells, but not Daudi nor 721.221, which is opposite to our data.⁶ The authors concluded that the tolerant status of TAP-deficient NK cells is maintained even after cytokine stimulation to avoid autoreactivity.⁸

Moins-Teisserenc et al⁷ found that four NK cell clones of one patient were not autoreactive, although they killed K562 targets. In contrast, a NK cell line from another case was autoreactive against B-LCL, killed the same cell type from another patient but was inhibited by normal B-LCL, presumably via the interaction of HLA class I molecules with specific NK cell IR, the latter being phenotypically and functionally normal in TAP deficiency.⁶

What makes this paper truly interesting and important is the observation that some skin lesions were massively infiltrated with activated NK cells.⁷ This suggests a direct involvement of autoreactive NK cells in the pathogenesis of the lung and skin lesions,⁷ whereas the self-aggressive peripheral blood NK cells have been stimulated in vitro and were not tested before activation. The initial hypothesis of de la Salle was that the insufficient clearance of viral infections (surprisingly not that severe in TAP-deficient patients) leads to bacterial colonization and superinfection followed by a chronic and deleterious overactivation of NK cells that cannot be inhibited by the insufficient levels of HLA class I molecules in the environment.

In this context, it is interesting to note that tissue NK cells have become a hot topic in recent years, again partly under the leadership of the KI. It is generally admitted that NK cells in various organs and tissues might not only migrate from peripheral blood, but that different NK cell phenotypes and even lineages might be organ specific.

Although only a bit more than 30 TAP-deficient patients have been described, the clinical presentation and their NK cells appear quite heterogeneous, as illustrated by the discordant findings of our group and those of Furukawa et al⁸ and Moins-Teisserenc et al⁷ We encountered a patient with TAP deficiency whose NK cells were cytotoxic ex vivo, who had very severe manifestations and died from cerebral vasculitis. In this case, the dogma of the unlicensed NK cells would not apply and might be explained by the clinical status of the patient.

In addition to TAP deficiency, two cases of human β2 microglobulin deficiency were presented by Ardeniz et al.⁹ In these patients, not only HLA class I expression is reduced, but also that of the CD1a, CD1b and CD1c molecules, of the FcRn receptor and presumably that of the HLA class I-related molecule MR1, involved in antibacterial defence (as all these structures need to bind β2m for a stable expression). Ex vivo NK cells were not cytotoxic towards K562, in accordance with mouse data (in assays with the appropriate mouse targets).⁹

Overall, NK cells from human HLA class I-deficient patients seem to behave as their mouse counterparts (hypo-responsive ex vivo, auto-aggressive upon cytokine-mediated activation). In both species, NK cells must be educated by the interaction of IR with their cognate MHC class I ligands to become functional, and, in the absence of this interaction, the cells remain hypo-responsive. Nevertheless, when they become stimulated in an infectious and inflammatory context, major auto-aggressive phenomena may occur.

A fundamental difference is that inbred mouse strains are genetically (and maybe even epigenetically) homogeneous, which is not the case when analysing biologic material from different, unrelated human beings. This may explain, at least in part, the discrepancies between our studies and those of Furukawa et al⁸ and Moins-Teisserenc et al.⁷ In addition, the former stimulated patient cells with cytokines alone for 60 hours,⁸ whereas we applied the method based on the co-culture of peripheral blood mononuclear cells with irradiated feeder cells (B-LCL) and IL-2. Moreover, inbred mice usually live in a pathogen-free environment, which might account for the absence of a clinical phenotype.

我们饶有兴趣地阅读了个人的反思。来自卡罗林斯卡学院 (KI) 的 Hans-Gustaf Ljunggren 介绍了通往基于自然杀伤 (NK) 细胞的癌症免疫疗法的途径。¹手稿包括一段关于小鼠主要组织相容性复合体 (MHC) I 类缺陷的段落，当我们发表了几篇关于等效实体人类白细胞抗原 (HLA) I 类缺陷的论文时，我们受到启发，简要提醒了 NK 细胞状态这些疾病。

Ljunggren 教授是最早描述 NK 细胞优先杀死具有低或不存在 MHC I 类分子的靶标的人之一，并假设存在 MHC I 类特异性抑制受体 (IR)，这将抑制 NK 细胞杀死周围的正常细胞。缺少自我假设）。²人类、大鼠和小鼠的几个组证明了此类受体的存在。

KI 的研究人员开始研究 MHC I 类分子遗传缺陷的小鼠，例如 β-2-微球蛋白 (β2m) 敲除 (KO) 动物。根据缺失的自我概念，预计 NK 细胞会杀死自体细胞，因为它们缺乏自身 MHC I 类分子的表达，但在基线情况下并非如此。来自这些动物的 NK 细胞对自体目标具有耐受性，并且无法在体外和体内进行缺失的自我识别。³这些观察结果在与抗原加工 (TAP) KO 小鼠相关的转运蛋白中得到证实。⁴

1994 年 de la Salle 等人⁵在近亲婚姻的两个兄弟姐妹中描述了一种相当于 TAP 缺陷小鼠的小鼠，他们患有上呼吸道和下呼吸道的慢性细菌感染。他们的血清 HLA I 类分型为阴性，通过流式细胞术评估的 HLA I 类分子的表达似乎强烈降低。鉴定了 TAP-2 基因中的常染色体隐性突变。有趣的是，来自患者的离体 NK 细胞对 K562（经典的人类 HLA I 类阴性 NK 细胞靶标）和自体细胞没有显示细胞毒活性。^5、6因此，这些患者的 NK 细胞 (a) 无法执行缺失的自我识别和 (b) 耐受自体 MHC I 类缺陷环境。然而，在细胞因子介导的激活后，它们杀死了几种癌细胞系（包括 K562）、自体 B 淋巴母细胞系 (B-LCL) 和皮肤成纤维细胞。⁶

Moins-Teissserenc 等人 (5) 和 Furukawa 等人。(1) 描述了额外的 TAP-1 和 TAP-2 缺陷患者，其中^{7、8 名}患者不仅患有呼吸道感染，而且还患有使人衰弱的肉芽肿性皮肤病变，甚至鼻软骨破坏。⁷关于他们的 NK 细胞，Furukawa 等人证实了我们的观察结果，即新鲜外周血单核细胞对 K562、Daudi、Molt4 和 HLA I 类阴性 B-LCL 721.221 没有细胞溶解作用，与正常效应细胞相反。⁸用白细胞介素 (IL)-2、IL-12 或 IL-15 激活后，K562 和 Molt4 被患者的 NK 细胞裂解，但 Daudi 和 721.221 均不裂解，这与我们的数据相反。⁶作者得出结论，即使在细胞因子刺激后，TAP 缺陷 NK 细胞的耐受状态仍能保持，以避免自身反应。⁸

Moins-Teisserenc 等人⁷发现一名患者的四个 NK 细胞克隆没有自身反应性，尽管它们杀死了 K562 目标。相比之下，来自另一个病例的 NK 细胞系对 B-LCL 具有自身反应性，杀死了来自另一名患者的相同细胞类型，但被正常 B-LCL 抑制，推测是通过HLA I 类分子与特定 NK 细胞 IR 的相互作用，后者在 TAP 缺乏症中表型和功能正常。⁶

使这篇论文真正有趣和重要的是观察到一些皮肤损伤被激活的 NK 细胞大量浸润。⁷这表明自体反应性 NK 细胞直接参与肺和皮肤病变的发病机制，⁷而自体外周血 NK 细胞已在体外受到刺激，并且在激活前未进行测试。de la Salle 的最初假设是病毒感染的清除不足（令人惊讶的是在 TAP 缺陷患者中没有那么严重）导致细菌定植和重复感染，然后是慢性和有害的 NK 细胞过度激活，这种过度激活无法被不足的水平所抑制环境中的 HLA I 类分子。

在此背景下，有趣的是，组织 NK 细胞近年来已成为热门话题，部分是在 KI 的领导下。人们普遍承认，各种器官和组织中的 NK 细胞不仅可能从外周血迁移，而且不同的 NK 细胞表型甚至谱系可能具有器官特异性。

虽然超过30只有一点点TAP缺陷患者已描述的那样，临床表现和他们的NK细胞出现相当异质的，由我们的组的和不一致的结果那些Furukawa等人的如图⁸和MOINS-Teisserenc等人⁷我们遇到了一名 TAP 缺乏症患者，其 NK 细胞在体外具有细胞毒性，表现非常严重并死于脑血管炎。在这种情况下，未经许可的 NK 细胞的教条将不适用，可能可以通过患者的临床状态来解释。

除了 TAP 缺乏症外，Ardeniz 等人还提出了两例人类 β2 微球蛋白缺乏症。⁹在这些患者中，不仅 HLA I 类表达降低，而且 FcRn 受体的 CD1a、CD1b 和 CD1c 分子以及可能参与抗菌防御的 HLA I 类相关分子 MR1 的表达也降低（正如所有这些结构需要结合 β2m 才能稳定表达）。根据小鼠数据（在具有适当小鼠靶标的测定中），离体 NK 细胞对 K562 没有细胞毒性。⁹

总体而言，来自人类 HLA I 类缺陷患者的 NK 细胞似乎表现得与它们的小鼠对应物一样（离体低反应性，对细胞因子介导的激活具有自动攻击性）。在这两个物种中，NK 细胞必须通过 IR 与其同源 MHC I 类配体的相互作用进行训练才能发挥功能，并且在没有这种相互作用的情况下，细胞保持低反应性。然而，当它们在感染性和炎症性环境中受到刺激时，可能会发生主要的自身攻击性现象。

一个根本的区别是近交小鼠品系在遗传上（甚至可能在表观遗传上）是同质的，但在分析来自不同的、不相关的人类的生物材料时，情况并非如此。这至少可以部分解释我们的研究与 Furukawa 等人⁸和 Moins-Teisserenc 等人的研究之间的差异。⁷此外，前者仅用细胞因子刺激患者细胞 60 小时，⁸而我们应用的方法基于外周血单核细胞与辐照饲养细胞 (B-LCL) 和 IL-2 的共培养。此外，近交小鼠通常生活在无病原体的环境中，这可能是缺乏临床表型的原因。