Expression of a CD3 epsilon transgene in CD3 epsilon(null) mice does not restore CD3 gamma and delta expression but efficiently rescues T cell development from a subpopulation of prothymocytes

A CD3 humanized mouse model unmasked unique features of T-cell responses to bispecific antibody treatment

ABSTRACT

T-cell bispecific antibodies (T-BsAbs) such as blinatumomab hold great promise for cancer immunotherapy. A better understanding of the in vivo immune response induced by T-BsAbs is crucial to improving their efficacy and safety profile. However, such efforts are hindered by the limitations of current preclinical models. To address this, we developed a syngeneic murine model with humanized CD3 and target antigen (CD20). This model enables the development of disseminated leukemia with a high tumor burden, which mirrors clinical findings in human patients with relapsed/refractory acute lymphoblastic leukemia. Treatment of this model with T-BsAbs results in cytokine release syndrome, with cytokine profiles and levels reflecting observations made in human patients. This model also faithfully recapitulates the dynamics of T-cell activation seen in human patients, including the temporary disappearance of T cells from the bloodstream. During this phase, T cells are sequestered in secondary lymphoid organs and undergo activation. Clinical correlative studies that rely primarily on peripheral blood samples are likely to overlook this critical activation stage, leading to a substantial underestimation of the extent of T-cell activation. Furthermore, we demonstrate that surface expression of the T-BsAb target antigen by leukemia cells triggers a swift immune response, promoting their own rejection. Humanizing the target antigen in the recipient mice is crucial to facilitate tolerance induction and successful establishment of high tumor burden. Our findings underscore the importance of meticulously optimized syngeneic murine models for investigating T-BsAb-induced immune responses and for translational research aimed at improving efficacy and safety.

GLP toxicology study of a fully-human T cell redirecting CD3:EGFRvIII binding immunotherapeutic bispecific antibody

We recently reported the development of a fully-human, CD3-binding bispecific antibody for immunotherapy of malignant glioma. To translate this therapeutic (hEGFRvIII-CD3- bi-scFv) to clinical trials and to help further the translation of other similar CD3-binding therapeutics, some of which are associated with neurologic toxicities, we performed a good laboratory practice (GLP) toxicity study to assess for potential behavioral, chemical, hematologic, and pathologic toxicities including evaluation for experimental autoimmune encephalomyelitis (EAE). To perform this study, male and female C57/BL6 mice heterozygous for the human CD3 transgene (20/sex) were allocated to one of four designated groups. All animals were administered one dose level of hEGFRvIII-CD3 bi-scFv or vehicle control. Test groups were monitored for feed consumption, changes in body weight, and behavioral disturbances including signs of EAE. Urinalysis, hematologic, and clinical chemistry analysis were also performed. Vehicle and test chemical-treated groups were humanely euthanized 48 hours or 14 days following dose administration. Complete gross necropsy of all tissues was performed, and selected tissues plus all observed gross lesions were collected and evaluated for microscopic changes. This included hematoxylin-eosin histopathological evaluation and Fe-ECR staining for myelin sheath enumeration. There were no abnormal clinical observations or signs of EAE noted during the study. There were no statistical changes in food consumption, body weight gain, or final body weight among groups exposed to hEGFRvIII-CD3 bi-scFv compared to the control groups for the 2- and 14-day timepoints. There were statistical differences in some clinical chemistry, hematologic and urinalysis endpoints, primarily in the females at the 14-day timepoint (hematocrit, calcium, phosphorous, and total protein). No pathological findings related to hEGFRvIII-CD3 bi-scFv administration were observed. A number of gross and microscopic observations were noted but all were considered to be incidental background findings. The results of this study allow for further translation of this and other important CD3 modulating bispecific antibodies.

A Rationally Designed Fully Human EGFRvIII:CD3-Targeted Bispecific Antibody Redirects Human T Cells to Treat Patient-derived Intracerebral Malignant Glioma

Purpose: Conventional therapy for malignant glioma fails to specifically target tumor cells. In contrast, substantial evidence indicates that if appropriately redirected, T cells can precisely eradicate tumors. Here we report the rational development of a fully human bispecific antibody (hEGFRvIII-CD3 bi-scFv) that redirects human T cells to lyse malignant glioma expressing a tumor-specific mutation of the EGFR (EGFRvIII).Experimental Design: We generated a panel of bispecific single-chain variable fragments and optimized design through successive rounds of screening and refinement. We tested the ability of our lead construct to redirect naïve T cells and induce target cell-specific lysis. To test for efficacy, we evaluated tumor growth and survival in xenogeneic and syngeneic models of glioma. Tumor penetrance following intravenous drug administration was assessed in highly invasive, orthotopic glioma models.Results: A highly expressed bispecific antibody with specificity to CD3 and EGFRvIII was generated (hEGFRvIII-CD3 bi-scFv). Antibody-induced T-cell activation, secretion of proinflammatory cytokines, and proliferation was robust and occurred exclusively in the presence of target antigen. hEGFRvIII-CD3 bi-scFv was potent and target-specific, mediating significant lysis of multiple malignant glioma cell lines and patient-derived malignant glioma samples that heterogeneously express EGFRvIII. In both subcutaneous and orthotopic models, well-engrafted, patient-derived malignant glioma was effectively treated despite heterogeneity of EGFRvIII expression; intravenous hEGFRvIII-CD3 bi-scFv administration caused significant regression of tumor burden (P < 0.0001) and significantly extended survival (P < 0.0001). Similar efficacy was obtained in highly infiltrative, syngeneic glioma models, and intravenously administered hEGFRvIII-CD3 bi-scFv localized to these orthotopic tumors.Conclusions: We have developed a clinically translatable bispecific antibody that redirects human T cells to safely and effectively treat malignant glioma. On the basis of these results, we have developed a clinical study of hEGFRvIII-CD3 bi-scFv for patients with EGFRvIII-positive malignant glioma. Clin Cancer Res; 24(15); 3611-31. ©2018 AACR.

Entire CD3ε, δ, and γ humanized mouse to evaluate human CD3-mediated therapeutics

T cell–mediated immunotherapy is an attractive strategy for treatment in various disease areas. In this therapeutic approach, the CD3 complex is one of the key molecules to modulate T cell functions; however, in many cases, we cannot evaluate the drug candidates in animal experiments because the therapeutics, usually monoclonal antibodies specific to human CD3, cannot react to mouse endogenous Cd3. Although immunodeficient mice transfused with human hematopoietic stem or precursor cells, known as humanized mice, are available for these studies, mice humanized in this manner are not completely immune competent. In this study we have succeeded in establishing a novel mouse strain in which all the three components of the Cd3 complex — Cd3ε, Cd3δ, and Cd3γ — are replaced by their human counterparts, CD3E, CD3D, and CD3G. Basic immunological assessments have confirmed that this strain of human CD3 EDG–replaced mice are entirely immune competent, and we have also demonstrated that a bispecific antibody that simultaneously binds to human CD3 and a tumor-associated antigen (e.g. ERBB2 or GPC3) can be evaluated in human CD3 EDG–replaced mice engrafted with tumors. Our mouse model provides a novel means to evaluate the in vivo efficacy of human CD3–mediated therapy.

Proteomic Signatures of Thymomas

Based on the histological features and outcome, the current WHO classification separates thymomas into A, AB, B1, B2 and B3 subtypes. It is hypothesized that the type A thymomas are derived from the thymic medulla while the type B thymomas are derived from the cortex. Due to occasional histological overlap between the tumor subtypes creating difficulties in their separation, the aim of this study was to provide their proteomic characterization and identify potential immunohistochemical markers aiding in tissue diagnosis. Pair-wise comparison of neoplastic and normal thymus by liquid chromatography tandem mass spectrometry (LC-MS/MS) of formalin fixed paraffin embedded tissue revealed 61 proteins differentially expressed in thymomas compared to normal tissue. Hierarchical clustering showed distinct segregation of subtypes AB, B1 and B2 from that of A and B3. Most notably, desmoyokin, a protein that is encoded by the AHNAK gene, was associated with type A thymomas and medulla of normal thymus, by LC-MS/MS and immunohistochemistry. In this global proteomic characterization of the thymoma, several proteins unique to different thymic compartments and thymoma subtypes were identified. Among differentially expressed proteins, desmoyokin is a marker specific for thymic medulla and is potentially promising immunohistochemical marker in separation of type A and B3 thymomas.

Co-potentiation of antigen recognition: A mechanism to boost weak T cell responses and provide immunotherapy in vivo

Adaptive immunity is mediated by antigen receptors that can induce weak or strong immune responses depending on the nature of the antigen that is bound. In T lymphocytes, antigen recognition triggers signal transduction by clustering T cell receptor (TCR)/CD3 multiprotein complexes. In addition, it hypothesized that biophysical changes induced in TCR/CD3 that accompany receptor engagement may contribute to signal intensity. Nonclustering monovalent TCR/CD3 engagement is functionally inert despite the fact that it may induce changes in conformational arrangement or in the flexibility of receptor subunits. We report that the intrinsically inert monovalent engagement of TCR/CD3 can specifically enhance physiologic T cell responses to weak antigens in vitro and in vivo without stimulating antigen-unengaged T cells and without interrupting T cell responses to strong antigens, an effect that we term as "co-potentiation." We identified Mono-7D6-Fab, which biophysically altered TCR/CD3 when bound and functionally enhanced immune reactivity to several weak antigens in vitro, including a gp100-derived peptide associated with melanoma. In vivo, Mono-7D6-Fab induced T cell antigen-dependent therapeutic responses against melanoma lung metastases, an effect that synergized with other anti-melanoma immunotherapies to significantly improve outcome and survival. We conclude that Mono-7D6-Fab directly co-potentiated TCR/CD3 engagement by weak antigens and that such concept can be translated into an immunotherapeutic design. The co-potentiation principle may be applicable to other receptors that could be regulated by otherwise inert compounds whose latent potency is only invoked in concert with specific physiologic ligands.

Negative Regulation of Humoral Immunity Due to Interplay between the SLAMF1, SLAMF5, and SLAMF6 Receptors

Whereas the SLAMF-associated protein (SAP) is involved in differentiation of T follicular helper (Tfh) cells and antibody responses, the precise requirements of SLAMF receptors in humoral immune responses are incompletely understood. By analyzing mice with targeted disruptions of the Slamf1, Slamf5, and Slamf6 genes, we found that both T-dependent and T-independent antibody responses were twofold higher compared to those in single knockout mice. These data suggest a suppressive synergy of SLAMF1, SLAMF5, and SLAMF6 in humoral immunity, which contrasts the decreased antibody responses resulting from a defective GC reaction in the absence of the adapter SAP. In adoptive co-transfer assays, both [Slamf1 + 5 + 6]^−/− B and T cells were capable of inducing enhanced antibody responses, but more pronounced enhancement was observed after adoptive transfer of [Slamf1 + 5 + 6]^−/− B cells compared to that of [Slamf1 + 5 + 6]^−/− T cells. In support of [Slamf1 + 5 + 6]^−/− B cell intrinsic activity, [Slamf1 + 5 + 6]^−/− mice also mounted significantly higher antibody responses to T-independent type 2 antigen. Furthermore, treatment of mice with anti-SLAMF6 monoclonal antibody results in severe inhibition of the development of Tfh cells and GC B cells, confirming a suppressive effect of SLAMF6. Taken together, these results establish SLAMF1, SLAMF5, and SLAMF6 as important negative regulators of humoral immune response, consistent with the notion that SLAM family receptors have dual functions in immune responses.

A PCR-Based Method to Genotype Mice Knocked Out for All Four CD3 Subunits, the Standard Recipient Strain for Retrogenic TCR/CD3 Bone Marrow Reconstitution Technology

The novel T-cell receptor (TCR)/CD3-retrogenic-reconstitution system represents a very useful strategy for studying TCR/CD3 signaling. Two retroviral vectors containing genes for all six subunits of the TCR/CD3 complex are used to transduce bone marrow precursors and reconstitute lethally irradiated recipient mice. Mice used in this system as bone marrow donors lack all four CD3 subunits (CD3γδɛζ^−/−). These mice are generated by crossing the strains CD3ζ^−/− and CD3γδɛ^−/−, the latter resulting from a knockout construct targeted to CD3ɛ that additionally silences the linked genes, CD3γ and CD3δ. Lacking mature T-cell function, CD3γδɛζ^−/− mice are immunocompromised animals often produced by heterozygous breeding strategies on the C57BL/6 background. As a more rapid and reliable means to identify CD3γδɛζ^−/− mice than previously described Northern and Southern blots, we designed polymerase chain reactions to distinguish knockout from wild-type CD3ɛ and CD3ζ alleles, facilitating the identification of CD3γδɛζ^−/− mice.

Progressive loss of CD3 expression after HTLV-I infection results from chromatin remodeling affecting all the CD3 genes and persists despite early viral genes silencing

Background

HTLV-I infected CD4⁺ T-cells lines usually progress towards a CD3^- or CD3^low phenotype. In this paper, we studied expression, kinetics, chromatin remodeling of the CD3 gene at different time-points post HTLV-I infection.

Results

The onset of this phenomenon coincided with a decrease of CD3γ followed by the subsequent progressive reduction in CD3δ, then CD3ε and CD3ζ mRNA. Transient transfection experiments showed that the CD3γ promoter was still active in CD3^- HTLV-I infected cells demonstrating that adequate amounts of the required transcription factors were available. We next looked at whether epigenetic mechanisms could be responsible for this progressive decrease in CD3 expression using DNase I hypersensitivity (DHS) experiments examining the CD3γ and CD3δ promoters and the CD3δ enhancer. In uninfected and cells immediately post-infection all three DHS sites were open, then the CD3γ promoter became non accessible, and this was followed by a sequential closure of all the DHS sites corresponding to all three transcriptional control regions. Furthermore, a continuous decrease of in vivo bound transcription initiation factors to the CD3γ promoter was observed after silencing of the viral genome. Coincidently, cells with a lower expression of CD3 grew more rapidly.

Conclusion

We conclude that HTLV-I infection initiates a process leading to a complete loss of CD3 membrane expression by an epigenetic mechanism which continues along time, despite an early silencing of the viral genome. Whether CD3 progressive loss is an epiphenomenon or a causal event in the process of eventual malignant transformation remains to be investigated.

Targeted in vitro and in vivo gene transfer into T lymphocytes: potential of direct inhibition of allo-immune activation

Background

Successful inhibition of alloimmune activation in organ transplantation remains one of the key events in achieving a long-term graft survival. Since T lymphocytes are largely responsible for alloimmune activation, targeted gene transfer of gene of cyclin kinase inhibitor p21 into T cells might inhibit their aberrant proliferation. A number of strategies using either adenoviral or lentiviral vectors linked to mono or bispecific antibodies directed against T cell surface markers/cytokines did not yield the desired results. Therefore, this study was designed to test if a CD3promoter-p21 chimeric construct would in vitro and in vivo transfer p21 gene to T lymphocytes and result in inhibition of proliferation. CD3 promoter-p21 chimeric constructs were prepared with p21 in the sense and antisense orientation. For in vitro studies EL4-IL-2 thyoma cells were used and for in vivo studies CD3p21 sense and antisense plasmid DNA was injected intramuscularly in mice. Lymphocyte proliferation was quantified by ³H-thymidine uptake assay; IL-2 mRNA expression was studied by RT-PCR and using Real Time PCR assay, we monitored the CD3, p21, TNF-α and IFN-γ mRNA expression.

Results

Transfection of CD3p21 sense and antisense in mouse thyoma cell line (EL4-IL-2) resulted in modulation of mitogen-induced proliferation. The intramuscular injection of CD3p21 sense and antisense plasmid DNA into mice also modulated lymphocyte proliferation and mRNA expression of pro-inflammatory cytokines.

Conclusion

These results demonstrate a novel strategy of in vitro and in vivo transfer of p21 gene to T cells using CD3-promoter to achieve targeted inhibition of lymphocyte proliferation and immune activation.

How TCRs bind MHCs, peptides, and coreceptors

Since the first crystal structure determinations of alphabeta T cell receptors (TCRs) bound to class I MHC-peptide (pMHC) antigens in 1996, a sizable database of 24 class I and class II TCR/pMHC complexes has been accumulated that now defines a substantial degree of structural variability in TCR/pMHC recognition. Recent determination of free and bound gammadelta TCR structures has enabled comparisons of the modes of antigen recognition by alphabeta and gammadelta T cells and antibodies. Crystal structures of TCR accessory (CD4, CD8) and coreceptor molecules (CD3epsilondelta, CD3epsilongamma) have further advanced our structural understanding of most of the components that constitute the TCR signaling complex. Despite all these efforts, the structural basis for MHC restriction and signaling remains elusive as no structural features that define a common binding mode or signaling mechanism have yet been gleaned from the current set of TCR/pMHC complexes. Notwithstanding, the impressive array of self, foreign (microbial), and autoimmune TCR complexes have uncovered the diverse ways in which antigens can be specifically recognized by TCRs.

Severe combined immunodeficiency caused by deficiency in either the delta or the epsilon subunit of CD3

We investigated the molecular mechanism underlying a severe combined immunodeficiency characterized by the selective and complete absence of T cells. The condition was found in 5 patients and 2 fetuses from 3 consanguineous families. Linkage analysis performed on the 3 families revealed that the patients were carrying homozygous haplotypes within the 11q23 region, in which the genes encoding the gamma, delta, and epsilon subunits of CD3 are located. Patients and affected fetuses from 2 families were homozygous for a mutation in the CD3D gene, and patients from the third family were homozygous for a mutation in the CD3E gene. The thymus from a CD3delta-deficient fetus was analyzed and revealed that T cell differentiation was blocked at entry into the double positive (CD4+CD8+) stage with the accumulation of intermediate CD4-single positive cells. This indicates that CD3delta plays an essential role in promoting progression of early thymocytes toward double-positive stage. Altogether, these findings extend the known molecular mechanisms underlying severe combined immunodeficiency to a new deficiency, i.e., CD3epsilon deficiency, and emphasize the essential roles played by the CD3epsilon and CD3delta subunits in human thymocyte development, since these subunits associate with both the pre-TCR and the TCR.

Regulation of pT alpha gene expression by a dosage of E2A, HEB, and SCL

The expression of the pT alpha gene is required for effective selection, proliferation, and survival of beta T-cell receptor (beta TCR)-expressing immature thymocytes. Here, we have identified two phylogenetically conserved E-boxes within the pT alpha enhancer sequence that are required for optimal enhancer activity and for its stage-specific activity in immature T cells. We have shown that the transcription factors E2A and HEB associate with high affinity to these E-boxes. Moreover, we have identified pT alpha as a direct target of E2A-HEB heterodimers in immature thymocytes because they specifically occupy the enhancer in vivo. In these cells, pT alpha mRNA levels are determined by the presence of one or two functional E2A or HEB alleles. Furthermore, E2A/HEB transcriptional activity is repressed by heterodimerization with SCL, a transcription factor that is turned off in differentiating thymocytes exactly at a stage when pT alpha is up-regulated. Taken together, our observations suggest that the dosage of E2A, HEB, and SCL determines pT alpha gene expression in immature T cells.

Initiation of TCR signaling: regulation within CD3 dimers

The number of possible T cell activation outcomes resulting from T cell receptor (TCR) engagement suggests that the TCR is able to differentially activate a myriad of signaling pathways depending on the nature of the stimulus. The complex structural organization of the TCR itself could underlie this diversity of responses. Assembly and stoichiometric studies have helped us to shed some light on the initiation of TCR signaling. The TCR is composed of TCR and CD3 dimers. Changes in the interaction between CD3 subunits within the CD3 dimers and in the interaction of these dimers with the TCR heterodimer could be the triggering mechanism that initiates the first activation events. One of the hallmarks of these early changes in TCR conformation is the induced recruitment of the adapter protein Nck to a proline-rich sequence of the cytoplasmic tail of CD3epsilon, but there may be others. According to our most recent observations, the TCR is organized in pre-existing clusters within plasma membrane microdomains, exhibiting a complexity above and beyond that of dimer composition complexity. How the presence of TCR in clusters influences TCR avidity and propagation of TCR signals is something that has yet to be investigated.

T-cell depletion and graft survival induced by anti-human CD3 immunotoxins in human CD3epsilon transgenic mice

BACKGROUND

Anti-CD3 immunotoxins are broad-spectrum immunosuppressive agents in a wide range of organ transplantation animal models with potential use in eliciting antigen-specific tolerance. However, the anti-CD3 immunotoxins used in animal studies do not cross-react with human T cells, limiting extrapolation to humans and hindering clinical development.

METHODS

Three anti-human CD3-directed immunotoxins, DT389-scFv(UCHT1), scFv(UCHT1)-PE38, and UCHT1-CRM9, were compared in vitro and in transgenic mice, tg(epsilon)600+/-, that have T cells expressing both human and murine CD3epsilon antigens.

RESULTS

These immunotoxins were extraordinarily potent in vitro against human or transgenic mouse T cells, with IC50 values in cellular assays ranging from pM to fM. Systemic administration of these immunotoxins dose-dependently depleted >99% of tg(epsilon)600+/- lymph node and spleen T cells in vivo. Depletion was specific for T cells. The loss of the concanavalin A-induced, but not the lipopolysaccharide-induced, splenic proliferative response from immunotoxin-treated animals further demonstrated specific loss of T-cell function. Immunotoxin treatment prolonged fully allogeneic skin graft survival in tg(epsilon)600+/- recipients to 25 days from 10 days in untreated animals. T-cells recovered to approximately 50% of normal levels after approximately 22 days in animals with or without skin grafts; T-cell recovery correlated with skin graft rejection. All three immunotoxins elicited >100 day median survival of fully allogeneic heterotopic heart grafts. By 100 days, T cells recovered to normal numbers in these animals, but the grafts showed chronic rejection.

CONCLUSION

These immunotoxins profoundly deplete T cells in vivo and effectively prolong allogeneic graft survival.

Improved binding of a bivalent single-chain immunotoxin results in increased efficacy for in vivo T-cell depletion

Anti-CD3 immunotoxins exhibit considerable promise for the induction of transplantation tolerance in pre-clinical large animal models. Recently an anti-human anti-CD3epsilon single-chain immunotoxin based on truncated diphtheria toxin has been described that can be expressed in CHO cells that have been mutated to diphtheria toxin resistance. After the two toxin glycosylation sites were removed, the bioactivity of the expressed immunotoxin was nearly equal to that of the chemically conjugated immunotoxin. This immunotoxin, A-dmDT390-sFv, contains diphtheria toxin to residue 390 at the N-terminus followed by VL and VH domains of antibody UCHT1 linked by a (G(4)S)(3) spacer (sFv). Surprisingly, we now report that this immunotoxin is severely compromised in its binding affinity toward CD3(+) cells as compared with the intact parental UCHT1 antibody, the UCHT1 Fab fragment or the engineered UCHT1 sFv domain alone. Binding was increased 7-fold by adding an additional identical sFv domain to the immunotoxin generating a divalent construct, A-dmDT390-bisFv (G(4)S). In vitro potency increased 10-fold over the chemically conjugated immunotoxin, UCHT1-CRM9 and the monovalent A-dmDT390-sFv. The in vivo potency of the genetically engineered immunotoxins was assayed in the transgenic heterozygote mouse, tgepsilon 600, in which the T-cells express human CD3epsilon as well as murine CD3epsilon. T-cell depletion in the spleen and lymph node observed with the divalent construct was increased 9- and 34-fold, respectively, compared with the monovalent construct. The additional sFv domain appears partially to compensate for steric hindrance of immunotoxin binding due to the large N-terminal toxin domain.

T Lymphocyte Development in the Absence of CD3ε or CD3γδεζ

CD3γ, δ, ε, and ζ proteins together with the pre-TCR α-chain (pTα) and a rearranged TCR β-chain assemble to form the pre-TCR that controls the double negative (DN) to double positive (DP) stages of thymopoiesis. The CD3 proteins are expressed before pTα and TCR β-chains in prothymocytes and are expressed intracellularly in precursor NK cells, suggesting that the CD3 complex may function independent of pTα and TCRβ. In this report, both the role of CD3ε exclusively, and the role of CD3 proteins collectively, in thymocyte and NK cell development were examined. In a mouse strain termed εΔP, a neomycin cassette inserted within the CD3ε promoter abolishes CD3ε and δ expression and also abolishes CD3γ expression in all but a small minority (≤1%) of prothymocytes. These prothymocytes became deficient in CD3ε alone upon reconstitution of CD3δ expression and were severely, but not completely, arrested at the DN stage, as small numbers of double positive thymocytes were detected. In de facto CD3γδεζnull mice generated by crossing the εΔP mice with CD3ζ−/− mice, thymopoiesis were arrested at the CD44−CD25+ DN stage as observed in RAG−/− mice, DJ and VDJ recombination at the TCRβ locus was functional, and normal numbers of NK cells were detected. Together, the findings demonstrate that during thymocyte development, the CD3 complex collectively is not essential until the critical CD44−CD25+ DN stage in which pre-TCR begins to function, whereas CD3ε is critical for the assembly of pre-TCR. Moreover, CD3 proteins are dispensable for NK cell development.