A human memory T cell subset with stem cell-like properties.

ABSTRACTHigh-dose chemotherapy may kill not only tumor cells but also immunocytes, and frequently induces severe lymphocytopenia. On the other hand, patients who recover from the nadir maintain immunity against infection, suggesting the existence of an unknown memory T-cell population with stress resistance, long-living capacity, proliferation and differentiation. Recently, the differentiation system of T-cell memory has been clarified using mouse models. However, the human T-cell memory system has great diversity induced by natural antigens derived from many pathogens and tumor cells throughout life, and profoundly differs from the mouse memory system constructed using artificial antigens and transgenic T cells. Here we report a novel human T-cell memory population, “young memory” T (TYM) cells. TYM cells are defined by positive expression of CD73, which represents high aldehyde dehydrogenase 1 (ALDH1) activity and CXCR3 among CD8+CD45RA+CD62L+ T cells. TYM proliferate upon TCR stimulation, with differentiation capacity into TCM and TEM and drug resistance. Moreover, TYM are involved in memory function for viral and tumor-associated antigens in healthy donors and cancer patients, respectively. Regulation of TYM might be very attractive for peptide vaccination, adoptive cell-transfer therapy and hematopoietic stem cell transplantation.

The immune system remembers in different ways. Immune responses can be generated to the diverse antigens which enter our body through barrier sites—to multifarious infectious pathogens each with specific cellular targets, modes of infection, and persistence. Immunological memory that results from these distinct insults is heterogeneous as manifested by memory T cells. Memory T cells persist in multiple subsets that differ in migration, functional potential, turnover, persistence, and tissue localization, among other things.1, 2 While memory T cells can direct protective responses to numerous pathogens—including viruses, bacteria, fungi, and parasites at the site of pathogen encounter, we lack vaccine strategies to specifically promote protective T-cell immunity. A challenge in the field has been to dissect mechanisms for promoting and maintaining T-cell memory given the enormous heterogeneity of this response. Studies in this area have therefore focused on defining the features of memory T cells that vary and the factors controlling them with more recent studies employing high-dimensional single-cell technologies. Nearly 25 years ago, Sallusto and Lanzavecchia identified subsets of human memory T cells differing in the expression of a lymph node homing receptor,3 which was also recapitulated in mouse infection and memory generation models.4, 5 These findings brought the notion of memory T-cell heterogeneity to the forefront—leading to the identification of memory T cells in diverse tissues remote from the original infection site,6, 7 suggesting a role for memory T cells in surveilling through multiple anatomic sites. However, we now know that while certain memory T-cell subsets can circulate throughout the body, a substantial fraction are non-circulating and are maintained as tissue resident memory T cells or TRM. The discovery of CD4+ and CD8+TRM in mouse infection models and their critical role in site-specific protection8-11 revealed an important spatial aspect of memory T-cell heterogeneity. These findings further indicated that studying memory T-cell functional roles in immunity would require direct examination of tissues—both in mice and in humans. The 10 review pieces in this issue from top scientists in the field discuss our current understanding of memory T-cell heterogeneity in the context of location and its distribution across the body. These reviews present how memory T-cell heterogeneity and tissue distribution are integrated into studies defining mechanisms for memory T-cell generation, functional responses to antigen, and memory T-cell maintenance to different pathogens and over age. They show how the field has moved forward revealing that tissue location is a major driver of memory T-cell heterogeneity and immune system variation. There have also been major conceptual breakthroughs in defining how long-term T-cell responses are maintained, including to acute versus persisting pathogens, and how these can alter over age. Collectively, the reviews in this issue discuss how tissue, antigen, age, and other factors contribute to memory heterogeneity, providing a state-of the-art synthesis of this complex field. Defining memory T-cell heterogeneity in terms of circulating and tissue resident subsets required elucidating the transcriptional changes involved in subset delineation. Ananda Goldrath provides a comprehensive overview of the key advances in defining the transcription factors which delineate TRM from circulating memory subsets, including those identified by her group and others.12 The Goldrath review further describes how TRM in specific sites exhibit tissue adaptations, and how they mediate secondary responses. Peter Szabo discusses the multiple axes of TRM heterogeneity with a focus on human T-cell memory.13 He describes studies from the past decade using multiple tissues from organ donors to elucidate the subset- and tissue-specific features of TRM14-16 and how these human studies reveal tissue-dependent pathways for TRM generation. He further discusses how systems immunology approaches have enabled precise dissection of the roles of lineages, subset, site, and age on the function and properties of memory T cells and their maintenance in tissues and circulation. Along the theme of computational approaches, José Borghans and colleagues elegantly describe how memory T-cell maintenance is measured and modeled and the challenges of assessing these critical aspects of memory T-cell dynamics throughout the body.17 They provide a clear and thorough analysis of the different ways in which memory T-cell longevity and turnover are measured, and that TRM are largely maintained in situ without influx of circulating, peripheral T cells.17 The evidence cited draws from studies in mice, humans, goats, and non-human primates, to different antigens, applying both experimental and computational modeling approaches.17 While the above reviews discuss memory T cells across tissues and their function and maintenance more broadly, a set of four reviews delves deeply into the role of memory T cells in specific sites. Jie Sun and Su Cheon analyze how TRM in the lung, while mediating optimal protective responses, can also predispose to immunopathology.18 They provide interesting examples of how TRM mediate lung damage in pneumonia and post-acute sequelae of respiratory infection, and further discuss pathways leading to overexuberant responses and how to target them.18 Jennifer Lund, Florian Hladik, and Martin Prlic investigate TRM in the female reproductive tract and how the changing local tissue environment of this organ in terms of hormonal influences, pregnancy, menstruation, and menopause affects resident immune cells.19 They further discuss new insights into the role of TRM in mediating protective responses and how these are adaptive for health during pregnancy. At the surface, Muzlifah Haniffa delves into the complex role of TRM in the skin and the different niches in which TRM are situated.20 Strobl and Haniffa synthesize tremendous work in the field that have elucidated the role of skin TRM in many types of skin diseases including psoriasis,21 fixed vitiligo,22 and alopecia.23 The last review in this section on tissue-specific T-cell immunity is by John Harty and colleagues who provide a comprehensive overview of T-cell immunity to malaria with a focus on T-cell responses in the liver.24 They discuss liver homing and residence of T cells in the unique liver microenvironment, their key role in protection and prospects for future development of vaccine-mediated protection to this endemic pathogen. The final set of reviews examines other crucial aspects of memory heterogeneity. Jenny Ning Jiang analyzes how T-cell repertoire and antigen specificity dictates responses to infection and cancer, and the role of cross-reactive T cells in these processes.25 These issues are important for T-cell-mediated protection and anti-tumor immunity.25 Annette Oxenius examines how T-cell memory is generated and maintained to pathogens that persist in the body in different forms including cytomegalovirus and chronic viruses.26 She discusses key adaptations of CD4 and CD8 T cells to persistent pathogens, pathways for their differentiation, and mechanisms for transcriptional control of these processes.26 Finally, we end the issue with new insights on memory T-cell aging by Arne Akbar, Daniel Gomez, and colleagues, who integrate new and previous results identifying senescent changes in memory T cells that lead to increased expression of NK markers and innate immune function.27 While these age-associated features can be detrimental to immune responses in promoting inflammation, they can also be beneficial by promoting rapid protection in the tissues.27 Together, this final set of reviews reveals aspects of T-cell memory that are unaffected by antigen nature and exogenous factors, and less dependent on tissue localization. To cap off this issue, I have generated an image representing memory T-cell heterogeneity using Dall-E, an AI-based image program to reflect the increasing use of computational pipelines and approaches that we now employ to understand the complexities of T-cell memory (Figure 1). To echo the theme of heterogeneity but also of diverse landscapes, I asked Dall-E to generate an image with cells in multiple colors and shapes (i.e., heterogeneous) in the style of the artist Henry Moore (1898–1996), a sculptor, printer, and painter whose work largely comprised representations of the human body, and how it fits into the landscape of the environment. Moore's works are abstract, depicting lying or standing figures with holes or spaces so that they blend into the background, much like memory T cells fit into the niches of a tissue. Henry Moore was also interested in memory and recording the past. He was an official wartime artist during World War II and drew detailed images of people in London sheltering in the underground (subway) stations during the wartime bombing (the Blitz). His poignant drawings of people wrapped in old blankets on the platforms, open-mouthed in exhausted sleep, transport you to a nearly forgotten time and place of human suffering. The images generated in Figure 1 depict cellular and spatial heterogeneity in different compartments reminiscent of immune memory, co-opting Moore's abstract style and spatial representations. As we emerge from the worldwide COVID-19 pandemic, our immunological memories have recorded the site and time of infection and the dissemination of viral antigens throughout our bodies. These widely distributed memory cell stores have the potential to protect us from further infection, morbidity and especially mortality. As the authors in this issue have so elegantly described, we know a great amount about memory T cells and what controls their function, how they can localize, migrate, and remain in tissues. But as with much scientific knowledge, we still need to connect all of the pieces to be able to predict with precision how memory T cells respond in every context. Given this substantial foundation, emerging new technologies for mechanistic studies in humans, and the talent, insight, and collaboration of the immunology community, we will get there in the future—and use this knowledge to protect for the next pandemic. I wish to thank Dr. Joshua Gray for help in the initial review of the manuscripts and helpful comments. There are no conflicts to report. N/A.

Post-thymic regulation of CD5 levels in human memory T cells is inversely associated with the strength of responsiveness to interleukin-15

Introduction: With the advent of immunotherapy (IT) against various cancers, its applications to other cancers have been extensively investigated. However, it has been a challenge to apply IT to chordomas, due to lack of clinically translatable in vivo models. Currently, there are no well-established murine chordoma cell lines that can be injected to syngeneic mice or no transgenic mouse models that develop chordomas spontaneously, which would allow us to study the interaction between murine chordomas and murine immune cells. Hence, we aimed to develop a humanized mouse model, where human immune cells are engrafted into immunodeficient mice, to study the interaction between human immune system and human chordomas. We also sought to utilize it to investigate synergistic effect between IT and radiation therapy (RT) against chordoma. Materials and Methods: Fifteen 10-12-week-old NSG mice, which lack mouse T-cells, B cells, and NK cells as well as functional mouse macrophages, were sublethally (1.5Gy) irradiated and then implanted with fetal thymic tissue and CD34+ stem cells that had been harvested from a fetus, whose HLA-types were partially-matched with those of the U-CH1 chordoma cell line. Reconstitution of immune cells in NSG mice was confirmed eight weeks post-transplantation, and then each animal (15 humanized NSG mice and 12 naïve NSG mice) was injected with U-CH1 cell suspension bilaterally and subcutaneously. Next, they were treated for 4 weeks as follows: A) control, isotype antibodies (Abs) injection (n=3), B) anti-human-PD-1 Abs (n=4, 10 mg/kg, 3 times/week for 4 weeks), C) RT + isotype Abs (n=3, unilaterally to the left-sided tumor, 8Gy x 4), D) anti-human-PD-1 Abs and RT (n=5), E) naïve NSG mice (n=6, without the engraftment of human immune cells) + isotype, and F) naïve NSG mice (n=6) + anti-human-PD-1 Abs. During and after the treatment, anti-tumor activities were monitored via tumor size, flow cytometry, qRT-PCR, and immunohistochemistry. Results: Eight weeks after stem cell engraftment, human peripheral blood mononuclear cells (PBMCs) of 43.8% among all PBMCs (human + mouse), human T-cells of 23.4% among human PBMCs, human CD8+ T-cells of 24.3% among human T-cells, and other lymphocytes such as B cells, macrophages, and NK cells were observed in peripheral blood of humanized mice via flow cytometry, which confirmed humanization. One week after the treatment, on the irradiated side, (D) demonstrated lowest tumor volume, highest number of human PBMCs, highest % of CD8+ human (cytotoxic) T-cells, highest % of CD45RO+CD4+ human (memory) T-cells, and lowest % of PD-1+CD8+ human (exhausted) T-cells in the tumors via flow cytometry, highest IFN-gamma in the tumors via qRT-PCR, and highest CD8+ human (cytotoxic) T-cells via immunohistochemistry, compared to the other five groups with statistical significance. Of note, on the nonirradiated side, a similar trend was observed with D) harboring the smallest tumor compared to the others (P=0.09), suggesting the abscopal effect. Finally, there were no statistically significant differences amongst (A) humanized NSG mice with isotype-control antibodies, (E) naïve NSG mice with isotype-control antibodies, and (F) naïve NSG mice with anti-PD-1 antibodies on either sides, indicating that HLA-partially-mismatched immune cells derived from the fetus donor were not able to eradicate U-CH1 chordoma cells. Conclusions: We demonstrated that this humanized mouse model could be a revolutionary platform to investigate IT against rare cancers such as chordomas, where murine equivalent cell lines are not available to date, which hinders us from utilizing syngeneic or transgenic mouse models to study IT. The direct synergistic effect between IT and RT against chordoma as well as the potential abscopal effect was observed, evidenced by lowest tumor volume and highest cytotoxic T-cells and memory T-cells. Citation Format: Wataru Ishida, Kyle L. McCormick, Aayushi Mahajan, Eric Feldstein, Michael Lim, Jeffrey N. Bruce, Peter D. Canoll, Sheng-fu L Lo. Investigating in vivo synergistic effect of checkpoint blockade and radiation therapy against chordomas in a humanized mouse model [abstract]. In: Proceedings of the Fourth CRI-CIMT-EATI-AACR International Cancer Immunotherapy Conference: Translating Science into Survival; Sept 30-Oct 3, 2018; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2019;7(2 Suppl):Abstract nr B165.

Objective To explore the characteristics of CD4+ memory T cell in children with influenza A (H1N1) vaccination. Methods According to the voluntary principle, 31 children with influenza A (H1N1) vaccination for over 47 months were selected.Their lymphocytes were isolated and joined with the influenza A (H1N1) vaccine for cultivation, which was taken as the experimental group, and the control group was treated with no vaccine.The flow cytometry was used to detect the expression of cell surface molecules. Results The results showed that the proportion of CD4+ T cells in PBMC was 29.85% in the experimental group and 39.00% in the control group; the proportion in the experimental group was lower than that in the control group.The proportions of CD4+ naive T cells in the experimental group and in the control group were both more than 70% (P>0.05); the proportion of CD4+ memory T cells in the experimental group was 28.54% and 25.52% in the control group(P>0.05). Memory T cells were divided into central and effector memory T cells.Detection results of two subsets: The proportions of CCR7 and CD62L single positive memory T cell subsets in the experimental group were 72.52% and 29.85%, respectively, their proportions in the control group were 84.0% and 93.44%; the proportions in the experimental group were significantly lower than those in the control group (P<0.05). Conclusion This study shows that CD4+ T cells are dominant cells in 31 children, and the proportion of naive T cells (CD45RA+ ) is high.H1N1 influenza vaccine can induce production of antigen-specific memory CD4+ T cells, but few in number; wherein, most of them are central memory CD4+ T cells, furthermore, the numbers of CCR7 and CD62L single positive memory T cells are small. Key words: Children; H1N1 influenza vaccine; Memory T cell; Immune memory

Immunological memory is a key feature of adaptive immunity. It provides the organism with long-lived and robust protection against infection. The important question is whether cyclophosphamide (CP), as immunosuppressive agent used in cancer therapy and in some autoimmune diseases, may act on the memory T-cell population. We investigated the effect of CP on the percentage of central memory T cells (TCM) and effector memory T cells (TEM) in the mouse model of CP-induced immunosuppression (8-10-week-old male C57BL/6 mice CP treated for 7 days at the daily dose of 50 μg/g body weight [bw], manifested the best immunosuppression status, as compared to lower doses of CP: 10 or 20 μg/g bw). The CP induced a significant decrease in the percentage of CD8+ (TCM), compared to nonimmunosuppressed mice. This effect was not observed in the case of CD4+ TCM population. The percentage of gated TEM with CD4 and CD8 phenotype was significantly decreased in CP-treated mice, as compared to the control ones. Taken together, the above data indicate that CP-induced immunosuppression in mice leads to a reduction in the abundance of central memory cells possessing preferentially CD8+ phenotype as well as to a reduction in the percentage of effector memory cells (splenocytes both CD4+ and CD8+), compared to the cells from nonimmunosuppressed mice. These findings in mice described in this article may contribute to the understanding of the complexity of the immunological responses in humans and extend research on the impact of the CP model of immunosuppression in mice and memory T-cell populations.

Alloantigen-Activated Human T-Cells Increase Extracellular Fatty Acid Uptake and Intracellular Lipid Metabolism during Xenogeneic Gvhd

Objective To illustrate the distribution of two memory T cells subsets CD45RO+ CD62L+ CCR7+ T cells (central memory T cells, TCM) and CD45RO+ CD62L-CCR7-T cells (effector memory T cells, TEM) in tumor infiltrating lymphocytes (TILs) of hepatocellular carcinoma (HCC) patients. Methods Polychromatic flow cytometry (PFC) was used to test the cell surface molecules of infiltrating lymphocytes in liver, including CD3, CD4, CD8, CD45RO, CD62L and CCR7. Results The proportions of CD4+ TCM subsets in CD4+ CD45RO+ Tm in HCC tissues, adjacent liver tissues and normal liver tissues were (0.25±0.17)%, (0.29±0.22)% and (0.37±0.33)%, and there were no significant differences (P>0.05). The proportions of CD4+ TEM subsets in CD4+ CD45RO+ Tm in HCC tissues, adjacent liver tissues and normal liver tissues were (89.7±7.4)%, (98.7±0.6)% and (95.5±3.7)%, respectively. Paired comparison showed that the proportions of CD4+ TEM subsets in HCC tissues were significantly lower than those in adjacent liver tissues and normal liver tissues (P<0.05). The proportions of CD8+ TCM subsets in CD8+ CD45RO+ Tm in HCC tissues, adjacent liver tissues and normal liver tissues were (0.41±0.31)%, (0.55±0.60)% and (0.26±0.25)%. Paired comparison showed that the proportions of CD8+ TCM subsets in adjacent liver tissues were significantly higher than those in normal liver tissues (P<0.05). The proportions of CD8+ TEM subsets in CD8+ CD45RO+ Tm from HCC tissues, adjacent liver tissues and normal liver tissues were (92.4±4.3)%, (98.1±0.7)% and (98.2±1.1)%. Paired comparison showed that the proportions of CD8+ TEM subsets in HCC tissues were significantly lower than those in adjacent liver tissues and in normal tissues (P<0.05). Conclusions Comparing to the TCM,TEM subsets including CD4+ and CD8+ , accounted for the majority of tumor infiltrating CD45RO+ Tm in HCC patients. The proportions of CD4+ TEM and CD8+ TEM subsets in CD45RO+ Tm from tumor tissues were significantly lower than those in non-cancerous tissues. However, the proportions of CD4+ TCM and CD8+ TCM subsets in CD45RO+ Tm from tumor tissues were the same as those from non-cancerous tissues. Key words: Central memory T cells; Effector memory T cells; Tumor infiltrating lymphocytes; Hepatocellular carcinoma

Human T-cell memory consists mainly of unexpanded clones

DNA Vaccines Encoding Ii-PADRE Generates Potent PADRE-specific CD4+ T-Cell Immune Responses and Enhances Vaccine Potency

A Comparison of Human CMVpp65-Specific Central Memory and Effector Memory CD8 T-Cells When Stimulated in-Vivo with IL-2 or IL-15/IL-15Rα Complex in a Murine Xenograft Model of Adoptive Cell Therapy

Combined mTOR Inhibition and OX40 Agonism Enhances CD8+ T Cell Memory and Protective Immunity Produced by Recombinant Adenovirus Vaccines

CRISPR Activation Screen to Optimize Chimeric Antigen Receptor (CAR) T Cell Immunophenotype

Idiosyncratic drug reactions are extremely severe and are not attributed to the normal pharmacology of a drug. Elucidating their toxicological mechanisms have been challenging. In recent years, idiosyncratic drug-induced liver injury (iDILI) has been linked with the immune system proposing that tissue damage may be mediated by the cytotoxicity of T-cells. Furthermore, several associations have been found with iDILI and gene polymorphisms which code for human leukocyte antigen (HLA). The hapten model of drug hypersensitivity proposes that a drug-modified peptide presented by HLA may interact with T-cells. Therefore, drug-protein adducts may be uniquely placed to interact with risk HLA alleles. However, this link has not yet been defined experimentally. The aims of this thesis were to define the role of T-cells in several forms of iDILI, to define the chemistry of drug-protein adducts and ultimately to link these adducts to risk HLA alleles and T-cell activation. The antibiotic treatment co-amoxiclav is the most common cause of iDILI and it has been linked with several HLA alleles including the DR15 haplotype. Three patients with co-amoxiclav-induced liver injury were recruited and PBMCs were isolated to examine the presence of drug-reactive T-cells in the memory T-cell population. Amoxicillin-reactive T-cells were characterised from all three patients displaying a mixture of T-cell phenotypes. Furthermore, T-cells were activated in a processing dependent manner suggesting the potential for hapten mediated T-cell activation. However, T-cells were not restricted to any specific HLA allele. Amoxicillin-protein adducts were detected, using mass spectrometric and proteomics methods in the serum of drug tolerant patients where the covalent modification of several lysine residues was detected on HSA. When incorporated into T-cell assays, amoxicillin-modified HSA was not able to activate patient T-cells. These reactions were therefore investigated at the peptide level to elucidate the interaction of amoxicillin-protein adducts with specific HLA alleles. Drug-modified peptides were designed containing lysine to bind amoxicillin and anchor residues to bind DR15 risk HLA alleles to study haptens in co-amoxiclav iDILI. Peptides were purified to remove free drug and were fully characterised using mass spectrometry. Positional derivatives were generated where the amoxicillin-modified lysine was placed in different positions on the peptide backbone. Upon examining the memory T-cell population from a co-amoxiclav patient who was positive for DR15, amoxicillin-modified peptide-specific T-cells were detected. T-cells responded with specificity for the location of amoxicillin modification on the peptide backbone. Furthermore, amoxicillin-modified peptides were restricted to risk HLA alleles of the DR15 haplotype. The drugs terbinafine and ticlopidine are associated with the HLA-A*33 serotype in iDILI patients, despite their unique pharmacological mechanism and chemistry. We sought to define the mechanistic potential of these drugs to activate the immune system and elucidate whether they follow similar or different toxicological mechanisms. Terbinafine can be bioactivated to the chemically reactive aldehyde TBF-A. We were unable to isolate TBF-A or terbinafine reactive T-cells; therefore, we sought to define the chemistry of TBF-A to be able to eventually incorporate drug-protein adducts into future T-cell assays. TBF-A was shown to bind to small molecules, whole proteins and model peptides via adducts with several amino acids including cysteine, histidine and tyrosine. To investigate T-cell responses to ticlopidine, naive T-cells were primed from healthy donors positive for the risk HLA allele. T-cells isolated from positively primed donors demonstrated ticlopidine-specific T-cell activation. Furthermore, T-cells were activated in a direct, processing independent manner and were restricted to the risk HLA allele. Several drugs can cause severe hypersensitivity reactions without the need for presentation via specific HLA alleles. Elucidating the chemistry of adducts in these haptens may help define why compounds such as the nitroso metabolite of sulfamethoxazole, SMX-NO, are such potent immunogens. Here, we sought to detect the presence of SMX-NO adducts in drug-tolerant patients. An in vitro characterisation of SMX-NO adducts revealed promiscuous binding to cysteine, lysine and tyrosine in addition to the detection of oxidised lysine residues. Similar adducts were also detected in patient sera. Herein, these data have examined several reactive drugs or metabolites and their associated drug-protein adducts. We have demonstrated that adducts can be presented as drug-modified peptides to activate T-cells in a highly specific HLA-restricted manner. Furthermore, the formation of these adducts may drive several co-signals required for the priming and activation of T-cells. These studies lay the foundations for future studies to elucidate precise protein targets implicated in drug hypersensitivity.

Surface expression of the IL-2 receptor α-chain (CD25) has been used to discriminate between CD4(+) CD25(HI) FOXP3(+) regulatory T (Treg) cells and CD4(+) CD25(NEG) FOXP3(-) non-Treg cells. However, this study reports that the majority of resting human memory CD4(+) FOXP3(-) T cells expresses intermediate levels of CD25 and that CD25 expression can be used to delineate a functionally distinct memory subpopulation. The CD25(NEG) memory T-cell population contains the vast majority of late differentiated cells that respond to antigens associated with chronic immune responses and are increased in patients with systemic lupus erythematosus (SLE). In contrast, the CD25(INT) memory T cells respond to antigens associated with recall responses, produce a greater array of cytokines, and are less dependent on costimulation for effector responses due to their expression of CD25. Lastly, compared to the CD25(NEG) and Treg-cell populations, the CD25(INT) memory population is lost to a greater degree from the blood of cancer patients treated with IL-2. Collectively, these results show that in humans, a large proportion of CD4(+) memory T cells express intermediate levels of CD25, and this CD25(INT) FOXP3(-) subset is a functionally distinct memory population that is uniquely affected by IL-2.