In This Issue

The Foxn1 gene is expressed in thymic epithelial cells, as well as in hair follicles. Disruption of this gene leads to the loss of T cells and results in hairless (nude) mice. In this issue, Larsen et al. (p. 686) tease out thymic-specific regulatory control of Foxn1. A potential regulatory element (RE) in the first intron of Foxn1 was identified based on chromatin accessibility, sequence conservation, and histone modifications. Two knockout mouse strains, Foxn1^ΔA/ΔA and Foxn1^ΔB/ΔB, were created by deleting either the 5′ or the 3′ half of the first intron of Foxn1. Although both strains had normal fur growth, Foxn1^ΔB/ΔB mice had no anatomical thymic defects and T cell populations comparable to those of wild-type mice. Foxn1^ΔA/ΔA mice, although furry, contained no apparent thymus and displayed a loss of T cell populations similar to nude mice, suggesting that the RE is specific for expression of Foxn1 in the thymus. To further characterize the critical genetic sequence in the RE, the researchers removed accessible chromatin regions within the Foxn1^A region (Foxn1^ΔC, Foxn1^ΔD, and Foxn1^ΔE). Of these three knockouts, only Foxn1^ΔD mice exhibited a lack of a thymus and peripheral T cells, in addition to loss of MAIT, NKT, Vγ4⁺, and Vγ6⁺ cells from the liver and lungs. By crossing the Foxn1^ΔD mice with nude mice, researchers were able to show that the RE functions in cis and is specific to the expression of Foxn1, only in the thymus and not in hair follicles. Finally, in contrast to nude mice, Foxn1^ΔD mice maintain a healthy body weight and are born in normal Mendelian ratios. This new mouse could provide a more robust model to study thymic selection of T cells.

The K562 cell line is currently used to sustain the proliferation and numerical expansion of NK and chimeric Ag receptor engineered NK cells for use in adoptive cell therapy. In this issue, Vidard et al. (p. 676) exploited the use of cytokines and K562-derived artificial APCs (aAPCs) to elucidate mechanisms driving NK cell proliferation and effector function. For four consecutive weeks, high expression, but not low expression, of CD137L by aAPCs could sustain NK cell expansion in the presence of cytokines without an impact on the cytotoxicity of NK cells. Consistent with these observations, NK cell proliferation in the presence of parental K562 cells was increased by the addition of agonistic anti-CD137 Abs. Interestingly, plate-bound anti-CD137 Ab lost the ability to trigger NK cell proliferation, indicating mandatory cell-to-cell contact. Finally, studies demonstrated that a collaboration between IL-15 (either membrane bound or soluble), IL-21, and CD137L was necessary to trigger potent NK cell expansion using aAPCs. Thus, this study demonstrates a “three-signal” model of cell-to-cell contact, CD137L activation, and cytokine signaling to efficiently expand NK cells, observations that may be used to improve current protocols to expand NK cells for use in adoptive cell therapy.

Machine learning can be a useful tool when trying to analyze large, complex data sets. In an effort to identify baseline immune predictors that can discriminate between high and low responders following influenza vaccination, Tomic et al. (p. 749) developed Sequential Iterative Modeling “OverNight” (SIMON). The new approach allows comparison of 128 different algorithms to identify the ones that best fit any given data distribution, and therefore maximizes predictive accuracies. To address missing data values between studies, the first step of SIMON used a novel algorithm to compute shared features and similarity between donors and therefore rapid generation of 34 unique datasets that may be used for further analysis. SIMON was applied to each of the datasets and identified 19 for additional modeling. Overall, the automated process improved the performance of the models in all 19 datasets by 30–90%. By integrating filtering and evaluation of model performance, SIMON was used to identify features that enhance knowledge about Ab generation following influenza vaccination. SIMON identified both known and novel immune cell subsets that correlated with a robust Ab production in response to influenza vaccination. In particular, the IL-17A–producing CD8⁺ effector memory T cells were shown to correlate with robust Ab responses. In conclusion, SIMON provides a new tool by which researchers can increase model accuracy for analyzing large, heterogeneous datasets to identify new targets for therapeutics.

Survivors of sepsis can experience long-term innate and adaptive immunoparalysis that renders them susceptible to secondary complications. In this issue, Danahy et al. (p. 725) assessed if tumor development/growth is negatively impacted by sepsis-induced immunosuppression. Sepsis in mice following cecal ligation and puncture (CLP) increased susceptibility to B16 melanoma growth for more than 100 d after sepsis induction. Whereas CD8 T cells can partially control tumor growth in this model, survivors of CLP-induced sepsis displayed a reduced frequency of CD8 tumor-infiltrating lymphocytes (TILs) and effector cytokine production concomitant with increased tumor burden. The reduced frequency of functional CD8 TILs was attributed to minimal expression of MHC class I on B16 cells in sepsis survivors, which created an unfavorable environment for CD8 T cell recognition/activation/function. The postseptic environment reduced the number of CD8 TILs with high expression of activating/inhibitory receptors PD-1 and LAG-3 and also reduced proliferation, IFN-γ production, and survival of CD8 TILs. Checkpoint blockade therapy (anti–PD-L1/anti–LAG-3) was administered to postseptic tumor-bearing mice but did not reduce tumor growth when administered after CD8 TILs had become reduced in frequency. Alternatively, checkpoint blockade administered before tumors were observed significantly reduced tumor growth to levels seen in sham-treated controls. Thus, sepsis-induced immunoparalysis can diminish CD8 T cell–mediated immunity and suggests that early cancer screening is essential for individuals surviving sepsis.