CD38 identifies pre-activated CD8+ T cells which can be reinvigorated by anti-PD-1 blockade in human lung cancer

Tumor Immune Estimation Resource (TIMER, cistrome.shinyapps.io/timer) is a new website that involves 10,897 samples across 39 cancer types from The Cancer Genome Atlas (TCGA) for estimating the level of immune infiltrates and it provides six major analytic modules to deeply excavate molecular characterization of tumor-immune interactions including Gene module, Survival module, Mutation module, SCNA module, Diff Exp module and Correlation module. We first analyzed CD38 expression in different types of cancer by using Diff Exp module. And then, we explore the clinical relevance of LUAD and LUSC via Survival module. Next, we find out the correlation between the level of CD38 expression and immune infiltrates, including B cells, CD4⁺ T cells, CD8⁺ T cells, neutrophils, macrophages and dendritic cells, via gene modules in diverse cancer types. The gene expression level converts into log2 RSEM.

Tumor (T, homogeneous cellularity, without foci of necrosis), paired normal lung tissues (N) and some fresh peripheral blood (PB) were obtained from patients with NSCLC who underwent surgical resection at the Second Affiliated Hospital, Zhejiang University School of Medicine. Autologous peripheral blood was collected before surgery. Normal autologous tissue was obtained from a macroscopically normal part of the excised pulmonary lobe, at least 5 cm away from the tumor. None of the patients had received radiotherapy or chemotherapy before operation.

A549 (human lung carcinoma) Cells were grown in Roswell Park Memorial Institute (RPMI) 1640 growth medium supplemented with 10% fetal bovine serum (FBS), 1% penicillin and streptomycin at 37 °C with 5% CO₂ and maintained at a confluence of 70–80%.

Freshly excised tissues were cut into small pieces and then digested in RPMI 1640 medium containing 2% FBS, type I collagenase (1 mg/ml), type IV collagenase (1 mg/ml) and hyaluronidase (10 ng/ml) for 1 h at 37 °C. PB lymphocytes collected after lysing on lysing solution (BD Pharm Lyse™).

The antibody CD3 (UCHT1), CD8a (RPA-T8), CD38 (HB-7), CD101 (BB27), PD-1 (EH12.2H7), CD103 (Ber-ACT8), IFN-γ (B27), TNF-α (MAb11), Granzyme B (QA16A02), perforin (dG9), CD69 (FN50), HLA-DR (L243) were purchased from Biolegend. We use mechanic dispersion and enzymatic digestion to prepare the single cell of normal and tumor tissues for extracellular staining of immune markers. For blocking nonspecific binding and stained with different combinations of fluorochrome-coupled antibodies, we preincubated fresh tissue cells (1 × 10⁶/ml) in a mixture of PBS, 2% fetal calf serum and 0.1% (w/v) sodium azide with FcgIII/IIR-specific antibody. And then, we followed the manufacturer’s protocol after 12 h incubation in the presence of Leukocyte Activation Cocktail (BD Pharmingen) to perform the intracellular staining. Fluorescence data were collected on a FACSCanto II system (BD Biosciences) and analyzed using FlowJo software (Tree Star).

To investigate the cytokine secretion of tumor-infiltrating CD8+ T cells, single cells of normal and tumor tissues were cultured in the presence of anti-CD3 (1 ug/ml, Biolegend) and anti-CD28 (1 ug/ml, Biolegend) or PMA and ionomycin or anti-PD-1 (10 ug/ml, Biolegend). After a while, single cells of normal and tumor tissues were collected for the perforin, Granzyme B, TNF-α and IFN-γ assay.

To visualize the tumor-killing power of CD3+ CD8+ CD38+ T cells, CD3+ CD8+ CD38+ T cells and CD3+ CD8+ CD38− T cells sorted by Aria II cell sorter (BD Biosciences) and 7-AAD was used to delete dead cells. The purity of all sorted cells was greater than 90%. The sorted cells were resuspended in a 96-well plate with 2 W medium per well. After standing for 24 h, the A549 cell line was added for co-cultivation according to E (effector cell):T (tumor cell line) = 2:1. After 6 h, use 7-AAD to detect the number of dead cells in A549.

Paraffin-embedded and formalin-fixed samples were cut into 5-μM sections, which were then processed for IF staining or IHC staining. First incubation with the CD103 (ABCAM) antibody, followed by HRP conjugated Goat Anti-Mouse IgG (Servicebio). Then, incubation with antibodies against human CD8, CD38 (ABCAM), followed by Cy5 conjugated Goat Anti-mouse IgG, FITC conjugated Goat Anti-Rabbit IgG (Servicebio). Images were acquired with a confocal microscopy (Zeiss LSM 710, Carl Zeiss, Dublin, CA).

Through analyzing the data of TIMER database, we found that B cells, CD8+ T cells, dendritic cells and CD38 high expression (Top 35%) were positive correlated with the survival of LUAD, and CD4+ T cells were positive correlated with the survival of LUSC (Fig. 1a). Moreover, CD38 expression was positive correlated with infiltrating levels of B cells, CD8+ T cells, CD4+ T cells, macrophages, neutrophils and dendritic cells in LUAD and LUSC, but negatively related to tumor purity (Fig. 1b). For further validation, we collected samples from 45 patients with NSCLC and analyzed the expression of CD38 in CD8+ T cells by flow cytometry (Fig. 1c, d). The information of the patients was shown in supplementary Table 1. In accordance with expectation, we found the expression level of CD38 in CD8+ T cells and the proportion of CD38+ CD8+ T cells in T cells were significantly increased in tumor (T) compared with paired normal lung tissues (N) and peripheral blood (PB) (Fig. 1e). These findings suggest that NSCLC patients with increased expression level of CD38 have a better clinical prognosis, and the CD38+ CD8+ T cells enriched in the tumor, are likely to be one of important candidates involved in anti-tumor-immune response.

To further investigate the cytotoxicity and cytokine secretion of CD38+ CD8+ T cells and CD38− CD8+ T cells in natural condition, we cultured single cell suspensions from peripheral blood (n = 6), adjacent normal lung tissues (n = 8) and tumor (n = 8) of NSCLC patients without TCR agonist in vitro and detected the secretion of cytotoxic molecules and cytokines through flow cytometry (Fig. 2a, b). There is no significant difference in the ability to secrete TNF-α between CD38+ CD8+ T cells and CD38− CD8+ T cells in tumor (T), paired normal lung tissues (N) and peripheral blood (PB) (Supplement Figure 1A, B). Tumor-infiltrating CD38− CD8+ T cells had a stronger IFN-γ secretion capacity than that in peripheral blood, but not CD38+ CD8+ T cells (Fig. 2c and Supplement Figure 1C). It is interesting that CD38+ CD8+ T cells in tumor had a stronger secretion of IFN-γ than CD38− CD8+ T cells, but not in normal tissues and peripheral blood (Fig. 2d and Supplement Figure 1D). There is no significant difference in perforin secretion between CD38+ CD8+ T cells and CD38− CD8+ T cells in all tissues (Supplement Figure 1E, F). Comparing with peripheral blood and tumors, the expression level of Granzyme B was higher in both CD38− CD8+ T cells and CD38+ CD8+ T cells (Fig. 2e). However, no matter which tissue it is derived from, the secretion capacity of Granzyme B was significant higher in CD38+ CD8+ T cells than CD38− CD8+ T cells (Fig. 2f). Furthermore, using a co-cultured experiment in vitro (Supplement Figure 2A, B, C, D), we found that tumor-infiltrating CD38+ CD8+ T cells had a stronger tumor-killing effect than CD38− CD8+ T cells (Fig. 2g, h). Taken together, despite the universal decreasing tendency of cytotoxicity and cytokine secretion of CD38+ CD8+ T cells and CD38− CD8+ T cells in tumor compared with normal tissue, our results found that tumor-infiltrating CD38+ CD8+ T cells are advantages in cytotoxicity, cytokine secretion and tumor-killing ability than tumor-infiltrating CD38− CD8+ T cells in the natural condition which may result from pre-activation.

Next, we investigated the cytotoxicity, cytokine secretion of CD38+ CD8+ T cells and CD38− CD38+ T cells in the fully activated condition induced by PMA and Ionomycin (Fig. 3a, b). We found that the TNF-α secretion in tumor-infiltrating CD38− CD8+ T cells was significantly decreased when compare with CD38− CD8+ T cells in paired normal tissues (Fig. 3c), but not in tumor-infiltrating CD38+ CD8+ T cells (Supplement Figure 3A). In peripheral blood, the TNF-α secretion capacity of CD38− CD8+ T cells was stronger than that of CD38+ CD8+ T cells (Fig. 3d), but not in normal tissues or tumors (Supplement Figure 2B). In terms of the secretion capacity of IFN-γ, CD38+ CD8+ T cells and CD38− CD8+ T cells in tumor were decreased in IFN-γ secretion than their counterparts in normal tissues (Fig. 3e). It was also interesting that the IFN-γ secretion of CD38+ CD8+ T cells from peripheral blood was weaker than that of CD38− CD8+ T cells, while the secretion of IFN-γ from normal tissues and tumors was stronger than that of CD38− CD8+ T cells (Fig. 3f). Then, we found tumor-infiltrating CD38+ CD8+ T cells and CD38− CD8+ T cells both were decreased in the secretion of Granzyme B compared with normal tissue and peripheral blood (Fig. 3g). Consistent with previous results, the Granzyme B secretion capacity of CD38+ CD8+ T cells was stronger than that of CD38− CD8+ T cells in all types of tissue (Fig. 3h). CD38+ CD8+ T cells and CD38− CD8+ T cells showed advantages in secrete perforin compared with their counterparts in both normal tissue and peripheral blood (Fig. 3i, Supplement Figure 2C). However, there was no difference between the perforin secretion between tumor-infiltrating CD38+ CD8+ T cells and tumor-infiltrating CD38− CD8+ T cells (Supplement Figure 2C). These results verified the advantages of tumor-infiltrating CD38+ CD8+ T cells in IFN-γ and Granzyme B secretion than tumor-infiltrating CD38− CD8+ T cells in fully activated condition.

In our study, we also found that tumor-infiltrating CD38+ CD8+ T cells and CD38− CD8+ T cells have significantly decreased of IFN-γ and Granzyme B secretion compared to adjacent normal tissues (Figs. 2e and 3e, g). We proposed that the functional impairment of CD38+ CD8+ T cells in tumor might be related to the occurrence of exhaustion. To verify our hypothesis, we analyzed the expression of PD-1 and CD101 in tumor-infiltrating CD38+ CD8+ T cells and CD38− CD8+ T cells by flow cytometry (Fig. 4a, b). We found that the expression of PD-1 significantly increased in tumor-infiltrating CD38+ CD8+ T cells compared with adjacent normal tissues and peripheral blood (Fig. 4c). While in CD38− CD8+ T cells, the expression of PD-1 in tumor and normal tissues was higher than that in peripheral blood, but not increased in tumor-infiltrating CD38− CD8+ T cells compared with CD38− CD8+ T cells in normal tissue (Fig. 4c). Moreover, the increased expression of PD-1 in CD38+ CD8+ T cells compared with CD38− CD+ T cells was only observed in tumor, but not in peripheral blood and normal tissues (Fig. 4d). In addition, CD3+ CD+ T cells and CD38− CD+ T cells derived from tumors and normal tissues had higher expression of CD101 than those derived from peripheral blood (Fig. 4e). Moreover, in all types of tissues, the CD101 expression of CD3+ CD+ T cells is higher than that of CD38− CD+ T cells (Fig. 4f).

Next, we compared the degree of activation of tumor-infiltrating PD-1+ CD3+ CD+ T cells with tumor-infiltrating PD-1+ CD38− CD+ T cells using CD69 and HLA-DR (Fig. 4g, i), which are the markers of T cell activation. It is interesting that tumor-infiltrating PD-1+ CD3+ CD+ T cells expressed higher levels of both CD69 and HLA-DR than tumor-infiltrating PD-1+ CD38− CD+ T cells (Fig. 4h, j). These findings suggest that CD3+ CD+ T cells in tumor of NSCLC had an exhausted phenotype which may result from the advantages in pre-activation.

The tissue-resident CD+ T cell, mainly identified by the CD103 marker, is an important subgroup exerting anti-tumor efficacy in tumor-immune microenvironment. In order to further understand the reasons for superior functional performance of CD3+ CD+ T cells, we analyzed the expression of CD103 on CD3+ CD+ T cells by flow cytometry and immunofluorescence (Fig. 5a, b). We found the number of CD3+ CD103+ CD+ T cells highly increased in tumors and were significantly higher than normal tissues and peripheral blood, while in CD38− CD+ T cells, the expression of CD103 in normal tissues and tumor were higher than that from peripheral blood, but there is no difference between the two (Fig. 5c, d). Moreover, immunofluorescence assay also showed the same results (Fig. 5b). Then, we activated tumor-infiltrating CD+ T cells through anti-CD3, anti-CD28, PMA and Ionomycin in vitro for 12 h (Fig. 5e). We found that CD3+ CD103+ CD+ T cells increased significantly after activation, and CD38− CD103− CD+ T cells significantly decreased in contrast. Meanwhile, the proportion of CD3+ CD103− CD+ T cells did not change (Fig. 5f). These results indicated that the increased CD103 expression in CD3+ CD+ T cells can be induced by activation and sustains their accumulation.

To further investigate the impact of CD103 expression on tumor-infiltrating CD3+ CD+ T cells, we stimulated tumor-infiltrating CD+ T cells through anti-CD3, anti-CD28, PMA and Ionomycin or anti-PD-1 in vitro for 12 h (Fig. 6a, b). We found that CD3+ CD103+ CD+ T cells secrete more IFN-γ compared with CD38− CD103− CD+ T cells without stimulation or with anti-CD3, anti-CD28 (Fig. 6c). And, CD3+ CD103+ CD+ T cells had a stronger ability to secrete TNF-α than both CD3+ CD103− CD+ T cells and CD38− CD103− CD+ T cells (Fig. 6d). About the Granzyme B expression, CD3+ CD103+ CD+ T cells and CD3+ CD103− CD+ T cells were higher than CD38-CD103− CD+ T cells (Fig. 6e). As for perforin, the secretion capacity in CD3+ CD103+ CD+ T cells was higher than CD3+ CD103− CD+ T cells and CD38− CD103− CD+ T cells (Fig. 6f). Intriguingly, CD3+ CD103+ CD+ T cells and CD3+ CD103− CD+ T cells can both be reinvigorated by anti-PD-1, and CD3+ CD103+ CD+ T cell showed stronger response (Fig. 6g–i). These results suggested that increased CD103 expression may sustain the cytotoxicity and cytokine secretion of tumor-infiltrating CD3+ CD+ T cells as well as response to anti-PD-1.