miRNA-based secretome therapeutics (SYN, TA1, IA1 and IA2) have been successfully produced in our laboratory based on allorecognition (± polymer-mediated immunocamouflage) based MLR.[3, 4, 25] As demonstrated in our previous studies, the effects of our acellular secretome products to pan T cell (PHA or anti-CD3/CD28) on CD3+ T cell proliferation (Fig. 2, center circle) as well as T regulatory (CD4+Foxp3+CD25+) and Th17 (CD4+IL17+) cells were distinct.[3, 4, 25] To further define the proliferation pattern within the CD3+ cells, CD4± and CD8± subsets and the CD:CD8 ratio were further analyzed (Fig. 2). As shown, resting PBMC demonstrated minimal proliferation and a CD4:CD8 ratio of 1.7 ± 0.1. In contrast, the pan T cell activators anti-CD3/CD28 and PHA induced massive CD3+ cell proliferation and altered the CD4:CD8 ratio (Fig. 2A). Despite both PHA and anti-CD3/CD28 being Pan T cell activators, there were differences in how these agents modulated the CD4/CD8 differentiation. PHA, but not anti-CD3/CD28, significantly increased the CD8+ population while simultaneously decreasing the CD4+ population resulting in a significant (p < 0.0001) decrease of CD4:CD8 ratio relative to resting PBMC (1.7 to 0.9). mAb activation also decreased the ratio but not as dramatically as PHA. Alloactivation, in comparison to the highly potent pan T cell activation, induced a more moderate proliferation of CD3+ cells relative to resting PBMC.[3, 4, 25] The reduction in proliferation arose consequent to < 10% of T cells within a population typically being capable of allorecognition (Fig. 2B).[38, 39] Alloactivation similarly decreased (p < 0.01) the CD4:CD8 ratio.
The secretome products showed significant variability (at Day 10) in their effects when used to activate resting PBMC. As expected, resting PBMC treated with the SYN secretome were virtually identical to the resting PBMC with regards to both proliferation, subset analysis and CD4:CD8 ratio. Similarly, the tolerogenic TA1 preparation showed minimal proliferation and no substantive changes in the subset differentiation or the CD4:CD8 ratio. Of note, the small increase in CD3+ proliferation (2.3 ± 0.1%) supports earlier observations showing increased proliferation/differentiation of regulatory T cells (Treg; CD4+Foxp3+) in TA1-treated PBMC.[21] However, in contrast to the SYN and TA1 products, IA1 and IA2 showed significant variation from both the resting PBMC and from each other (Fig. 2C). IA1, derived from a control MLR, significantly increased the total CD3+ cell proliferation and decreased the relative abundance of CD4+ cells (38.8 ± 3.4% versus 56.0 ± 1.5% for resting PBMC). Consequent to this change, the IA1 CD4:CD8 ratio was significantly reduced (1.0 ± 0.1; p < 0.001) suggestive of a pro-inflammatory state and similar to that noted with the pan T cell activators. Interestingly, the cancer cell (HeLa) stimulated biologic IA2 while inducing a similar level of CD3+ cell proliferation, showed dramatically different phenotype distribution (Fig. 2C). In contrast to IA1, which reduced the CD4:CD8 ratio, IA2 significantly increased the ratio relative to both IA1 and the resting PBMC (2.1 ± 0.4, 1.0 ± 0.1, and 1.7 ± 0.1; respectively) consequent to a decrease in CD8+ cells. These findings further support our previous study suggesting that the anti-HeLa effects of IA2 treated PBMC were distinct from that of IA1 treated PBMC.[4, 25]
Interestingly, both IA1 and IA2 activation also resulted in a significant expansion (p < 0.05) of double negative (CD4−CD8−) T cells. These double negative cells, while poorly understood, have been implicated in both inflammation and as regulatory cells.[40] Under certain activating stimuli, CD4−CD8− T cells can display the phenotype of effector cells that are capable of producing pro-inflammatory cytokines e.g., IFN-γ, IL-17A, and the phenotype of suppressor cells that secrete immune regulatory cytokine (e.g., IL-10). Consequently, CD4−CD8− T cells have been described in promoting neuroinflammation after ischemic stroke, [41] as well as preventing autoimmune diseases and graft-versus-host disease (GvHD).[42–45] Interestingly, a recent study revealed the potential role of CD4−CD8− T cells in limiting alloreactive immune response by suppressing the CD4 T cell functionality.[46] Surprisingly perhaps, TA1 treatment reduced the CD4−CD8− population despite previous research documenting its potent tolerogenic activity in a murine model of Type 1 diabetes.[3]
In contrast, double positive (CD4+CD8+) T cells were reduced in the TA1, IA1 and IA2 (0.7 ± 0.0, 3.0 ± 0.5, and 1.2 ± 0.1%, respectively) activated cells relative to Pan T cell activation (PHA: 7.0 ± 2.3%; and Anti-CD3: 5.7 ± 0.5%), and alloactivation (6.0 ± 0.7%). The biologic functions of CD4+CD8+ T cells remain unclear with, perhaps most, studies reporting a pro-inflammatory role in cancers though some evidence of an immunosuppressive role have been reported.[47–50] CD4+CD8+ T cells are increased in urologic cancers patients and found to be high type-2 cytokine producers favoring a Th2 response in vitro while inhibiting Th1 cells known to play a crucial role in anti-tumor immunity.[50] Hence, the decrease in the CD4+CD8+ cells post secretome (TA1, IA1 and IA2) activation, relative to Pan T cell or allo- activation, may have clinical benefit and be indicative of the secretomes more ‘restrained’ immunoactivation.
To determine how these differential T cell activation strategies (i.e., Pan T cell, allorecognition and secretome) affected resting human PBMC, the intracellular expression of 84 miRNA involved in immunopathology pathways were previously assessed.[3, 4, 25] To further investigate the differential effects of the three T cell activation strategies on resting PBMC, the 'relative pattern of miRNA expression' was examined using subset of thirteen differentially expressed miRNA (Fig. 3A; miRNA were selected via clustergram heatmap[4] and/or log2 fold change analysis). The putative/described functions for these miRNA are summarized in Supplementary Table 1S.[3, 4, 25] However, it is important to note that the ‘putative’ functions of the distinct miRNA can vary significantly depending on the biological (e.g., prostate versus T cell) model used. Moreover, consequent to the low genetic fidelity of a single miRNA, we propose that the most informative approach is the analysis of the differential activation strategies on the relative 'pattern of miRNA expression' across a number of miRNA (Fig. 3).
As shown, the Pan T cell activators PHA and anti-CD3/CD28 demonstrated significant similarities in their 'patterns of miRNA expression' and CD3+ proliferation (Fig. 3B). While these Pan T cell activators are commonly used as activation surrogates for allorecognition and/or to enhance cytokine expression levels prior to flow analysis (anti-CD3/CD28), comparison of the miRNA expression profiles show significant variances from the allorecognition expression profile (Control MLR; Fig. 3B). Surprisingly, minimal differences within the group of 13 miRNA were noted between the control- and mPEG- MLRs despite significant differences in T cell proliferation. However, in the context of the overall 84 miRNA screened, distinct patterns are quite apparent (Supplemental Table 2S and Scott et al.[25]). In contrast to both Pan T cell activators and allorecognition, secretome (IA1, IA2 and TA1) activation gave rise to dramatically reduced levels of proliferation and subtler changes, though highly distinct, changes in miRNA expression (Fig. 3). The IA1 secretome (Proliferation of 12.2 ± 1.2%) yielded a distinct miRNA expression profile from those of the Pan-T cell activators; though a muted similarity in the peaks and troughs is discernable. In contrast, the tolerogenic TA1 secretome induced minimal proliferation (2.3 ± 0.1%) and produced a miRNA pattern that varied significantly from the pro-inflammatory and proliferation inducing Pan T cell and IA1 activators (Fig. 3). Indeed, previous clustergram heatmap analysis showed that miRNA expression induced by TA1 resembled resting and SYN treated PBMC but induced a potent Treg-mediated tolerogenic effect both in vitro and in vivo.[3, 4, 21, 25] Interestingly, the HeLa-PBMC manufactured IA2, while exerting a proliferative effect (10.7 ± 0.5%) similar to IA1, varied significantly from IA1 miRNA pattern (Fig. 3). Importantly, the differential patterns of expression (Fig. 3B) of the IA1, IA2 and TA1 miRNA were associated with differential biological effects on the naive PBMC with TA1 inducing systemic tolerance (in vitro and in vivo) and IA1 enhancing PBMC-mediated inhibition of cancer cell growth while IA2 exhibiting direct toxicity (apoptosis) of cancer cells.[3, 4] Importantly, the distinct expression patterns of miRNA between the Pan T cell activators (PHA and anti-CD3/CD28), alloactivation the secretome products induced activation translated into dramatically different biological responses.[3, 4, 25]
Consequent to our interests in using the IA1 secretome to activate resting PBMC to more efficiently cancer cells while preventing adverse events (e.g., cytokine storms), we more directly compared the miRNAs expression profile of IA1 to pan-T cell (anti-CD3/CD28; Fig. 4A), TA1 (Fig. 4B) and IA2 (Fig. 4C) activation via volcano plot analyses. Volcano plot analyses visualizes the differential miRNA data based on log scale changes and allows for statistical comparison of the expression of discreet miRNA between two samples (e.g., IA1 versus IA2) – but largely misses out on the overall PATTERN of changes seen with clustergram heatmaps[4] and Log2 fold change. As noted in Fig. 4A, distinct differences are noted between IA1 and anti-CD3/CD28. IA1 significantly (p < 0.05) upregulated the expression of miR-125b-5p and miR-451a relative to anti-CD3/CD28, while miR-18a-5p, miR-17-5p, miR-20a-5p and miR-135b-5p were downregulated. Similar to Fig. 3 multiple other miRNA were also differentially expressed between IA1 and anti-CD3/CD28 activation though they did not reach significance in the volcano plot analyses (though if compared to resting PBMC they are different). Interestingly, the miRNA expression profiles between IA1 and TA1 were not statistically significantly different (Fig. 4B), though, as also seen in Fig. 3, miR-298, miR-214-3p, miR-302a-3p and miR-206 were over-expressed in IA1 relative to TA1. Finally, the expression of miR-149-5p and miR-18b-5p were significantly (p < 0.05) upregulated in PBMC treated with IA1 when compared to the same donor PBMC treated with IA2 (Fig. 4C).
In sum, previous clustergram,[4] and our current studies utilizing log2 fold change and volcano plot analyses demonstrate the differential activation strategies yielded dramatically different miRNA expression profiles that in turn resulted in significant differences in T cell activation and subset differentiation. In order to better understand these differences, an integrative Venn diagram analysis was done using all three sets of data (Fig. 5) in order to differentially compare the pan T cell, allorecognition and secretome activation. As demonstrated, Pan T cell activation using PHA and anti-CD3/CD28 yielded similar, though not identical, changes in miRNA expression (solid circles = over expression; dashed circles = reduced expression; overlap are miRNA in common). For further comparison purposes, we averaged the miRNA expression profile and proliferation rates of PHA and anti-CD3/CD28 to represent the efficacy of pan T cell activation strategy. In contrast to Pan T cell activation, the miRNA changes induced by allorecognition were much more discreet (relative to resting PBMC) and highly limited when compared to the Pan T cell activators. Moreover, allorecognition resulted in a significant reduction in cell proliferation (Pan T: 86.3% versus 30.9% for Allorecognition). Similar to the allorecognition response, the allo-derived IA1 secretome also reduced the miRNA response pattern relative to Pan T cell activation and, not surprisingly, was similar to the pattern of expression observed in the alloresponse but with the increased expression of miR-298 and decreased expression of miR-206 and miR-214-3p. While some miRNA are in common to the Pan T cell, Allo, IA1 and IA2 pro-inflammatory responses (overlaps in Venn diagrams), some of these (e.g., miR-155-5p) are also implicated in the tolerogenic TA1 and mPEG-Allo responses. This again argues that the ‘pattern of miRNA expression’ (Fig. 3B) encompassing increases, decreases and static levels of multiple, rather than a specific (single or small number), of miRNA is crucial.
Importantly, the biomanufaturing process is of importance. This is most obvious in comparing the MLR vs mPEG-MLR and IA1 vs IA2 miRNA patterns. While IA1 and IA2 stimulate similar proliferative effects (12.2 ± 1.2 and 10.7 ± 0.5%, respectively), their impacts on CD4/CD8 differentiation (Fig. 2) were vastly different and, as observed in our previous study, IA1 and IA2 exhibited distinct biological activities and anti-cancer mechanisms.[3, 4, 25] Indeed, as shown in Fig. 5, PBMC pre-treated with IA1 and IA2 induced entirely different miRNA expression that resulted in vastly different responses to cancer cells. IA2 but not IA1, increased the expression of cellular miR-29b-3p which has been shown to promote cellular apoptosis in cancers. [51, 52] Consistently, miR-181a-5p, an oncogene in various tumors suppressing apoptosis, was downregulated by IA2, further supporting our previous findings that IA2 inhibited cancer cell proliferation via an apoptosis-associated mechanism.[53–55]
In conclusion, these studies demonstrate that pan T cell activators, alloactivation and secretome-based therapeutics induced differential patterns of miRNA expression in leukocytes, which governs/reflects significant differences in cell proliferation, differentiation and immunological activity. Pan T cell activators induced massive miRNA alteration profiles and T cell proliferation relative to resting cells. Allo-MLR demonstrated a more discriminatory alteration of miRNA expression relative to pan T cell activators, while mPEG-MLR diminished allorecognition related miRNA expression. Importantly, IA1 and TA1 secretome derived from allo- and mPEG-MLR respectively, exerted similar miRNA pattern of change and immunomodulatory efficacies to its origin MLR response. In contrast, resting cell generated secretome SYN had minimal effects on recipient resting PBMC. Of interest, the HeLa-MLR derived IA2 therapeutic exhibited distinct alterations to the leukocyte miRNA expression profile, suggesting an apoptosis-associated immunomodulatory and anti-cancer pathway.