Regulatory activity of azabisphosphonate-capped dendrimers on human CD4⁺ T cell proliferation enhances ex-vivo expansion of NK cells from PBMCs for immunotherapy

Natural Killer cells constitute a heterogeneous and multi-functional population of the innate immune system. Although the CD56^dim/bright functional dichotomy has been revised recently [1], NK cells are generally divided in two subsets that differ in their anatomic distribution, cytotoxic potential and ability to proliferate and produce cytokines [2, 3]. NK cells initially-obtained their name due to their natural cytotoxicity against tumor cells requiring no prior sensitization, unlike T cells [4]. It is well established that the cytotoxicity of NK cells relies notably on their ability to sense the decrease/absent expression of MHC-I molecules on their target ("missing-self model") [5, 6]. In humans, this sensing is controlled by a set of inhibitory receptors belonging to the Killer immunoglobulin-like receptor (KIR) family and/or the heterodimer CD94/NKG2A: each receptor having variable specificity for allotypic variants of MHC-I molecules [7]. The NK cell repertoire of inhibitory receptors is qualitatively and quantitatively variable between humans due to the inherited set of genes coding for these receptors, but also within the same individual, due to the stochastic expression of these genes [8]. This has important implications particularly during the treatment of acute leukemias which require a Stem Cell Transplantation (SCT). Indeed, alloreaction mediated by NK cells could occur between haploidentical individuals presenting a functional mismatch in the NK cell repertoire towards recipients MHC-I ligands. In this context, NK cell alloreactivity has been shown to increase prognosis by enhancing anti-tumor activity (GvL effect) and decrease side effects of immune reconstitution (GvHD) by depleting recipients' DCs [9, 10]. In mice, infusion of alloreactive NK cells in the context of SCT also induces potent antitumor effects [9, 11] and such therapeutic approaches are now realistic in humans [12]. More generally, adoptive transfer of ex-vivo expanded NK cells constitutes a promising approach in immunotherapy of cancer [13, 14]. Unfortunately, NK cell expansion remains tedious to achieve, using protocols with purification steps, clonal dilution and/or monoclonal antibodies limiting the outcome of NK cell-based immunotherapy [15]. Dendrimers are versatile tree-like branched synthetic polymers with very promising medical applications such as chemotherapeutic agent delivery [16]. More remarkably, it was shown that a N-acetyl-glucosamine-coated poly-amido-amine (PAMAM) dendrimer stimulates an antitumor immune response involving enhancement of the functions of CD4 T cells and NK cells [17]. A mannosylated dendrimer of the same PAMAM family conjugated to ovalbumin (OVA) has been shown to induce, in vitro and in vivo, a very potent immune response against OVA highlighting their adjuvanticity [18]. We have recently reported that a group of nanosized synthetic phosphonate-capped dendrimers (especially 3a-G1) activate human monocytes toward an anti-inflammatory and immunosuppressive pathway [19‐21]. We also described an innovative protocol using dendrimer 3a-G1 that allows high yield expansion human NK cells from PBMCs [22]. Expanded NK cells are fully functional and can efficiently lyse a broad spectrum of tumor cell lines. Prospecting the transfer from bench to clinic of such expanded NK cells, we had to decipher the origin of this expansion process. Here, we show that 3a-G1 driven expansion of NK cells from PBMCs is not occurring through a direct activation of the NK cell reservoir but actually through the regulation of CD4⁺ T cell expansion. Ultimately, we found that our protocol prevents the IL-2 related expansion of CD4⁺/CD25⁺/CD127^low/FoxP3⁺ regulatory T cells. Given the fact that regulatory T cells might affect NK cell functions in vivo [23, 24], this last finding supports the use of our expansion protocol for NK cell-based adoptive immunotherapy of cancers.

In this report, we elucidate the origin of the enrichment and subsequent expansion of NK cells from human PBMCs using 3a-G1 phosphonate-capped dendrimers [22]. Therefore, we focused our analysis on the first two weeks of culture although the expansion procedure requires 4 weeks to provide suitable amounts of cells for clinical purposes. Such amplified NK cells are perfectly cytotoxic against the K562 cell line but also a broad range of other tumor cell line. Although this has not been checked systematically, we did found that mid-term amplified NK are also cytotoxic against the K562 cell line and that 3a-G1 doesn't affect their cytotoxicity when compared with untreated cells [see Additional file 1]. Contrary to expectation, we could not demonstrate any significant activation of proliferation of pure NK cells exposed to 3a-G1. Conversely, we showed that during the first week of culture, 3a-G1 mainly acts by inhibiting CD4⁺ T cell proliferation without affecting NK cells. In terms of cell expansion, we found that NK cells are normally competing with CD4⁺ T cells when PBMCs are exposed to interleukin-2 and that 3a-G1 cancels this competition. Therefore, the decreased CD4⁺ T cell representation results in more nutrients and cytokines for the expansion of NK cells. We propose that the higher proliferation status of NK cells when PBMCs are exposed to 3a-G1 (Fig. 1) is mainly due to an increase in the availability of IL-2 that has not been consumed by proliferating T cells. Supporting our hypothesis, other investigators have described the use of anti-CD3 antibodies and IL-2 as a method for the in vitro expansion of human NK cells from PBMCs [27]. No clues were provided about the origin of this process but it suggests that targeting T cells to some extent sustains the expansion of NK cells from PBMCs. Interestingly; we demonstrated that like such antibodies, 3a-G1 dendrimers specifically interacts with CD4⁺ T cells. We believe that this interaction might drive the inhibition of CD4⁺ T cell proliferation observed not only among PBMCs but also when pure CD4⁺ T cells were stimulated with anti-CD3/CD28 coated beads. Molecular determinants are still needed regarding the mode of action of 3a-G1 but given its structural features, it is tempting to speculate that 3a-G1 could act by triggering Sphingosine 1-phosphate (S1P) receptors. Indeed, there is some evidence that S1P regulates T cell proliferation [28]. Interestingly, the phosphate moiety was shown to be important for this effect. To address that point, we are now concentrating our effort in the synthesis of a biotin analogue of 3a-G1 to perform pull-down experiment on CD4⁺ T cell protein extracts with the aim of identifying by proteomics the molecular determinants of 3a-G1 regulatory activity. Furthermore, Miller and colleagues have described the importance of monocytes in the expansion of human NK cells from IL-2 treated PBMCs [29]. We have shown that depleting monocytes from PBMCs prevents CD4⁺ T cell proliferation. In agreement with Miller's report, we also found that NK cells are less able to proliferate when monocytes are depleted from PBMCs. Therefore, monocytes are supporting the ex-vivo expansion of both cell types. Interestingly, we showed that monocytes rapidly engulfed phosphorus-containing dendrimers and consequently become activated [19, 20]. We have addressed the particular mode of activation of these monocytes highlighting an immune-suppressive phenotype on mixed leukocyte reaction [21] that could sustain the inhibition of T cell proliferation although we have shown here, using anti-CD3/CD28 microbeads, that monocytes are not required for regulatory activity of phosphonate-capped dendrimers. Again, Miller and colleagues showed that CD5⁺ and CD8⁺ cell depletion led to higher NK cell expansion yield providing support that T cells constitute a barrier for the expansion of NK cells. IL-2 stimulation of PBMCs was shown to elicit absolute expansion of NK cells and CD56⁺ T cells, e.g. NK-T cells, γδ T cells and some αβ/CD8⁺ T cells [30]. The combination of IL-2 and 3a-G1 in our hands also led to a generally slightly higher representation of γδ-T cells (data not shown) but we were never able to detect any NKT (Vα 24⁺) cell or CD8⁺ T cell expansion under our conditions. In contrast, we found that a proportion of CD4⁺ T cells that became activated under IL-2 stimulation were presenting a regulatory T cell phenotype e.g. CD25⁺/FoxP3⁺/CD127 ^low , the best up to date combination to characterise regulatory T cells [26]. Such in vitro induction of T regulatory activity by stimulated human CD4⁺/CD25^- has already been described [31]. In vivo, regulatory T cells play an important role in maintaining peripheral tolerance and preventing auto-immunity but they could also affect anti-tumor immunity by notably acting on NK cell activity [23, 24]. Then, the presence of regulatory T cells during the process of NK cell expansion by 3a-G1 would have had a highly detrimental effect. Interestingly, the inhibition of CD4⁺ T cell activation by 3a-G1 is global and also affects the accumulation of these phenotypically related regulatory T cells. Although it can't be excluded that the presence of few remaining regulatory T cells could have a detrimental effect for the activity of infused NK cells in vivo, it does not affect the cytotoxic property of the expanded NK cells in vitro against classical tumor cell lines [22].

On the wave of tetramer technology [32], this project was initiated to use dendrimer plasticity to chemically build a platform of bi-phosphate entities that would promote γδ-T cell expansion [33]. This contemporary attempt of building and testing a sophisticated hypothesis actually ended with unexpected results. Indeed, it turned to favour the expansion of NK cells another subset of cytotoxic lymphocytes and we show here that this is happening by selectively inhibiting the activation and proliferation of CD4⁺ T lymphocytes. Having set the expansion protocol using good manufacturing practice (GMP)-compliant components, we are now planning to translate from bench to clinic the use of such ex-vivo amplified NK cells as a conditioning treatment for patients undergoing leukemia therapy. The therapeutic relevance of our method does have some limitation as we did observe variation in NK cell expansion between donors at the term of the amplification process [22]. However, deciphering the molecular determinants of phosphonate-capped dendrimer activity in the regulation of T cell proliferation could also lead to further applications in the treatment of pathologies where T cell proliferation is undesirable, such as cutaneous T-cell lymphoma and/or auto-immune diseases for which efficient treatments are still needed [34].