Adjuvants and myeloid-derived suppressor cells: Enemies or allies in therapeutic cancer vaccination

Abstract

Adjuvants are a critical but largely overlooked and poorly understood component included in vaccine formulations to stimulate and modulate the desired immune responses to an antigen. However, unlike in the protective infectious disease vaccines, adjuvants for cancer vaccines also need to overcome the effect of tumor-induced suppressive immune populations circulating in tumor-bearing individuals. Myeloid-derived suppressor cells (MDSC) are considered to be one of the key immunosuppressive populations that inhibit tumor-specific T cell responses in cancer patients. This review focuses on the different signals for the activation of the immune system induced by adjuvants, and the close relationship to the mechanisms of recruitment and activation of MDSC. This work explores the possibility that a cancer vaccine adjuvant may either strengthen or weaken the effect of tumor-induced MDSC, and the crucial need to address this in present and future cancer vaccines.

Abbreviations

MDSC

myeloid-derived suppressor cells

Mo-MDSC

monocytic MDSC

G-MDSC

granulocytic MDSC

Treg

regulatory T cells

TAM

tumor-associated macrophages

TLR

Toll-like receptors

NK

natural killer

GM-CSF

granulocyte macrophage colony-stimulating factor

ARG1

arginase 1

NOS2

inducible nitric oxide synthase

APC

antigen-presenting cells

DC

dendritic cells

CTL

cytotoxic T lymphocytes

Introduction

After many years of intense immunological research and strong pre-clinical and clinical findings, the scientific community has reached a consensus that immune system has the inherent capacity to recognize and eradicate cancer.¹ This in turn has resulted in the development of multiple different immunotherapeutic approaches, which unfortunately only very few have achieved a real clinical benefit to human patients.² It has more recently become clear that these disappointing clinical immunotherapy results are in large part due to the significant immune suppression induced by the cancer itself. It is now well established the tumors induce specific types of immune cells that accumulate in the tumor microenvironment, and support tumor growth and metastasis by (at least in part) negatively impacting effective antitumor T cell responses. Among the immune cells that mediate this immunosuppressive tumor environment are regulatory T cells (Treg), tumor-associated macrophages (TAM), type 2 helper CD4⁺ (Th2) and MDSC.³ Therefore, one of the major directions of development in vaccine immunotherapy is combination with therapies that target either the suppressive cell populations and/or the signals involved in their recruitment and function.⁴

As noted above, evidence now demonstrates that MDSC are a major mediator of tumor-induced immunosuppression, and that they closely interact with the other immunosuppressive cell populations. MDSC are generated in response to a variety of tumor-derived cytokines and growth factors, and are a heterogeneous population of myeloid cells at different stages of differentiation. These cells are directly involved in many pro-tumoral processes, and also impair T cell and natural killer (NK) cell responses – in particular inhibiting CD8⁺ T cell activation and effector function.⁵ As a result, high levels of circulating MDSC have been correlated with poor prognosis and disease progression in patients, most compellingly demonstrated in colorectal and breast cancer.^6,7 Consequently, MDSC are becoming a major therapeutic target in order to directly abrogate their pro-tumor effect as well as to improve the efficacy of concurrent immunotherapy strategies.⁸ In this regard, even the most effective immunotherapeutic approaches currently used clinically, such as cancer vaccines, adoptive T cell therapy, and immune checkpoint blockade, may be significantly augmented if combined with therapies to overcome MDSC suppressive pressure.

The real impact, on the life expectancy of cancer patients, produced by therapeutic antitumor vaccines in particular, is also still far from being fully reached.⁴ Although cancer vaccines have impressive antitumor activity in animal models, only one vaccine (Sipuleucel-T (Provenge®), for metastatic prostate cancer) has been approved by the FDA, - with an extension of overall survival by only 4 months.⁹ Thus how cancer vaccines can be made to be more effective in human patients is the most significant challenge faced by the field. One of the crucial components of vaccines is the adjuvant, which is essential to boost the immune system in order to achieve the specific antitumor response driven by the antigen in the vaccine formulation.¹⁰ For this reason, choosing the right adjuvant is very important for vaccine design, however only relatively recently their mechanisms of action have started to be elucidated.¹¹ Additionally, cancer vaccines most overcome tumor-induced immunosuppression, and adjuvants are the vaccine component that could play a major role in achieving this.¹² It is important to note that this is not a challenge faced by most infectious disease vaccines (with the notable exception of HIV), and thus adjuvants for cancer vaccines may require unique properties that are not found in the traditional adjuvants used for infectious diseases. Accordingly to Olivera Finn, adjuvants used for cancer vaccines fall into three different categories: (1) bacterial products or Toll-like receptor (TLR) agonists, (2) cytokines and growth factors and (3) immunostimulatory delivery systems.¹¹ In Table 1 we have compiled and classified into these three categories all the cancer vaccine adjuvants being tested in clinical trials registered with the United States National Cancer Institute (www.clinicaltrials.gov) and the Public Register of Cuban Clinical Trials (http://rpcec.sld.cu). In this review we will explore the possibility, based on relatively little published data, that the immunostimulatory signals produced by these and other relevant adjuvants are positively or negatively impacting on the suppressive capacity of MDSC.

Table 1.

Adjuvants and immunomodulators currently used in cancer vaccines´ clinical trials registered at the NCI and the PRCCT

	Adjuvant	Localization	Type of cancer vaccine formulation	Phase	NCI or PRCCT reference
TLR-ligands	CpG-	Non-Hodgkin lymphoma	Mantle cell lymphoma vaccine	II	NCT00490529
	BCG	Bladder, melanoma	HS-410 and PANVAC	II/I, II	NCT02010203, NCT02015104
			CSF-470 vaccine (combined with GM-CSF)	III/II	NCT01729663
	Imiquimod	Glioblastoma, melanoma	peptide vaccines (combined with GM-CSF)	II/I	NCT02078648, NCT01191034
	Resiquimod	Glioblastoma, melanoma	Tumor lysate-pulsed DC and peptide vaccines (combined with Montanide ISA-51)	II/I, II	NCT01204684, NCT02126579, NCT00960752
	PolyI:C	Glioblastoma, NSCLC, non-Hodgkin lymphoma	Tumor lysate-pulsed DC and peptide vaccines or Immunomodulator	II/I, II	NCT01920191, NCT01720836, NCT01976585, NCT01204684
	VSSP	Solid tumors, prostate, breast	Peptide, protein and NGcGM3 ganglioside vaccines (combined with Montanide ISA 51)	I, II/III	RPCEC00000102, RPCEC00000111, RPCEC00000070
	MPL	NSCLC, ovarian, melanoma	NY-ESO-1 peptide vaccine	II/I	NCT01584115
	AS-15 (containing QS-21, MPL, CpG)	Bladder, melanoma	peptide vaccines (combinated with IL-2 or PolyI:C)	II	NCT01435356, NCT01437605
Cytokines and growth factors	GM-CSF	Glioblastoma, breast, NSCLC, melanoma, myeloma, non-Hodgkin lymphoma, pancreas, prostate, AML	Allogenic GM-CSF secreting cells, peptide and protein vaccines or immunomodulator	II/I, II, III	NCT01480479, NCT01479244, NCT01355393, NCT02019524, NCT01570036, NCT01579188, NCT01875653, NCT01349569, NCT01926639, NCT01088789, NCT01322490, NCT01784913, NCT00153582
	IL-12	Melanoma, glioblastoma	Immunomodulator	II/I, II	NCT01236573, NCT01307618, NCT01502293, NCT01213407
	IL-2	Melanoma, CLL, NSCLC, breast	IL-2 secreting cancer cells and immunomodulator	II/I, III/II, III	NCT01604031, NCT00279058, NCT01383148, NCT01256801
Delivery systems	Iscomatrix	Lungs, esophagus, pleura and mediastinum malignancies	H1299 cell lysate vaccine	II/I	NCT02054104
	Montanide ISA-51	Melanoma, pancreas, colon, cervix, NSCLC	Peptide and protein vaccines	II/I, III	NCT00108875, NCT00673777, NCT01950156, NCT00674258, NCT01308294, RPCEC00000161
	Alum	NSCLC	Racotumomab anti idiotipic vaccine vaccine	III	RPCEC00000179

General Aspects of MDSC

MDSC are a heterogeneous population of myeloid cells with immunoregulatory activity that include immature precursors of monocytes, macrophages, granulocytes and dendritic cells (DC).^13,14 This population of myeloid progenitors can be identified in mice by the co-expression of CD11b and Gr1 molecules, and are currently classified in two main subsets: monocytic MDSC (Mo-MDSC) and granulocytic/polymorphonuclear MDSC (G-MDSC), which differ in phenotype and biological properties.⁵ Murine CD11b⁺Ly6C^hiLy6G⁻ Mo-MDSC are highly immunosuppressive and employ antigen-independent mechanisms to suppress T cell functionality, whereas CD11b⁺Ly6C^lowLy6G⁺ G-MDSC are less immunosuppressive and exert their effect in an antigen-specific manner.^8,15-17 Human G-MDSC exhibit a LIN⁻CD11b⁺HLA-DR⁻CD33⁺ phenotype and can express CD15 and/or CD66b. In comparison, human Mo-MDSC are CD33⁺CD14⁺HLA-DR^low/−.⁸

MDSC limit the functional capacity of CD4⁺ and CD8⁺ T cells via four basic mechanisms: (1) affecting the availability of essential amino acids required for T cell growth and differentiation, (2) generating oxidizing molecules, (3) interfering with migration, function and viability of lymphocytes (including affect the cytotoxicity of NK cells in a cell-cell contact dependent fashion¹⁸) and (4) inducing the development or expanding the pre-existing Treg populations.⁸ As part of this latter suppressor activity, MDSC can control Treg homeostasis causing an increase in the Treg levels in both cancer patients and tumor-bearing mice. This effect is associated with de novo differentiation of Treg from naïve T CD4⁺ cells, as well as via trans-differentiation of Th17 cells into Foxp3⁺Treg.¹⁹ All these observations serve as the rationale for development of therapeutic agents aimed at reducing the negative effect of MDSC on the effector immune cells, especially for cancer treatment. Along these lines, it has been shown that the administration of 25-hydroxy vitamin D3 or all-trans retinoic acid reduce MDSC levels by increasing their differentiation toward mature myeloid cells in cancer patients.²⁰ Additionally, the use of the drugs sildenafil and sunitinib, as well as gemcitabine and 5-fluorouracil, has been shown to decrease the number and/or suppressive function of MDSC in several human cancers.^20-23

Although most of the studies describing MDSC phenotype and biological function have been made in tumor-bearing hosts, recent findings demonstrate that immunosuppressive MDSC are also recruited during non-tumor-related inflammatory responses in which T cells are activated, either in acute or chronic settings. In fact, it appears that MDSC are expanded in inflammatory sites after tissue injury, burns or bacterial and virus infection to prevent tissue damage by exacerbated T cell responses.^19,20 Recent data suggest that MDSC can also accumulate at the vaccination sites and are capable of suppressing T cell proliferation.²⁴ These findings raise a potential connection between a cancer vaccine's adjuvant and the recruitment of MDSC or the modulation of their suppressive function. A theoretical two-signal model that has been recently proposed to explain the expansion of MDSC²⁵ could help in understanding this interesting relationship. In this model, the first signal determines the aberrant patterns of proliferation and differentiation of the myeloid progenitors that lead to the accumulation of immature myeloid cells. This process is regulated by several soluble factors such as granulocyte macrophage colony-stimulating factor (GM-CSF), granulocyte colony-stimulating factor (G-CSF), macrophage colony-stimulating factor (M-CSF), interleukin (IL)-6 and vascular endothelial growth factor (VEGF); which signals through the Signal Transducer and Activator of Transcription (STAT) 3 and STAT5. Thus, adjuvants based on cytokines and growth factors, for instance GM-CSF, could supply this first signal for expanding MDSC. The second signal is required to activate the suppressor function of MDSC, such as up-regulation of arginase 1 (ARG1) and inducible nitric oxide synthase (NOS2), as well as the production of immunosuppressive cytokines. In this regard, pro-inflammatory molecules such as interferon gamma (IFN-γ), IL-1β, IL-13 and the TLR ligands could provide this second signal to activate MDSC activity, by signaling through STAT1 and NF-κB. Interestingly, the most studied vaccine adjuvants are TLR ligands due to their role in activating DC and consequently effector T cells. Therefore, this review will examine the evidence accumulated so far that suggest a complex interaction between the different kinds of cancer vaccine's adjuvants and MDSC, which we believe has been previously under-appreciated but cannot be ignored if effective vaccine-adjuvant formulations are to be developed for cancer immunotherapy.

Modulation of MDSC with Adjuvants that are Agonists of Pattern Recognition Receptors

A common approach for the development of vaccine adjuvants has been to focus on products mimicking or containing pathogen-associated molecular patterns (PAMP), which can be recognized by the pattern recognition receptors (PRR) expressed in antigen-presenting cells (APC). As it naturally occurs during infection, signaling through PRR induces increased co-stimulation and secretion of pro-inflammatory cytokines by DC, both required signals for the maximal activation of T cells.²⁶ TLR are the best characterized family of PRR, and many adjuvants recognized by these innate receptors have been described. Lipopolysaccharide (LPS) from Gram-negative bacteria and monophosphoryl lipid A (MPL) from Salmonella minnesota, both agonists of TLR4, are well known examples of these kinds of adjuvants.^11,27 Other adjuvants in this class are bacterial oligonucleotides containing unmethylated CpG sequences (CpG) that signals through TLR9^28,29 and the synthetic agonist of TLR3 polyinosinic-polycytidylic acid (polyI:C).^26,30 Signaling through the majority of TLR promotes Th1 polarization and cytotoxic T lymphocytes (CTL) responses,^31,32 hence there is a particular interest in this type of adjuvants for cancer immunotherapy. Several preclinical studies and clinical trials are testing the therapeutic effect of TLR agonists in cancer, but thus far only MPL, imiquimod (a TLR7 ligand) and Bacillus Calmette–Guérin (BCG, that acts as a mixed TLR2/TLR4 agonist) have been licensed by the FDA for clinical use in cancer patients.²⁷

However, additional properties of these adjuvants need to be further characterized for they successful use in cancer immunotherapy. As previously mentioned, MDSC are not only recruited by tumors but accumulate during any inflammatory condition as a negative counter-balancing mechanism to prevent tissue damage by activated T cells.^19,33-36 Therefore, immunization with TLR agonist adjuvants, which resembles bacterial or viral infection, may induce the expansion of MDSC. Accumulating evidence in the last years support this idea, most of them obtained in tumor-free mice. Morecki et al demonstrated that CpG treatment increases the number of MDSC in the spleen and blood of mice, which efficiently suppresses the allogeneic stimulation of T cells and the development of graft-versus-host disease (GVHD).³⁷ Also complete Freund's adjuvant (CFA) expands a population of murine MDSC able not only to inhibit IFN-γ production by T cells but also to induce their apoptosis in a nitric oxide (NO)-mediated process.³⁸ BCG-recruited MDSC suppress T cell activation through IL-1β-mediated NO secretion.²⁴ Repeated injections of LPS increase the frequency of MDSC that produce IL-10 and utilize heme oxygenase-1 (HO-1) to inhibit T cell proliferation and cytokine production.³⁹ A nanoparticulated adjuvant and TLR2/TLR4 agonist named VSSP (Very Small Size Proteoliposomes) expands MDSC populations that have a mild suppressive activity due to the activity of NOS.⁴⁰ In fact, according to the model proposed by Condamine and Gabrilovich to explain MDSC differentiation, TLR ligands are a second pro-inflammatory signal causing, through NK-κB activation, the upregulation of ARG1, NOS2 and the release of immunosuppressive cytokines.²⁵ Another pro-inflammatory cytokine with a similar effect to TLR agonists during MDSC differentiation is IL-1β²⁵ and it has been described that aluminum hydroxide adjuvants activates Nalp3 inflammasome, caspase-1 and induces IL-1β release.^41,42 Hence, it is likely that also the widely employed alum adjuvant could also induce enhanced MDSC suppressive function.

Previous work has suggested that the MDSC which accumulate during acute parasitic infections suppress T cell activation through NO secretion, whereas MDSC expanded in chronically infected hosts show increased activity of ARG1 and enhanced generation of reactive oxygen species (ROS).^43,44 As noted above, MDSC induced due by several different adjuvants depend on NO secretion for their suppressive activity.^24,38,40 In contrast, tumor-induced MDSC employ both NOS2 and ARG1 as effectors of their potent inhibitory function.^45-47 It has been described that the activity of NOS2 switches to the generation of ROS when is competing with ARG1 for the same substrate (L-arginine).⁴⁸ Therefore, it seems to be that MDSC expanded as a consequence of TLR agonists treatment resemble those generated during acute infections, whereas the sustained expansion of MDSC observed during the chronic inflammation associated with cancer is comparable to MDSC recruited in chronic infections. Based on their described suppressive mechanisms and biological function, MDSC associated with chronic inflammation may be more suppressive than their counterparts induced by an acute inflammatory state. In support of this idea, it has been shown that MDSC expanded by the VSSP adjuvant in healthy mice are significantly less suppressive than tumor-induced MDSC, in a per cell-based comparison.⁴⁰

This evidence indicates that cancer vaccine adjuvants that signal through PRR will expand MDSC, but this is not necessarily a reason not to use them. The more relevant question is whether these adjuvants reinforce or ameliorate the immunosuppression generated by existing tumor-induced MDSC. A few recent studies have addressed this matter. We have found that VSSP increases the frequency of splenic MDSC in tumor-bearing mice, but reduces the suppressive function of the overall population and impairs their migration to the tumor site.⁴⁰ VSSP also accelerates the differentiation of both splenic and tumor-infiltrating MDSC toward mature DC.^40,49 As a consequence, VSSP potentiates an effective CTL response in mice with high numbers of MDSC,⁴⁰ indicating that it is possible to reduce tumor-induced immunosuppression with adjuvants that somewhat paradoxically expand the number of MDSC. Shirota et al have reported that intratumoral administration of CpG also decreases the ability of Mo-MDSC to inhibit T cells, and promotes their differentiation toward M1 macrophages with tumoricidal capability.⁵⁰ However, the optimal dosage of CpG for using in cancer vaccines should be carefully selected, since high doses of this adjuvant induce a tolerogenic response in mouse plasmacytoid DC and in vitro-generated human DC.⁵¹ These authors demonstrated that tolerogenic signaling through TLR9 in DC requires physical interaction with Toll/IL-1 receptor domain-containing adaptor inducing IFN-β (TRIF), which promotes the secretion of TGF-β and the expression of indoleamine 2,3-dioxygenase 1.⁵¹

It has been also shown that polyI:C provokes the conversion of tumor-induced MDSC into mature macrophages, but this process may require IFN-α produced by plasmacytoid DC.⁵² Recently, Lee et al demonstrated that the TLR7/8 agonist resiquimod differentiates tumor-induced MDSC in vitro toward macrophages and DC.⁵³ In contrast, in vitro incubation of bone marrow-derived MDSC with LPS and IFN-γ increases NO production, thus potentiating the suppressive activity of MDSC, and impairs their differentiation into DC.⁵⁴ In addition, LPS was unable to differentiate tumor-induced MDSC to mature APC in vitro or to change their suppressive activity, in the same experimental setting where VSSP effectively generated DC from MDSC.⁴⁹ All together these observations suggest that even though the signaling pathways triggered by TLR engagement are quite limited, the final outcome of the interaction between TLR ligands and MDSC is diverse. This underscores that experimental evaluation of this interaction needs to be performed before moving forward in the development of cancer vaccine formulations.

Interaction Between MDSC and Cytokines Used as Adjuvants

Cytokines and growth factors are also used as adjuvants for cancer vaccines on the basis of their ability to stimulate the immune system, either by promoting the differentiation, activation and/or recruitment of APC - or by directly affecting the T cell compartment. However, many cytokines and growth factors have dual opposing roles. Therefore, it is essential to know under which conditions the cytokine adjuvants can enhance or impair immune responses. Moreover, as mentioned above, MDSC are composed of immature precursors of several types of APC.^13,14 Consequently, cytokines and growth factors involved in the common pathways of recruitment and differentiation of both APC and MDSC can potentially tip the balance between suppressive and stimulatory cells.

Among the cytokines used as adjuvants to activate APC in cancer vaccines, GM-CSF is one that has shown the capability either to activate or suppress the immune system. On the one hand, GM-CSF was the first cytokine reported to promote DC differentiation and expansion in vitro and in vivo.^55,56 Thus it has been successfully used to stimulate tumor-specific immune responses, for example, when combined with irradiated tumor cell vaccines in several murine models and clinical trials.⁵⁷ On another hand, studies with transplantable tumors have shown that GM-CSF secretion by the tumor lines results in alterations in both the number and subpopulations of MDSC.⁵⁸ In fact, in this setting GM-CSF is the main factor driving the generation of these cells.^5,59 In MDSC, this cytokine signals mainly through STAT5, which not only induces the proliferation of immature myeloid cells but further prevents MDSC differentiation into mature APC.⁶⁰ Thus, it remains an open question what the best strategy is for using GM-CSF, that would allow for its beneficial immunostimulatory effects while overcoming the expansion of MDSC induced by this cytokine.

A possible solution has come in the meta-analysis of the experience of using GM-CSF as a cancer vaccine's adjuvant. This analysis, based on several clinical studies, suggested that GM-CSF enhances the efficacy of vaccination when used in low repeated doses (40–80 μg for 1-5 d), whereas higher doses (100 and 500 μg) have the opposite effect.⁶¹ Thus, the final immunological outcome of GM-CSF administration may be determined mainly by its dosage, which highlights the importance of an appropriate design of the treatment schedules for the vaccines employing GM-CSF as an adjuvant. However, additional hypotheses and further studies need to be done in order to develop a less empirical way to manipulate the immune system using GM-CSF. Currently the ongoing clinical studies with NeuVax in breast cancer patients will be highly illustrative in this respect.

The common pathways of differentiation of MDSC and some APC can be exploited also from a different angle. Paradoxically, the unavoidable presence of MDSC on tumor-bearing individuals could be used by vaccinologists to their advantage. As discussed above for the adjuvants based on the TLR ligands, there are signals that can drive the differentiation of the tumor-recruited MDSC into mature myeloid cells, with the subsequent beneficial increment of mature immunostimulatory APC. One cytokine used as vaccine adjuvant which promotes the differentiation of MDSC into APC and promotes a robust immune response is IL-12.^62,63 IL-12 is an important inflammatory cytokine able to induce a potent immune response by enhancing the IFN-γ secretion and functionality of NK and NKT cells,^63-65 as well as via Th1 polarization.^66,67 Consistent with this, IL-12 has demonstrated potent antitumor effect in preclinical models, by stimulating immune effector cells.^63,65 The impact of IL-12 on the modulation of MDSC is being currently studied, as an additional mechanism that could explain the efficacy of IL-12-based cancer therapy. In fact, accumulating evidence demonstrates that IL-12 induces MDSC differentiation toward a phenotype characteristic of APC.^68-70 For example, Steding et al showed that treatment of tumor-bearing mice with an IL-12-containing adenovirus reduces the percentage of MDSC within the tumor, up-regulates the expression of surface markers associated with a mature DC phenotype on these cells, and increases the percentage of functional CD8⁺ T cells, all resulting in the improved overall survival.⁶⁹ Additionally, adoptive transfer of tumor-specific CD8⁺ T cells engineered to secrete functional single-chain IL-12 reprograms MDSC to secrete IFN-γ within the tumor, inducing tumor regression.⁶⁸ Moreover, this programmatic change in tumor MDSC enabled the direct recognition of cross-presented cognate antigen by CD8⁺ T cells.⁶⁸ These authors later demonstrated that IL-12 delivery up-regulates the expression of Fas receptor (CD95) on MDSC, macrophages and DC in the tumor stroma, which is crucial for the maintenance of effector memory T cells in the tumor as well as for the antitumor effect of the transferred IL-12-producing CD8⁺ T cells.⁷¹ Interestingly, a connection has been suggested between the presence of MDSC and the efficacy of IL-12 therapy. In an orthotropic glioma model, MDSC depletion abrogates the survival benefit of IL-12 immunogene therapy.⁷⁰

In additional to IL-12, other cytokines that directly interact with T cells have been used as vaccine adjuvants, including IL-7, IL-15 and most importantly IL-2.^72,73 The cytokines IL-15^74,75 and IL-7⁷⁶ are being developed as adjuvants because of their important function in the maintenance and development of effector and memory T cells,⁷⁷ but are currently only in preclinical studies. In contrast, treatment with high doses of IL-2 has been extensively used clinically in cancer patients.^74,78,79 IL-2 has central roles in the generation, activation, and homeostasis of T lymphocytes and NK cells,⁸⁰ as well as in the regulation of Th1, Th2, and Th17 differentiation.⁸¹ This cytokine has induced clinical responses in patients with metastatic melanoma⁷² and renal cell carcinoma (RCC),⁷³ and is a standard treatment for metastatic RCC. However, IL-2 also drives expansion of Treg, which, in turn, dampens antitumor immunity.⁸²

Recent evidence has shown that these three T cells growth factors can modify the number of MDSC. However, their effect on MDSC function has not been addressed in any definitive way, and no generalizations can be made at this point. An intuitive consequence of the expansion and activation of T cells after treatment with T cells growth factors could be the recruitment and activation of the suppressive function of MDSC. However, Habibi et al has demonstrated that in vivo intralesional injection of IL-15 and IL-7 after radiofrequency thermal ablation (RFA) increases immune responses and inhibits tumor development, in comparison to RFA only. They showed that this response was possibly mediated by the reduction of MDSC within tumor microenvironment.⁸³ In contrast, Ochoa et al showed that IL-2 treatment increases the percentage of MDSC able to express high levels of ARG1 in RCC patients who received a previous therapy with bevacizumab.⁸⁴

Related to IL-2, there is one issue that should be also taken into consideration, and it is the effect on Treg expansion and this close relation with MDSC. The interaction between MDSC and Treg in cancer is well documented. It has been shown that MDSC induce the expansion of Foxp3⁺CD25⁺ Treg in vitro and in vivo.⁸⁵ In fact, in a Lewis lung cancer model, depletion of MDSC causes a significant decrease of tumor-infiltrating Treg, tumor growth reduction and prolonged survival of tumor-bearing mice.⁸⁶ Conversely, Treg depletion promoted the conversion of MDSC to a less immunosuppressive phenotype.^87,88 In the case of immunotherapy with IL-2 alone, the effect on MDSC has not been explored, however given the documented effect on Treg development, the consequent expansion of MDSC seems highly probably. Nonetheless, it has been shown that, in combination with an agonist antibody to CD40, IL-2 successfully reduces Treg and MDSC within the tumor microenvironment in mice with renal adenocarcinoma.⁸⁹ The latter effect is mediated by the induction of Fas-ligand (FasL) in these cells.

The Potential Influence of Vaccine Delivery Systems on Tumor-Induced Immunosuppression

A different category of adjuvants has been designed for antigen delivery to a chosen processing pathway in the APC. In general, there are three antigen-processing pathways in specialized APC such as DC. The more classical are referred as the endogenous pathway, in which cytosolic antigens are processed by the proteasome and the antigenic peptides loaded onto MHC I molecules at the ER,^90,91 and the exogenous pathway involving the proteolytic degradation of internalized exogenous antigens in endocytic compartments followed by the binding of peptides onto MHC II in these vesicles.^92,93 In a few specialized subsets of DC, the extracellular antigens can be processed and presented on MHC I molecules, a process known as cross-presentation, which is essential for the activation of CTL against exogenous antigens.⁹⁴ Vaccine antigens are commonly processed through the exogenous pathway in DC, leading to CD4⁺ T cell activation and antibody production.¹¹ However, for cancer vaccines it is essential to generate effective CTL responses during immunization, a process that requires the delivery of exogenous antigens into the cytosol and its further processing by components of the endogenous pathway for cross-presentation.⁹⁴ To accomplish this, microparticles (poly _D,L-lactic-co-glycolic acid microspheres), virus-like particles and immunostimulatory complexes (ISCOMs),^95-99) as well as heat-shock proteins,¹⁰⁰ have been used to efficiently deliver vaccines’ antigens to the cytoplasm and facilitate MHC I presentation. Other GM3-containing nanoparticles previously mentioned (VSSP) are able to induce cross-presentation of the accompanying antigen by DC and thus potentiates the Th-independent activation of CTL.¹⁰¹

Recently Ugel et al demonstrated that tumor antigens are also cross-presented by splenic MDSC, an important process for inducing cross-tolerance of tumor-specific CTL.¹⁰² This finding highlights the importance of evaluating whether these delivery systems intended to facilitate presentation of vaccines’ antigens in MHC I by DC could also potentiate cross-tolerance by tumor-induced MDSC, which may offset the immune activating effects of these types of cancer vaccines. We have examined this issue for the VSSP nanoparticles. We find that VSSP treatment abrogates the cross-presentation of a model tumor antigen by splenic Mo-MDSC and G-MDSC subsets, indicating an effect of the adjuvant on the mechanism of antigen cross-presentation occurring in these cells.⁴⁹ It is noteworthy that when the tumor antigen is administered together with VSSP it is not cross-presented by MDSC, whereas in the same experimental setting a vaccine using polyI:C fails to prevent MDSC cross-presentation of the same antigen. More importantly, concomitant treatment with VSSP also inhibits cross-presentation of the tumor antigen when using vaccine containing polyI:C as the adjuvant.⁴⁹ These results suggest that VSSP nanoparticles could be used to abrogate MDSC cross-presentation of tumor-antigens when employed as a vaccine's adjuvant, or as immunomodulator together with other vaccines. Altogether, these data demonstrated that a more comprehensive understanding of the effect of delivery systems on antigen cross-presentation/cross-tolerance by MDSC is needed.

Another form of antigen delivery is its trapping and slow release by the formation of a depot at the site of injection. This is probably the oldest mechanism of action described for adjuvants.¹⁰³ Alum-based salts and oil in water emulsions are 2 of the more widely used adjuvant systems, and both produce a depot effect.¹⁰ This kind of adjuvants are also included in numerous cancer vaccines formulations.^104-106

Recently, Overwijk et al published an important preclinical study proposing that the depot effect of incomplete Freund´s adjuvant (IFA) acts to trap effectors T cells at the injection site. This phenomenon does not only promote a diversion of the effector cells away from the tumor, but also induces dysfunction and deletion of these cells at the vaccination site.¹⁰⁷ The paper suggests that this effect could be a plausible explanation for why many vaccinated patients have increased numbers of circulating tumor-specific T cells without having significant tumor regression.¹⁰⁸ In this scenario, MDSC have also been shown to play an important role. At the site of injection, IFN-γ production induces the accumulation of both G-MDSC and Mo-MDSC. Moreover, the recruited Mo-MDSC up-regulate FasL and the immunosuppressive Programmed cell Death Ligand 1 (PD-L1).¹⁰⁶ This phenomenon of MDSC recruitment, together with the induction of T cell apoptosis at the vaccination sites, creates an immunosuppressive environment that prevents not only the effective antitumor response but also result in hyporesponsiveness to subsequent vaccination.

More than one hundred clinical trials have been completed or are ongoing in the United States that use immunodominant peptides formulated in IFA to augment CTL responses. For these vaccines approaches, we believe that non-persistent adjuvants may be the most effective. Alum salt-formulated cancer vaccine formulations should be less susceptible to this phenomenon of CD8⁺ T cells trapping, since most of them are designed to induce good antibody responses.¹⁰ However, alum has also been used for stabilizing, within the vaccine, other adjuvants such as IL-12¹⁰⁹ or MPL¹¹⁰ and the antigen, to induce cellular responses. These vaccine formulation approaches should be carefully explored for the possible implications of the depot effect and probably other forms of co-localization of the antigen and the adjuvant needs to be considered.

Concluding Remarks

A close examination of the elements discussed in this review supports the point that we cannot make untested assumptions regarding the interaction among adjuvants and MDSC. More importantly, the expansion of MDSC by adjuvants may not always be bad. Some of the adjuvants can differentiate tumor-induced MDSC into DC, other to macrophages, while others reinforce MDSC's undifferentiated and suppressive state (Fig. 1). And importantly, some can exert opposite effects depending on the administered dose. Therefore, each individual adjuvant under development needs to be tested in pre-clinical studies and clinical trials before reaching a decision about its suitability for usage in cancer vaccines. Although many compounds have been studied for the therapeutic targeting of MDSC, finding cancer vaccine adjuvants that are able to modulate the suppressive function of tumor-expanded MDSC could be a better choice due to their simultaneous capability to activate DC- resulting in induction of maximal effector function of tumor-specific CD4⁺ and CD8⁺ T cells in the absence of MDSC immunosuppression. Moreover, the majority of the studied TLR ligands and some cytokines differentiate tumor-induced MDSC toward APC, which represents a desirable tipping of the balance to an increase of immunostimulatory APC with the concomitant loss of immunosuppressive MDSC.

Figure 1.

Schema of possible different interactions of cancer vaccine adjuvants with MDSC and potential outcomes on T cell activation. (A) After vaccine inoculation, adjuvants interact with tissue resident DC promoting their migration, activation and antigen presentation to T cells into the lymphoid organ. Additionally, adjuvants can also interact directly with DC or T lymphocytes inside the draining lymph node or spleen. Activated antigen-specific T cells most then migrate to inflamed tissue, which ideally would be the tumor and exert anti-tumor response. However, adjuvants that induce antigen depot could induce local inflammation into the site of injection that will trap effector T cells and recruit MDSC avoiding effective anti-tumor response. (B) Adjuvants can also directly induce MDSC and/or interact with those highly suppressive cells recruited by tumors. The interaction of the adjuvant with the tumor-induced MDSC can also modify tumor-specific antigen presentation by these cells at both the tumor and distant sites. (C) The interactions between adjuvants and MDSC have unpredictable consequences. A negative effect may be the activation and augmentation of the classical suppressive mechanisms of MDSC, their differentiation into M2 macrophages and their expansion of Treg populations. This outcome will reinforce the suppressive tumor microenvironment and will negatively impact the patient's outcome. In contrast, a positive outcome may be the inhibition of the suppressive capacity of MDSC and/or the induction of MDSC differentiation into DC and M1 macrophages, with the subsequent activation of anti-tumor T cell responses.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

Funding provided by the Center of Molecular Immunology.

Author Contributions

Fernández A, Oliver L, Alvarez R, Fernández LE, Lee KP and Mesa C drafted the review and wrote the paper.