Low dose gemcitabine-loaded lipid nanocapsules target monocytic myeloid-derived suppressor cells and potentiate cancer
immunotherapy
Maria Stella Sasso
a, Giovanna Lollo
b,c, Marion Pitorre
b,c, Samantha Solito
a, Laura Pinton
a, Sara Valpione
d, Guillaume Bastiat
b,c, Susanna Mandruzzato
a,d, Vincenzo Bronte
e, Ilaria Marigo
d,*,1, Jean-Pierre Benoit
b,c,**,1aSection of Oncology and Immunology, Department of Surgery, Oncology and Gastroenterology, University of Padova, 35128 Padova, Italy
bLUNAM UniversitedMicro et Nanomedecines Biomimetiques, F-49933 Angers, France
cINSERM U1066, IBS-CHU, 4 Rue Larrey, F-49933 Angers Cedex 9, France
dVeneto Institute of Oncology IOV-IRCCS, 35128 Padova, Italy
eImmunology Section, Department of Medicien, University of Verona, 37135 Verona, Italy
a r t i c l e i n f o
Article history:
Received 21 December 2015 Received in revised form 15 April 2016
Accepted 18 April 2016 Available online 22 April 2016
Keywords:
Lipid nanocapsules Gemcitabine
Myeloid-derived suppressor cells Adoptive T cell therapy
a b s t r a c t
Tumor-induced expansion of myeloid-derived suppressor cells (MDSCs) is known to impair the efficacy of cancer immunotherapy. Among pharmacological approaches for MDSC modulation, chemotherapy with selected drugs has a considerable interest due to the possibility of a rapid translation to the clinic.
However, such approach is poorly selective and may be associated with dose-dependent toxicities. In the present study, we showed that lipid nanocapsules (LNCs) loaded with a lauroyl-modified form of gemcitabine (GemC12) efficiently target the monocytic (M) MDSC subset. Subcutaneous adminis- tration of GemC12-loaded LNCs reduced the percentage of spleen and tumor-infiltrating M-MDSCs in lymphoma and melanoma-bearing mice, with enhanced efficacy when compared to free gemcitabine.
Consistently,fluorochrome-labeled LNCs were preferentially uptaken by monocytic cells rather than by other immune cells, in both tumor-bearing mice and human blood samples from healthy donors and melanoma patients. Very low dose administration of GemC12-loaded LNCs attenuated tumor-associated immunosuppression and increased the efficacy of adoptive T cell therapy. Overall, our results show that GemC12-LNCs have monocyte-targeting properties that can be useful for immunomodulatory purposes, and unveil new possibilities for the exploitation of nanoparticulate drug formulations in cancer immunotherapy.
©2016 Elsevier Ltd. All rights reserved.
1. Introduction
Anticancer adoptive T cell therapy (ACT) was proven to be effective in a series of clinical trials on different cancer types (i.e.
melanoma, cervical cancer, lymphoma, leukemia, bile duct cancer, and neuroblastoma), resulting in objective clinical responses in patients who failed conventional treatment options[1]. ACT con- sists in the infusion ofex-vivoexpanded, tumor-specific cytotoxic T cells, which may either originate from naturally occurring tumor- infiltrating lymphocytes or be generated by genetic engineering of autologous T cells[1]. Immuno-depleting preparative regimens based on chemotherapy, either alone or in combination with total body irradiation, are administered prior to ACT to increase the clinical response[2e4]. These regimens are known to support the Abbreviations:ACT, Adoptive T cell therapy; DCs, Dendritic cells; GemC12,
gemcitabine-C12; GemC12-LNCs, LNCs loaded with GemC12; GemHCl, gemcitabine hydrochloride; IV, intravenous; LNC, lipid nanocapsules; MDSCs, myeloid-derived suppressor cells; M-MDSC, monocytic myeloid-derived suppressor cell; M4, mac- rophages; OVA, ovalbumin; PMN-MDSC, polyorphonuclear (granulocytic) myeloid- derived suppressor cell; SC, subcutaneous; TERT, telomerase reverse transcriptase;
PBMCs, peripheral blood mononuclear cells.
*Corresponding author.
**Corresponding author. LUNAM Universite dMicro et Nanomedecines Bio- mimetiques, F-49933 Angers, France.
E-mail addresses:ilaria.marigo@ioveneto.it(I. Marigo),jean-pierre.benoit@univ- angers.fr(J.-P. Benoit).
1 Authors contributed equally.
Contents lists available atScienceDirect
Biomaterials
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / b i o m a t e r i a l s
http://dx.doi.org/10.1016/j.biomaterials.2016.04.010 0142-9612/©2016 Elsevier Ltd. All rights reserved.
expansion and function of transferred T cells by multiple mecha- nisms, including the reduction of immunosuppressive cell pop- ulations such as T regulatory cells and myeloid-derived suppressor cells (MDSCs)[1]. However, currently employed preparative treat- ments cause a severe suppression of the immune system and are associated with serious side effects, including hematologic toxic- ities and grade 3 and 4 non-hematologic toxicities [2,3,5]. This toxicity increases the risks associated to the therapy and limits its applicability to highly selected patients with good performance status. The development of more selective and less aggressive preconditioning treatments is expected to reduce ACT toxicity and extend its applicability, ultimately resulting in improved clinical results.
MDSCs are critical players of tumor-induced immunosuppres- sion in both mouse models and cancer patients. This cell population accumulates in the spleen and cancers of tumor-bearing hosts, where it suppresses T cell activation, proliferation and cytotoxic function [6e8]. In mice, MDSCs include a monocytic (M-MDSCs) and a granulocytic (PMN-MDSCs) fraction, phenotypically defined as CD11bþLy6GLy6Chigh and CD11bþLy6GþLy6Cint/low cells, respectively[9]. Human MDSCs similarly comprise monocytic and granulocytic subsets and a more immature population lacking lineage marker expression [9,10]. Interestingly, although PMN- MDSC are often more abundant in mouse tumor models[8], M- MDSCs were found to be more potent suppressors of T cell response when compared with PMN-MDSCs on a per cell basis[6,7,11e13].
Based on these observations, M-MDSC-targeted depletion appears to be an attractive strategy to relief tumor-induced immunosup- pression and improve the efficacy of cancer immunotherapies, including ACT.
Here we propose an innovative approach to target M-MDSCs for cancer immunotherapy consisting in low dose administration of PEGylated lipid nanocapsules (LNCs) loaded with a lauroyl- modified gemcitabine molecule. LNCs were previously developed using a solvent-free method based on a phase-inversion tempera- ture process and have a lipoprotein-like, extremely stable structure (physical stability>18 months)[14].
Gemcitabine (marketed as gemcitabine hydrochloride for in- jection, GemHCl) is a nucleoside analogue active against different human solid tumors, including pancreatic, ovarian, breast and non- small cell lung cancer[15e18]. In addition to its use in conventional cancer chemotherapy, gemcitabine (60e120 mg/kg) has been explored as immunomodulatory agent to reduce MDSC percentage in mouse tumor models[13,19e23]. Although different encapsu- lated gemcitabine formulations have been so far developed and tested for safety and tumor-directed toxicity in murine cancer models [24e28], the interaction between gemcitabine-loaded nanoparticles and immune cells and their potential exploitation in cancer immunotherapy has never been explored. Here we investigated the interaction of GemC12-LNCs with primary mouse and human immune cells, and evaluated the immunomodulatory activity of this formulation and its combination with adoptive T cell therapy in lymphoma and melanoma mouse models. GemC12- LNCs were found to have monocyte-targeting proprieties that can be useful to deplete M-MDSCs and attenuate tumor-associated immunosuppression.
2. Materials and methods 2.1. Chemicals
Labrafac®WL 1349 (Labrafac, caprylic-capric acid triglycerides) was provided by Gattefosse S.A. (Saint-Priest, France). Kolliphor® HS15 (Kol, formerly Solutol®HS15; mixture of free polyethylene glycol 660 and polyethylene glycol 660 hydroxystearate) were
supplied by BASF (Ludwigshafen, Germany). Gemcitabine base was provided by Carbosynth (Berkshire, United Kingdom). Deionized water was obtained from a Milli-Q plus system (Millipore, Paris, France). Span®80 (Span 80), Tween®80 (Tween 80), dodecanoic anhydride, sodium chloride, urea and acid lauric were purchased from Sigma (St Quentin-Fallavier, France). Ethanol, dichloro- methane and methanol were purchased from Fischer Scientific (Loughborough, United Kingdom). 4-(N)-lauroyl gemcitabine (GemC12) was synthetized as previously described[29].
2.2. Preparation of the formulations
LNC formulation was based on a phase inversion process already described by Ref.[30]. An initial emulsion constituted by the oil (0.886 g of Labrafac), surfactant (0.691 g Kol and 0.179 Span 80) water and NaCl (0.729 g water and 0.032 g NaCl) was heated up to 75C under magnetic stirring and then cooled to 45C. This tem- perature cycle was repeated three times. During the last tempera- ture decrease, at the phase inversion temperature (55 C), an irreversible shock was induced by dilution with 1.515 g of pure water. Finally, a gentle magnetic stirring was maintained during 5 min. To obtain GemC12-loaded LNCs, 44 mg of GemC12 were solubilized in the initial mixture of Labrafac and Span80 with the addition of 100ml of acetone[30]. The sample was stirred during 1 h in a water bath at 60. Then, the same process described for the blank systems was followed. DiD-loaded LNCs were obtained adding 150ml of DiD-acetone solution (5 mg/ml) in the oily mixture of the LNC.
2.3. Determination of GemC12 encapsulation efficiency
To determine the encapsulation efficiency (EE) of GemC12 into LNCs, the suspension wasfirstly placed in a dialysis tube (cutoff of 100 kDa) and dialyzed against 4000-mL of water during 24 h. The total amount of drug was estimated by dissolving dialyzed LNC into methanol solution (1:36 v/v). This sample was analyzed at the UPLC. The UPLC system consists of an Acquity UPLC-Waters system equipped with a C18 column (1.7100 mm, Acquity UPLC Beh C18). Mobil phase consists of methanol (0.343 mL/min-flow rate).
The detection wavelengths were 248 and 266 nm. Peak heights were recorded and processed on MPower software (Waters S.A., Saint-Quentin en Yvelines, France). Drug concentration was calcu- lated from linear titration curve, using Gem-C12 methanol solu- tions at concentration range from 1 to 100mg/mL, as previously reported[29].
2.4. Physico-chemical characterization of developed LNCs
Hydrodynamic diameter (Z-Average), polydispersity index (PI) and Zeta Potential of blank, DiD and GemC12 loaded lipid nano- capsules were determined using a Zetasizer®Nanoserie DTS 1060 (Malvern Instruments S.A., Worcestershire, United Kingdom). Hy- drodynamic diameter and PI of the formulations were determined by photon correlation spectroscopy (PCS). Samples were diluted to an appropriate concentration in deionized water and each analysis was carried out at 25C with an angle detection of 173. The zeta potential values were calculated from the mean electrophoretic mobility values, as determined by laser Doppler anemometry (LDA). For LDA measurements, samples were diluted with KCl 1 mM and placed in an electrophoretic cell.
PCS and LDA analysis were performed in triplicate using a NanoZS®(Malvern Instruments, Malvern, UK).
2.5. Mice, tumor cell lines and human samples
C57BL/6, congenic CD45.1 (Ly5þ) mice and OT-I transgenic mice (C57BL/6-Tg (TcraTcrb)1100Mjb/Crl) were purchased from Charles River Laboratories and maintained under specific pathogen-free conditions in the animal facilities of the Istituto Oncologico Ven- eto (Padova, Italy) and University animal facility (SCAHU) in Angers (France). OT-I/CD45.1 F1 mice were obtained by crossbreeding OT-I and CD45.1 mice. Female C57BL/6 (WT) and female/male OT-I/
CD45.1 transgenic mice were used between 8 and 12 weeks of age. Experiments involving animals were performed according to the national guidelines and approved by the local ethics committees.
The EG7-OVA lymphoma and B16-F10 (indicated in the text as B16) melanoma cell lines were obtained from the American Type Culture Collection and cultured in DMEM 10% FBS (Gibco) supple- mented with 2 mML-glutammine, 10 mM HEPES, 20mM 2b-Mer- captoethanol, 150 U/ml streptomycin, 200 U/ml penicillin. G418 was added to EG7-OVA cell cultures to maintain OVA expression.
Cultures were maintained at 37C with 5% CO2. To establish tu- mors, EG7-OVA cells (0.5 106 cells/mouse) or B16 cells (0.25106cells/mouse) were injected subcutaneously on the left flank of C57BL/6 mice, and tumor growth was monitored every 2 days by digital calipers.
Peripheral blood from melanoma patients was received from the Melanoma Oncology Unit of the Veneto Institute of Oncology (IOV-IRCSS). Patients had a diagnosis of active metastatic mela- noma and were treated with anti-CTLA-4 antibodies (Ipilimumab).
Blood was collected at a minimum of 3 months after treatment completion (standard 4 cycles), corresponding to a drug-free in- terval superior to 4 half-lives of treatment. All patients had an initial stabilization of the metastatic disease after ipilimumab, allowing for treatment-free interval, and blood samples were taken at the time of disease relapse, before new treatment start. The project was approved by the Ethics Committee and all patients gave their informed consent.
2.6. In vitro induction of MDSCs from bone marrow cells
Tibias and femurs from healthy C57BL/6 mice were removed using sterile techniques and BM wasflushed. Red blood cells were lysed with an ammonium chloride-potassium solution. To obtain in vitro-derived MDSCs, 1.5106cells/well were plated into 6 well plates (Falcon, BD Biosciences) in 2 ml of medium supplemented with 40 ng/ml GM-CSF and 40 ng/ml IL-6 (Peprotech Inc.). Cells were maintained at 37C in 5% CO2-humidified atmosphere for 4 days. Blank LNCs, GemC12-LNCs or free GemC12 were added to the culture at day 0 at the following concentrations: [GemC12] 0.11mg/
ml, [LNC] 4.35mg/ml. Culture medium was RPMI 1640 (Euroclone) supplemented with 2 mML-glutamine, 10 mM HEPES, 20mM 2b- Mercaptoethanol, 150 U/ml streptomycin, 200 U/ml penicillin, and 10% heat inactivated FBS (Biochrom).
2.7. Preparation of cell suspensions from tumors and lymphoid organs
Mice were euthanized and spleens, tumors and axillary and inguinal lymph nodes were collected. Tumors were cut in small pieces and incubated with a digestive solution composed of colla- genase IV (1 mg/ml), hyaluronidase (0.1 mg/ml), DNase (0.03 KU/
ml) at 37C for 1 h. At the end of digestion, tumor cells were collected,filtered to remove cell clumps and debris and washed in complete medium twice to remove all digestive solutions prior to subsequent use. Spleens and lymph nodes were mechanically dis- aggregated andfiltered with 100mmfilters.
2.8. Cytofluorimetric analysis
For the analysis of mouse samples, red blood cells were removed from cell suspension using an ammonium chloride-potassium lysing solution and cells were washed with cold PBS and incu- bated 10 min at þ4 C with purified anti-FcgR antibody (clone 2.4G2) to minimize non-specific antibody binding. Antibodies of interest were added to cell suspensions following FcgR blocking and incubated 20 min atþ4C in the dark. Employed antibodies were: anti Ly6G (Gr-1) APC (clone RB6-8C5), anti-CD11b PE-Cy7 or CD11b PerCP-Cy5.5 (clone M1/70), anti-LY6G APC-Cy7 or APC (clone 1A8), anti-LY6C eFluor 450 or FITC (clone HK1.4), anti- CD45.1 PE (clone A20), anti-CD8 PerCP-Cy5.5 (clone 53e6.7), anti- CD3 FITC (clone 145-2C11), anti-CD115 PE (clone AFS98), anti-CD25 APC (clone PC61.5), anti-Foxp3 PE (clone NRRF-30) (all from eBio- science); anti-I-A/I-E PerCP-Cy5.5 (clone M5/114.15.2) anti-CD11c BV421 or APC (clone HL3), anti-CD4 APC or PE or PerCP-Cy5.5 (clone RM4-5) (all from BD Biosciences), anti-F4/80 FITC (AbD Serotec, clone CI:A3-1). Aqua LIVE/DEAD®dye (Invitogen) was used to analyze cell viability. For intracellular staining of INF-gthe BD Citofix/Cytoperm Kit was used, according to manufacturer's in- struction. INF-gwas stained using anti-INF-gFITC (clone XMG1.2, BD Biosciences). For intranuclear staining of Foxp3 antigen the Biolegend FOXP3fix/perm buffer set was employed, according to manufacturer instruction. Hematopoietic stem and precursors cells in the bone marrow were defined as follows: ST-HSC: Lineagec- KitþSca1þCD34þFlk2; LT-HSC: Lineage-c-KitþSca1þCD34Flk2 MPP: Lineagec-KitþSca1þCD34þFlk2þ; CMP: Lineagec-KitþS- ca1CD34þCD16/CD32; GMP: Lineage-KitþSca1CD34þCD16/
CD32þ; MEP: Lineagec-KitþSca1CD34CD16/CD32. Employed antibodies were: CD16/32 FITC (clone 93), Sca-1 AF700 (clone D7), CD127 APC-efluor780 (clone A7R34), CD34 efluor 450 (clone RAM34), All from eBioscience; CD117 PE-Cy7 (clone 2B8), CD135 PE (clone A2F10.1), Lineage Antibody cocktail APC (REF 558074) all from BD bioscience.
Human cells were incubated with FcReceptor (FcR) Blocking Reagent (Miltenyi Biotec) to saturate FcR and then labeled with the following monoclonal antibodies: CD3 ECD (Beckman Coulter;
clone UCHT1), CD14 FITC (BD Pharmingen; clone M5E2), CD19 FITC (BD Pharmingen; clone HIB19), CD56 FITC (BD Biosciences; clone NCAM16.2), CD11 b PE or AF700 (Beckman Coulter; clone IM2581U), CD16-FITC (Miltenyi Biotec; clone VEP13), CD33 PECy7 (BD Biosciences; clone P67.6), CD15-V450 (BD Biosciences; clone MMA), IL4Ra-PE (R&D Systems; clone 25463). Human immune cell subsets were defined based on morphologic gating and on surface antigen staining as follows: T cells (CD3þ), monocytes (CD14þ), B cells (CD19þ), NK cells (CD56þ), neutrophils (CD11bþ/CD16þ), eo- sinophils (CD11bþ/CD16), granulocytes (CD15þ). IL4Raþ CD14þ human M-MDSC were defined as described in Ref.[10]. Flow data were acquired with a BD LSRII or BD FACS Calibur instrument and analyzed with FlowJo (Tree Star, Inc.) software.
2.9. T cell suppression assays
CD11bþ cells were isolated from tumors with anti-CD11b MicroBeads (Miltenyi Biotec). OT-I/CD45.1 splenocytes derived from the spleen of OT-I mice were labeled with carboxyfluorescein succinimidyl ester (CFSE, CFSE-Cell Trace Kit, Invitrogen Molecular Probe) according to manufacturer's instructions. A mixed leucocyte peptide culture (MLPC) was prepared by mixing g-irradiated C57BL/6/CD45.2 splenocytes with OT-I/CD45.1 CFSE-labeled sple- nocytes in order to obtain 1% OT-I T CD8þlymphocytes in thefinal culture. MPLC was plated in flat-bottom 96-well plates (0.6106cells/well) and 1mg/ml H-2 Kb-restricted OVA peptide (OVA257e264, SIINFEKL, synthesized by JPT, Peptide Technologies)
was added to the culture to stimulate OVA-specific OT-I T CD8þ lymphocytes. CD11bþcells from tumor masses orin vitro-derived MDSCs were added at degreasing percentages with respect to MLPC culture (24%, 12%, 6%, 3% and 1.5%). Cultures were maintained at 37C with 5% CO2in RPMI medium supplemented with 10% FBS Superior (Biochrom), 2 mM L-glutammine, 1 mM Na-pyruvate, 150 U/ml streptomycin, 200 U/ml penicillin, 20 mM 2b- mercaptohetanol.
After 3 days of culture, cells were collected, washed with cold PBS and stained with anti-CD45.1 PE (clone A20, eBioscience) and anti-CD8 PerCP-Cy5.5 (clone 53e6.7, e Bioscience). Flow data were acquired with a FACSCaliburflow cytometer (BD Biosciences) and analyzed with FlowJo (Tree Star, Inc.) software. The percentage of cells in 0e7 generation of proliferation was determined based on CFSEfluorescence within the CD8þCD45.1þgate, using the FlowJo software. Proliferation index reported inFig. 1is the total number of divisions divided by the number of cells that went into division and was calculated as a tool of FlowJo software. The % of suppression (Fig. 4) was calculated as 100e% of proliferation. % of proliferation was obtained by summing the % of T cells in generation 3e7 and by normalizing with respect to control MLPC without CD11bþcells (assumed to be 100% of proliferation).
For T cell suppression assays using human samples, peripheral blood mononuclear cells (PBMCs) were isolated from peripheral blood of melanoma patients by density gradient centrifugation on Ficoll-Paque PLUS (GE Healthcare-Amersham), as previously described[31]. Monocytes were obtained by depletion of CD3þ/ CD19þ/CD56þ lymphocytes from PBMCs by immunomagnetic beads (Miltenyi Biotec). Monocytes were diluted at 106cells/ml and treated with 2.5mg/ml GemC12-LNCs or left untreated and incu- bated at 37C for 1.5 h. LNC-treated monocytes were washed and their ability to inhibit the proliferation of an allogenic mixed leu- cocyte reaction (MLR) was tested. To perform allogenic MLR, PBMCs from a healthy donor were stained with 0.5mM CellTrace™Violet Cell Proliferation Kit (Invitrogen) and cultured at a 1:1 ratio with a pool of gamma-irradiated PBMCs from three different healthy do- nors. Monocytes from melanoma patients were added as third part to the culture at a 1:1 ratio with CellTrace TM-labeled PBMCs. Cells were cultured in a round-bottom 96-well plate at 37C and 5% CO2
in arginine-free-RPMI medium (Biological Industries), supple- mented with 150 mM arginine, 10% FBS, 10 U/ml penicillin and streptomycin and HEPES. After 7 days of culture, cells were har- vested and stained with anti-CD3ε PE-Cy7 (Beckman Coulter) antibody. Data acquisition was performed on LSRIIflow cytometer (BD Bioscience) and samples were analyzed by FlowJo software (Tree Star, Inc.)
2.10. DiD-LNC uptake studies on human blood
For LNC uptake studies on human blood, peripheral blood specimens from three healthy donors were lysed to remove red blood cells, with a hypotonic solution of ammonium chloride po- tassium. Cells were plated (3106cells/well) into a 24-well tissue culture plate (Falcon, BD Bioscience) in IMDM (Iscove's Modified Dulbecco's Medium, Gibco Invitrogen, California, USA) supple- mented with 10% FBS (Fetal Bovine Serum, Gibco), 10 mM HEPES, 150 U/ml streptomycin, 200 U/ml penicillin, in the presence of DiD- loaded LNCs for 5 h. After incubation, cells were harvested, washed and analyzed byflow cytometry.
2.11. Inhibition of LNC uptake by LY294002
Total spleen cells were isolated from EG7 tumor-bearing mice and pre-incubated 45 min at 37C either with LY294002 (Sigma Aldrich) or with vehicle (DMSO) as negative control, diluted in
complete or serum-free medium. Cells were subsequently incu- bated 4 h either in complete cell culture medium or in serum-free medium with 5 mg/ml DiD- LNCs. At the end of the incubation period, cells were extensively washed and the percentage of DiD- positive cells within different cell subsets was determined by flow cytometry.
2.12. Drug administration
GemC12-loaded LNCs, Gemcitabine hydrochloride and free GemC12 were administered either SC, on the right flank (flank opposite to the tumor mass), or intravenously in the tail vein. All treatments were administered at 11 mg/kg GemHCl or molar equivalent GemC12 dose (either free or encapsulated in LNCs).
Gemcitabine hydrochloride was purchased from Accord Healthcare (Italy) and diluted at the required concentration with 0.9% sterile NaCl. Forin vivoexperiments, GemC12 was dissolved in ethanol, Tween®80 and water (87.6/5.5/6.9 v/v/v) to form a micellar solu- tion. Untreated control mice received injections of sterile 0.9% NaCl solutions. Drug treatments were administered at day 8 following tumor injection, unless differently indicated. At day 8, tumor sur- face was ~50 mm2 for EG7-OVA tumors and ~20 mm2 for B16 tumors.
2.13. Adoptive T cell therapy (ACT)
OVA-specific CD8þCD45.1þcytotoxic T cells were prepared by culturing OT-I/CD45.1 splenocytes in presence of 1 mg/ml OVA257e264(SIINFEKL) peptide and 20 IU/ml of IL-2 (Novartis). Cell cultures were maintained 7 days before in vivo injection. OVA- specific T cells were injected IV on the tail vein (0.25106cells/
mouse). The number of injected OTI T cells was selected on the basis of titration studies as maximum number inducing a non-complete response in treated animals.
TERT-specific T cells were originally obtained from a mixed- leukocyte peptide culture set up with splenocytes obtained from vaccinated mice [32] and cultured in presence of 0.1 mg/ml TERT198e205peptide (VGRNFTNL) and 20 IU/ml IL-2. TERT-specific T cells were injected IV on the tail vein (5 106 cells/mouse).
Following TERT-specific T cell injection, mice were intraperitoneally injected with 50mg of FGK45.5 antibody against CD40 and then rubbed with Imiquimod after the administration of the TERT198e205
peptide (200mg) subcutaneously at the tail base, in order to boost T cell response[33].
2.14. Statistical analysis
Values are reported as means±standard errors (SE) or standard deviations (SD). Survival experiments are reported as Kaplan- Meyer curves and significance was determined with log-rank test with post-test correction for pairwise multiple comparisons by the Holm-Sidak method. Student's t-test was performed on groups with normal distribution, as evaluated by Shapiro-Wilk normality test. For non-normal data, group comparison was performed using the Mann-Whitney Rank Sum Test. Values were considered sig- nificant with p<0.05 and are indicated as *p<0.05; **p<0.01 and
***p<0.001.
3. Results
3.1. Physicochemical characterization of LNC formulations
GemC12-LNCs were formulated as described in Ref.[29]. 4-(N)- lauroyl-modified gemcitabine (gemcitabine-C12, GemC12) was obtained by adding a single 12-carbon alkyl chain to the amine
group of gemcitabine molecule. This modification confers to the molecule an amphiphilic structure allowing encapsulation into LNCs at the water-oil interface (Fig. 1)[29]. In experiments using LNCs loaded with the DiD dye, DiD amphipathic molecules were encapsulated in place of GemC12 and similarly localized at the water-oil interface (Fig. 1)[34]. In addition, lauroyl modification of gemcitabine is known to increase drug stability by protecting it from enzymatic deamination [29,35]. LNC formulations (blank, DiD-loaded and GemC12-loaded) showed a size ranging from 55 to 67 nm (Z-average) and a slightly negative net surface charge (zeta potential around3 mV). The polydispersity index (PI) was lower than 0.1 for all these formulations, indicating a monomodal and narrow size distribution (Fig. 1).
3.2. GemC12-LNCs deplete the monocytic-macrophagic fraction of in vitro-derived MDSCs
Since gemcitabine was reported to be particularly effective in depleting mouse MDSCs, as compared with other chemothera- peutics[13,20], we sought to determine the efficacy of GemC12- LNCs as MDSC-targeting agent. To this end, wefirst assessed the effects of GemC12-LNCs on in vitro-derived MDSCs, obtained by culturing mouse bone marrow cells in presence of IL-6 and GM-CSF, as previously described[36]. In thisin vitrosystem, both GemC12- LNCs and free GemC12 caused a dramatic depletion of monocytes- macrophages (Gr1int and Gr1low-neg cell fraction), while leaving granulocytic cells (Gr1high) unaffected (Fig. 2A). Consistently, a considerably lower number of living cells were recovered in drug- treated cultures as compared with not treated controls (Fig. 2B).
The cytotoxic activity of GemC12-LNCs was completely due to the drug cargo, since blank nanocarriers did not alter cell phenotype nor reduced cell viability (Fig. 2A and B). MDSC cultures lacking the monocytic-macrophagic component were less immunosuppressive as compared with control cultures, when assessed on a per cell basis (Fig. 2C), in line with previous reports showing a greater immunosuppressive activity of M-MDSCs with respect to PMN- MDSCs[6,7,11e13].
Thesein vitrodata supported the use of GemC12-LNCs as ther- apeutic agent to target M-MDSCs, prompting us to investigate the
in vivoimmunomodulatory activity of this formulation.
3.3. Low dose GemC12-LNCs selectively reduce M-MDSCs in tumor and spleen
Subcutaneous (SC) injection of nanoparticulate formulations has been associated with the formation of depots from which particles are gradually absorbed in the systemic circulation, resulting in a slower clearance as compared to intravenous (IV) injection[37,38]. Of note, the small LNC size was expected to favor their absorption by SC lymphatic drainage, since small nano- particles (below 100 nm) are known to be transported through the interstitial flow into lymphatic capillaries[39,40]. Therefore, we investigated the use of the SC route for LNC administration, with the intent of obtaining a sustained drug release and, possibly, an enhanced efficacy, as compared to standard IV injection.
Biodistribution studies were performed using LNCs loaded with the DiD dye, without any drug cargo. Mice bearing established EG7- OVA subcutaneous lymphomas were injected either IV (on the tail vein) or SC (on theflank opposite to tumor injection site) with DiD- loaded LNCs, and a semi-quantitative measurement of DiD fluo- rescence was performed on explanted organs at sequential time points. As shown in Supplementary Fig. S1, DiD-loaded LNCs reached peripheral tissues following both IV and SC injection, although DiDfluorescence was generally lower after SC injection as compared with IV. The IV administration resulted in a marked LNC accumulation in the liver and in a rapidfluorescence decrease from 24 to 72 h in all explanted organs (Supplementary Fig. S1).
Conversely, following SC injection, DiD signal increased with time in the spleen and showed little variation in the other organs, sug- gesting a slow LNC absorption and a more gradual distribution and elimination as compared to IV injection (Supplementary Fig. S1).
DiD tracking allows to qualitatively evaluate LNC distribution in peripheral tissues, but it is not predictive of tissue drug concen- tration over time, since thefluorescent probe and the drug may have very different elimination kinetics following the release from nanocarriers. For this reason, we further compared the IV and SC LNC administration by directly assessing the impact of GemC12- loaded LNCs on MDSC frequency. EG7-OVA tumor-bearing mice Fig. 1.Structure and physicochemical characteristics of GemC12-LNCs.A. Chemical structure of lauroyl-modified gemcitabine (GemC12).B. graphical representation of GemC12- and DiD-loaded LNCs.C. Size curve profiles of blank and GemC12-loaded LNCs as assessed by photon correlation spectroscopy.D. Table showing the physicochemical characteristics of blank, GemC12-loaded and DiD-loaded LNCs. EE: encapsulation efficiency of GemC12 or DiD; PI: polydispersity Index. Mean±SD; n¼3.
were treated either with 11 mg/kg of GemHCl IV or with an equivalent molar dose of GemC12-LNCs, either injected IV or SC.
The frequency of spleen and tumor-infiltrating immune pop- ulations was determined byflow cytometry 24 h following treat- ment. Myeloid cell subsets were defined as reported in Supplementary Fig. S2, while CD4þand CD8þT cells were gated as CD3þCD4þand CD3þCD8þcells, respectively. The administration of GemC12-LNCs resulted in a marked reduction in splenic and tumor-infiltrating M-MDSCs, while other myeloid subsets showed minor variations (Fig. 3A and B). Remarkably, the strongest M- MDSC decrease was observed in mice treated SC with GemC12- LNCs, as compared with both IV GemC12-LNCs and GemHCl in the tumor (Fig. 3A), and with IV GemC12-LNCs in the spleen (Fig. 3B). In mice treated SC with GemC12-LNCs and IV with GemHCl, the percentages of PMN-MDSCs and macrophages in tu- mors were slightly increased, probably as a consequence of M- MDSC frequency reduction (Fig. 3A). In the same groups, splenic DCs were only weakly decreased, while PMN-MDSCs were signifi- cantly reduced only in mice treated with GemHCl (Fig. 3B). None of the treatments affected the frequency of either splenic or tumor- infiltrating T cells (Fig. 3A and B). Since SC injection of GemC12-
LNCs was associated to an enhanced M-MDSC reduction as compared with IV injection, we selected the SC route for further experiments.
3.4. Low dose GemC12-LNCs transiently reduce M-MDSCs and relieve tumor-associated immunosuppression
We next evaluated the duration of M-MDSC reduction following SC injection of GemC12-LNCs, GemHCl or free GemC12 at very low dose (11 mg/kg GemHCl or molar equivalent GemC12 and GemC12- LNCs). To this end, we measured the frequency of MDSCs and macrophages at sequential time points following treatment. Within the tumor site, M-MDSCs were strongly reduced up to 48 h following GemC12-LNC administration (Fig. 4A). Remarkably, GemC12-LNCs caused a stronger and prolonged M-MDSC reduction as compared with free GemC12 and GemHCl (Fig. 4A). PMN-MDSCs exhibited no significant variations over the considered time points, except for a moderate frequency increase in the GemC12-LNC group at 24 h, likely compensating for the decreased M-MDSC percentage (Fig. 4A). Tumor macrophages had a slight tendency to increase at the 72-h time point, notably in the GemHCl group, Fig. 2.Cytotoxic activity of GemC12-LNCs on in vitro-derived MDSCs. Mouse bone marrow cells were cultured in presence of GM-CSF and IL-6 to induce MDSC generation. Blank-LNCs, GemC12-LNCs or free GemC12 were added at day 0. Cells were recovered after 4 days of culture and tested for surface marker expression and immunosuppressive activity.A.
Representativeflow cytometry plots of MDSCs showing expression of CD11b and Gr-1 markers (gated on living cells).B. Ratio of numbers of recovered cells at day 4 on plated cells at day 0, normalized with respect to untreated control cultures. Means±SE of 3 independent experiments.C. Immunosuppressive activity of MDSCs. MDSCs were recovered and co- cultured at decreasing percentages with CFSE-labeled, antigen-activated CD8þT cells. Histograms represent T cell countversusCFSEfluorescence intensity. For each histogram, T cell proliferation index and the percentage of T cells that underwent 0e2 cell divisions are reported. Data for co-cultures with 24% and 12% MDSCs are shown and are representative of 3 independent experiments.
although the reason of this variation was unclear (Fig. 4A).
In the spleen, M-MDSCs were reduced for 48 h following either GemC12-LNC or GemHCl administration (Fig. 4B). Of note, at the 24-h time point, M-MDSC frequency was significantly lower in the GemC12-LNC group as compared with the other treatment groups (Fig. 4B). At 24 h, PMN-MDSCs were significantly decreased only in the GemHCl group, suggesting a lower selectivity of free GemHCl for monocytic cells with respect to GemC12-LNCs (Fig. 4B). In all treated animals, spleen PMN-MDSCs were reduced at the 72-h time point (Fig. 4B). Such delayed PMN-MDSC reduction might be interpreted as a secondary effect of spleen M-MDSC depletion, since in mice this latter population has been reported to include precursor cells with the ability to differentiate in both monocytes and granulocytes[13,41], although this hypothesis would require further investigation. Importantly, at the time of analysis, tumor sizes were comparable among the different groups (Fig. 4C), thus excluding the possibility of differences in myeloid cell expansion due to an altered tumor growth. Consistently with the lack of toxicity on tumor macrophages (Fig. 4A), administered treatments did not reduce the frequency of resident macrophages in the spleen (red pulp macrophages, Fig. 4B) and in the liver (Kupffer cells, Supplementary Fig. S3).
In both the spleen and tumor site myeloid cell subsets were not increased at 1 week from GemC12-LNC injection (Supplementary Fig. S4), indicating the absence of a delayed rebound MDSC expansion, differently from what reported with higher gemcitabine doses[42].
T cell frequencies in both the spleen and the tumor were not affected at all considered time points (data not shown). In addition, neither GemC12-LNC nor GemHCl administration resulted in the
reduction of spleen and tumor-infiltrating T regulatory cells (Fig. 4D). In summary, low dose GemC12-LNCs strongly reduced tumor-infiltrating M-MDSCs with improved efficacy as compared with GemHCl. Moreover, no substantial differences were observed between free GemC12 and GemHCl, suggesting that the enhanced efficacy of GemC12-LNCs over GemHCl could not be attributed to the use of a modified gemcitabine molecule, but rather to factors related to the nanocarriers.
We reasoned that the short duration of M-MDSC reduction following low dose GemC12-LNC and GemHCl injection, could be due to the rapid reconstitution of this cell population from bone marrow hematopoietic precursors, indicating a minimal bone marrow toxicity of the administered treatments. To investigate this hypothesis, we quantified the number of hematopoietic stem and precursor cells in the bone marrow 24 h following treatment administration. As shown inFig. 4E, the number of hematopoietic stem and precursor cells in the bone marrow was not reduced in mice receiving GemC12-LNCs, as compared to controls, except for a reduction in the number of common myeloid progenitors (CMPs).
In addition, only GemHCl but not GemC12-LNCs exerted a signifi- cant toxicity on granulocyte-macrophage progenitors (GMPs, Fig. 4E).
The suppression of bone marrow hematopoietic function is a common adverse event of anticancer chemotherapy, which results in the reduction of circulating mature blood cells with a nadir at 7e14 days and a recovery time of 21e28 days from treatment administration[43]. The transient M-MDSC reduction induced by low dose GemC12-LNC administration occurred in absence of a substantial impairment of bone marrow hematopoietic activity, thus supporting the minimal toxicity of this treatment.
Fig. 3.GemC12-LNCs selectively reduce tumor and spleen M-MDSCs. Mice bearing EG7-OVA tumors were treated with either GemC12-LNCs (injected SC or IV) or GemHCl (IV) and examined 24 h later. Data are presented as percentages of myeloid cell subsets and T cells in the tumor (A) and in the spleen (B). M4: macrophages. Means±SE, n¼4 mice per group. *p<0.05, **p<0.01, ***p<0.001. Student'st-test. Statistical comparison is between each treatment group and the untreated controls, unless differently indicated by lines.
To assess whether the targeted depletion of M-MDSC by GemC12-LNCs was paralleled by an attenuation of MDSC- dependent immunosuppression in the tumor microenvironment, we isolated total MDSCs (CD11bþ cells) from tumor masses and tested their ability to impair T cell proliferationex vivo. Purified CD11bþcells were co-cultured with CFSE-labeled antigen-activated T cells, and T cell proliferation was measured based on CFSE dilu- tion. As shown inFig. 4F, CD11bþcells isolated from mice treated with either GemC12-LNCs or GemHCl had a reduced ability to suppress T cell proliferation as compared with control CD11bþcells from untreated mice. Of note, treatment with GemC12-LNCs was associated with a stronger reduction of MDSC suppressive activity as compared with GemHCl, consistently with the more intense MDSC depletion.
3.5. LNCs are efficiently uptaken by M-MDSCs in tumor-bearing mice and human blood
The effective M-MDSC-targeting by GemC12-LNCs could be due,
at least in part, to an efficient LNC uptake by this cell population. To investigate this hypothesis, we injected EG7-OVA tumor-bearing mice with DiD-loaded LNCs and determined the percentage of LNC-containing cells (DiDþ) within different splenic and tumor- infiltrating immune cell subsets. As expected, the greatest LNC internalization was observed in tumor-infiltrating M-MDSCs and macrophages and in splenic M-MDSCs (Fig. 5A). Lower percentages of DiDþcells were found within PMN-MDSCs in both sites, and red pulp macrophages and DCs in the spleen (Fig. 5A). The uptake of DiD-LNCs by CD4þand CD8þT cells was negligible (Fig. 5A).
We next investigated whether leucocytes in human blood had a similar LNC uptake pattern as compared to mouse cells. Human peripheral whole blood from healthy donors was incubated with DiD-loaded LNCs, and the percentage of DiDþcells within different immune cell subsets was determined byflow cytometry. As shown inFig. 5B, human monocytes had a far higher rate of LNC uptake as compared with other leucocytes, including T and B cells, NK cells, neutrophils and eosinophils, even at very low LNC concentrations.
The extremely efficient LNC internalization by human blood Fig. 4.GemC12-LNCs transiently reduce M-MDSCs and relieve tumor-associated immunosuppression.A,B. Frequency of M-MDSCs, PMN-MDSCs and macrophages (M4) in the tumor (A) and M-MDSCs, PMN-MDSCs and red pulp M4in the spleen (B) at 24, 48 and 72 h following the administration of indicated treatments. Means±SE, n¼4 mice per group.C.
Tumor surface (mm2) at the time of the assessment.D. Treg (CD3þCD4þCD25þFOXP3þ) frequency in the tumor mass and in the spleen 24 h post GemHCl or GemC12-LNC injection.
Means±SE, n¼4 mice per group.E. Number of hematopoietic stem and precursor cells in the bone marrow 24 h following treatment administration. LT-HSC: long term he- matopoietic stem cells. ST-HSC: short term hematopoietic stem cells. MPP: multipotent progenitors. CMP: common myeloid progenitors. GMP: granulocyte-macrophage pro- genitors. MEP: megakaryocyteeerythrocyte progenitors. Means±SE, n¼4 mice per group.F.Ex-vivomeasurement of myeloid cell (CD11bþ) suppressive activity 24 h following drug treatment. OT-I/CD45.1 splenocytes were labeled with CFSE and cultured in presence of the SIINFEKL OVA peptide andg-irradiated CD45.2 splenocytes as feeder cells, to induce antigen-specific activation of OT-I CD8þT cells. CD11bþcells were isolated from tumor masses and added to the splenocyte culture at decreasing percentages to test their ability to impair T cell proliferation. CFSE dilution upon T cell division was measured byflow cytometry after 3 days of co-culture and used to calculate the % of suppression.
Means±SE of 3 independent experiments. *p<0.05 **p<0.01 ***p<0.001, ns: not significant. Student'st-test. Statistical comparison is between each treatment group and the untreated controls, unless differently indicated by lines.
monocytes supported the possibility of using GemC12-loaded LNCs to target M-MDSCs in cancer patients. Thus, we subsequently evaluated the targeting proprieties of GemC12-LNCs using fresh blood samples from melanoma patients. Following blood incuba- tion with DiD-loaded LNCs, maximum LNC internalization was observed within the IL4RaþCD14þM-MDSC fraction[10]and in the overall monocyte population (CD14þ cells) (Fig. 5C, left panel), consistently with the uptake pattern observed in blood cells from heathy donors. To investigate the ability of GemC12-LNCs to relieve M-MDSC-mediated immunosuppression, we isolated PBMCs from melanoma patients' blood samples and enriched them of CD14þ monocytes by immunomagnetic depletion of CD3þ, CD19þ and CD56þcells. Enriched CD14þcells were then either pre-incubated with GemC12-LNCs or left untreated, washed, and co-cultured with T lymphocytes activated by alloantigens in a mixed leuco- cyte reaction (MLR) assay, to measure their ability to suppress T cell proliferation.Fig. 5D shows the results of this functional assay using PBMCs from 4 different patients. Not all tested CD14þcells had a detectable immunosuppressive activity, consistently with previous reports showing an inter-patient variability in MDSC expansion [31]. However, for 2 out of 4 patients (patient number 1 and 2, Fig. 5D) we observed a reduction in the percentage of proliferating T cells following addition of CD14þcells to the MLR culture. In these two patients, treatment of CD14þcells with GemC12-LNCs resulted in the recovery of T cell proliferation, indicating an inhibitory effect of GemC12-LNCs on immunosuppressive monocytic cells. In the other two patients, where immunosuppression was absent, LNCs had no activity.
3.6. M-MDSCs internalize LNCs by macropinocytosis
Monocytes/macrophages are professional phagocytes, with enhanced ability to uptake particulate materials as compared with non-myeloid cells. Thus, we reasoned that immune cell-specific mechanisms could be involved in the extremely efficient LNC up- take by these cells, including macropinocytosis, and immune receptor-mediated endocytosis and phagocytosis.
Macropinocytosis was previously proposed as the mechanism responsible for the internalization of gold nanoparticles by primary human blood monocytes and monocyte-derived macrophages[44]
and of poly-propylene sulfide nanoparticles by primary, bone marrow mouse monocytes [45]. To investigate whether this mechanism could be similarly involved in LNC internalization by M-MDSCs, we evaluated the efficiency of DiD-LNC uptake in pres- ence of the pan-phosphoinositide 3 kinase (PI3K) inhibitor LY294002, which is known to inhibit macropinocytosis process [46e48]. Splenocytes from EG7 tumor-bearing mice were incu- bated for 4 h with 5 mg/ml DiD-LNCs following 45-min pre- incubation with LY294002, and the efficiency of DiD internaliza- tion was assessed byflow cytometry. As shown inFig. 6A, LNC uptake by M-MDSCs was reduced by LY294002 in a dose- dependent manner. A slighter inhibitory effect was observed also on dendritic cells (whose uptake was nonetheless much lower than M-MDSCs), while all other cell subsets showed a very limited nanocarrier internalization. LY294002 has been reported to inhibit both macropinocytosis and Fcg receptor-mediated phagocytosis, while it exerts limited effects on scavenger-receptor dependent endocytosis[46,47]. Nanocarriers are known to bind plasma pro- teins by multiple attractive forces, thus also resulting in particle opsonization by IgG and complement activation[49,50]. In our study,in vitrouptake experiments were performed using cell cul- ture medium containing heat-inactivated fetal bovine serum (FBS), thus excluding a role of complement activation in LNC recognition.
However, since IgG in FBS might still cross-react with mouse Fcg receptors and induce Fcg receptor-mediated phagocytosis, we
further evaluated LNC uptake by M-MDSC in serum-free medium.
As shown inFig. 6B, maximum nanocarrier uptake was not signif- icantly reduced in total absence of serum (compare withFig. 6A).
Furthermore, LY294002 strongly inhibited LNC internalization also in serum-free medium, indicating that LNC uptake was not dependent on IgG-mediated opsonization and phagocytosis. Thus, M-MDSCs were found to internalize LNCs at least in part through macropinocytosis. The high LNC uptake by M-MDSCs and mono- cytes might hence depend on a higher macropinocytic activity of these cells as compared with other immune cell populations.
3.7. Preconditioning with low dose GemC12-LNCs enhances ACT efficacy
The relief of tumor-induced immunosuppression is known to improve the efficacy of immunotherapeutic treatments, including ACT[13,51,52], by creating a more favorable environment for T cell recruitment, proliferation and function. We hence speculated that M-MDSC depletion by low dose GemC12-LNCs could be used to precondition tumor-bearing mice prior to ACT in order to enhance antitumor efficacy.
Since M-MDSC depletion by GemC12-loaded LNCs had a maximum 48-h duration, wefirst investigated whether repeated LNC administrations after T cell transfer could be useful to extend the time window of attenuated immunosuppression and foster T cell expansion upon adoptive transfer. Three different LNC admin- istration schedules were tested in combination with ACT. The fre- quency of transferred T cells (tracked by the CD45.1 congenic marker) in lymph nodes and tumor masses, and the IFNgproduc- tion upon antigen re-stimulation were evaluated. ACT protocol consisted in the IV infusion ofin vitropre-activated T cells specif- ically recognizing the OVA model antigen expressed by EG7-OVA tumors. The frequency of overall CD8þCD45.1þ cells and CD8þCD45.1þIFNgþ cells in both lymph nodes and tumors was dramatically reduced when mice were treated with one or two LNC injections after T cells infusion (schedules II and III, Fig. 7A), as compared to mice receiving either ACT alone or a single GemC12- LNC dose prior to ACT (schedule I, Fig. 7A). These results indi- cated the existence of a deleterious effect of GemC12-loaded LNC administration in the days immediately following T cell infusion, likely due to the direct drug cytotoxicity on transferred T cells.
Conversely, following one single GemC12-LNC injection 24 h before ACT (schedule I, Fig 7A), the frequency of overall and IFNg-pro- ducing CD8þCD54.1þT cells was not changed in lymph nodes and tended to increase in tumors, as compared with control mice receiving ACT alone. Interestingly, in mice receiving ACT alone or ACT plus a single dose of GemC12-loaded LNCs, tumor-infiltrating CD8þCD54.1þT cells produced IFNg even without re-stimulation with their cognate peptide, thus suggesting that these cells were already functionally activatedin vivo(Fig. 7A). Based on these re- sults, we selected the schedule I (Fig. 7A) as the best LNC admin- istration schedule in combination with ACT.
Since we previously observed a very low level of DiD-LNC up- take by endogenous T cells (Fig. 5A), we wondered whether antigen-activated T cells transferred in ACT protocols might have an enhanced LNC internalization as compared to endogenous T cells, which likely mostly comprise resting cells. To investigate this hy- pothesis, we injected mice with DiD-loaded LNCs following treat- ment with GemC12-LNCs plus ACT, and we measured the frequency of DiDþ cells within either the endogenous or the transferred (CD45.1þ) T cells (Fig. 7B). As shown in Fig. 7B, we observed a significantly higher frequency of DiDþ cells within transferred CD8þCD45.1þT cells as compared with endogenous CD8þCD45.1-T cells, although still below the values reached by monocytic cells and macrophages (Fig. 5A). The increased ability of activated T cells
Fig. 5.LNC uptake by mouse and human immune cells.A. DiD-loaded LNCs were SC administered to mice bearing EG7-OVA tumors at day 8 post tumor injection, and cell uptake was analyzed 24 h later. Data are reported as percentages of DiDþcells within indicated cell populations. Means±SE, n¼4 mice.B. Hemolyzed whole human blood from 3 different healthy donors was incubated for 5 h with decreasing concentrations of DiD-loaded LNCs (LNC concentrations are reported in thefigure legend). The percentage of DiDþcells in
to internalize LNCs may contribute to the reported sensitivity of transferred T cells to GemC12-LNC cytotoxicity. Besides, since activated lymphocytes are a highly cycling cell population, they are likely to be particularly sensitive to the cytotoxic activity of anti- proliferative drugs, such as gemcitabine.
Having defined the best administration schedule of GemC12- LNCs in combination with ACT, we next evaluated the therapeutic efficacy of the combined protocol in the EG7-OVA tumor model (Fig. 8A and B). As expected, ACT with OVA-specific T cells signifi- cantly improved mouse survival even without any preconditioning treatment (Fig. 8B). However, combination of GemC12-loaded LNCs with ACT resulted in a further extended survival as compared to mice receiving ACT alone (Fig. 8B). Remarkably, the survival of mice preconditioned with GemHCl, either administered SC or IV, did not significantly differ from that of control mice receiving only ACT (Fig. 8B). As control, low dose administration of either GemC12- LNCs or GemHCl, without ACT, did not improve mouse survival (Supplementary Fig. S5).
With the intent of validating the abovefindings in a different therapeutic model, we investigated the efficacy of GemC12-LNCs in combination with the adoptive transfer of cytotoxic T cells targeting mouse telomerase reverse transcriptase (TERT) protein. As opposed to ovalbumin, TERT is an endogenous tumor antigen, overexpressed by several mouse and human tumors[32]. TERT-specific T cells recognize their cognate peptide with lower affinity as compared with OVA-specific T cells, and therefore a preconditioning regimen is strictly required for the efficacy of TERT-targeting ACT[32], as for most clinical ACT approaches[1]. The efficacy of GemC12-LNCs plus TERT-targeting ACT was evaluated using the B16 melanoma model, which is a more aggressive and less immunogenic tumor as compared with the EG7-OVA cell line. B16 tumor-bearing mice were injected SC with either GemC12-LNCs or GemHCl when tumor surface reached ~20 mm2 (day 8 post tumor injection), and received TERT-specific T cells the day after (Fig. 8C and see Materials and Methods for detailed therapeutic protocol). As reported in Fig. 8D, low dose GemC12-LNCs and GemHCl administration pro- duced similar changes in myeloid cell frequencies as previously observed in the EG7-OVA model. Of note, in the B16 model, only GemC12-LNCs caused a significant reduction of tumor-infiltrating M-MDSCs (Fig. 8D). As expected, TERT-targeting ACT was not able to prolong mouse survival in absence of any preconditioning treatment (Fig. 7E). Remarkably, combination of GemC12-LNC administration with ACT resulted in a significantly extended sur- vival, while either treatment with low dose GemC12-LNCs without ACT or combination of GemHCl plus ACT were ineffective (Fig. 8E).
4. Discussion
In the present study we investigated the use of an innovative encapsulated gemcitabine formulation, i.e. GemC12-loaded lipid nanocapsules (GemC12-LNCs,Fig. 1), as MDSC-targeting drug for cancer immunotherapy and demonstrated its therapeutic efficacy in combination with adoptive T cell therapy. GemC12-LNCs exten- sively depleted monocytic-macrophagic MDSCs inducedin vitro from mouse bone marrow cells, while leaving granulocytic MDSCs unaffected (Fig. 2). In thisfirst set of experiments, GemC12-LNCs showed equivalent efficacy to free GemC12. However, the activity
of encapsulated versus free drug formulations is known to be deeply affected by pharmacokinetic parameters (e.g. drug absorp- tion, distribution, vascular residence, interaction with plasma proteins[53]) which cannot be predicted only fromin vitrostudies, thus we further investigated the activity of GemC12-LNCs in tumor- bearing mice. In the EG7-OVA mouse tumor model, very low dose GemC12-LNCs (molar equivalent of 11 mg/kg GemHCl) strongly decreased the frequency of M-MDSCs in the spleen and tumor, while other immune cell subsets (including PMN-MDSCs, macro- phages, DCs, and T cells) were not reduced or showed only minor variations (Fig. 3). Importantly, M-MDSCs are endowed with a strong immunosuppressive function [6,7,11e13], supporting the importance of selectively targeting this cell population in cancer immunotherapy.
We observed an enhanced M-MDSC reduction associated with the SC administration of GemC12-LNCs, as compared with IV in- jection (Fig. 3). As shown by biodistribution studies (Supplementary Fig. S1), the SC injection was associated to a slower LNC distribution to peripheral tissues, indicating a gradual ab- sorption from the injection site that may result in a prolonged cell exposition to low drug doses. Although the maximal drug plasmatic concentration reached following SC administration is likely much lower as compared with IV injection (consistently with the reduced hepatic accumulation observed in biodistribution studies), our re- sults indicate that this low concentration is sufficient to trigger M- MDSC-directed cytotoxicity. Conversely, gemcitabine use in con- ventional chemotherapy is primarily aimed at inducing tumor- directed cytotoxicity, which requires higher drug doses and a higher plasmatic drug exposure that are commonly achieved by IV infusion[35].
We showed that GemC12-loaded LNCs produce a more intense and prolonged reduction of tumor-infiltrating M-MDSCs than both free GemC12 and GemHCl (Fig. 4A and B). Of note, the transitory effect produced by GemC12-LNC administration (48 h) was likely due to the limited drug toxicity on bone marrow hematopoietic stem and precursor cells (Fig. 4E), allowing a rapid replenishment of the peripheral myeloid cell pool starting from these precursors.
Suppression of bone marrow hematopoietic function and conse- quent hematologic toxicity (i.e. the prolonged reduction of circu- lating blood cell elements) may occur following systemic chemotherapy and are generally dose-dependent[43]. At the doses commonly employed in clinics gemcitabine administration is associated with low frequency grade 3 and 4 hematologic toxicities [54]. Consistently, here we showed a mild reduction of bone marrow hematopoietic stem and precursor cells following low dose GemC12-LNC and GemHCl SC injection, which was indicative of a very limited hematologic toxicity. In addition, we previously investigated the hematopoietic and liver toxicity of GemC12-LNCs in mouse models, following either SC or IV administration at a higher dose than the dose employed in the present study (40 mg/kg [55]). Remarkably, even at this higher dose, either IV or SC GemC12- LNCs induced mild platelet reduction, and no reduction of circu- lating granulocytes and red blood cells, nor detectable hepatic toxicity[55]. Thus, GemC12-LNC low dose administration appears to be associated with a very favorable safety profile.
The selective targeting of M-MDSCs by GemC12-LNCs reported in this study likely resulted from the combination of a strong LNC
each immune cell subset was determined after 5 h of incubation. Means±SE of 3 independent experiments.C. Hemolyzed whole blood from melanoma patients was incubated for 1.5 h with 100mg/ml DiD-loaded LNCs and the percentage of DiDþcells within the indicated cell subsets was determined byflow cytometry (right panel). Left panel shows the percentages of indicated cell subsets on whole blood leucocytes. Means±SE, n¼3.D. PBMCs from melanoma patients were enriched for monocytic (CD14þ) cells by immuno- magnetic depletion of CD3þ, CD19þand CD56þcells. Enriched CD14þcells were pre-incubated for 1.5 h with GemC12-LNCs (equivalent to 2.5mg/ml gemcitabine), washed, and then plated at 1:1 ratio with alloantigen-activated lymphocytes in a mixed leucocyte reaction (MLR) assay. T cell proliferation was tracked by Cell Trace®dye dilution, following 7 days of co-culture. Results of functional assays using PBMCs from 4 different patients are reported. Histograms show the percentage of proliferating lymphocytes (left peak, lower Cell Trace®fluorescence intensity) and non-proliferating lymphocytes (right peak, higher Cell Trace®fluorescence intensity) within the CD3þT cell gate.
uptake by M-MDSCs (Fig. 5A) and a high sensitivity of this cell population to gemcitabine cytotoxicity. M-MDSCs appear to be particularly sensitive to drugs active on cycling cells (as gemcita- bine), at least in part due to their high proliferation rate [13].
Conversely, in our model tumor macrophages were not reduced following GemC12-LNC administration (Fig. 3 and 4), despite their strong LNC uptake (Fig. 5A), possibly because of a slower prolifer- ative activity as compared to M-MDSCs. Similarly, also resident macrophages populations in the spleen and liver, were not sensitive to low dose GemHCl and GemC12-LNC administration (Fig. 4and Supplementary Fig.S3).
We showed that mouse M-MDSCs internalize LNCs mainly through macropinocytosis (Fig. 6). The rapid LNC uptake by this cell subset may hence depend on and higher macropinocytic activity of M-MDSCs as compared with most other leucocytes. Consistently with this hypothesis, other works previously reported an enhanced uptake of diverse nanoparticulate formulations by either mouse or human monocytes, with respect to other leucocytes[44,45,56e58], suggesting an intrinsic ability of monocytic cells to rapidly inter- nalize nano-sized materials of different chemical composition, size and shape.
The break in tumor-associated immunosuppression induced by M-MDSC depletion can be useful to foster the activity of tumor- specific T cells in immunotherapy approaches. In the present study, we specifically investigated the use of GemC12-loaded LNCs as supporting treatment in ACT protocols. Although endogenous T cells were not affected by GemC12-LNC administration (Fig. 3and data not shown) and displayed a very low level of LNC uptake (Fig. 5A), we reported a dramatic decrease in the frequency of adoptively transferred T cells when GemC12-LNCs were adminis- tered following ACT (Fig. 7A). The cytotoxic activity of GemC12- loaded LNCs on transferred T cells may be in part due to the high proliferative activity of these cells, which are strongly activated in vitroprior to the infusion in the tumor-bearing hosts. Besides, we reported that activated T cells transferred in vivo also have an enhanced rate of LNC internalization as compared with endoge- nous T cells (Fig. 7B).
Based on these results, we evaluated the therapeutic efficacy of an ACT protocol in which GemC12-loaded LNCs were administered as preconditioning treatment prior to T cell transfer, but not following adoptive T cell infusion. Remarkably, GemC12-LNCs (but not GemHCl) significantly increased the efficacy of ACT using OVA- specific T cells (Fig. 8A and B). OVA-targeting ACT exerts strong antitumor efficacy even without any supporting treatment, in both the EG7 and other OVA-expressing tumor models (Fig. 8B and data not shown). The OVA antigen is a high avidity antigen that might
mimic the non-self epitopes generated by tumor mutations[59,60].
Conversely, self (nonmutated) tumor antigens are usually weaker as compared with the OVA model antigen, since they have to escape mechanisms of central and peripheral tolerance before inducing an antigen-specific immune response[61]. Thus, we further investi- gated the efficacy of GemC12-LNCs in combination with an ACT protocol targeting the weak TERT tumor antigen. Mice bearing established B16 melanomas showed a modest but significant sur- vival increase following treatment with GemC12-LNCs plus TERT- targeting ACT, while either ACT alone or combined with GemHCl administration were ineffective (Fig. 8E). Although in the B16 model the impact of GemC12-LNCsþACT on mouse survival was not striking, the efficacy of this treatment is still remarkable due to the fast growth rate of the employed tumor model, the late start of treatments (day 8) and the weakness of the targeted tumor antigen.
In both the EG7 and B16 models, the survival differences be- tween the GemC12-LNCþACT and the GemHClþACT groups were not statistically significant. However, the administration of encap- sulated, but not free, gemcitabine combined with ACT resulted in a clearly improved survival with respect to untreated mice (B16 model) or mice treated with ACT alone (EG7 model), indicating an enhanced treatment potency associated to drug encapsulation.
Besides ACT, antibody-based therapies targeting negative reg- ulatory receptors in T cells, such as CTLA-4 and PD1, were proven to be highly effective for the treatment of selected human cancer types, including melanoma[62]. Interestingly MDSC-targeting ap- proaches have already been shown to synergize with antibody- based immunotherapies[63]. Thus, although the combination of GemC12-LNCs with anti-PD1/anti-CTLA-4 antibodies was not spe- cifically investigated in this study, GemC12-LNC-mediated MDSC depletion might also be useful in combination with antibody-based therapies.
Gemcitabine was previously reported to reduce MDSC fre- quency in mouse models following a single administration of 60e120 mg/kg[13,19e23]. Along with gemcitabine, other chemo- therapeutic drugs are known to reduce MDSC frequency, usually affecting both the monocytic and the granulocytic component (i.e.
sunitinib, sorafenib, fludarabine, 5-fluorouracyl and docetaxel [13,20,64]). In addition, PEG-polypropilene sulfide polymeric micelle loaded with 6-thioguanine were recently reported to reduce the frequency of either Mor PMN-MDSCs (depending on the tissue site) and to increase the efficacy of OVA-specific ACT, although their effectiveness in ACT targeting weak tumor antigens was not evaluated [65]. In the present study, we showed that GemC12-loaded LNCs, administered at a considerably lower dose than previously reported for free gemcitabine, can be effectively Fig. 6.LNC uptake by mouse M-MDSCs is inhibited by LY294002.A. Total splenocytes from EG7 tumor-bearing mice were pre-incubated 45 min with growing concentrations of LY294002 and then incubated 4 h in complete cell culture medium in presence of 5mg/ml DiD-loaded LNCs. The percentage of DiD-positive cells within different cell subsets was determined byflow cytometry.B. Splenocytes were treated as described in A, but maintained in serum-free culture medium. Means±SE, n¼3. Experiments were repeated twice with similar results.
used to attenuate MDSC-dependent immunosuppression and improve cancer immunotherapy, by specifically targeting the highly immunosuppressive M-MDSC subset. In addition, the use of a chemotherapeutic agent such as gemcitabine, already widely employed in clinical oncology and with a well-known toxicity
profile[15e18], may further facilitate the clinical translation of this approach.
Fig. 7.Setting of an optimal administration schedule of GemC12-LNCs in combination with ACT.A. C57BL/6 CD45.2 mice were injected SC with GemC12-LNCs (molar equivalent of 11 mg/kg GemHCl) either at day 8 (schedule I), at day 8 and at day 12 (schedule II), or at day 8, 10 and 12 (schedule III) following EG7-OVA tumor injection. ACT was given at day 9 and consisted in the intravenous infusion ofin vitro-activated, OVA-specific CD8þCD45.1þT cells. Mice were examined at day 15 and cells from lymph nodes and tumors were cultured for 16 h, either in presence or absence of the specific OVA peptide, beforeflow cytometry staining. Data are reported as percentages of CD8þCD45.1þT cells and CD8þCD45.1þIFNgþT cells in lymph nodes and in tumors after GemC12-LNC treatment according to schedule I, II or III plus ACT. Means±SE, n¼5 mice per group. *p<0.05
**p<0.01, Mann-Whitney Rank Sum Test. Statistical comparison is between each group and the control group receiving ACT alone, unless differently indicated by lines.B. DiD- loaded LNCs were administered SC to C57BL/6 CD45.2 mice, bearing established EG7-OVA tumors, following treatment with GemC12-LNCs plus ACT. The percentage of DiDþ cells within endogenous CD4þT cells (CD3þCD4þCD45endogenous CD8þT cells (CD3þCD8þCD45.1) and transferred OVA-specific CD8þT cells (CD3þCD8þCD45.1þ) was determined byflow cytometry 24 h after DiD-LNC administration. Means±SE, n¼5 mice per group. *p<0.05 **p<0.01, Mann-Whitney Rank Sum Test.
