Phosphatidylserine-targeting antibodies augment the anti-tumorigenic activity of anti-PD-1 therapy by enhancing immune activation and downregulating pro-oncogenic factors induced by T-cell checkpoint inhibition in murine triple-negative breast cancers

The murine TNBC cancer line E0771 was purchased from CH3 BioSystems, LLC (Buffalo, NY, USA). EMT-6 was obtained from ATCC (Manassas, VA, USA). All cells were maintained in RPMI-1640 (GIBCO) supplements with 10 % v/v heat-inactivated fetal bovine serum (GIBCO, Waltham, MA, USA).

PS-targeting antibody chimeric mch1N11 is a mouse IgG2a-kappa with human variable heavy and light chain regions; C44 is an IgG2a isotype. Purified anti-mouse PD-1 (clone RMP1-14) was obtained from BioXcell (West Lebanon, NH, USA). All in-vivo antibodies tested negative for endotoxin with a limit of detection of 0.5 EU/ml of stock antibody. Antibodies for flow cytometer analysis were CD45 (clone 30-F11), CD3e (clone 17A2), CD4 (clone GK1.5), CD8a (clone 53–6.7), and CD25 (clone PC61.5), all obtained from Bioscience (San Diego, CA, USA). Antibodies for immunohistochemistry were anti-PD-L1 antibody (clone MIH6, catalog number ab80276; Abcam, Cambridge, MA, USA) and anti-PD-1 antibody (clone RMP1-14, catalog number ab63477; Abcam).

E0771 and EMT-6 tumors were imbedded in OCT medium and frozen in a dry-ice isopentane bath. Tissue sections were cut (5 μm) and stained for PD-L1 and PD-1 positive cells. Tissue samples were stained with secondary antibodies labeled with peroxidase or alkaline phosphatase followed by counterstain with hematoxylin and analyzed by bright-field microscopy. Slides were digitized using a Lumenera infinity digital camera and version 6.4 software (Lumenera, ON, Canada).

Female C57BL/6 or Balb/c mice 4–6 weeks old were purchased from Charles River Laboratories (Wilmington, MA, USA). All studies were approved by the Institutional Animal Care and Use Committee at the University of California, Irvine, USA.

EMT-6/E0771 tumor cells were suspended at 1 × 10⁷/ml in Matrigel (50 % v/v) in PBS and 0.1 ml was injected into the 9/10 mammary fat pad (E0771, C57BL/6; EMT-6, Balb/c). Tumor volumes (V) were calculated using the formula:

where L is the length, W is the width, and H is the height of the tumor. The percent tumor growth inhibition (% TGI) was calculated using the formula:

where T is the mean tumor volume of the treated group at the end of study and C is the mean tumor volume of the control group at the end of study. For tumor rechallenge studies, animals with no palpable tumor were injected with E0771 cells under the same initial dosing conditions but on the opposing mammary fat pad (4/5). The tumor rechallenge response endpoint was expressed as tumor growth delay and the difference in time (days) was calculated between the growth delay of the treated group and the naïve control group. All treatment was administered via intraperitoneal injection in 100 μl volumes twice weekly (C44 control, 10 mpk; mch1N11, 10 mpk; anti-PD-1 2.5 mpk; and mch1N11 + anti-PD-1, 10/2.5 mpk respectively). Doses were selected though preliminary maximum tolerated dose (MTD) studies (data not presented), and no toxicity/weight loss was encountered in the data presented.

Spleens were obtained from naïve nontumor-bearing mice that were untreated, single, or combination treated, or from E0771 tumor-bearing mice treated with C44, or from animals with regressed E0771 tumors following treatment with mch1N11 and anti-PD-1. Spleens were harvested on day 12 following tumor implantation or from nontumor animals following a matching treatment regimen. Single-cell preparations of splenocytes were resuspended in RPM1-1640 supplemented with 10 % FCS containing antibiotics at 1 × 10⁶ cells/ml and 100 μl added, in triplicate, to wells of EliSpot microplates coated with anti-mouse IFNγ IgG, in the absence or presence of 1 × 10⁵ irradiated (15,000 rad) E0771 cells to determine tumor-specific stimulation. Plates were incubated for 48 h at 37 °C and spots were developed using anti-mouse IFNγ IgG–HRP conjugate followed by peroxidase substrate. Spots were counted using an automated EliSpot plate reader.

Tumors were excised from mice and physically dissociated and digested in 1 mg/ml collagenase (Sigma, St. Louis, MO, USA), 0.1 mg/ml hyaluronidase (Sigma, St. Louis, MO, USA), and 200 units/ml DNase type IV (Sigma, St. Louis, MO, USA) for 1.5 h at 37 °C and passed through a 70 μm sieve filter (Falcon, Corning, NY, USA). Cells were collected, treated with ACK lysis buffer to remove red blood cells, washed twice with PBS, resuspended in FACS staining buffer, and stained with antibodies for 20 min at 4 °C.

E0771 RNA was prepared from six tumors for each treatment group shown in Fig. 2a at study end (day 26) by Direct-zol™ RNA mini prep kit (ZymoResearch, Irvine, CA, USA). Gene expression was directly measured via counts of corresponding mRNA in each sample using an nCounter (NanoString, Seattle, WA, USA) GX murine PanCancer Immune Profiling Panel, which is a multiplex assay for 770 genes involved in the murine inflammatory response [47]. The nCounter system allows for direct detection and counting of nucleic acid via reporter probes appended with multiple fluorophore barcodes and biotinylated capture probes that attach to microscopic beads, which are then affixed to lanes in a translucent cartridge and read in an optical scanner. Batches of 12 separate samples (six from each treatment group) at one time were prepared as per the manufacturer’s instructions, with 100–300 ng of total RNA hybridized with probes at 65 °C for 16–18 h before being placed into the automated nCounter Prep in which samples were affixed to cartridges. Cartridges were then immediately placed into the nCounter Digital Analyzer optical scanner and read at a goal resolution of 550 fields of view, which is the maximum resolution for this instrument. Analysis was performed using the nSolver 2.6 analysis software with immune cell types based upon classifications by Newman et al. and Bindea et al. [48, 49]. Probes used to classify cell types were as follows; T cells (CD2, CD3e, CD3g, CD6g), regulatory T cells (Tregs) (FoxP3), macrophages (CD68,CD163,CD84), Th1 cells (CD38, Ctla4, IFN-g, LTA, Stat4, Tbx1), Th2 cells (Cxcr6, Birc5, Gata3 Pmch, Stat6), and dendritic cells (CCl17, Cdl7, Ccl22, Cd1d2).

Statistical analysis was performed using Student’s t test and expressed as median with SD, with the exception of analysis for Kaplan–Meier survival curves which utilized Mantel–Cox analysis. p < 0.05 was considered significant.

In-vitro PD-L1 expression on E0771 and EMT-6 cells was examined by flow cytometry. As shown in Fig. 1a, each BC cell line had basal PD-L1 expression, with E0771 cells having slighter higher levels. When cells were stimulated with IFNγ, levels of PD-L1 increased substantially in each cell line compared with untreated cells, with levels in EMT-6 cells reaching higher expression levels than those observed in E0771 cells. Tumors derived from implanted E0771 and EMT-6 cells also had detectable expression of PD-1 and PD-L1. As shown in Fig. 1b, PD-1 and PD-L1 were detectable in both E0771 and EMT-6 tumor samples by immunohistochemistry, thus supporting the evaluation of combination therapy by antibodies targeting PS and the PD-1/PD-L1 axis in these murine models.

To determine the therapeutic effect of PS and PD-1 targeting alone and in combination, mice were implanted orthotopically with cells from the murine TNBC line E0771. When tumors reached approximately 100 mm³ (day 10), treatments commenced with twice-weekly dosing for a total of six doses (3 weeks) by intraperitoneal injection. Treatment by the PS-targeting antibody, mch1N11 (10 mpk), resulted in 55 % TGI compared with control treatment (C44) (Fig. 2a, upper panel). Treatment by anti-PD-1 antibody (2.5 mpk) resulted in a TGI of 71 % compared with control treatment. Combinational treatment of mch1N11 (10 mpk) with anti-PD-1 (2.5 mpk) antibody significantly suppressed E0771 growth by 90 % compared with control (Fig. 2a, upper panel). Comparison of the final tumor volumes in each treatment group demonstrated that, in addition to significantly suppressing the growth of E0771 compared with control treatment (p < 0.0001), combination treatment by mch1N11and anti-PD-1 significantly inhibited tumor growth more than treatment by either single arm alone (Fig. 2a, lower panel, p = 0.0339 to anti-PD-1 treatment). In the EMT-6 BC model, unlike the results observed in the E0771 model, mch1N11 (10 mpk) and the anti-PD-1 blocking antibody (2.5 mpk) had no growth inhibitory effects as single agents (Fig. 2b, upper panel). However, similar to the results from combination treatment in the E0771 model, combining PS-targeting and anti-PD-1 therapies together was capable of significantly impeding tumor growth in the EMT-6 model (TGI = 57 %, p = 0.005 to control treatment, Fig. 2b upper and lower panels). No loss of weight was noted in any of the treatment groups in the E0771 or EMT-6 models (data not shown).

To determine whether treatment by mch1N11 and anti-PD-1 therapy prolonged overall survival in a BC model, mice inoculated with E0771 tumors were treated with control antibody (C44, 10 mpk), mch1N11 (10 mpk), anti-PD-1 antibody (10 mpk), or the combination of mch1N11 plus anti-PD-1 antibody (10 mpk/10 mpk). As shown in Fig. 3 and summarized in Table 1, all control mice had tumors that progressed with treatment by the C44 isotype antibody and survived until day 25 of treatment, when the end point was reached due to excessive tumor burden. All mice receiving mch1N11 antibody also progressed while receiving treatment, but were able to survive for an additional 7 days beyond control mice. Mice receiving anti-PD-1 therapy alone at 10 mpk survived significantly longer than those receiving control or mch1N11 treatments, having a median survival time of 55 days. Mice receiving the combination of mch1N11 with anti-PD-1 antibody experienced a significant increase (p = 0.0155) in overall survival compared with single anti-PD-1 treatment, with six of 10 mice having complete tumor regression by day 25, compared with two of 10 mice in the anti-PD-1 group, accompanied by no tumor reoccurrence 30 days after the study end (day 60) in all animals with complete regression (Table 1).

To elucidate whether mice that had complete tumor regression in the mch1N11 plus anti-PD-1 treatment group (Fig. 3 and Table 1) were resistant to E0771 rechallenge, the six mice that experienced complete tumor regression and six control naïve mice were inoculated with E0771 cells in the 4/5 mammary fat pad and tumor growth was measured. After 10 days, the naïve mice had an average tumor volume of 134 ± 5.3 mm³ compared with 32.8 ± 4.6 mm³ in the E0771 rechallenged group (Fig. 4a). On day 17, only three of the six mice in the rechallenged group had detectable tumors with an average volume of 20.8 ± 10.5 mm³, compared with 199.8 ± 39.5 mm³ in control mice. On day 26 all mice in the control group reached maximum study tumor volume limits (1656 ± 210 mm³) while the tumors continued to regress in the rechallenged group, with only two mice having detectable tumors (average 9.2 ± 8.2 mm³). By day 35 no tumors were detected in the rechallenge group (Fig. 4a), and no tumor growth was noted for an additional 14 days (data not shown).

To ascertain whether the mice that experienced complete tumor regression with mch1N11 and anti-PD-1 treatment and were subsequently resistant to E0771 rechallenge (Fig. 4a) had an enhanced tumor-specific immune response, we evaluated IFNγ production in splenocytes cultured with and without irradiated E0771 cells by EliSpot (Fig. 4b). Splenocytes from nontumor-bearing control animals (no treatment and C44 treated) had similar IFNγ production when cultured in media alone (naïve, 0.40 ± 0.23; naïve + C44 treated, 0.27 ± 0.27). When cultured in the presence of irradiated E0771, there was a slight increase in IFNγ production in naïve mice treated with the C44 control antibody compared with nontreated mice (naïve, 0.13 ± 0.13; naïve + C44 treated, 2.0 ± 0.46). In nontumor-bearing mice that received mch1N11 and anti-PD-1 treatment, there was a slight increase in IFNγ production when cultured in media alone (9.0 ± 0.81); however, this signal did not increase when irradiated E0771 cells were cocultured with splenocytes from combination-treated animals (10.9 ± 1.71). Splenocytes from tumor-bearing mice that received C44 control antibody had a signal of 10.7 ± 1.7 when grown in media alone, and IFNγ production increased to 60.1 ± 4.8 spots per well with coculturing with irradiated E0771 cells. In the animals with regressed tumors following mch1N11 and anti-PD-1 treatment, and further resistance to E0771 rechallenge (Fig. 4a, b), IFNγ EliSpots were significantly higher than observed under previous treatments and conditions. Splenocytes from these animals cultured in media alone had a baseline signal of 347 ± 7.9 spots per well, suggesting mch1N11 and anti-PD-1 treatment increases the number of functional T cells in the spleens of combination-treated animals, perhaps leading to the observed tumor regression and subsequent rejection of tumor recolonization with E0771 reinoculation. The addition of irradiated E0771 cells to these splenocytes elevated the signal to 614.7 ± 9.7 spots per well, demonstrating that E0771 antigens are capable of further stimulating IFNγ production in T cells from mch1N11 and anti-PD-1-treated animals (Fig. 4b).

The anti-tumor response induced by mch1N11 in combination with anti-PD-1 in mice bearing the TNBC model E0771 tumors suggests that treatment by PS and PD-1-targeting antibodies may induce TIL changes within the tumor microenvironment. Examination of excised E0771 tumors at the study end in Fig. 2a demonstrated that treatment by mch1N11 increased the presence of CD45⁺ cells from a median value of 11.6 % in control (C44) tumors to 18 % (p = 0.0179 to control) with PS targeting, while anti-PD-1 therapy increased CD45⁺ levels to 21 % (Fig. 5a). In combination treated mice, the addition of mch1N11 to anti-PD-1 therapy significantly increased CD45⁺ levels to 25.7 % (p < 0.0001 to C44 control, p = 0.0139 to anti-PD-1 single treatment). Examination of CD8⁺ and CD3⁺ levels (CD8⁺, CD3⁺, and CD4⁺ cells were all subpopulations of CD45⁺ cells) showed a similar trend in response to treatments. Single mch1N11 treatments increased CD8⁺/CD3⁺ levels to 8.9 %/7.7 % respectively over control levels (CD8⁺ = 5.7 %, CD3⁺ = 2.7 %). Single anti-PD-1 therapy increased CD8⁺ and CD3⁺ levels to 14.6 % and 11 % respectively, while the combination of PS with anti-PD-1 therapy elevated CD8⁺ levels to 15.2 % and CD3⁺ levels to 12.8 %. Examination of the CD4⁺ to CD8⁺ T-cell ratio demonstrated that while each single arm treatment group decreased the percentage of CD4⁺ to CD8⁺ cells, combinational therapy further significantly decreased this ratio (p = 0.023), suggesting combined treatment was capable of further reducing the CD4 T-cell helper populations and increasing the effector CD8⁺ cells in the tumor microenvironment (Fig. 5a). FACS examination of the EMT-6 in-vivo study (Fig. 2b) demonstrated that, unlike our observations in the E0771 model, single-arm treatments had little observable effect on altering TILs (Fig. 5b). C44 control CD45⁺ levels had a median value of 36.8 %, while mch1N11 had a value of 34.7 % and anti-PD-1 of 33.4 %. Only in the combination group was an increase in CD45⁺ noted (46.0 %, p =0.023). In CD8⁺ populations, C44 control tumors were 12.0 %, with mch1N11 of 11.80 % and anti-PD-1 exhibiting 7.6 %. The combination treatment group did increase to 20.9 %, but the results were not significant (p = 0.112). Similar results occurred in CD3⁺ populations. C44 control contained 3.9 % CD3⁺ cells, while mch1N11 showed 2.6 % and anti-PD-1 treatment showed 2.3 %. Again only the combinational treatment arm showed a significant increase to 18.4 % (p = 0.012, Fig. 5b). Examination of the ratio of CD4⁺ to CD8⁺ T cells in EMT-6 tumors demonstrated that all of the treatment arms slightly increased the percentage of CD4 cells to CD8 cells; however, none of the changes were statistically significant for control treatment (C44 control, 0.9 %; mch1N11, 2.3 %; anti-PD-1 3.2 %; and mch1N11 + anti-PD1, 2.1 %).

Combined, our results in the E0771 model demonstrate that while each single arm treatment is capable of increasing CD45⁺, CD8⁺, and CD3⁺ populations in E0771 tumors, combining PS-targeting and anti-PD-1 antibodies increases the percentage of TILs over that observed with single treatments. Contrary to these results, only the combination of the PS blocking antibody treatment with anti-PD-1 treatment in the EMT-6 model was capable of inducing TILs, reflecting the TGI data in Fig. 2a, b.

To further capture and understand changes in the immune landscape that accompany PS-targeting and anti-PD-1 therapies, RNA isolated from tumors obtained at the study termination (shown in Fig. 2a) were subjected to mRNA expression analysis utilizing NanoString™ PanCancer Immune Profiling Panel analysis based on classifications described by Newman et al. and Bindea et al. [48, 49]. Differences in the expression patterns of genes associated with CD45⁺, T cells (probes; CD2, CD3e, CD3g, CD6g), and tumor-associated macrophages (probes; CD68, CD163, CD84) showed that while single treatments were capable of increasing the levels of these cell types, most notable for CD45⁺ and T cells, over control (C44) levels, combination treatment further enhanced levels of these cell types compared with each single treatment (Fig. 6a). Increased Treg (probe; Foxp3) levels occurred in all treatment groups compared with the control, with anti-PD-1 therapy having slightly less overall levels compared with treatments that included a PS-blocking antibody. Examination of Th1 (probes; CD38, Ctla4, IFN-g, LTA, Stat4, Tbx1) and Th2 (probes; Cxcr6, Birc5, Gata3 Pmch, Stat6) immunoprofiles showed that distinct changes occurred with each type of treatment. Th1 levels were increased only when PD-1 therapy was included, either as a single treatment or in combination with PS-blocking antibodies, while Th2 levels increased over control levels with each single treatment, and combination therapy brought Th2 levels down to those observed in control treatment mice (Fig. 6a). Examination of dendritic cell markers (probes; CCl17, Cdl7, Ccl22, Cd1d2) demonstrated that PS treatment did not increase levels over control, while anti-PD-1 alone and when combined with mch1N11 both increased levels similarly, suggesting that anti-PD-1 treatment increased dendritic cell levels. Finally we examined levels of PD-L1 and PD-1. Both single-arm treatments similarly increased PD-L1 levels; however, combination treatment increased PD-L1 mRNA levels over that observed in each single arm (Fig. 6a). PD-1 levels increased slightly with mch1N11 compared with control, while treatment with anti-PD-1 alone or in combination similarly induced PD-1 mRNA levels in E0771 tumors. These data suggest that combinational treatment with PS-targeting and anti-PD-1 antibodies increase immune cell infiltration (CD45, T cells, macrophages) over each single-arm treatment and that treatment with PS and anti-PD-1 is capable of inducing PD-L1 levels which may further serve as available targets for intervention with anti-PD-L1/PD-1 therapeutics.

Numerous secreted cytokines and immunomodulatory proteins are expressed in the tumor microenvironment that are associated with immune system activation or suppression [50, 51]. We examined the levels of cytokine IFNγ, IL-10, and TNF mRNA in tumors following treatments with single or a combination of mch1N11 and anti-PD-1 antibody. IFNγ levels did not change with mch1N11; however, anti-PD-1 single treatment and mch1N11 and anti-PD-1 antibody combination increased IFNγ levels, with the anti-PD-1-alone group having the highest levels (Fig. 6b). Levels of IL-12 and TNF increased compared with control in both single-arm treatment groups, with combination treatment further enhancing expression of each of these immune promoting factors over that observed in either single treatment arm (Fig. 6b). Further examination of immunosuppression associated cytokines, including IL-4, IL-9, IL-10, IL-13, IL-17, and TGFβ, revealed the complex changes associated within each of our treatment groups (Fig. 6c). Levels of IL-4 and IL-9 were unchanged from control levels in the mch1N11 and combinational mch1N11 and anti-PD-1 treatment groups; however, anti-PD-1 treatment alone dramatically induced IL-4 and IL-9 levels (Fig. 6c). IL-10 levels were also unaffected by mch1N11 as a single agent; however, anti-PD-1 therapy alone increased levels the most, with the addition of PS antibody to anti-PD-1 therapy reducing IL-10 levels. Similar induction of the immunosuppressing/tumor-promoting cytokines IL-17 and IL-13 was noted with anti-PD-1 treatment, with combination treatment reducing levels below those observed in C44 control mice. The only immunosuppressive/pro-oncogenic cytokine assayed that showed an increase in mice receiving mch1N11 therapy over that observed in the anti-PD-1 single treatment arm was TGFβ. While all treatments increased TGFβ levels over those observed in control tumors, the combination of PS-targeting and anti-PD-1 antibodies induced levels higher than observed in each single treatment arm (Fig. 6c).