PLK1 inhibition upregulates PD-L1 expression
We and others have previously reported that PLK1 inhibition or knock-down results in a cell cycle arrest in G2/M phase leading to cancer cell death.14,19−22 Herein, we found that PLK1 knockdown also increases PD-L1 surface expression in both human (A549) and murine (LLC-JSP) lung cancer cell lines. As shown in Fig. 1A, 85% knockdown of PLK1 mRNA (by siRNA against PLK1) resulted in 2.5-fold increase in PD-L1 mRNA expression in A549 cell line compared with untreated cells at 2 days after siRNA treatment. This was then confirmed at the surface protein level in A549 (Fig. 1B) and LLC-JSP (Fig. 1C) lung cancer cell lines at 3 days after siRNA treatments.
Mitotic kinase inhibitor (MKI)-induced PD-L1 expression
Following on the discovery that PLK1 inhibition results in PD-L1 upregulation, we sought to determine whether this holds true for inhibition of other mitotic kinases. We screened three mitotic kinase small molecule inhibitors against PLK1 (volasertib), Aurora kinase A (alisertib), and CHK1 (AZD7762) in human and mouse lung cancer cell lines. As shown in Supplementary Fig. 1, treatment of human NSCLC cells (A549 and H460) and a murine lung cancer cell (LLC-JSP) with volasertib, alisertib, or AZD7762 all reduced cancer cell viability and upregulated surface PD-L1 levels. These results establish the link between mitotic kinase inhibition and PD-L1 upregulation, and also demonstrate the potency of volasertib over other mitotic kinase inhibitors. In agreement with our results using PLK1 siRNA (Fig. 1) or volasertib (Supplementary Fig. 1), onvansertib (a newer PLK1 small molecule inhibitor) also reduced cell viability and upregulated PD-L1 expression in A549 cells (Supplementary Fig. 2).
PLK1 inhibition leads to PD-L1 upregulation through activation of the mitogen-activated protein kinase (MAPK) pathway
Previous studies have shown the involvement of the MAPK pathway and the NF-kB transcription factor in PD-L1 upregulation after paclitaxel (mitotic inhibitor chemotherapy) treatment.23,24 Specifically, Gong et al. reported that paclitaxel-induced PD-L1 expression was MAPK-dependent in colorectal and liver cancers,23 while Peng et al. found that paclitaxel induces PD-L1 upregulation via NF-kB in ovarian cancer.24 To decipher the mechanism of PLK1 inhibition-induced PD-L1 expression, we probed downstream signaling pathways after PLK1 inhibition with volasertib. We found that volasertib led to activation of MAPK pathway, specifically with increase in phosphorylated ERK1/2 expression, and volasertib also led to an increase in phosphorylated NF-kB expression (Supplementary Fig. 3). In NSCLC, Jong et al. reported that the MAPK pathway plays a key role in PD-L1 regulation through increasing transcriptional activity and stabilizing PD-L1 (CD274) mRNA.25 Thus, we hypothesized that volasertib induced PD-L1 expression through the MAPK pathway, which is a known regulator of PD-L1 expression.26 To investigate this, NSCLC cells (H460) were treated with volasertib, ERK1/2 small molecule inhibitor SCH772984, or both drugs, and PD-L1 expression was assessed by flow cytometry. As shown in Fig. 2A, inhibition of ERK1/2 could negate the elevated PD-L1 expression induced by volasertib, as cells treated with both volasertib and SCH772984 showed similar PD-L1 expression level to vehicle control-treated cells. Moreover, since PLK1 inhibition also led to increased expression of phosphorylated NF-kB (a known transcription factor for PD-L1), NSCLC cells were treated with NF-kB inhibitor SC75741 in combination with volasertib, and PD-L1 expression was again assessed with flow cytometry. As shown in Fig. 2B, SC75741 did not alter PD-L1 expression level, suggesting that PLK1 inhibition-induced PD-L1 upregulation is not dependent on NF-kB in NSCLC cells.
Combination of PLK1 inhibition and PD-L1 blockade reduces tumor growth and prolongs survival in mice
Based on our finding that PLK1 inhibition upregulated PD-L1 protein, we investigate whether the PLK1 inhibitor volasertib and a PD-L1 monoclonal antibody synergize in vivo. We used a LLC-JSP cell line to develop a flank tumor model in immune-competent mice similar to a previous report.27 Mice with tumors (> 60 mm3) were treated i.p. with volasertib, PD-L1 antibody, or the combination of volasertib and PD-L1 antibody as shown in Fig. 3A. The combination treatment (volasertib + PD-L1 antibody) significantly reduced tumor growth (Fig. 3B) and prolonged survival of mice (Fig. 3C) better than each monotherapy, which did not provide a statistically significant survival effect compared with the vehicle-treated mice.
PD-L1 antibody conjugated nanoparticles for delivery of PLK1 inhibitor volasertib (ARAC)
Our data show for the first time the benefit of combining PD-L1 antibody and PLK1 inhibitor to enhance therapeutic impact. However, each drug alone carries significant toxicity risks that may limit the clinical translation of this combination. To overcome this limitation, we investigated a targeted delivery approach utilizing our polymer modified mesoporous silica nanoparticle (NP) platform. The same nanoparticle platform has been proven effective for targeted delivery of siRNA to breast tumors and lung tumors in our prior work.14,16,17,28 Volasertib (iPLK1) was loaded onto mesoporous silica nanoparticle (MSNP) core prior to surface modification with polyethylenimine (PEI), polyethylene glycol (PEG), and PD-L1 antibody (Fig. 4A). The final composition (p-iPLK1-NP or ARAC) contained 14.3% PEI and 10% PEG (by thermogravimetric analysis (TGA); Supplementary Fig. 4C), with 0.5% volasertib (by UV-Vis absorbance at 330 nm), and 4.0% PD-L1 antibody (by BCA assay) (by weight of MSNP). Volasertib loaded nanoparticles, with PD-L1 antibody (p-iPLK1-NP; ARAC) or without (iPLK1-NP), had a hydrodynamic size of 90 nm (Fig. 4B), and rigid core size of 50 nm by TEM (Supplementary Fig. 4A). In agreement with our prior studies, we found that the nanoconstruct can be stored stably at -80 oC and retain efficacy and hydrodynamic size as fresh material (Supplementary Figs. 4B). Regarding drug release, we found that volasertib was preferentially released from the nanoconstruct in endo/lysosomal solution (pH 4.5) over PBS (pH 7.4) (Supplementary Fig. 4D). As shown in Supplementary Fig. 5, ARAC significantly reduced cell viability of human (A549 and H460) and mouse (LLC-JSP) lung cancer cells in a dose-dependent manner, while the bare nanoparticle (NP) showed minimal toxicity. Moreover, treatment of LLC-JSP cells with iPLK1-NP significantly reduced cell viability more than the free volasertib counterpart treated at the same dose (Fig. 4C). Similar results were obtained with B16-F10 melanoma cells and 4T1 breast cancer cells (Supplementary Fig. 6) suggesting broad efficacy of the material. In vitro, PD-L1 antibody (see ARAC vs. iPLK1-NP; Fig. 4D) had no role in cancer killing (since there were no T cells in the system) or enhancing the delivery (since all nanoparticles were taken up by cells within 3 days regardless of having PD-L1 antibody or not). In agreement with previous finding using PLK1 siRNA (Fig. 1) or free inhibitors, treatment with iPLK1-NP resulted in significant increase in surface PD-L1 expression of the surviving cells at 2 days (about 60% cells were dead), but not at 2 hrs (Fig. 4E) since PLK1 inhibition effects (i.e. cell cycle arrest and death) had not yet transpired. However, PD-L1 antibody on the nanoparticles was as effective as free PD-L1 antibody, given at a 30-fold higher dose, at reducing surface staining of PD-L1 level (see ARAC at 2 hrs, Fig. 4E), which may be attributed to blockade by the therapeutic antibody or PD-L1 internalization along with the nanoparticles (Supplementary Fig. 7). This is owing to the high local concentration of antibodies on the nanoparticles that the cells encountered. Nanoparticles with dense PD-L1 antibodies (approximately 2 x 103 antibodies per particle) can bind multiple PD-L1 ligands on cell surface at once and endocytose with PD-L1 (termed receptor-meditated endocytosis).29 Furthermore, Fig. 4F suggests that PD-L1 upregulation in the surviving population upon treatment with iPLK1-NP is not via selection process (e.g., death of cells with low PD-L1 first) since untreated cells (PBS) did not have any high PD-L1 population), but rather is due to signaling effects of PLK1 inhibition. Indeed, PLK1 inhibition also led to upregulation of PD-L1 expression in melanoma and breast cancer cells with varying PD-L1 baseline expression (Supplementary Fig. 6).
Uptake, feed-forward delivery, and specificity of ARAC nanoconstruct
To study the uptake of the nanoconstruct, NSCLC cells were treated with PD-L1 antibody-conjugated nanoparticles (p-NP) carrying a fluorescent dye (Dy677) and incubated at 4oC for 1 hr or 37oC for 1–4 hr prior to imaging. As shown in Supplementary Fig. 7, most p-NP was effectively internalized into NSCLC cells (H460) upon 1 hr incubation at 37oC, while most p-NP stayed on the cell membrane without any significant cellular uptake upon 1 h incubation at 4oC. This suggests that active endocytosis, as opposed to passive uptake,30 is the primary uptake mechanism of p-NP. Further, higher PD-L1 staining is observed intracellularly upon 1 hr incubation at 37oC compared with 1 hr incubation at 4oC, suggesting that PD-L1 is internalized with the nanoparticles. Moreover, since most p-NPs are intracellular at 1 hr, the observed lower surface PD-L1 expression reported in Fig. 4E (at 2 hr time-point) is primarily due to PD-L1 internalization and not due to the nanoparticles’ blocking the PD-L1 staining antibody on the cell membrane.
While ARAC initially engages PD-L1 upon binding and internalization (as shown in Fig. 4E and Supplementary Fig. 7), surviving cells have upregulated PD-L1 due to the signaling effects of PLK1 inhibition. In this context, upregulated PD-L1 is used as the homing target for subsequent ARAC, leading to cancer targeting in a feedforward manner (i.e., higher targeting with increased doses of the treatment). To investigate the feedforward targeting of ARAC, we used 4T1 murine cancer cells which express low baseline PD-L1 levels (lowest of any cancer cell line tested in our study). ARAC led to the upregulation of PD-L1 in 4T1 cells 4 days post treatment (Supplementary Fig. 8A). We then assessed the cellular uptake of ARAC in control 4T1 cells (with low PD-L1) and ARAC-treated 4T1 cells (with upregulated PD-L1). As shown in Supplementary Fig. 8B, after 1 hour exposure, ARAC was preferentially taken up by the PD-L1 high cells vs. PD-L1 low cells by nearly 4-fold, demonstrating the selectivity and feed-forward targeting by ARAC. We also evaluated the cell killing selectivity by comparing viability of murine cancer cells (LLC-JSP, 4T1, B16-F10) vs. bone marrow-derived dendritic cells (BMDC) after treatment with ARAC. As shown in Supplementary Fig. 8C, ARAC led to significant cell killing in cancer cells but minimal killing in dendritic cells. Similar to Fig. 4D, PD-L1 antibody has no effect on enhancing the delivery in this setting since all nanoparticles are taken up by cells within 3 days regardless of PD-L1 expression. Thus, the treatment selectivity to cancer cells over BMDC cells is due to cancer dependence on PLK1, as previously reported.31
ARAC induces anti-tumor immune response in a bilateral lung cancer tumor model
To assess the anti-tumor immune effect of ARAC, we utilized a bilateral flank tumor model. C57BL/6 mice were injected with 100,000 and 40,000 LLC-JSP cells on the right and left flank, respectively. At day 12 post injection, mice were grouped (n=7) and the right flank (local) tumors were injected with PBS, p-NP (NP with PD-L1 antibody), iPLK1-NP (NP with volasertib), or ARAC as shown in Fig.5A.Growth of local (treated) and distant (untreated) tumors were monitored. Treatments with ARAC significantly reduced growth of local tumors compared with p-NP or iPLK1-NP (Fig.5B). We did not observe significant tumor reduction with iPLK1-NP since the volasertib administered in this study (0.125 mg/kg intratumoral for 3 doses) was much lower than the efficacious dose range reported for volasertib in xenograft tumors (e.g. 10–40 mg/kg once or twice a week systemically).32–34 Similarly, the PD-L1 antibody dose on p-NP (20 µg for 3 doses) is lower than the efficacious dose (200 µg i.p. once a week) reported in this model. As shown in Fig.5C, a delay in the onset of distal tumors was also observed in ARAC group, suggesting that a systemic anti-tumor immune response was generated. Further, ARAC significantly prolonged survival of mice vs. saline or single drug NPs (Fig.5D). In a separate study, mice were inoculated with 250,000 and 100,000 LLC-JSP cells for bilateral tumors and treated as shown in Fig.5A with ARAC or saline. Tumors and tumor-draining lymph nodes were harvested one day post last treatment for immune profiling with flow cytometry. As shown in Fig.5E, PD-L1 surface expression was reduced in ARAC-treated tumors in both hematopoietic (CD45+, primarily immune cells) and nonhematopoietic (CD45-, primarily cancer cells) cell populations, confirming the efficacy of PD-L1 antibody on the nanoconstruct. ARAC also significantly enhanced proliferation (Ki67+) of effector CD8+T cells in the local tumor-draining lymph nodes (Fig.5F). Longer time-point (than one day post dosing) may be needed to see significant changes in distant tumors to decipher the observed delayed onset of the distant tumors in Fig.5C. Importantly, ARAC-treated tumors had significantly higher population of total immune cells (CD45+) and CD8+T cells, and significantly higher ratio of CD8+/Treg ratio as compared with control tumors (Fig.5G).
To evaluate ARAC systemically, we developed an experimental metastatic lung tumor model by intravenous injection of LLC-JSP cells (200,000 cells), which developed tumors mainly in the lungs (confirmed at sacrifice). Mice were randomly grouped and treated intravenously (i.v.) via tail vein with saline, free drugs (volasertib + PD-L1 antibody at same dose or 5-fold higher dose than dose on ARAC), ARAC, or ARAC plus anti-CD8 antibody (Fig. 6A). Mice treated with ARAC survived significantly longer than those treated with saline or free drugs at same dose (***p < 0.001 vs. saline; **p < 0.01 vs. free drugs (1x)) (Fig. 6B) and slightly better than those treated with the 5-fold dose of the free drug combo (*p < 0.05 vs. saline). Thus, delivery with ARAC could effectively reduce required dose of drug by at least 5-fold. Moreover, ARAC’s efficacy was confirmed to be immune-mediated as CD8 + T cell depletion by anti-CD8 antibodies abolished the prolonged survival of ARAC-treated mice (Fig. 6C). Furthermore, treatment with ARAC did not cause any weight loss, demonstrating its favorable safety in mice (Fig. 6D).
ARAC efficacy in ICI-refractory KLN-205 syngeneic tumor model
We further evaluated ARAC’s in vivo efficacy in an ICI-refractory tumor model, KLN-205.35,36 We compared ARAC efficacy to dual ICI treatments of PD-1 and CTLA-4 antibodies (which was recently granted FDA approval for metastatic NSCLC). DBA/2 mice were inoculated with 500,000 KLN-205 cells on the right flank. When tumors developed (~ 40 mm3), mice were grouped (n = 7) and treated i.v. with 50 mg/kg ARAC (for 4 doses on days 0, 3, 9, 12), i.p. with ICIs (PD-1 and CTLA-4 antibodies – 200 µg/dose and 100 µg/dose respectively – for 6 doses on days 0, 3, 9, 12, 21, 28), or saline. As shown in Fig. 7A, ARAC treatments significantly reduced tumor growth vs. saline (**p < 0.01) and vs. ICIs (**p < 0.01), while ICIs had no effect on tumor growth vs. saline. Moreover, mice treated with ARAC survived significantly longer than saline or ICI-treated mice (Fig. 7B).
Safety of nanoparticle platform in cynomolgus monkeys
To date, we found our NP platform to meet required safety criteria: (1) Low cytotoxicity of multiple organ cells (< 10% cell death),28 (2) Great blood compatibility,16 (3) Not triggering adverse immune response of blood immune cells (PBMC),16 (4) Excellent safety after 7 doses given systemically to mice over 1 month by not causing adverse effects to body weight, serum biomarkers, and histology of kidney and liver,18 (5) Good maximum tolerated dose (MTD not reached at 4-fold of efficacious dose),18 and (6) effective clearance as MSNP is soluble to benign silicic acid37,38 at serum pH and cleared in urine.39,40 Herein, we report preliminary toxicology study of the NP platform co-delivering a PLK1 siRNA (as an alternative for volasertib) and PD-L1 antibody (avelumab), namely p-siPLK1-NP, in non-human primates (NHP). We aim to replace volasertib with PLK1 siRNA as we progress towards clinics since the NP construct has shown outstanding efficacy in delivery siPLK1 to tumors in lungs14,17 and we previously found that cancer was less prone to develop resistance to siRNA than to small molecule inhibitor counterparts.41 As shown in Supplementary Fig. 9, we also confirmed that PD-L1 antibody-conjugated NP could effectively deliver PLK1 siRNA and reduce cell viability of NSCLC cells. To study the toxicity of this nanoconstruct, cynomolgus monkeys (n = 3) received intravenous infusion of 5.6 mg/kg bare NP, 6 mg/kg p-siPLK1-NP (estimated efficacious dose42) and 18 mg/kg p-siPLK1-NP (3-fold efficacious dose), with a one week washout period in between dosing. Clinical signs, body weights, food consumption, dermal observations, clinical pathology parameters (hematology, coagulation, clinical chemistry, and cytokine secretion), gross necropsy findings, organ weights, and histopathologic examinations were evaluated. There were no test article-related clinical observations or effects on body weight (Supplementary Table 1), food consumption, or coagulation (Supplementary Table 3). Dermal observations of erythema and edema were noted at 6 and 24 hours post dose but resolved at 48 hrs post dose (Supplementary Table 6). For hematology, nonadverse decreased white blood cell (WBC), neutrophil, and lymphocyte counts were observed in 2 of 3 monkeys at the highest dose (18 mg/kg p-siPLK1-NP) on day 2 post dose but were resolved by day 7 post dose, indicating recovery (Supplementary Table 2). For clinical chemistry, nonadverse increased aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels were noted on day 2 following administration of NP alone, 6 mg/kg p-siPLK1-NP, and 18 mg/kg p-siPLK1-NP. However, for each group, AST and ALT levels of day 7 were similar to predose values, indicating recovery (Supplementary Table 4). Further, key cytokines (IL-1β, IL-6, IFN-γ, TNF-α, IFN-α, and MCP-1) were monitored in serum collected at 0, 6, 24, and 48 hours. Even in monkeys receiving the highest dose (18 mg/kg), only MCP-1 and IFN-α were mildly to moderately elevated in 2 out of 3 monkeys at 6 hrs, which subsided at 24 hrs without intervention (Supplementary Table 5). Increase of these cytokines right after treatment may be part of therapeutic actions (e.g., IFN-α suggests induction of innate immunity by siRNA;43,44 PD-L1 inhibition induced higher production of MCP-1).45 Terminal euthanasia and necropsy was performed one week following the 18 mg/kg dosing to assess gross pathology, organ weights, and histopathology. A few observations were noted that were considered incidental and of the nature commonly observed in this species and age of monkeys. Thus, there were no test article-related effects on survival, organ weights, gross pathology, or histopathology (Supplementary Tables 7–9). In conclusion, the nanoconstruct was found to be safe and well tolerated in NHP at up to 3-fold anticipated efficacious dose.