Even though therapeutic dendritic cell-based vaccines are immunogenic, clinical responses to these therapies are often minimal. To address this issue, Maurer et al. focused on comprehensively examining the characteristics of DCs used in such vaccines to determine whether patient-specific characteristics affect efficacy.
As a DC source, the researchers turned to samples available from a phase I clinical trial of a DC vaccine in patients with late-stage melanoma for whom clinical outcomes were available. The genetically-engineered DC vaccine (AdVTMM2/DC) used recombinant adenovirus that expresses three shared melanoma antigens (tyrosinase, MART-1, and MAGE-A6) to infect monocyte-derived DCs (iDCs) that had been matured using IFNγ and LPS (mDCs).
Gene and protein expression profiling was performed on DCs to identify biomarkers involved in immunogenicity and positive clinical response to DC-based vaccination. First, the phenotype of the DCs was assessed by determining the expression of HLA-DR, CD11c, CD40, CD80, CD86, and CCR7. None of these canonical markers correlated with immune or clinical outcomes, nor with overall (OS) or progression-free survival (PFS). IL-12p70 (an essential cytokine for promotion of type 1 immunity) and IL-10 also did not correlate with outcome, in contrast to findings in other reports.
The researchers then assessed 11 costimulatory and immune checkpoint molecules on these cells, and only PD-L1, PD-L2, and CTLA-4 increased upon maturation and after adenoviral transduction. PD-1 and Lag3 were expressed at baseline in healthy and patient DCs, but their expression decreased after maturation. The only molecule that was expressed differently between patients and healthy donors was ICOSL. In both healthy donor and patient DCs, the surface expression of ICOSL decreased upon maturation and viral transduction. However, patient DCs had a lower expression of ICOSL at baseline that also was further reduced when matured. This was not correlated with clinical outcomes, but it did negatively correlate with IL-10 production in patient mDCs and AdVTMM2/DCs.
When looking at mRNA expression levels of ICOSLG, no significant differences between iDCs, mDCs, and AdVTMM2/DCs were found. However, baseline RNA levels did correlate with better outcomes and longer OS and PFS. When comparing patients based on a better clinical outcome, IL12B expression was positively associated with ICOSLG expression in AdVTMM2/DC. Further support for the potential role of ICOSL in activating T cell immunity was found in public datasets, where ICOSLG expression correlated with favorable outcomes and CD8+ T cell infiltration at metastatic sites in cutaneous melanoma.
Based on the findings that the RNA levels of ICOSLG did not change, but that the surface protein expression decreased upon maturation, the researchers evaluated whether the DCs shed ICOSL by assessing soluble ICOSL (sICOSL). In vitro, patient DCs indeed had more ICOSL in the culture supernatant upon maturation than healthy donor DCs. The sICOSL concentration of patient DCs positively correlated with clinical outcomes of therapy, longer OS and PFS, and expression of Th1-polarizing chemokines (MIP-1a and CXCL9). In both the patient and healthy donor AdVTMM2/DC cultures, sICOSL was also detected 24 hours after viral transduction.
The researchers then determined the role DC ICOSL expression has on the generation of immune responses. ICOSL surface expression on patient AdVTMM2/DC correlated positively to their in vivo ability to induce tyrosinase-specific CD8+ T cell responses. Increased sICOSL also positively correlated with overall vaccine-induced CD8+ T cell responses. Both surface and soluble ICOSL expression were also associated with improved OS in patients.
To determine whether the expression of ICOSL was a vital factor for the initiation of T cell activation or recall responses, the researchers studied an in vitro model in which T cells were primed or boosted using CMV pp65 antigen. Cultured mDCs were pulsed with a CMV pp65 peptide pool with or without antibody blockade of ICOSL. When ICOSL was blocked, CD4+ and CD8+ T cells expressed less CD107a and produced significantly less granzyme B. Recall responses were not affected by ICOSL blockade.
Maurer et al. next determined the molecular pathways affecting the ICOSL surface expression on DCs. All three studied DC types – iDCs, mDCs, and the adVTMM2/DCs – were assessed for gene expression profiles. Patient mDCs had dysregulated metabolic, toll-like receptor, IL-8, and NF-κB signaling pathways. Since noncanonical NF-κB has previously been implicated in the regulation of ICOSL on murine B cells, that pathway was further assessed in this context. Microarray data showed significant increases in NFKBID, an inhibitor of the NF-κB signaling pathway, which were not detected in healthy donor mDCs, indicating involvement of canonical NF-κB signaling. To confirm the involvement of this pathway, the researchers performed cellular imaging and found that healthy donor DCs had an increase in NF-κB p65 translocation from the cytoplasm to the nucleus upon maturation, which was absent in patient DCs.
To further assess the NF-κB pathway, inflammatory and Th1-polarizing molecules were studied in the supernatant after DC maturation. Patients had lower levels of soluble TNFα, CD40L, and CXCL9, but patients with better therapy responses had higher levels of these molecules and CXCL11. The authors confirmed direct involvement of the NF-κB pathway in ICOSL expression by showing binding of NF-κB p65 to the ICOSLG promoter region in healthy donor DCs using ChIP analysis. Furthermore, when parthenolide, which directly modifies the p65 subunit to inhibit NF-κB signaling, was added to the cultures, the ICOSL surface protein expression levels were significantly lower.
The microarray data also showed that patient DCs expressed the metalloproteases ADAM10 and ADAM17, and upon maturation, ADAM10 expression reduced, while ADAM17 expression increased. When metalloproteases were inhibited using Tapi-2 during DC maturation, DCs increased the expression of ICOSL on their surface, while sICOSL levels in the supernatant decreased. These data suggest that both NF-κB signaling and metalloproteases are involved in the regulation of ICOSL on DCs.
These results from Maurer et al. provide insights into mechanisms of dysregulation and functional defects in dendritic cells in patients with metastatic melanoma and provide directions that could help develop more effective therapeutic vaccines. It will be of particular interest to determine how patient responses are affected using a method in which ICOSL is induced or forcibly expressed on DC vaccines.
Write-up by Maartje Wouters, image by Lauren Hitchings
This week, first author Deena Maurer and lead author Lisa Butterfield answered our questions.
What prompted you to tackle this research question?
DM: This study is a follow-up to the data presented from a recent phase I DC vaccine-based clinical trial led by Dr. Lisa Butterfield at UPMC Hillman Cancer Center. The clinical trial revealed that in advanced-stage melanoma patients (n =35), classical DC markers important for DC function, such as CD86, CCR7, and HLA-DR, did not correlate with favorable clinical outcomes in patients. Moreover, levels of IL-12p70, a Th1-skewing chemokine, also did not correlate with clinical outcomes or survival rates observed in patients. These surprising data led us to identify critical molecules and signaling pathways that correlated with favorable clinical outcomes and survival rates observed in late-stage patients. Our goal for this study was to identify target molecules and critical signaling pathways that could be manipulated to develop improved vaccines implementing DCs for the treatment of advanced-stage melanoma patients.
LB: There have been over 200 clinical trials testing vaccination with dendritic cells. We know they are the most critical cell type for promoting effective immunity, but we need to know more about what specifically they need to do or express to be most effective in vivo. Doing an unbiased profiling of the DC vaccine cells and correlating those data to the in vivo effects in cancer patients was our approach to answer that question.
What was the most surprising finding of this study for you?
DM: The microarray analysis performed on patient cells ex vivo revealed that transcript levels of ICOSLG correlated with clinical outcomes and survival rates observed in patients at baseline, and that NF-κB was dysregulated. This is interesting because these data suggest that patient monocytes are intrinsically defective. The clinical associations observed with ICOSL may be imprinted due to dysregulation in patient precursor cells. It will be important to profile patient monocytes and investigate epigenetic changes within these cells.
LB: I was surprised that blockade of only one of many costimulatory and immune regulatory molecules had a significant effect, which strengthens the conclusions that ICOSL is important. I had expected that the continued presence of CD80, CD86, and other molecules in those experiments would have been potent enough to overshadow ICOSL’s impact.
What was the coolest thing you’ve learned (about) recently outside of work?
DM: During the shutdown, I’ve had a lot more quality time for hikes with my two dogs, Biscuit & Wishbone. I started to notice that their paws would smell like Doritos. After doing some research and speaking with their vet, I learned that dogs naturally sweat through their paws and sometimes get “Frito Feet” due to bacteria built up. Needless to say, the more active the dog, the more likely he/she will have stinky paws. In order to avoid this, my dogs' paws regularly get cleaned and booties are always an option!
LB: I moved to San Francisco a year and a half ago and have been exploring the city on foot since the shutdown. It’s an amazingly beautiful city filled with hills and streets so steep that they require stairs to climb. At the top of each and every one of these hills are incredible views of the city, the bay, the bridges and the ocean. I knew it was a wonderful city in many ways, but now know much more.