SV-BR-1-GM, a Clinically Effective GM-CSF-Secreting Breast Cancer Cell Line, Expresses an Immune Signature and Directly Activates CD4+ T Lymphocytes

Abstract

Targeted cancer immunotherapy with irradiated, granulocyte-macrophage colony-stimulating factor (GM-CSF)-secreting, allogeneic cancer cell lines has been an effective approach to reduce tumor burden in several patients. It is generally assumed that to be effective, these cell lines need to express immunogenic antigens coexpressed in patient tumor cells, and antigen-presenting cells need to take up such antigens then present them to patient T cells. We have previously reported that, in a phase I pilot study (ClinicalTrials.gov NCT00095862), a subject with stage IV breast cancer experienced substantial regression of breast, lung, and brain lesions following inoculation with clinical formulations of SV-BR-1-GM, a GM-CSF-secreting breast tumor cell line. To identify diagnostic features permitting the prospective identification of patients likely to benefit from SV-BR-1-GM, we conducted a molecular analysis of the SV-BR-1-GM cell line and of patient-derived blood, as well as a tumor specimen. Compared to normal human breast cells, SV-BR-1-GM cells overexpress genes encoding tumor-associated antigens (TAAs) such as PRAME, a cancer/testis antigen. Curiously, despite its presumptive breast epithelial origin, the cell line expresses major histocompatibility complex (MHC) class II genes (HLA-DRA, HLA-DRB3, HLA-DMA, HLA-DMB), in addition to several other factors known to play immunostimulatory roles. These factors include MHC class I components (B2M, HLA-A, HLA-B), ADA (encoding adenosine deaminase), ADGRE5 (CD97), CD58 (LFA3), CD74 (encoding invariant chain and CLIP), CD83, CXCL8 (IL8), CXCL16, HLA-F, IL6, IL18, and KITLG. Moreover, both SV-BR-1-GM cells and the responding study subject carried an HLA-DRB3*02:02 allele, raising the question of whether SV-BR-1-GM cells can directly present endogenous antigens to T cells, thereby inducing a tumor-directed immune response. In support of this, SV-BR-1-GM cells (which also carry the HLA-DRB3*01:01 allele) treated with yellow fever virus (YFV) envelope (Env) 43-59 peptides reactivated YFV-DRB3*01:01-specific CD4+ T cells. Thus, the partial HLA allele match between SV-BR-1-GM and the clinical responder might have enabled patient T lymphocytes to directly recognize SV-BR-1-GM TAAs as presented on SV-BR-1-GM MHCs. Taken together, our findings are consistent with a potentially unique mechanism of action by which SV-BR-1-GM cells can act as APCs for previously primed CD4+ T cells.

Keywords: GVAX; SV-BR-1-GM; antigen-presenting cells; targeted immunotherapy; therapeutic cancer vaccine; whole-cell vaccine.

Figures

Figure 1

Development of SV-BR-1-GM. (A) The SV-BR-1-GM cell line was derived from SV-BR-1 breast cancer cells following stable transfection with CSF2 [encoding human granulocyte–macrophage colony-stimulating factor (GM-CSF)]. The SV-BR-1 cell line itself was established from a chest wall lesion of a metastatic breast cancer patient (16, 17). The depicted developmental stages of SV-BR-1-GM represent samples used for this study rather than provide a comprehensive overview of all lots generated thus far. From an original master cell bank (MCB), several “clinical product” (CP) lots for actual or potential clinical use were established. “RES Lot II” refers to a research sample type and “ECB” to an engineering cell bank. RNA for gene expression analysis was extracted from cells taken directly from cryogenic vials (“cryo”) or following a culturing step (“culture”). “CP Lot VIII” was studied both as a presumptive static culture (“1d”) and as an expanded culture (“3d”), whereby for the former, samples were harvested on the first day (t = 0), and for the latter, on the third day (t = 2, i.e., 2 days later) of a time-course assessing GM-CSF secretion (outlined in Figure S5 in Supplementary Presentation S1 in Supplementary Material). Although no culture expansion took place from seeding to t = 0 (“static culture”), cell numbers of cultures harvested at t = 2 were 1.7–3.2 times the seeding cell numbers (“expanded culture”). (B) Culture morphology of SV-BR-1-GM, as exemplified by 40× and 100× original magnifications of a culture derived from the SV-BR-1-GM Lot 11 bank following 24 h of serum starvation. Of note, SV-BR-1-GM cells may grow in monolayers but can also grow as minimally adherent, sphere-like, structures, especially when seeded at very low densities. (C,D) Quality control (QC). (C) Hierarchical clustering of SV-BR-1-GM samples based on their microarray gene expression profiles. Normalized gene expression levels of samples belonging to the same sample type were averaged (arithmetic mean) prior to clustering. (D) Only samples with a RINe value of at least 7.5 were used for this study. Note that the CP Lot V cryo sample clustered separately and did not pass the minimal variability QC metric (see Materials and Methods) and was thus excluded from additional analyses.

Figure 2

Microarray-based transcript levels of immunostimulatory factors expressed in SV-BR-1-GM cells. 111 genes with known immunostimulatory roles were identified in published reports (–83) (Data Sheet S3 in Supplementary Material) and their microarray-based RNA expression levels determined. The 22 genes shown presented with transcript levels >1.5 times the background cutoff value (median quantile-normalized “expression” level, see Materials and Methods) in each of the SV-BR-1-GM samples. “Relative Expression Values” refers to quantile-normalized mRNA levels. ILMN_… refer to the Illumina probe sequence identifiers (PROBE_ID) of the probes yielding the expression levels shown.

Figure 3

nCounter-based transcript levels of immunostimulatory factors expressed in irradiated and nonirradiated SV-BR-1-GM cells. SV-BR-1-GM cells from 4- and 24-h-old clinical formulations were seeded in 6-well plates and cultured in full medium. Cells were harvested after 1 day and 3 days of culturing then subjected to nCounter-based assessment of the transcript levels of a set of 24 immune-related genes (Immune Signature candidates). For the 24-h series, also nonirradiated cells resuspended in Lactated Ringer’s solution (LRS) and processed in parallel were included. In a separate experiment, transcript levels of the Immune Signature candidates were measured from nonirradiated human mammary epithelial cells (HMECs) and SV-BR-1-GM cells cultured in parallel. Shown are background-subtracted, normalized transcript levels of HLA-A, -B, and -DRB3, whereby the probes (CodeSets) were designed to distinguish between HLA-A*01:01(A) and HLA-A*24:02(B), HLA-B*35:08(C) and HLA-A*55:01(D), and HLA-DRB3*01:01(E) and HLA-DRB3*02:02(F); and of IL6(G) and KITLG(H). (I) Comparison of the expression levels of the 24 immune-related genes between SV-BR-1 and SV-BR-1-GM cells by the nCounter approach. Comparing the 4-h formulations cultured for 1 day, almost identical expression profiles were obtained. Note that the values of CSF2 (encoding GM-CSF) shown indicate background or minute transcript levels of endogenous CSF2 since the nCounter ProbeSet employed did not match in sequence with the exogenous CSF2 expressed by SV-BR-1-GM. For (A–H), values shown are arithmetic means of background-subtracted, normalized transcript levels from triplicate wells ± SDs or from three aliquots per formulation ± SDs for the d0 samples. d0 denotes cells obtained from the original 4- or 24-h-old formulations and lysed without culturing. GM refers to cultured, nonirradiated SV-BR-1-GM cells not subjected to the formulation process. For (I), values shown are arithmetic means of background-subtracted, normalized transcript levels from triplicate (SV-BR-1-GM) or duplicate (SV-BR-1) wells. Values for all genes of the Immune Signature candidates are shown in Data Sheet S3 in Supplementary Material.

Figure 4

Cytokines secretion by irradiated and nonirradiated SV-BR-1-GM cells. (A) SV-BR-1-GM cells from 4- and 24-h-old clinical formulations were seeded in 6-well plates and cultured in full medium. Culture supernatants (SNs) were harvested after 1 day and 3 days of culturing then assessed for cytokine release. Note the substantially reduced levels of granulocyte–macrophage colony-stimulating factor (GM-CSF) and interleukin (IL)-6 for the 24 h compared to the 4 h old formulation. Values shown are arithmetic means from triplicate wells ± SDs, expressed as pg cytokine/1 million viable cells (at time of seeding)/24 h. For KITLG, one of the 24 h, 3d wells was excluded as the obtained cytokine levels were substantially higher than those of the other two wells. For IL-8, ELISAs for the 4 and 24 h samples were conducted on different days. (B) Cytokine levels in the Lactated Ringer’s solution (LRS) fractions of the formulations from (A). For IL-10, data are only shown for the 24 h sample. Values shown are arithmetic means from technical duplicates, expressed as pg cytokine/1 million total cells. (C) Nonirradiated human mammary epithelial cells (HMECs) and SV-BR-1-GM cells were cultured in parallel then assessed for cytokine release 24 h after replacing the culture medium. Note that the IL-6 and IL-8 levels from the nonirradiated SV-BR-1-GM cells shown here are substantially lower than those from the irradiated SV-BR-1-GM cells shown in (A). Values shown are arithmetic means from triplicate wells ± SDs, expressed as pg cytokine/1 million viable cells/24 h, whereby cell viability was determined on the day of medium change (initiation of cytokine accumulation) from cells cultured in parallel wells.

Figure 5

HLA-DR expression on irradiated and nonirradiated SV-BR-1-GM cells. SV-BR-1-GM cells were stained with a FITC-conjugated mouse monoclonal anti-HLA-DR antibody (clone L243) recognizing an epitope from HLA-DRα only present when complexed with an HLA-DRβ chain. Shown are arithmetic means ± SDs of technical triplicates (SV-BR-1-GM) or duplicates (A549) from (i) the percentages of HLA-DR positive cells (left axis) and (ii) the per-cell HLA-DR signal intensities normalized to the geometric mean of the signal intensity by the A549 negative control and reference cell line (value of A549 = 1) (right axis) (86, 87) Non-IRR, non-irradiated; IRR, irradiated.

Figure 6

Anti-HLA-DRβ3 antibody staining of a tumor specimen from the strong clinical responder. To assess whether the strong clinical responder (16), referred to as subject A002 in this article, presented with tumor expression of HLA-DRβ3, paraffin-embedded A002 tumor material was stained with a rabbit polyclonal antibody raised against an N-terminal region of human HLA-DRβ3. As demonstrated, immunoreactivity was apparent.

Figure 7

Cancer/testis antigen (CTA) expression in SV-BR-1-GM cells. RNA expression levels of 279 confirmed or putative CTAs (Data Sheets S4 in Supplementary Material) were compared between SV-BR-1-GM, other established breast cancer cell lines, and several normal human breast cell types. Quantile-normalized and log2-transformed microarray-based RNA expression levels are displayed according to a global gradient color scheme. Red means a higher expression level than white, and white means a higher expression level than blue. Only CTAs with a maximum representative expression value among all samples >1.5 times the background cutoff value were included.

Figure 8

Cancer/testis antigen selectively expressed in SV-BR-1-GM cells. Quantile-normalized microarray-based RNA levels of PRAME(A), KIF2C(B), CEP55(C), and PBK(D) “Relative Expression Values” refers to quantile-normalized mRNA levels. Values shown are arithmetic means. Error bars indicate SDs, except for STROMAL, NCL, BASAL, ALDH neg, ALDH POS, ERBB3 neg. for which standard error of the mean (S.E.M.) values are shown.

Figure 9

SV-BR-1-GM cells act as antigen-presenting cells (APCs). SV-BR-1-GM cells were cultured and serum-starved for 24 h then coincubated with yellow fever virus (YFV) Envelope (Env) 43–59 peptides (35) known to bind to HLA-DR complexes with an HLA-DRB3*01:01-based β chain and a YFV-DRB3*01:01-specific CD4+ T cell clone (A). (A) T cell clone after staining with YFV Env p8/DRB3*01:01 tetramers, as assessed by flow cytometry. Almost all T cells are both YFV Env p8/DRB3 and CD4 positive. (B) After 72 h of coculturing, T cell activation was assessed by determining the levels of secreted interferon (IFN)-γ. Values shown are arithmetic means from technical triplicates ± SDs, normalized to the mean IFN-γ level obtained from the YFV peptide-treated non-DRB3 PBMC reference wells. Background IFN-γ levels obtained from T cells treated with peptides in the absence of APCs (SV-BR-1-GM or PBMCs) were subtracted. (C) IFN-γ levels without background subtraction and normalization of a part of the experiment represented by panel (B). Values shown are arithmetic means of the Europium emission values at 615 nm from technical triplicates ± SDs. (B,C) one-tailed Student’s t-tests were employed to assess significance, with * referring to 0.01 ≤ p < 0.05, ** to 0.001 ≤ p < 0.01, and *** to p < 0.001.