Kaiso (ZBTB33) subcellular partitioning functionally links LC3A/B, the tumor microenvironment, and breast cancer survival

Abstract

The use of digital pathology for the histomorphologic profiling of pathological specimens is expanding the precision and specificity of quantitative tissue analysis at an unprecedented scale; thus, enabling the discovery of new and functionally relevant histological features of both predictive and prognostic significance. In this study, we apply quantitative automated image processing and computational methods to profile the subcellular distribution of the multi-functional transcriptional regulator, Kaiso (ZBTB33), in the tumors of a large racially diverse breast cancer cohort from a designated health disparities region in the United States. Multiplex multivariate analysis of the association of Kaiso's subcellular distribution with other breast cancer biomarkers reveals novel functional and predictive linkages between Kaiso and the autophagy-related proteins, LC3A/B, that are associated with features of the tumor immune microenvironment, survival, and race. These findings identify effective modalities of Kaiso biomarker assessment and uncover unanticipated insights into Kaiso's role in breast cancer progression.

Fig. 1. Profiles of the subcellular distribution of Kaiso (ZBTB33) show that nuclear and cytoplasmic Kaiso are differentially correlated with breast cancer subtype and hormone status.

a Representative subcellular patterns of Kaiso expression in breast cancer tissues detected by anti -Kaiso immunohistochemical staining. Shown is a range of high (upper panels) versus low (bottom panel) nuclear and cytoplasmic protein expression. b Comparison of the distribution of digitally determined H-scores for nuclear versus cytoplasmic Kaiso enrichment. The difference in distribution is shown in contrast to the similarities in the distribution of nuclear versus cytoplasmic androgen receptor (AR). c Cytoplasmic Kaiso is differentially enriched in ER⁻ breast cancers compared to nuclear Kaiso and is significantly enriched in the more aggressive breast cancer subtypes, LumB, HER2⁺, and TNBC. d Nuclear and cytoplasmic levels of Kaiso (H-score) do not show significant differences (t-test) based on race (also see Supplementary Fig. 1). e Kaiso (ZBTB33) mRNA abundance (median) is predictive of poor overall breast cancer survival as demonstrated in two independent publicly available breast cancer cohorts. f Comparison of nuclear and cytoplasmic levels of Kaiso in breast cancer patients to levels detected by RNA-seq in (N = 131) patients demonstrates that Kaiso (ZBTB33) mRNA levels do not correlate with either nuclear or cytoplasmic levels of Kaiso. LumA Luminal A, LumB Luminal B, HER2⁺ human epidermal growth factor receptor 2 positive, TNBC triple-negative BC, ER estrogen-receptor status, NHW non-Hispanic White, NHB non-Hispanic White.

Fig. 2. Both nuclear Kaiso and cytoplasmic Kaiso are predictive of poor breast cancer survival.

a Analysis of the association between subcellular Kaiso distribution and survival demonstrates that high cytoplasmic Kaiso is much more predictive of poor survival compared to nuclear Kaiso. Nuclear, cytoplasmic, and total Kaiso H-scores predict survival in both b high risk (lymph node-positive) and c low risk (lymph node-negative) breast cancer patients, where total Kaiso score is most predictive of survival in both low and high-risk breast cancer patients. (HR could not be calculated for cytoplasmic Kaiso in low-risk patients because no deaths were recorded in that risk group). NHW non-Hispanic white; NHB non-Hispanic black. Optimized cut-offs were determined by the method of maximally selected rank statistics (Supplementary Fig. 3 and Supplementary Table 2).

Fig. 3. Quantitative comparison of digitally scored functional and predictive biomarker abundance reveals associations between cytoplasmic Kaiso and the autophagy-related antigen, LC3A/B, that correlate with subtype and survival.

a Unsupervised hierarchical clustering of the nuclear, membrane and cytoplasmic biomarker H-scores for each of (N = 555) patients is shown in correlation with patient clinicopathologic and demographic attributes (below). b Kaplan–Meier survival analysis of specific antigen expression clusters (identified by color code) in a demonstrating associations between cytoplasmic Kaiso, LC3A/B expression, and overall survival in TNBC. c Representative sample of LC3A/B immune-histochemical staining in breast cancer tissues. Arrowhead indicates subcellular puncta noted in the cytoplasm of multiple sections. d Kaplan–Meier survival analysis shows that high LC3A/B cytoplasmic staining is associated with poor overall breast cancer survival. e Like cytoplasmic Kaiso, LC3A/B staining is highly correlated with the more aggressive breast cancer subtypes LumB, HER2, and TNBC with the strongest association with TNBC. f LC3A/B shows a trend of higher expression in NHB versus NHW patients and is significantly more expressed in patients with ER- breast cancer. g Correlation between LC3A/B protein, nuclear Kaiso, cytoplasmic Kaiso, and the RNA levels for LC3A/B, (MAPL1LC3A, MAPL1LC3B), and Kaiso (ZBTB33) showing the strongest correlation between LC3A/B and MAPL1L3B RNA in addition to ZBTB33 RNA and MAPL1LC3B RNA. Spearman correlation is shown in red. p-value for Spearman correlation is shown in blue.

Fig. 4. Breast cancer cells depleted of Kaiso are enriched for autophagy terms and show a defect in LC3A/B maturation.

a Differential gene expression analysis of wild type TNBC cell line MDA-MB-231 versus MDA-MB-231 cells depleted of Kaiso by RNAi shows a significant overlap of differentially expressed genes with an autophagy-related gene list (also see Supplementary Fig. 6, and Supplementary Table 3) (p-value for the significance of the overlap is provided by the hypergeometric test). b Gene set enrichment analysis shows significant enrichment for autophagy terms in MDA-MB-231 cells that differentially express Kaiso (Supplementary Data 4). c Analysis of autophagocytic flux in WT versus MDA-MB-231 depleted of Kaiso by RNAi reveal a significant defect in autophagy demonstrated by decreased autophagocytic puncta formation in cells depleted of Kaiso. d Quantitative GFP-LC3 immunofluorescent analysis of LC3 puncta formation (average puncta per cell before (left) and after (right) addition of chloroquine in MDA-MB-231 cells expressing GFP-LC3). e Immunoblot analysis of LC3A/B lipidation (GFP-LC3ii formation) in MDA-MB-231 transfected with scramble control RNAi versus 3 different short hairpins against Kaiso. f Quantitative densitometer scan (N = 4 biological replicates each with N = 3 technical replicates). (*) = p-value < 0.05, (***) = p-value < 0.001 (t-test) for each unique Kaiso RNAi hairpin. g Immunofluorescent colocalization of Kaiso (green) and LC3A/B (red) in MCF-7 and MDA-MB-231 cells. h Quantitative profiling of Kaiso and LC3A/B colocalization in both the cytoplasm and nucleus (overall) of MCF-7 and MDA-MB-231 cells.

Fig. 5. Kaiso and LC3A/B show extensive colocalization in both the nucleus and cytoplasm of patient tumors.

a Immunofluorescent staining of Kaiso (green) and LC3A/B (red) in TNBC breast cancer (upper panel) and Luminal B breast cancer (lower panel). b Colocalization maps showing relative colocalization of Kaiso and LC3A/B in patient tumors.

Fig. 6. Patients stratified by nuclear and cytoplasmic Kaiso are variably enriched in cell stress and immune response pathways and differentially predict survival based on genetic ancestry.

a Volcano plot of differential gene expression of patients stratified by nuclear Kaiso. (Right) Gene set enrichment analysis of patients stratified by cytoplasmic Kaiso (Supplementary Data 5 and 6). b Volcano plot of differential gene expression of patients stratified by cytoplasmic Kaiso (right) Gene set enrichment analysis of patients stratified by nuclear Kaiso. c Population-specific composition of a representative portion (N = 131) of the breast cancer cohort, by genetic ancestry-based ancestry informative markers extracted from the RNA-seq data. Indicated colors reflect the percent admixture of each genetic population. d Forest plot analysis of overall survival hazard-based optimized cut-off for cytoplasmic Kaiso for the total population, NHW patients, NHB patients, and median cut-off for the total population. e Forest plot analysis of overall survival hazard-based optimized cut-off for LC3A/B in the total population, NHW patients, NHB patients, and the median cut-off for the total population. f Forest plot analysis of overall survival hazard based on optimized cut-off for nuclear Kaiso in the total population, NHW patients, NHB patients, and the median cut-off for the total population. g Forest plot analysis of overall survival hazard based on optimized cut-off for nuclear Kaiso in TNBC patients using optimized cut-off for NHW patients, NHB patients, and the median cut-off for the total TNBC population.

Fig. 7. Elevated cytoplasmic Kaiso, LC3A/B, and race are associated with an immune-suppressive tumor microenvironment.

a Representative multi-spectral quantitative immunofluorescence (mQIF) of the tumor microenvironment of a breast cancer TMA core, stained for pan-cytokeratin (Cy7, cyan); PDL-1 (Cy5, red); CD8 (FITC, green); CD68 (TRITC, carmine). b Coordinate map for nearest-neighbor analysis of tumor and stromal immunophenotypes. c Nearest-neighbor analysis showing the frequency distribution of immune cell proximities to tumor-associated with expression quartiles (Q1–Q4) of cytoplasmic Kaiso, LC3A/B, nuclear Kaiso, and race (white versus black).