Bone Marrow Stromal Cells Enhance DNA Damage Signaling Independent Of Stat3 Activation In Pediatric AML

Despite aggressive treatments, death from chemoresistant disease still occurs for almost half of children with AML. It is clear that the bone marrow microenvironment protects a subset of cells from chemotherapy, but the mechanisms of resistance remain unknown. We recently reported that patients whose AML blasts activated Stat3 in response to two bone marrow-derived ligands, G-CSF and IL-6, had a significantly superior survival rate, compared to patients whose blasts failed to activate Stat3 (Redell, et al, 2013, Blood121:1083). In this followup study of a subset of these patients, we further investigated the basis for the failure of the Stat3 response, as well as mechanisms of environment-mediated chemotherapy resistance. Our hypotheses were A) Stat3 responsiveness to G-CSF/IL-6 reflects the cell’s general ability to respond to ligand stimulation, rather than a specific defect in a Stat3 pathway; and B) Stat3 responsiveness correlates with other functional responses to stromal exposure, including survival, proliferation, and chemotherapy resistance. Second samples from 27 patients were studied by FACS analysis of intracellular markers. To study stroma-induced changes, AML cells were co-cultured on stromal cells overnight, or briefly stimulated with stroma-conditioned medium (CM). We used HS5 cells, which secrete high levels of many soluble factors, and HS27A cells, which secrete very few factors. The stromal cell lines express mOrange for exclusion by FACS.

First, primary AML cells were stimulated with G-CSF (10 ng/ml) or IL-6 + soluble IL-6R (5 + 10 ng/ml) for 15 min, then pY-Stat3 (PE) and pERK1/2 (AlexaFluor488) were measured by FACS. Data were acquired on the LSRII (BD) and analyzed with FCSExpress4 (DeNovo). Ligand responses were expressed as the fold change in mean fluorescence intensity over unstimulated cells (ΔMFI). We found a strong correlation between G-CSF-induced pY-Stat3 ΔMFI and pERK1/2 ΔMFI (R=0.858, p<0.001), suggesting that failure to activate Stat3 in unresponsive samples was not due to a Stat3-specific block. As expected, HS5 CM strongly activated Stat3 in samples that also responded to G-CSF and IL-6, while HS27A CM did not activate Stat3 in any samples. We compared responses to G-CSF v. HS5 CM, and found a strong correlation for pY-Stat3 ΔMFI (R=0.886, p<0.001) and for pERK1/2 ΔMFI (R=0.907, p<0.001). Samples that failed to respond to G-CSF also failed to respond to a combination of multiple soluble factors, supporting the idea that signaling function or dysfunction is a generalized property of AML cells.

To further investigate functional changes induced in AML cells by stromal cells, we co-cultured blasts from 17 samples overnight with HS5 or HS27A cells. We labeled cleaved PARP (AlexaFluor647) to measure viable cells (%cPARP-) in each condition. Within the viable AML cell population, we measured the proliferative fraction (Ki67-FITC+ or Ki67-AlexaFluor700+) and the fraction with active DNA damage signaling (γH2AX-AlexaFluor488+). We found a small increase in the %viable for co-cultured cells v. cultured alone (p<0.05 for HS27A v. alone; Wilcoxon Signed Rank Test) and in the %Ki67+ (p<0.005 for HS27A v. alone). Moreover, increases were seen in γH2AX, with 5.6 ±6%+ in cells cultured alone, v. 7.9 ±6% in HS5 co-cultures (p<0.05) and v. 11.5 ±9% in HS27A co-cultures (p<0.001). For 10 samples, cell number was sufficient for additional overnight cultures with etoposide (up to 10 mM). Both stromal cell lines enhanced DNA damage signaling, reflected by an increase in γH2AX from 7 ±10% in etoposide-treated cells cultured alone, to 18 ±19% in HS5 co-cultures (p<0.05) and 19 ±16% in HS27A co-cultures (p<0.005). To test whether HS5 CM-induced pY-Stat3 or pERK1/2 were related to these parameters, we examined bivariate correlations. We found no correlation in most cases, but an inverse relationship between both ΔMFIs and %viable after etoposide (pY-Stat3: R=-0.733, p<0.05 and pERK1/2: R=-0.683, p<0.05). In other words, samples characterized by signaling failure tended to have better survival after etoposide treatment. This result is consistent with our prior study showing that patients whose blasts failed to activate Stat3 had poorer outcome than those whose blasts activated Stat3 normally. In summary, we found that environment-induced Stat3 signaling did not promote chemoresistance, and that stromal cells may instead confer chemoresistance by enhancing DNA damage signaling.