Dynamic, large-scale profiling of transcription factor activity from live cells in 3D culture

Abstract

Background: Extracellular activation of signal transduction pathways and their downstream target transcription factors (TFs) are critical regulators of cellular processes and tissue development. The intracellular signaling network is complex, and techniques that quantify the activities of numerous pathways and connect their activities to the resulting phenotype would identify the signals and mechanisms regulating tissue development. The ability to investigate tissue development should capture the dynamic pathway activity and requires an environment that supports cellular organization into structures that mimic in vivo phenotypes. Taken together, our objective was to develop cellular arrays for dynamic, large-scale quantification of TF activity as cells organized into spherical structures within 3D culture.

Methodology/principal findings: TF-specific and normalization reporter constructs were delivered in parallel to a cellular array containing a well-established breast cancer cell line cultured in Matrigel. Bioluminescence imaging provided a rapid, non-invasive, and sensitive method to quantify luciferase levels, and was applied repeatedly on each sample to monitor dynamic activity. Arrays measuring 28 TFs identified up to 19 active, with 13 factors changing significantly over time. Stimulation of cells with β-estradiol or activin A resulted in differential TF activity profiles evolving from initial stimulation of the ligand. Many TFs changed as expected based on previous reports, yet arrays were able to replicate these results in a single experiment. Additionally, arrays identified TFs that had not previously been linked with activin A.

Conclusions/significance: This system provides a method for large-scale, non-invasive, and dynamic quantification of signaling pathway activity as cells organize into structures. The arrays may find utility for investigating mechanisms regulating normal and abnormal tissue growth, biomaterial design, or as a platform for screening therapeutics.

Figure 1. Reporter constructs to normalize for transfection efficiency during structure formation in hydrogels.

(A) MCF-7/WS8 cells growing within Matrigel hydrogels without DNA complexes began as individual cells at day 1 (top left panel) and formed multi-cellular spheroids by day 3 (top right panel), a process that was not inhibited by 0.5 µg DNA lipoplexes (bottom panel). Scale bar, 200 µm. (B–D) Cells were transfected with reporter genes for EGFP and GLuc in parallel (0.5 or 0.05 µg DNA total with an pEGFP:pGLuc mass ratio of 9∶1) within hydrogels. Percentage of EGFP positive cells (B) and activity of GLuc (C) increased with DNA amount at both 24 and 48 h. Values are means ± s.d. from two independent experiments carried out in triplicate. GLuc activity correlated with %EGFP-positive cells (D).

Figure 2. Dual-luciferase reporter constructs to measure TF activity.

(A–D) MCF-7/WS8 cells within hydrogels were transfected in parallel with pFLuc reporter constructs containing enhancer elements for specific TFs (AP1, ER, and p53), vector control (TA), or not transfected (NC). For wells with reporter constructs, a constituently active pGLuc construct was delivered to normalize for transfection efficiency. (A) Cells were extracted and lysed after 48 h to determine FLuc production, which was normalized to transfection efficiency. (B–D) Bioluminescence imaging was used to non-invasively quantify FLuc production after 48 h from cells growing within Matrigel. (B) Pseudo-color mapping demonstrates localized luminescence output from cells in hydrogels seeded in alternating wells of a 96-well plate. (C) Raw FLuc signals were significantly greater than bioluminescence noise from NC. (D) FLuc signals from bioluminescence imaging were normalized to transfection efficiency and normalized TF activities for all three factors were significantly greater than TA. Values are means ± s.d. from at least three independent experiments carried out in triplicate.

Figure 3. Dual-luciferase constructs to capture changes in TF activity in response to E₂ treatment.

(A,B) MCF-7/WS8 cells transfected with pFLuc and pGLuc reporter constructs and seeded within Matrigel were treated with E₂ for 24 h. TF activity was measured by extracting cells from hydrogels and lysing (A) or by bioluminescence imaging (B). (C) MCF-7/WS8 cells were grown on 2D polystyrene, transfected, and treated with E₂ for 24 h. TF activity was measured by lysing cells. Values are means ± s.d. from at least three independent experiments carried out in triplicate. NS indicates no significant difference with treatment.

Figure 4. Large-scale arrays to monitor TF signaling in parallel.

(A) Pseudo-color mapping indicated localized luminescence from wells throughout a portion of a 96-well plate randomly seeded with 16 conditions. (B) Quantification of TF activity at 24 h by bioluminescence imaging for 28 TFs. Values are means ± s.d. from at least three independent experiments carried out in triplicate. NS indicates not significantly greater than the TA control.

Figure 5. Dynamic large-scale activities of TFs in response to E₂ treatment.

(A) TFs had variable activities as functions of treatment and time, as indicated by a heat map representation of log₂ transformation of the means. Vehicle control indicated by ‘−’ and E₂ treatment indicated by ‘+’. (B) TFs that had significant changes with treatment or time are presented on individual temporal plots, with log₂ transformations of the TF activity means plotted. Dashed lines indicate vehicle control and solid lines indicate E₂ treated. Horizontal lines represent comparisons at different time points for each condition, with 6 h to 24 h (top left), 24 h to 48 h (top right) and 6 h to 48 h (bottom). NS indicates no significant difference. A significant difference with E₂ treatment is indicated with the p-value at the significant time point. For example, CRE had significant differences in activity between 6 and 24 h (P = 4e-6) and between 6 and 48 h (P = 1e-13) for untreated cells, and significant differences for all temporal comparisons for E₂ treated cells. ER had significantly greater activity at 24 and 48 h (P = 0.007, P = 0.013), but no significant differences with time. Values are log₂ transformed means from at least three independent experiments carried out in duplicate for each condition.

Figure 6. Dynamic large-scale activities of TFs in response to activin A treatment.

(A) TFs had variable activities as functions of treatment and time, as indicated by a heat map representation of log₂ transformation of the means. Vehicle control indicated by ‘−’ and activin A treatment indicated by ‘+’. (B) TFs that had significant changes with treatment or time are presented on individual temporal plots, with log₂ transformations of the TF activity means plotted. Dashed lines represent vehicle control and solid lines represent activin A treatment. Statistical comparisons are described in legend of Figure 5. Values are log₂ transformed means from at least three independent experiments carried out in duplicate for each condition.