Reconstitution of Morphogen Signaling Gradients in Cultured Cells

Abstract

Development of multicellular organisms depends on the proper establishment of signaling information in space and time. Secreted molecules called morphogens form concentration gradients in space and provide positional information to differentiating cells within the organism. Although the key molecular components of morphogen pathways have been identified, how the architectures and key parameters of morphogen pathways control the properties of signaling gradients, such as their size, speed and robustness to perturbations, remains challenging to study in developing embryos. Reconstituting morphogen gradients in cell culture provides an alternative approach to address this question. Here we describe the methodology for reconstituting Sonic Hedgehog (SHH) signaling gradients in mouse fibroblast cells. The protocol includes the design of morphogen sending and receiving cell lines, the set-up of radial and linear gradients, the quantitative time-lapse imaging, and the data analysis. Similar approaches could potentially be applied to other cell-cell communication pathways.

Introduction

Establishing precise patterns of gene expression in space and time is a key feature in the development of multicellular organisms. Such patterns are controlled in part by morphogen gradients. Morphogens are signaling molecules that emanate from a source and diffuse through space creating concentration gradients (1). Morphogen ligand gradients then trigger graded signaling responses in the field of receiving cells, which differentiate into discrete cell types based on the quantitative signals. Features of the morphogen gradients, such as the amplitude, lengthscale, timescale, and robustness to environmental and genetic perturbations provide the crucial information to allow for complex tissue patterning. These quantitative features are determined by the key biochemical parameters in the morphogen pathway, such as the diffusion and degradation rates of the morphogen, as well as the architecture of the pathway, such as the signal transduction logic and feedback loops (2–4). Understanding how the extra- and intra-cellular contexts of cells regulate these features of morphogen gradients is important for revealing the principles underlying the spatial organization of tissues. Addressing this question requires direct visualization of the morphogen ligand or signaling gradients in space and time, as well as genetic manipulation of the extra- and intra-cellular environment in a spatio-temporally precise manner, both of which remain technically challenging in developing embryos.

Here, we describe a new approach for quantitatively studying morphogen signaling gradients outside an embryo, using the Hedgehog (HH) pathway as an example (5). Within the HH pathway, mammalian Sonic Hedgehog (SHH) is the most studied ligand and a classic example of a long-range morphogen. During embryo development, SHH is responsible for patterning tissues such as the neural tube, limb, and the gut (6, 7). In the absence of the ligand, the Patched (PTCH) receptor represses the intracellular signaling cascade (8, 9). In this “signal-OFF” state, Glioma-associated Oncogene Homolog (GLI) transcription factors are processed from a full-length protein to a transcriptional repressor (10). Binding of SHH to PTCH relieves the negative regulation and subsequently, GLI proteins are processed into transcription activator to induce gene expression. Using this pathway, we will demonstrate how morphogen signaling gradients can be reconstituted in a petri dish, and how the spatio-temporal dynamics of the signaling gradient can be quantitatively analyzed.

1.1. General Method

In living organisms, specific cells secrete morphogens which other cells, or even themselves, respond to. Our synthetic system relies on the same principle; we created cell lines that either produce morphogen ligands (“sender cells”) upon induction with a chemical or respond to the ligand (“receiver cells”) by turning on a transcription-based fluorescent reporter. By co-culturing senders and receivers in different spatial arrangements, signaling gradients of radial or linear geometries can be created in a petri dish, to recapitulate some aspects of the patterning systems found in nature (11, 12).

1.2. Strengths

The reconstituted system provides several strengths. First, genetic manipulation in cultured cells can be achieved easily and precisely. For example, we were not only able to rewire the SHH pathway to eliminate or employ a key negative feedback loop, but also able to tune the strength of the negative feedback, a key parameter that determines the gradient robustness to variations in the morphogen production rate (5). Second, the effects of a gene on signal sending vs. receiving can be easily disentangled. This is because signaling senders and receivers can be manipulated separately and patterned in defined spatial arrangements. Third, the signaling gradients can be quantitatively measured in space and time in the reconstituted system. Morphogens often exist at low concentrations, making it challenging to quantify the ligand directly. Instead, the reporter system is established on the capability of cells to detect low levels of morphogens and induce gene expression, which ultimately determines cell fates. Finally, this system allows for the isolation of morphogen signaling from other developmental processes that could occur simultaneously in a developing embryo and complicate the interpretation of mutant phenotypes. While simplified, this cell-based reconstituted system still relies on the complexities of cellular processes, such as protein degradation, ligand uptake, and extracellular matrix dynamics, and thus is suitable for studying morphogen-mediate intercellular communication.

1.3. Weaknesses

Reconstitution allows for precise control of the morphogen pathways and isolation of the gradient formation process, which is valuable and necessary in many cases. However, it does not recapitulate the entire spectrum of complexity that exists in a developing tissue. First, developing embryos may have quantitative or qualitative differences in the expression levels of morphogen pathway components, such as receptors or co-receptors. However, the ease of genetic manipulation in cultured cells enables the possibility of reconstituting the pathway complexity piece by piece and precisely tuning the level of gene expression. Second, cells within both the sender and receiver populations exhibit cell-to-cell variability, such as transgene expression levels, motility, and sensitivity to contact inhibition when the cultures reach confluency. Therefore, to deduce a biologically meaningful conclusion, it is crucial to take into account the cell-to-cell variability when designing cell lines and experiments, such as checking multiple clonal populations and averaging across multiple gradient measurements. Third, cell proliferation has been implicated to play a role in establishing gradients in certain contexts, but cell division is relatively limited under our current culture condition (13). Therefore, without further modification, it cannot be used to assess the contribution of cell division to patterning. However, we expect that incorporating these additional cellular behaviors into the reconstituted system is possible by choosing the right cell types and growth conditions.

Materials

2.1. Parental Cell Lines

Identifying a proper cell line that is able to recapitulate the signaling pathway is essential. For receivers, a cell type where all the necessary components of the pathway are or can be expressed, excluding the ligand, reduces the number of components that need to be reconstituted. The chosen cell line also needs to have the structural or behavioral features relevant to the signaling pathway. In the case of the SHH pathway, the presence of cilia is needed for proper signaling activation in mammalian cells (14). Additionally, while the sender and receiver cells do not necessarily have to be the same cell type, the cell lines must be able to be co-cultured stably in the same condition. With these criteria, NIH3T3 cells, an immortalized mouse embryonic fibroblast cell line, were chosen for creating SHH sender and receiver cells (15). NIH3T3 cells can respond to SHH without differentiation, and do not produce any HH ligands naturally. They are subject to contact inhibition and enter quiescence at high confluency, which promotes the formation of cilia and competency of activating SHH signaling pathway (16). We also observed that NIH3T3 cells can stay at 100% confluency for over 3 days without major cell death (5).

2.2. Cell Culture Materials

2.2.1. NIH3T3 Cell Culture Media

• High glucose DMEM (Thermofisher 11960044) supplemented with the following:

• 10% Cosmic Calf Serum (GE Healthcare SH30087.03)

• 1x Penicillin-Streptomycin-Glutamine (Gibco 10378016)

• 1 mM sodium pyruvate (Gibco 11360070)

2.2.2. NIH3T3 Imaging Media

• FluoroBrite™ DMEM (Thermofisher A1896701) supplemented with the following:

• 10% Cosmic Calf Serum (GE Healthcare SH30087.03)

• 1x Penicillin-Streptomycin-Glutamine (Gibco 10378016)

• 1 mM Sodium Pyruvate (Gibco 11360070)

2.2.3. General Tissue-Culture Supplies

• Trypsin-EDTA (0.05%) (Gibco 25300054)

• Dulbecco’s phosphate-buffered saline (1x DPBS) without calcium or magnesium (Gibco 14190250)

• Corning® Costar® TC-Treated Multiple Well Plates

2.2.4. Transfection Kit and Chemicals

• Lipofectamine LTX (Thermofisher 15338030)

• Blasticidin S Solution (InvivoGen ant-bl-1)

• Hygromycin B Gold Solution (InvivoGen ant-hg-1)

• (Z)4-Hydroxytamoxifen (4-OHT) (Sigma H7904) dissolved in DMSO at a concentration of 500 μg/mL.

2.2.5. Fluorescent Dye Control

• Fluorescein Sodium Salt (Millipore Sigma F6377)

2.2.6. Materials for establishing and imaging gradients

• Ibidi cell culture inserts (Ibidi 80209) for setting up linear gradients

• 24-well Imaging plates with glass-bottom or imaging-compatible polymer coverslip plates (e.g. Ibidi 82406)

2.3. Microscopy

This method is suited for quantitatively measuring the spatio-temporal dynamics of morphogen signaling gradients using time-lapse imaging. We recommend an inverted widefield microscope with multiple filter cubes for identifying senders and receivers separately and autofocus capability for long-term movies over 48 hrs. We use Nikon Ti2-E equipped with the Perfect Focus System and the OKO Labs environmental enclosure to keep cells at 37° C with 5% CO₂ and proper humidity. 10x or 20x objectives provide sufficient resolution.

Methods

3.1. Morphogen Producing and Detecting Cell Lines

To create the SHH sender cells, an inducible gene expression system of two plasmids under the control of 4-OHT was used (Figure 1A). One plasmid constitutively expresses a GAL4 transcription factor fused to a mutant estrogen receptor ERT2, and a blue fluorescent protein mTurquoise2 fused to histone H2B (17, 18). The two coding sequences are separated by a viral 2A self-cleaving peptide (T2A) to produce separate protein products under the control of a single PGK (3-phosphoglycerate kinase) promoter (19). In the absence of 4-OHT, the ERT2-GAL4 fusion protein is found in the cytoplasm where it is unable to induce gene expression. Once 4-OHT is introduced, GAL4 moves into the nucleus where it binds to an upstream activating sequence (UAS) sequence that controls the expression of Shh. This enables a rapid induction of gene expression that is sensitive to dose and temporal control. In addition, the nuclear localization of mTurquoise2 allows for easy identification of sender cells.

Figure 1.

Engineering clonal stable sender and receiver cell lines. A) The constructs used for creating senders and receivers based on a piggyBac transposase system. B) Procedures for transfecting constructs and selecting for clonal populations. Constructs are transfected into wild-type NIH3T3 cells together with a plasmid overexpressing piggyBac transposase. Cells stably integrated with the desired plasmids are selected with antibiotics. Clonal populations with desired features (e.g. high expression level of SHH) were further sorted. C) Co-culture of senders and receivers is used for producing radial or linear gradients that can be quantitatively imaged and analyzed.

Receiver cells lines are created through stable integration of a fluorescent reporter that is controlled by GLI proteins, which are transcription factors downstream of the SHH signaling pathway. The reporter was constructed based on the reporter mice made by Balaskas et al (20, 21). Specifically, 8 tandem copies of GLI-binding sites (GBS) taken from the enhancer of FoxA2 gene, a natural target of SHH in the neural tube, was placed upstream of a minimal CMV promoter, and together they drive the expression of a yellow fluorescent protein mCitrine that is fused to H2B (H2B-mCitrine) (Figure 1A) (20–22). Increasing concentration of recombinant SHH induces increasing mean intensity of mCitrine within the receiver population (5).

Both sender and receiver cell lines were created using a piggyBac transposon system (System Biosciences) where plasmids carrying genes of interest are co-transfected with a plasmid expressing piggyBac transposase (Figure 1B). The piggyBac transposase recognizes inverted terminal repeats (ITRs) that flank the genes of interest, cut the ITRs and insert the genes into the genome at TTAA DNA sequences (23). For reproducibility, cells with plasmids stably integrated were selected using antibiotics, and clonal cell lines were isolated and used throughout the study. To achieve stable integration, each plasmid carries a unique antibiotic resistance gene that can be selected for by treating the cells with the corresponding antibiotic 48 hrs after transfection. Because of the high efficiency of integration and the low false-positive rate of antibiotic selection in the piggyBac system, at least two constructs carrying different antibiotic resistance genes can be integrated into the genome of the same cell simultaneously, greatly shortening the time required for cell line construction. In such cases, the cells are first treated with the “harsher” antibiotic (that cells are more sensitive to) for two days to eliminate the un-transfected cells as quickly as possible (see Note 1), before being switched to the secondary antibiotic selection in fresh media for another two days, or until the negative control cells have died off. To select clonal cell population, either limiting dilution or fluorescence-activated cell sorting (FACS) can be used to ensure only one cell can populate a single well in a 96-well plate. After about 2 weeks, healthy clones with desired properties can be selected through qPCR, immunofluorescence staining, flow cytometry, imaging, or whichever method best suited for the underlying biology (see Note 2).

3.2. Radial Gradients

This experimental set up is both versatile and simple to execute (Figure 2A).

Figure 2.

Reconstituting and quantifying radial gradients. A) Experimental setup. Sender and receiver cell lines are mixed at a 1:2000 ratio to establish the geometry in which a single sender is surrounded by receivers and two senders are far apart from each other. After induction of SHH expression with 4-OHT, signaling response in receiver cells can be measured based on mCitrine expression. B) Representative time-lapse images of a radial gradient. Gradients begin to form at ~12 hours and can reach distances of ~100 μm. C) Quantification of radial gradients. Concentric circles around the sender are used to create bands of constant width, within which the average mCitrine fluorescence intensity is measured (inset). The spatio-temporal dynamics of signaling gradient formation is plotted as a function of distance from the sender (mean of 9 gradients).

1. Grow sender and receiver cells to confluency in a 6-well plate (see Note 3).

2. Wash the wells with 500 μL 1x DPBS.

3. Add 200 μL Trypsin-EDTA (0.05%) to each well and incubate the cells for 5 minutes or until the cells detach from the plate.

4. Add 1 mL culture media to each well to neutralize the Trypsin-EDTA and gently pipet to dissociate the cells into single-cell suspension (See Note 4).

5. Spin down both sender and receiver populations at 500 xg for 5 minutes to pellet the cells and remove the Trypsin-EDTA.

6. Aspirate the supernatant and resuspend the cells in 1 mL culture media by gentle pipetting.

7. Count the sender and receiver populations (see Note 5).

8. Thoroughly mix senders and receivers at a ratio of 1:2000.

9. Dispense 500,000 cells into each well on a 24-well imaging plate with minimal disturbance as to keep the cells at an even density across the well. The cell number should ensure 100% confluency to minimize cell division after attachment (see Note 6).

10. Once the cells have attached (4–5 hours to overnight), check the well to ensure single, isolated senders in a field of receivers.

11. Dilute 4-OHT to 100 ng/mL in fresh imaging media (see Note 7).

12. Aspirate the regular culture media and add 1 mL of the imaging media containing 4-OHT to induce SHH production in the senders.

13. Set up time-lapse imaging immediately or keep the plate in the incubator and take images at desired time points (e.g. ~50 hours post 4-OHT addition).

3.3. Linear Gradients

To create gradients in which senders and receivers share a linear boundary, removable adhesive culture inserts are used to confine senders in a rectangular region (Figure 3A).

Figure 3.

Reconstituting and quantifying linear gradients. A) Experimental setup. A culture insert is used to plate sender cells within a rectangular region, and receiver cells are then used to cover the remainder of the plate. Gradients can be analyzed starting from the sender-receiver boundary. B) Representative time-lapse images of a linear gradient. Initial mCitrine expression begins at ~10 hrs and after 50 hrs the gradient extends further than 200 μm. White dashed line indicates the sender-receiver boundary. C) Quantification of linear gradients. Slices of the gradient parallel to the sender-receiver boundary are analyzed and the average mCitrine intensity within each slice across space and time is plotted (mean of 7 gradients).

1. Prepare sender and receiver cell populations the same as steps 1–7 in the above radial gradient protocol.

2. Place the Ibidi insert in a 24-well plate well using forceps so that one of the chambers is centered in the middle of the well (the other chamber will not be used).

3. Rinse the centered chamber with 100 μL culture media to prime all surfaces for sender cell placement.

4. Seed 100,000 sender cells suspended in 100 μL culture media into the centered chamber and keep the plate in an incubator for 4–5 hours.

5. Once the cells have settled, rinse the Ibidi chamber with 100 μL culture media three times to remove unattached cells and prevent the formation of satellite gradients.

6. Remove the Ibidi insert carefully by pulling it perpendicular from the surface of the well to minimize boundary disturbance (see Note 3).

7. Wash the entire well with 500 μL culture media three times to get rid of any loosely attached cells.

8. Dispense 500,000 receiver cells into the well.

9. Incubate the plate overnight to give receiver cells sufficient time to form confluent layers and enter quiescence (see Note 5).

10. The following morning, replace the culture media with 4-OHT diluted to 100 ng/mL in fresh imaging media (see Note 8).

11. Set-up time-lapse imaging immediately or keep the plate in the incubator and take images at desired time points (e.g. ~50 hours post 4-OHT addition).

3.4. Time-lapse Imaging

To quantify the spatio-temporal dynamics of the signaling gradients, the plates were imaged on a Nikon Ti2 fluorescence microscope. The microscope is equipped with a fully enclosed environmental chamber that is kept at 37°C, 5% CO₂ and the appropriate humidity. Images were collected using a 10x objective and filter cubes for mTurquoise2 (Ex436/Em480) and mCitrine (Ex500/Em535) in 30-minute intervals over a 50-hour period (Figure 2B, 3B). SHH signaling gradients start to appear at ~12 hrs in the radial geometry and at ~10 hrs in the linear geometry (Figure 2C, 3C).

3.5. Imaging Controls

Several factors, such as media autofluorescence and non-uniform illumination within the field of view, can confound the quantification of the true signal from the SHH reporter. Therefore, two types of controls were measured side by side with the gradient samples.

1. To measure the autofluorescence of the media, add 1 mL of fresh imaging media to an empty well in the same 24-well imaging plate.

2. To quantify the non-uniform illumination within the field of view, add 1 mL of fresh imaging media containing fluorescein at a final concentration of 10 nM to an empty well in the same 24-well imaging plate.

3. Image 5 random positions in each of the control wells using the same microscope settings as those used for the experiment and use the mean of the 5 positions for the data analysis.

3.6. Image Analysis

We have developed customized MATLAB codes for analyzing both radial and linear gradients. In both cases, the raw data is pre-processed in two steps; first by subtracting the media-only background signal and second, by dividing the mCitrine signals by the fluorescein control to account for non-uniform illumination. These pre-processed images were then analyzed as radial or linear gradients.

For radial gradients, single individual senders are segmented and tracked based on their nuclear-localized mTurquoise2 using a program developed by Bintu L (24). From these clusters of identified points, the XY positions of the senders are extracted, which represents a single pixel in the center of the sender nucleus. The list of sender positions over time are used to center the analysis of radial gradients and account for cell migration. Radial gradients are analyzed by building concentric circles around the sender cell position at each time point (Figure 2C, inset). The averaged signal intensity within each ring corresponds to the activation level of the signaling reporter. The actual gradient starts outside the sender cell boundary and is identified as the peak in the mCitrine signal. The peaks then can be used to align and compare the individual quantified gradients to each other. The 8xGBS-CMV promoter drives a basal level of mCitrine expression, which is independent from signaling response. The level of the basal expression depends on the copy number of the reporter integrated into the genome and their genomic locations, and thus the basal expression level often correlates with the maximum expression level induced by SHH. Therefore, we calculate the signaling activation based on the fold change in mCitrine intensity over the basal level of mCitrine intensity. Single gradients from the same experimental sample can be averaged together and plotted as a function of to time and distance from the sender (Figure 2C). An alternative method for image analysis is the Fiji Radial Profile plugin where a circle can be drawn encompassing the gradient and centered on the single sender cell. The intensity of the pixels is associated with the degree of receiver cell activation. However, the Fiji Radial Profile plugin only works for static images.

The analysis of the linear gradients depends on quantifying the receiver response in intervals from a defined linear boundary. To automatically detect the linear boundary between senders and receivers, all pixels in the mTurquoise2 channel are first binarized based on the background threshold, with pixels above the threshold identified as being associated with senders and assigned “1”, and pixels below the threshold as “0”. By adding up all the pixels within each column that runs parallel to the orientation of the sender-receiver boundary, a sender density profile can be calculated as a function of space, with x=1 having the highest sender density, and x=1024 having the lowest sender density (each image has 1024 × 1024 pixels) (see Note 9). The boundary can then be defined as where the sender density drops below 10% of the maximum sender density. The mCitrine intensity of pixels within each column parallel to the defined sender-receiver boundary were then averaged to quantify the gradient as a function of distance (Figure 3C, inset). Similar to the radial gradients, the averaged mCitrine intensities were normalized to the basal level of expression. By applying this method to all timepoints, the spatio-temporal dynamics of signaling gradients can be reconstructed. These values again can be averaged across multiple gradients or displayed separately in respect to time and distance (Figure 3C).

All the image analysis codes are archived and publicly available at http://doi.org/10.5281/zenodo.3772886.