The Ticking Clock: Why 3D Cell Cultures Are a Moving Target

Groundbreaking Research Reveals Continual Change in HepG2 Spheroids Over Time

The quest to find better drug screening models is crucial. For years, scientists relied on flat, two-dimensional (2D) cell cultures, but these often fail to mimic the complexity of human tissue, leading to high failure rates in drug development. The shift to three-dimensional (3D) spheroid cultures, tiny, spherical clumps of cells, was adopted as a powerful solution, offering a more realistic environment, particularly for modelling the liver with HepG2 cells.

However, the question remained: just how stable are these 3D models over an extended period? Our co-authored study, published in the journal Scientific Reports on 25 May 2021, answers this, revealing a surprising and important truth about the duration of culture.

The Publication Details

The article, titled "Continual proteomic divergence of HepG2 cells as a consequence of long-term spheroid culture," was a collaborative effort by: Andrea Antonio Ellero, Iman van den Bout, Maré Vlok, Allan Duncan Cromarty, and Tracey Hurrell.

The Core Finding: Continuous Change, Not Stability

We hypothesised that while 3D culture changes the cells, the new phenotype might not be static; it could be continually evolving. To test this, we used high-resolution proteomic methods to track the abundance of over 4,800 proteins within HepG2 spheroids over 28 days.

Divergence is Rapid and Ongoing: The protein profile (proteome) of the HepG2 spheroids became distinctly different (divergent) from the standard 2D monolayer cells after 14 days in culture and critically, it continued to change over the following two weeks (Days 21 and 28).
The Magnitude of Change: The number of proteins showing statistically significant differences compared to the starting monolayer population kept increasing with time. For instance, by Day 28, nearly 58% of the detected proteome had changed. Even comparing Day 14 to Day 28 revealed a change in over 41% of the proteins, confirming that the cell state is a continuously moving target.
Hepatic Proteins are Modulated: We saw progressive increases in hepatic markers like Albumin (ALB) and α-fetoprotein (AFP).

Structural and Analytical Assurance

The most compelling evidence for this transformation was found in the proteins responsible for the physical cell structure and communication, backed by stringent analytical methodology.

Structural Remodelling

Extracellular Matrix (ECM) Remodelling: The cells progressively deposited and altered the composition of the ECM. We saw increases in structural and regulatory proteins like coagulation factor XIII (F13A1) and tissue inhibitor of metalloproteinase 3 (TIMP3).
Strengthening Cell-Cell Links: Cell-cell junction and adhesion proteins were continually modulated. For instance, Cx32, a protein that forms gap junction channels for direct cell-cell communication, was progressively upregulated throughout the time course.

Technical Deep Dive: Quantifying Proteins with TMT

To precisely compare protein changes, we used Tandem Mass Tag (TMT) labelling.

This process involved breaking proteins into peptides, uniquely labelling each sample group with a TMT tag, and then combining them into one mixture.
In the mass spectrometer, fragmentation using Higher-Energy Collisional Dissociation (HCD) released a unique reporter ion for each original sample.
Measuring the intensity of these ions precisely quantified the relative protein abundance, proving the continual proteomic divergence.

The Challenge: Too Many Samples, Too Few Tags

Imagine you have 13 different coloured paints (your protein samples) but only six spaces on your palette (the TMT tags) to mix them and compare their intensity. If you mix the first six paints in Set 1 and the next six paints in Set 2, you can't accurately compare Paint 1 to Paint 13 because they were measured on different days, on different palettes, with different lighting.

The Problem: The TMT system allows highly accurate relative comparison (i.e., this protein is twice as high as that protein) only when samples are mixed and measured in the exact same run. Running samples separately would introduce technical variations, ruining the critical long-term comparisons needed for the 28-day study.

The Solution: The Universal Reference “Bridge”

The researchers overcame this by creating a single, universal reference sample to act as a “bridge” between all the separate measurement runs.

Creation of the “Bridge”: They created a Pooled sample by taking a small, equal amount of protein from all 12 original experimental samples (D0, D14, D21, D28 replicates). This Pooled sample represented the average of the entire experiment.
Dedication of a Channel: In every single TMT run they conducted, one of the six available tags was always dedicated to this Pooled reference sample.
The Comparison:
- Run 1 (TMT Set A): Compares D0-Replicate 1, D14-Replicate 1, D21-Replicate 1, etc., all against the common Pool/Bridge.
- Run 2 (TMT Set B): Compares D0-Replicate 2, D14-Replicate 2, D21-Replicate 2, etc., all against the common Pool/Bridge.
Final Normalisation: Because every sample's measurement was tied back to the exact same universal reference (the Bridge), the researchers could then normalise all the data together. This allowed them to accurately compare any D14 sample from Run A to any D28 sample from Run B, achieving the comprehensive 13-sample comparison despite the 6-tag limit.

They essentially used the Pool as a reliable internal yardstick to stitch the results of the separate TMT runs together.

The Crucial Impact on Drug Testing and Toxicology

Our finding that the HepG2 spheroid proteome continues to evolve dramatically has immediate and major consequences for how these models are used:

Eliminating the Dynamic Baseline: If a researcher tests a drug on Day 14 and another tests it on Day 28, they are essentially using two biologically different model systems. The culture time itself introduces significant variability that undermines the core purpose of a reproducible assay.
Safety and Efficacy Implications: The continuous shifts in hepatic proteins directly impact the predictability of toxicity. Inconsistencies in preclinical screening models are a large contributor to drug candidate failure.
Validation is Essential: Before being accepted as a reliable platform, 3D culture models must be thoroughly characterised to determine the precise, reproducible culture period that is "fit-to-purpose" for a given experiment.

In short, the success of 3D modelling requires treating culture duration not as a constant, but as a critical and dynamic parameter that defines the very biology of the model.