**Abstract**

Culturing cells in 3D is often considered to be significantly more difficult than culturing them in 2D. In practice, this is not the case: the situation is that equipment needed for 3D cell culture has not been optimised as much as equipment for 2D. Here we present a few key features which must be considered when designing 3D cell culture equipment. These include diffusion gradients, shear stress and time. Diffusion gradients are unavoidably introduced when cells are cultured as clusters. Perhaps the most important consequence of this is that the resulting hypoxia is a major driving force in the metabolic reprogramming. Most cells in tissues do not experience liquid shear stress and it should therefore be minimised. Time is the factor that is most often overlooked. Cells, irrespective of their origin, are damaged when cultures are initiated: they need time to recover. All of these features can be readily combined into a clinostat incubator and bioreactor. Surprisingly, growing cells in a clinostat system do not require specialised media, scaffolds, ECM substitutes or growth factors. This considerably facilitates the transition to 3D. Most importantly, cells growing this way mirror cells growing *in vivo* and are thus valuable for biomedical research.

**Keywords:** clinostat, functionality, hypoxia, spheroids and organoids, *in vivo* mimetic, culture time, minimisation of infections, direct observation, media change

#### **1. Introduction**

Cells are pre-programmed to carry out certain functions – this represents their potential. At the same time, they are sensitive to physical or biological changes in their surrounding environment and will modify their function accordingly – and this becomes their constantly changing actuality. Cells *in vivo* are surrounded by an extracellular matrix (ECM) and are actively communicating with other cells. This forms part of an active environment which modifies and integrates their activities.

In the early days of cell culture in the 1950's the focus was to get cells to propagate rapidly and reliably in flasks, facilitated by the destruction of their ECM. With the realisation that cells grown in 3D conditions are more mimetic of human cell

biology, the focus has changed away from getting the cells to propagate to getting the cells to function (physiologically). These two conditions mark the extremes in a spectrum of cellular activity [1].

Bearing this in mind, there are a number of factors that should be considered when changing this focus and transitioning from the classical 2D cell culture to 3D cell culture. These factors not only indicate which 3D culture systems could be expected to be advantageous over others but also indicate which should generate data that is more representative of the *in vivo* performance of such cells.

Perhaps, the most significant difference between 2D and 3D culture is the establishment of longer diffusion gradients for the majority of cells and thus the cells will experience significantly different levels of oxygen, CO₂, nutrients and waste products.

Related to the diffusion gradients is the amounts of various compounds in the environment around cells: cells in 2D are typically exposed to levels of for example O₂ and glucose which are not seen in the intact, healthy organism.

Another very significant difference will be the establishment of channels of communication between cells that are not only juxtapositioned but also further away. In 2D, immortal cells are typically passaged roughly every week (and shorter for faster growing cells). At the end of this cycle, cells are usually treated with enzymes or cocktails (containing trypsin, collagenases or other compounds) that damage proteins protruding from the plasma membrane and which dissolve or fragment the ECM. Similar cocktails may be used to produce cell suspensions from tissue biopsies, or these biopsies may be pressed through a mesh to 'liberate' cells. All these treatments release proapoptotic factors, damage cells and have a significant impact on gene expression. Cells will attempt to repair this damage and recover, but need time to do so. So, the final factor to consider is time.

While there are numerous publications illustrating that 3D cell culture can mimic functionalities of human tissues, perhaps one of the most graphic is shown in **Figure 1** where a freshly extirpated human liver biopsy and 29 day old C3A spheroids have been biosynthetically labelled with [35S]-methionine and their proteins extracted and run on high resolution two-dimensional gels (IPG-SDS). Notice that, not only are the proteins expressed in very similar amounts, but also that their post-translationally modifications are very similar.

For many purposes in medical research, what is needed is a model system that accurately reflects what happens in the living organism - more often than not a human being – to shed light on many different processes, whether normal physiology or what goes wrong in disease, or infections by microorganisms, or the effects of compounds during treatment or poisoning.

3D cell culture promises to offer what is needed, but the field is still relatively new and many of the products used are small modifications of existing products that have been available for many years and as such, many have not been ideally suited to the purpose.

#### **2. A purpose-built system for culturing cells**

For these reasons, when we started to design some equipment specifically designed to support 3D cell culture. In doing so, we used four aims to guide the process. These were:

1.Use *in vivo* functionality as a yardstick

*A Purpose-Built System for Culturing Cells as* In Vivo *Mimetic 3D Structures DOI: http://dx.doi.org/10.5772/intechopen.96091*

#### **Figure 1.**

*Human biopsy tissue (ca. 0.5 mm3 ) and spheroids (29 day-old, ca. 1 mm3 ) were biosynthetically labelled for 20 hrs with [35S]-methionine. Some of the proteins are named for reference: ACTB, Actin beta; ALBU, Albumin; ALDH2, Aldehyde dehydrogenase 2; APE, Apolipoprotein E; CCND1, Cyclin D; HSPA8, Heat shock protein 8; HSPH1, Heat shock protein H1; HYOU1, Hypoxia up-regulated protein 1; PSMA5, Proteasome subunit alpha type-5; SAHH, S-adenosyl-L-homocysteine hydrolase; TUBB5, Tubulin beta chain 5; VCL, Vinculin; YWHAH, 14–3-3 protein eta. The tissue was homogenised, freeze dried, redissolved in lysis buffer and analysed by 2DGE according to [2]. Images were collected using AGFA phosphorimager plates and reader. [ 35S]-methionine labelled (A) human liver and (B) C3A spheroids.*


The main requirements that we addressed were: diffusion gradients, shear stress and time.

#### **2.1 Diffusion gradients**

Atmospheric oxygen (21%) provides a partial pressure (pO₂) of about 145 mm Hg (ca. 190 μM) to cells grown in 2D cultures. This is considerably more than the partial pressure of oxygen measured in tissues (11% to 0.1%) which should be considered as normoxic for cell culture [3]. Thus, cells grown in 2D cultures are exposed to unphysiological, hyperoxic conditions.

In the human body cells are usually located within 200 μm of a capillary [4] (corresponding to only 10 to 40 cell layers thick). Because cells are actively consuming oxygen, there will be a diffusion gradient into the cell. This is not problematic in 2D because the cultures are typically only one cell layer thick, but it becomes a challenge in 3D because there the spheroids can become tens or hundreds of cell layers thick. Despite that there may be a preferential transport of oxygen through cells and tissues by hydrophobic channelling within membranes, suggesting that oxygen diffusion within cells and tissues may be faster than through water, there will clearly be a limit [5].

In seminal work, Sutherlands group clearly demonstrated that the pO₂ in the centre of a spheroid fell to 0% when the radius was greater than 250 μm, corresponding quite well with the *in vivo* measurements [6]. The oxygen diffusion gradient started considerably outside of spheroids (ca. 100–200 μM) and continued to fall to the centre of the spheroid. They showed that this was an oxygen diffusiondepletion zone surrounding the spheroid and that in it the partial oxygen pressure fell by one third.

A flow of media past a spheroid significantly reduced this zone and had a very significant effect on the oxygenation of the spheroid. This allowed the spheroids to become larger before their cores reached anoxia. The beneficial effect of flow was almost completely negated if the spheroid was resting on a gas impermeable surface (e.g. glass or a gas-impermeable membrane) [6]. Spheroids appear to have a large capacity to adapt and significantly reduce their consumption when the supply of either O₂ or glucose or both is restricted [7–9].

Interestingly, hepatocytes, (which express haemoglobin *in vivo*), when grown as spheroids, massively overexpress (x30) haemoglobin, presumably to assist with oxygen transport into the oxygen depleted core [1]. Based on the changes in protein expression seen as cells recover their tissue mimetic metabolic equilibrium, it has been proposed that one of the strongest forces driving this recovery process which re-establishes the *in vivo* phenotype is hypoxia (affecting proteins like HIF 1 alpha (the main hypoxia sensor) and HYOU1 (a stress-responsive protein - see **Figure 1**)) [9]. The oxygen gradient formed has also been shown to have a marked effect on the formation of hepatic zonation during the differentiation of human embryonic stem cells and thus there will be a gradient of differences in metabolic enzymes into the spheroid [10].

Exactly the same arguments apply to CO₂: it has been demonstrated that CO₂ (as HCO3 − ) diffuses through spheroids of many cell types essentially as if the cells are not there [11]. Cells in clusters increase their aerobic respiration and decrease

#### *A Purpose-Built System for Culturing Cells as* In Vivo *Mimetic 3D Structures DOI: http://dx.doi.org/10.5772/intechopen.96091*

oxidative phosphorylation as they reprogram to a more anabolic based metabolism. This reduces their need for O₂ and their production of CO₂. Although this was first noted by Warburg in relation to cancer [12], it probably more strongly reflects the effects of 'mis-culture' of tumour cells in 2D rather than a metabolic style reflective of tumours *in vivo*.

Interestingly the rate of glucose diffusivity through spheroids of different cell types has been shown to differ by a factor of up to 4. The diffusion into a spheroid is quite rapid and an equilibrium is established after about 1 hour (single cells reach this equilibrium within a minute) [13]. These differences may be correlated to how tightly connected the cells become. C3A spheroids have been shown to rapidly deplete normoglycaemic media (5.5 mM) of glucose within 8 hours, converting much of it to glycogen. The cells were then able to reconvert the glycogen to glucose and survive for the next 40 (or 64 hrs) hrs until the next media exchange [9].

Diffusion gradients will also apply to nutrients and waste products. For example, NH3 is produced by transamination followed by deamination, from biogenic amines and purine and pyrimidine bases. NH3 (as NH4OH) is a smaller and less lipophilic molecule and thus its diffusivity five times slower than CO₂ through cells than through pure water at 37 °C making it more difficult to 'escape' [11].

The conclusions are clear for 3D cell culture: without a vasculature, cell clusters should not be too big in order to avoid anoxia (and ensuing necrosis) and should be irrigated on all sides to diminish the depletion zone and accelerate gas, substrate and metabolite exchange.

#### **2.2 Shear stress**

There is a growing appreciation that the mechanical properties and cell mechanics, play an important role in gene expression and cell development. The concept that is emerging is that cell types which experience shear stress *in vivo*, (usually fluid movement induced) actually need the stress to differentiate correctly (and retain their differentiation) and that shear stress is detrimental to cells that are not naturally exposed *in vivo* [14].

Thus, in some cases shear stress is positive: see-saw shaking of induced pluripotent stem cell (iPSC) constructs for 17 days promoted cell aggregation, and induced significantly higher expression of chondrogenic-related marker genes than observed in static cultures [15]. A platform rocking at 7 ° with a 3 second cycle results in an average shear stress of about 0.01 Pascal [16]. These shear forces are however distributed unevenly – both spatially and temporally during the motion of the container (bag, or flask) [17]. A similar, ultra-low shear stress is also seen in clinostat cultures. In this case though, the shear stress is distributed essentially homogeneously spatially and temporarily throughout the culture [9].

Fluid-induced shear stress (ca. 0.02–0.06 Pa) in microfluidic devices *in vitro* increased the mechanical properties of neocartilage [18] and have been shown to be beneficial for several epithelial or endothelial cell layers (as seen in ducts, blood vessels and the kidney) [19, 20].

Stirred tank suspension bioreactors and orbital shakers are used widely [21–23] but both result in significantly higher shear forces (0.3–0.66 Pa and 0.6–1.6 Pa respectively) and are considered to be in the critical/lethal range for mammalian cells [24].

Since most cells are found in tissues which experience very little shear stress, equipment that is designed to cultivate cells from these tissues should expose cell clusters to as little shear stress as possible. It is easier to reintroduce shear stress if needed, than to struggle to remove it from a system not designed to be shear stress free.

### **2.3 Time**

As mentioned above, enzymatic treatment of tissues or cells in 2D damages both the ECM and surface located proteins (including their modifications). This raises a number of questions.

The first is as to whether this damage can be repaired, i.e. can the 3D cultures recover the metabolic or physiological properties that they exhibited *in vivo* in the living organism. *In vivo* performance thus becomes an important benchmark – even though it may be very difficult to measure.

If the damage cannot be repaired, then the question raised is whether this failure is due to a limitation of the cells used, whether several cell types are needed, whether the procedures used prevent the recovery, or whether it is a true limitation that the performance cannot be replicated *in vitro*. Answering this question could therefore require a great deal of research. Fortunately, it appears that often the damage can be repaired.

If it can, the next question is how long do cells need to repair this damage?

A final question is that if the damage can be repaired, then once the cells have recovered, how stable is the 3D culture, i.e. for how long can the 3D culture be used?

The answer to these questions depends very much on the origin of the cells. For immortal cells, there are now numerous publications suggesting that these cells need to grow as spheroids for at least 2 weeks to recover their *in vivo* functionality [25–27]. Stem cells, including induced pluripotent stem cells appear to need a wide variety of times – some need only a few days while others never fully recover *in vivo* functionality [27, 28]. As a general rule primary cells seem to retain (most of) their functionality but use some of their mortal time to re-establish partial tissue organisation [29–31].

One observation that appears to be true for the majority of cells, irrespective of their origin, is that they grow considerably much slower as spheroids or organoids as they do in 2D cultures. Thus, doubling times in 3D may be as long as every 50 or 100 days rather than the 1 to 2 days seen in 2D [32, 33]. This makes sense: tissues and tumours *in vivo* do not double in volume every day or even every week. In other words, cells grown in 2D *in vitro* do not represent either their parental tissue or the tumour from which they were isolated. For example, the HepG2 cell line has a doubling time of 1.4 days [8] compared to 135 days for an average hepatocellular carcinoma [34] and approximately 327 days for liver hepatocytes [35].

Starting with isolated cells in culture, it is necessary therefore to anticipate that the cells need to re-adapt to being in clusters once again. There is a lot that needs to occur in such a readaptation process: for example, isolated cells do not have tight junctions and so their import and export pumps will have mixed: and need to be 're-deployed' to different regions of the plasma membrane once tight junction has been re-established [36]. While this is critically important for most cells, the specialisation of pumps in the liver cells is exquisite and intricate [37]. Redeployment of pumps might not be an active process – the cell after all was not designed to work in 2D cultures. Hence, the cell may have to rely on protein turnover to degrade the misplaced pumps and on membrane delivery processes to establish the pumps on the correct sides of the tight junctions. And there are innumerable other redeployments (changes to the cytoskeleton [1], organelle organisation, gene and protein expression, epigenetic marking [38], post-translational modifications and metabolic reprogramming [8, 9]) to complete.

The take home message here is that researchers have to be much more patient and expect that they will need to maintain 3D cultures for extended periods of time *A Purpose-Built System for Culturing Cells as* In Vivo *Mimetic 3D Structures DOI: http://dx.doi.org/10.5772/intechopen.96091*

(weeks or months). With this in mind, these extended periods will place a premium on cultures that are highly reproducible and preferably also high yield. Cultures that reach and maintain a dynamic equilibrium for weeks or months will be advantageous in that they will mimic the homeostasis seen in tissues and will provide a large window of utility. Within this window, it will be possible to for example collect multiple samples from the same culture (like collecting biopsies from the same animal at different times) or perform repeated treatment studies [39]. These samples could be used for the same assay at different times or multiple different assays at the same time, or both if there is sufficient biomass available. But the stability of the biological process needs to be documented throughout this window before perturbation experiments can be initiated [32, 40].

Maintaining cultures for extended times greatly increases the risk of infections, and thus requires great care. This can be facilitated if all aspects of a 3D culture system have to be considered so that the culture is well protected from external contamination.

### **3. A clinostat incubator**

So – what does a cell culture system that addresses all of these issues look like? First it is a CO₂ incubator. The incubator constructed can maintain a steady temperature both over time and within its volume. The inclusion of a powerful fan to mix the air within the incubator ensures no differences in temperature or gas partial pressures (and speed temperature recovery after the door has been opened). Measurements show that the difference in temperature between the top of the chamber and the bottom are less than 0.2 °C.

**Figure 2** illustrates that the incubator quickly reaches running conditions after about 15 minutes when it is first switched on and that it can maintain a steady temperature and CO₂ % (in the figure, over 12 hrs).

When the door is opened, the controlling software switches off the fan, heaters and the UV-C lamp (if active) and closes the CO₂ valve. Closing the door reactivates these functions. **Figure 3** illustrates the effect of opening the door wide open (90 °) for respectively either 30 seconds or two minutes showing that running conditions are re-established by about 4 or 6 minutes.

Note that the internal temperature in the culture vessel falls by a maximum of only 0.1 °C while the door is open illustrating that the cultures are not exposed to any cold shock which would affect their gene expression.

#### **Figure 2.**

*% CO₂ and temperature levels at (A) initial start up and (B) during a 12 hour period. Measurements were collected using the internal sensors.*

#### **Figure 3.**

*The effect on CO₂ % and temperature levels of opening the door to 90 ° for either 30 seconds or 2 minutes (as indicated by the red bars). C) Illustrates the effect on the temperature inside a culture vessel of changing the media in the 6 cultures (open door for 30 seconds to take out the culture vessel, 1 minute to change the media, 30 seconds to return the vessel, 1 minute to prepare the sterile bench for the next vessel – All repeated 6 times) Measurements were collected using the internal sensors and an external thermocouple inside a culture vessel.*

#### **4. Clinostat technology**

In order to reduce the diffusion depleted zone to a minimum, we have adopted the clinostat technology. Introduced for cell culture more than 20 years ago, this technology has been used to culture hundreds of different types of cells and tissues (cell lines, stem cells, primary cells) as well as bacteria and viruses and produced some excellent research. The major limitation of the initial equipment was that it was difficult to use (for example the culture chamber could not be opened but had to be accessed via luer lock ports).

Basically, a clinostat keeps cell clusters in suspension by rotating the culture vessel in a vertical direction (like a wheel). At the right speed, the uplift caused by the effect of liquid viscosity between the vessel and the clusters is balanced by the effect of gravity (these systems are often referred to as 'simulated microgravity' because the clusters appear to float or be maintained in a 'stationary orbit' (relative to the culture vessel)). Thus, the incubator that has been built has been equipped with clinostat motors. These have to run very smoothly so that there is no vibration (which would otherwise shake the clusters apart). **Figure 4** A) would reveal 50/60 Hz mains 'noise' effects while B) would reveal high frequency noise.

To retain 'continuity' with previously published data, we have maintained the basic geometry of the culture chamber.

*A Purpose-Built System for Culturing Cells as* In Vivo *Mimetic 3D Structures DOI: http://dx.doi.org/10.5772/intechopen.96091*

The next requirement is to be able to open the culture vessel easily for the purpose of introducing or removing cell clusters, for changing the media, or adding compounds or collecting samples.

**Figure 5** illustrates a culture vessel with a top access port for media exchange, a front access port for collecting individual cell clusters, and front window that can be removed to reveal a petri-dish like 10 mL culture chamber.

Changing the growth media is illustrated in the video (https://bit.ly/2PkiE9m). The culture chamber is removed from the incubator and the spheroids are allowed to sink to the bottom. The top port plug is removed and 90–95% of the media is sucked away using a sterile syringe and long needle. Fresh media is introduced in the same way, making sure to overfill the growth chamber so that any bubbles are driven into the 'drip-cup' around the top port. The plug is then replaced, the drip cup emptied,

#### **Figure 4.**

*Variations in the rotational speed of a clinostat motor. (A) Long term variations during 1 second: RPM (± STD) 5.03 ± 0.210; 10.03 ± 0.277; 30.21 ± 0.278; 100.38 ± 0.224. (B) Sort term variations during one thousandth of a second. RPM (± STD) 5.12 ± 0.002; 9.82 ± 0.004; 30.13 ± 0.008; 100.66 ± 0.007. The Permanent Magnet Synchronous Motor (PMSM) used for the clinostat was loaded with a dummy inertia block to mimic the weight of the full culture vessel and run with an Odrive V3.6 motor drive and controller software with anticogging feature. A TLE5012B encoder and digital oscilloscope (Picoscope 2200) was used to measure rpm, Measurements were made each microsecond.*

**Figure 5.**

*A culture vessel designed to provide easy access to the culture chamber. (A) Exploded view of parts. (B) Front view showing spheroids in the culture chamber (white spots).*

washed with 70% ethanol for sterility and the culture vessel replaced in the clinostat incubator. The whole procedure takes less than 40 seconds, resulting in minimal stress for the cell clusters (a video of this can be seen on CelVivo's website). Small samples of the media can be collected at any time using the same approach.

In this design (**Figure 5**), the gas exchange membrane has been moved from its 'cylinder end' position to a circumferential 'side' position whilst still retaining essentially the same area to allow rapid gas exchange between the culture chamber and the humidification chamber. Relocating the gas exchange membrane allows the culture chamber to be illuminated from the back and observed from the front (using the cameras), facilitating inspection of the clusters without disturbing the culture, or even opening the incubator door. This also helps to minimise infection risks.
