1 Introduction

1.1 Shortcomings of Contemporary Cell Culture

Cell culture is a system that seeks to multiply cells extracted from living tissues in order to study them or to build new cell communities in vitro. So, first and foremost, the extracted cells need to stay alive. Cells come from a tissue context comprising neighboring cells and their microenvironment, the extracellular matrix (ECM). Cells are removed abruptly out of their optimal and complex tissue context (Cigognini et al. 2013) and are seeded usually on onto polystyrene, glass, or other surfaces. Robbed of their niche, cells must settle on a foreign surface, where they are expected to adhere, spread, and then divide by mitosis. At this point, it might be important to note that animal cells adhere to the proteins attached to the plastic dishes rather than the plastic itself. Indeed, plastic surfaces are treated to achieve fast binding of adhesive proteins, usually from serum, that mediate the actual cell adherence (Ramsey et al. 1984). Typical mediator proteins from serum are fibronectin and vitronectin, as they engage integrin receptors that all animal cells express on their surface (Hayman et al. 1985). Fibronectin is both an ECM and a serum protein. Serum fibronectin is a circulating biosynthetic product of the liver of the serum donor, while cellular fibronectin is made by the adhering cells (the vast majority of anchorage-dependent cells, if not all, are capable of producing one form of fibronectin or another) (Stenman and Vaheri 1978; Owens and Cimino 1982). As cells adhere to the serum proteins, they will add their own synthesized fibronectin to this. For example, human cells grown in the presence of fetal bovine serum will form a hybrid matrix of bovine and human fibronectin. Thus, the cells coat foreign surfaces with their own adhesive matrix proteins to facilitate their own anchorage. Obviously, in vitro animal culture is a highly artificial system designed for optimal multiplication of cells, convenience, visual inspection, and optical studies of cell cultures. Getting cells to survive outside a tissue context is rather challenging, and decades of research and development have gone into the formulations of culture media that are commercially available to date. However, convenience issues seem to be getting in the way of redesigning cell culture despite scientific evidence that current cell culture systems are deficient on several levels. We will review these shortcomings only briefly (for more detailed reviews, please see Chen et al. 2011; Park et al. 2015), but it is important to be aware of them. Some of these shortcomings are not easily remedied, while others can be, as we will demonstrate in this chapter.

For a start, let us consider the cell attachment surface. Current surfaces are made of glass (still the best for optical imaging and short-term culture) (Mather and Roberts 1998), polystyrene (Curtis et al. 1983), and polycyclic olefins (Niles and Coassin 2008). These surfaces offer no intrinsic purchase for anchorage-dependent cells, other than perhaps strong electrostatic interactions with the cell membrane. Most of the 213 cell types of the human body are anchorage-dependent and would not thrive in suspension, unless they are transformed or cancer cells (Benecke et al. 1978). Therefore, surfaces are treated to make them negatively charged and hydrophilic so that polar regions of the proteins can stick to them and, in turn, mediate cell attachment via presentation of peptide sequences to cell receptors. Such well-described sequences are the RGD motifs that engage integrins (Ruoslahti 1996). Without the presence of those “sticky” proteins, cells will often not attach to plastic, as the experience of seeding cells in the absence of serum teaches us (Barnes and Sato 1980). Classical adhesion-mediating proteins are ECM proteins, which are secreted and immobilized outside of cells as supramolecular assemblies (Grinnell and Feld 1979). Cells have transmembrane receptors that bind to these ECM assemblies, and therefore, commercial ECM extracts or synthetic ECM-derived peptides contain integrin recognition sequences (e.g., nectins, laminins) (Aplin et al. 1998; Lutolf and Hubbell 2005). These peptides are used to coat cell culture plastic or come pre-coupled to plastic. In this regard, the current concept of “plastic adherence” for the isolation and characterization of mesenchymal stromal cells as a distinguishing criterion from hematopoietic progenitor cells from the bone marrow (Dominici et al. 2006), is quite intriguing. In reality, this is a selection of serum-protein-adherent cells or of cells particularly attracted to certain plastic surface charges. Emulating this on a more synthetic basis are peptides derived from the nectin family of proteins that are directly coupled to plasticware. One might wonder how different batches and qualities of serum or peptide coatings might decide over the very first selection of mesenchymal progenitor cells for subsequent expansion and later functionality and therapeutic application for tissue repair. Research in this direction might uncover interesting findings on the natural history of cultured mesenchymal stem cells.

This brings us to the missing microenvironment. Cells seeded onto culture vessels do not find an intricate ECM to settle in. No member of the family of collagens, laminins, nectins, microfibrillar proteins, and proteoglycans is present to welcome freshly seeded cells and help them anchor. Moreover, cell fate is highly dependent on the surrounding microenvironment. Indeed, a particular set of ECM proteins could constitute a specific niche that would, for example, maintain stem cells, while a different ECM composition could favor differentiation into other cell types. The ECM is even more complex than it seems, since various matrix proteins can also bind different growth factors that promote cell growth and/or differentiation (Chen et al. 2007; Lin and Bissell 1993). To compensate for this lack of ECM diversity, artificial matrices meant to emulate extracellular matrix are continuously developed and published in biomaterials journals with the intention to supersede nature (Lutolf and Hubbell 2005; Kyburz and Anseth 2015). However, most coating materials commercially available for plastic ware are mostly ECM extracts and polymers of animal origin, either cell lines or tissue extracts. These products do not come cheap and, in many cases, would face regulatory hurdles for cell therapeutic intentions. Conceptually, these products reflect a disregard of the fact that the microenvironments the cells were harvested from was originally made by these very cells. Obviously, freshly seeded cells are deprived of a fitting microenvironment to start with; however, we also need to understand that they are hobbled by the culture conditions in the first place.

Culture media as such constitute another level of challenge in animal cell culture models. Although the development of culture medium spans several decades, it has remained essentially the same system: cells are fully immersed in a seawater equivalent supplemented with carbon sources (e.g., glucose, pyruvate, glutamine, and other amino acids), vitamins, varying micronutrients, and a carbon dioxide-based buffer system (unless HEPES is being used) (Bettger and McKeehan 1986; Eagle 1955). So-called high-glucose media are on the market, because they allegedly promote cell proliferation. It might be worthwhile to consider that high-glucose medium in essence reflects a diabetic condition, and therefore uncontrolled nonenzymatic glycation of cell cultures might occur in the long run and not reflect physiology (Brownlee 2001; Ahmed 2005).

The focus of this book chapter is that, except for the urothelium of the urinary bladder, the inner lining of blood vessels, and perhaps the synovia of joints, cells in vivo are not exposed to large amounts of water, unless in a pathological situation like an edema. Despite all the additives mentioned above, most culture media cannot sustain viability of cultured cells, at least not at the desired proliferation rates. Therefore, an enigmatic component is added in the range of 0.5–20% v/v – fetal bovine serum (FBS) (Van der Valk et al. 2010; Arigony et al. 2013; Honn et al. 1975). FBS contains an array of hormones and stimulatory factors (Zheng et al. 2006), along with the abovementioned adhesive components: fibronectin and vitronectin. The global supply of FBS is limited; it comes from different countries with considerable batch to batch variations, leading to different proliferation and functionalities of cells cultured in those respective serum batches (Gstraunthaler et al. 2013). While endothelial cells might be continuously exposed to serum (plasma) proteins in some way, most other cell types might never see serum components in their entire life span, unless under pathological circumstances, such as leaky vasculature during inflammation or a wound. Current discussions to remove fetal bovine serum from culture media pivot therefore on reproducibility, concerns on transfer of xenogeneic pathogens and immunology, as well as issues of animal welfare related to the harvest of the biological product (Van der Valk et al. 2004). However, an often lackluster performance of cells in serum-free preparations in combination with high costs has prevented a deeper penetration of serum-free culture media at present in research laboratories.

Growing cells in the classical incubator is performed in the dark, at a humid atmosphere of 5% carbon dioxide, atmospheric oxygen (20%), and at 37 °C. Firstly, the inside of a dark incubator might not provide sufficient electromagnetic stimulation (light) for cell types harvested from the body surface (corneal epithelium, dermal and epidermal cells, such as keratinocytes and melanocytes). More importantly, oxygen levels in arterial blood range from 9.5% to 13% and drop, in actual tissues, to as low as 0.5% (Mohyeldin et al. 2010). Atmospheric CO2 is 0.04% but reaches 4.6% and 6% in tissues (Hoffman et al. 1998; Bolevich et al. 2016). It follows that cells are routinely grown in hyperoxic and thus stressful conditions, possibly leading to faster aging of cell cultures or cell lines that genetically drift due to the incorporation of mutations associated with reactive oxygen species (ROS). This hyperoxic environment may also affect cell fate since numerous stress-related pathways are involved in cell differentiation, notably by activating stress-activated protein kinases such P38 and JNK kinases (Lee and Choi 2003). Temperature wise, 37 °C might be ever only reached in the liver bed. Elsewhere, temperatures are typically lower by 2–3 °C (Werner and Buse 1988; Birnie and Grayson 1952). Although shear forces will be present in bioreactors with flow and cell cultures on microcarriers in suspensions that are continuously stirred (Ratcliffe and Niklason 2002), the absence of mechanical stimulation in conventional cell culture must also be mentioned here. Reintroducing macromolecular crowdedness into culture media, by ways of adding macromolecules, could not only remedy the wateriness of the culture medium but also solve the issue of insufficient microenvironment formation.

2 The Long and the Short of Macromolecular Crowding

The first application of macromolecular crowding in cell culture goes back 30 years. Firstly polyethylene glycol (4 kDa, 5% w/v), polyvinylpyrrolidone (30–40 kDa, 6.5% w/v), and dextran T-40 (40 kDa, 5% w/v) (Bateman et al. 1986) and later dextran sulfate (500 kDa, 0.01% w/v) (Bateman and Golub 1990) were shown to increase the conversion of procollagen to collagen. The authors showed an increased deposition of collagen in the ECM by fibroblasts in order to highlight the activity of bone morphogenetic protein 1, then known as procollagen C proteinase (Bateman et al. 1986). Thus, a method had been described to increase collagen deposition in vitro. It was also used to study collagen deposition by fibroblasts from a patient with osteogenesis imperfecta (Cetta et al. 1993), and in a cell-free system, dextran sulfate or polyethylene glycol was shown to speed up the activity of the same enzyme purified from chick embryo tendons (Hojima et al. 1994). In these publications, the terms “excluded volume” and “macromolecular crowding” were used, respectively, but the mechanism leading to an increased procollagen cleavage and collagen deposition remained unclear; a protein precipitation and aggregation step of some kind were discussed at that time. In the last 10 years, we and others have been able to gain more mechanistic insights into the effects of MMC in cell culture.

All considerations of macromolecular crowding start with the notion that the interior of a cell is crammed with macromolecules. In addition, eukaryotic cells contain a cytoskeletal meshwork that adds confinement to this macromolecular system. These considerations have been reflected in numerous review articles (Minton 2001; Ellis 2001). But when we take a closer look at the extracellular space in multicellular organisms, we find a comparable situation: the immediate vicinity of cells is also crammed with macromolecules, many of the immobilized, thereby adding confinement to the microenvironment. Therefore, any molecule secreted into the extracellular space will encounter a crowded environment in vivo.

Another important point to consider is that, in chemical reactions, reactants are modeled as point masses. This means that the actual sizes of the reaction partners are negligible and are therefore indeed neglected. In biochemical reactions, however, size does matter. In comparison with their respective substrate, enzymes can be huge, for example, glucose (180 Da) and glucose oxidase (160 kDa) (Woolridge et al. 1986), or are dwarfed by it like DNA polymerases working on a DNA strand, or exhibit similar masses such as bone morphogenetic protein 1 (88 kDa) (Lee et al. 1997) and procollagen (100 kDa) (Bornstein 1974). This means that the size of a macromolecular crowder might matter substantially with regard to the sizes of the molecules partaking in a biochemical reaction that are to be modulated by MMC.

3 Paradigm Shift: Make Your Own Matrix Scaffold In Vitro

The ECM of tissues is composed of fibrillar proteins and a gel-like ground substance made of dozens of glycoproteins and proteoglycans that are secreted by and immobilized around matrix-producing cells (Theocharis et al. 2016). It is important to distinguish between the biosynthesis of ECM molecules, their secretion, and their extracellular assembly into water-insoluble structures – the deposited ECM (Mouw et al. 2014). Clearly, secretion does not necessarily follow biosynthesis, and deposition does not necessarily follow secretion.

The biosynthesis of ECM molecules depends on signals that are received by the cell, such as growth factors or breakdown products of ECM. The secretion of a variety of collagens, however, is greatly aided by the presence of vitamin C (Murad et al. 1981). The extracellular assembly that follows successful secretion is a supramolecular aggregation of partners forming polymeric fibrils, or mesh-like structures. The basic polymeric structures are represented by fibronectin, collagens, laminins, and elastic microfibrils (Theocharis et al. 2016). Some collagens are forming fibrils, such as types I, II, and VII, while other collagens use these already formed fibrils as assembly points without forming fiber systems, such as type IX and XI collagens (Gelse et al. 2003). In analogy, elastin binds to fibrillin-containing microfibrils. Others collagens like collagens III and V mix to form heterotypic fibrils with collagen I and form the bulk of the connective tissue of the dermis (Birk et al. 1990; Fleischmajer et al. 1990). These biopolymeric assemblies represent molecular landing sites for a variety of ligands that in turn fasten growth factors to the ECM. Examples here are latent transforming growth factor beta-binding protein (LTBP) binding to fibronectin or elastic microfibrils and capturing latent TGFbeta and negatively charged proteoglycans associated with fibrillary structures that capture positively charged growth factors, like FGF and VEGF. Finally, enzymatic and nonenzymatic modifications of these supramolecular assemblies lead to the maturation of the built cell microenvironment . This includes later remodeling, involving breaking down structures while making way for new ones and changes in shape, structure, and composition of this ECM (Royce and Steinmann 2003).

Earlier work suggests that macromolecular crowding can aid in enzymatic reactions by stabilizing complexes of enzyme and substrate, thus increasing enzymatic activities (Zhou et al. 2008). Interestingly, additional effects are seen through the deposition, stabilization, and remodeling of ECM and are dependent on a variety of enzymes (Fig. 1). To select a few, the intracellular posttranslational modification of collagen that is crucial to its thermo-stabilization and therefore successful release into the extracellular space is catalyzed by prolyl hydroxylase (Notbohm et al. 1992). The extracellular proteolytic conversion of procollagen to collagen is controlled by bone morphogenetic protein 1 (also known as procollagen C proteinase) activity (Fig. 1). The extracellular covalent crosslinking of collagens (and elastin) is effected by the copper-dependent enzyme lysyl oxidase and the calcium-dependent transglutaminase 2 (Wang and Griffin 2012). Remodeling, removing, and reshuffling of the ECM are performed by matrix metalloproteinases and other proteolytic enzymes (Bonnans et al. 2014). The activation of these ECM-modifying enzymes sometimes requires their own cleavage by proteolytic enzymes. Therefore, multiple potential targets for MMC with regard to enzyme-mediated modifications in the extracellular space arise (for comprehensive overview, see Royce and Steinmann 2003).

Fig. 1
figure 1

Enzymatic key activities in the formation and maturation of ECM that are targets of MMC. Left: extracellular collagen deposition is boosted by accelerating bone morphogenetic protein 1 (BMP1), also known as procollagen C proteinase. The enzymatic activity shortens procollagen to collagen triple helices by removing globular C-termini. This trimming process now allows the assembly of collagen molecules to form insoluble fiber systems. This enzymatic process is rate-limiting for collagen deposition and slow in standard culture (blue columns). Therefore, the acceleration of this step via MMC (orange column) represents a substantial advantage in cellular microenvironment formation. (Reprinted from Lareu et al. 2007b, with permission from the publisher, Elsevier Center: Enzymes play a crucial role in the stabilization of deposited ECM). Shown here are high molecular weight double bands (red box) representing collagen alpha chains that have been covalently crosslinked by lysyl oxidase (LOX) activity. (Reprinted from Chen et al. 2011, with permission from Elsevier). Right: the remodeling of ECM involves the proteolytic removal of already deposited matrix components and often is accompanied by the deposition of new matrix. Therefore, not only the quantity but also the composition of the ECM will change as some ECM components will recede while others emerge. Here, matrix metalloproteinase 2 (MMP2) has been demonstrated via zymography to be strongly associated with matrix during adipogenesis under MMC. (Reprinted from Ang et al. 2014 with permission from Mary Ann Liebert Inc.)

Full size image

The first ECM-relevant observation with macromolecular crowding focused on the procollagen-converting enzyme bone morphogenetic protein 1/procollagen C proteinase. Macromolecular crowding induces faster conversion of procollagen to collagen, leading to stronger presence of cleaved collagen in the pericellular matrix and a correspondingly weaker presence of uncleaved procollagen in the culture medium (Bateman et al. 1986; Bateman and Golub 1990). In a cell-free system, the proteolytic activity of procollagen C proteinase on procollagen was found to be accelerated under MMC (Hojima et al. 1994). The addition of crowders to the culture medium was also shown to increase the conversion of procollagen to collagen in cell culture ; immunochemical analysis demonstrated a substantial increase of the clipped-off collagen I C-propeptide trimer under crowding (Chen et al. 2011). Moreover, the authors demonstrated that procollagen C proteinase enhancer (PCPE) , an allosteric regulator of procollagen C proteinase, was also proteolytically converted into an active enhancer of procollagen C proteinase activity and a smaller fragment that is speculated to have anti-collagenolytic properties, similar to tissue inhibitors of metalloproteinases (TIMP) (Mott et al. 2000). Interestingly, PCPE also activates the crosslinking enzyme lysyl oxidase (Trackman 2005). Our work demonstrated that MMC accelerated more than one converting enzyme in a cascade process. We also speculated that the binding of PCPE to procollagen C proteinase would be faster and tighter under crowding as well, which would have a tremendous effect, as PCPE is known to increase the catalytic activity of procollagen C proteinase by one order of magnitude (Lareu et al. 2007a) (Fig. 1).

Once procollagen is cleaved to collagen, the collagen molecules assemble as trimers in an orderly fashion, first forming nuclei and then extending and elongating into microfibrils that grow in length and girth. In parallel, crosslinking enzymes stabilize these meshworks and assist to immobilize them and protect them against proteolytic attack (Bornstein 1974). Lysyl oxidase crosslinks have especially been described and investigated extensively in various connective tissues ranging from tendon, skin, and bone to cornea and blood vessels (Kielty and Grant 2003). Transglutaminase 2 also plays a role in covalently crosslinking collagen assemblies (Wang and Griffin 2012; Raghunath et al. 1999; Raghunath et al. 1996; Zeugolis et al. 2010) as well as affixing ligands to ECM assemblies such as latent transforming growth factor binding proteins (Raghunath et al. 1998; Taipale et al. 1996) (Fig. 1).

However, ECM also requires remodeling, when tissue changes arise. Obviously, growth of a tissue and repair processes after injury require the addition and the rearrangement of the ECM. One classic example is the cascade of events that occurs after the fracture of a bone (Alford and Hankenson 2006; Damsky 1999). Firstly, the adjacent ends of the fractured bone will be partially resorbed via decalcification and removal of collagen fibers, and then a collagenous soft tissue occurs between the borders of the fracture, the callus, which then finally gets fully calcified. Damaged and denatured ECM will be proteolytically removed, and new ECM needs to be deposited and welded to pre-existing structures. In this case, the connective tissue is replaced with a similar connective tissue. Cell differentiation processes also require remodeling not only in terms of shape but also in terms of composition. The formation of adipose tissue, for example, requires a drastic change of progenitor cells from spindle shaped to spherical as demanded by the intracellular accumulation of lipid droplets (Nakajima et al. 1998; Gregoire et al. 1998). In consequence, the ECM, having to accommodate different cell shapes, will switch from a collagen I- and fibronectin-rich ECM to an ECM dominated by collagen IV, invoking the reduction of fibronectin. This sequence of events in adipogenesis involving proteolytic remodeling of the ECM was first proposed in culture in murine pre-adipocyte 3T3-L1 cells (Lilla et al. 2002; Selvarajan et al. 2001) and was confirmed later in human cells by differentiating human bone-derived mesenchymal stromal cells under MMC in vitro (Ang et al. 2014). Here, proteolytic enzymes that are specific to ECM components and assembly thereof come into play. We have shown that MMP-2 is more closely associated with the ECM under crowding conditions and that MMP activity can be accelerated under crowding conditions in a cell-free system (Ang et al. 2014) (Fig. 1).

4 Not to Be Overlooked: Collagens and Ascorbic Acid

We have shown that MMC greatly accelerates enzymatic processes that facilitate collagen matrix formation and deposition, which are normally tardy in standard cell culture. However, it cannot be stressed enough that any of these enzymes can only work with ECM components that are available in the pericellular space. This is where ascorbic acid (Vitamin C) comes in, as it is responsible for the efficient secretion of procollagen (Graham et al. 1995). Ascorbic acid is a crucial cofactor for the enzymes prolyl hydroxylase and lysyl hydroxylase, both major posttranslational modifiers of procollagen in the endoplasmic reticulum. While hydroxylation of the frequent lysyl residues in collagen plays a role later for collagen crosslinking, hydroxylation of the prolyl residues in procollagen immediately determines its thermal stability and thus its folding state (Yamauchi and Sricholpech 2012). The term procollagen refers to a triple helix composed of three procollagen alpha chains. In the endoplasmic reticulum, three such procollagen alpha chains form a procollagen triple helix. The assembly of the three alpha chains in a triple helix starts at the C-terminus, in a zipper-like fashion, toward the N-terminus. This process takes roughly a quarter of an hour for one triple helix (Raghunath et al. 1994). During folding, several enzymes posttranslationally modify the nascent triple helix. Because a folded triple helix is not easily accessible to these enzymes, they are “surfing” along the procollagen alpha chains ahead of the folding front to fulfill their tasks. All biochemical processes are in equilibrium. This means for collagen triple helices that they are continuously folding and unfolding (“breathing”). As only a completely folded procollagen can leave the endoplasmic reticulum, the equilibrium at body temperature must be shifted to the fully folded state. Here, prolyl hydroxylation comes into play by generating hydroxyprolines that help locking the triple helix in the folded state. In the absence of ascorbic acid, collagen biosynthesis can still occur but under-hydroxylated procollagen populations will be retained within the cell because they are not properly folded. It follows that ascorbic acid is crucial for an efficient collagen secretion (Barnes 1975). Vitamin C cannot be synthesized by humans, making it an essential vitamin that must be ingested. Nutritional Vitamin C deficiency, known as scurvy, leads to a paucity of extracellular collagen. This has dramatic effects on wound healing, blood vessel integrity, and periodontal health, all pointing to severe impairment of ECM maintenance (Peterkofsky 1991). It was estimated that British troops “even in World War I had more scurvy than gunshot casualties” (Kiple 2002) and that prior to the twentieth century, more soldiers and sailors deployed in warfare succumbed to scurvy than died in battle or accidents (McDowell 2013).

While scurvy might now be less frequent in modern societies, it is often involuntarily reproduced in ascorbic acid-free cell cultures where we find cells intracellularly retaining their procollagen, while the little that is secreted extracellularly hardly forms a pericellular matrix. The sensitivity to under-hydroxylation appears to vary between collagen and cell types and may depend on the percentage of hydroxylated prolyl residues per procollagen alpha chain and the specific thermostability pattern of the collagen type in question. For collagen I, a minimum of 120 hydroxylated prolyl residues is required to achieve thermostability (Royce and Steinmann 2003). While collagen I and VII may be the most sensitive to vitamin C deficiency in vitro, observations with collagen IV deposition suggest that it might be less vulnerable, but no systematic studies exist here. While MMC facilitates ECM deposition, it also depends on full collagen secretion to unlock its greatest benefits in building a “homemade” ECM microenvironment for cells in culture. Other ECM components are not dependent on the role of ascorbic acid such as fibronectin or elastic microfibrillar components. Therefore, an ECM built in the absence of vitamin C would preferably contain non-collagenous components and therefore present with a skewed stoichiometry. This, however, also would offer an opportunity of modulating ECM compositions using MMC.

5 Macromolecular Crowding in Macromolecular Assemblies

Collagen gels can be generated in vitro exploiting the intrinsic self-assembly process of collagen triple helices. Solutions of collagen containing pretrimmed procollagen can be made or bought. Collagen molecules are prevented from fibril formation by keeping them at very low pH, for example, in 1 M hydrochloric acid or 0.5 M acetic acid. When this solution is brought to neutral pH, the collagen assembly process begins rapidly, resulting in the formation of a gel made of collagen fibrils in a haphazard arrangement. This way of making collagen hydrogels is very popular, and collagen gels are used for a variety of assays and culture systems (Collin et al. 2011). The process of gel formation, which is essentially a process of fibrillogenesis, can be followed via turbidimetry. Light is shone through a cuvette containing the collagen solution; when fibrillogenesis occurs, the light is scattered because of the forming fibrils that render the solution turbid. Collagen gel formation is described in turbidimetry by two phases. Firstly, multiple foci form, consisting of a few collagen triple helices forming nuclei. This is called the lag phase, and while there is no apparent signal in turbidimetry, the formation of numerous nuclei is happening. Then, the solution starts getting turbid rapidly because of the nuclei that are now elongated into linear structures forming proper fibrils. This is the growth phase. The absorption curve shows a steep slope during this phase and a plateau when all collagen triple helices are spent in the process of fibrillogenesis (Dewavrin et al. 2015) (Fig. 2). When we first experimented with negatively charged macromolecular crowders (e.g., dextran sulfate), gel formation would occur so rapidly that turbidimetry was impossible to do (unpublished). We then studied the effects of single neutrally charged crowders to capture changes of fibril formation quantitatively (Dewavrin et al. 2015, 2014). MMC appeared to shorten the lag phase of collagen fibrillogenesis by creating a greater number of elongation nuclei in a shorter time window while the rate and extent of elongation (visible as the slope) remained unchanged. However, a plateau was reached faster, which suggested that crowding leads to the formation of more but thinner collagen fibrils (Fig. 2). These cell-free in vitro data on collagen fibrillogenesis were later confirmed in monolayer cell culture systems with the basement membrane collagen IV in the adipogenic differentiation of human bone marrow-derived stromal cells and progenitors from the stromal vascular fraction of human subcutaneous fat. Here, brown fat cells generated under macromolecular crowding were embedded in a collagen IV cocoon consisting of thin fibrils, whereas non-crowded cultures showed coarse and thick collagen bundles framing but not enveloping cells (Lee et al. 2016) (Fig. 3).

Fig. 2
figure 2

Effects of mono-crowding using 400 kDa on the kinetics of collagen I gel formation. The starting material is an acidic solution containing single collagen triple helices. Upon neutralization of this solution, the collagen fiber assembly starts and a collagen gel forms. Left panel: phases of collagen assembly as assessed by turbidimetry at 313 nm: No apparent signal is obtained before and during nucleation (lag time). The phase of elongation of nuclei (fiber growth) becomes visible with an increased absorption and a final A313 plateau, denoting the conclusion of collagen assembly. Right panel: the generation of an increased number of nuclei under MMC leads to the formation of more but finer collagen fibers. (Reprinted from Dewavrin et al. 2014, with permission from Elsevier)

Full size image
Fig. 3
figure 3

MMC enhances formation of Col IV basement membrane architecture during adipogenic differentiation. Human bone marrow-derived stem cells were differentiated into brown adipocytes under standard culture conditions (left panels) and under mixed macromolecular crowding using Ficoll (right panels). Z-stack images were obtained for each condition through confocal microscopy. The Z-project images show nuclei (blue), Col IV (red), and lipid droplets (green). The inserts and cross-sections are the reconstructed 3D images of the selected cells boxed in yellow to show the pericellular distribution of Col IV. Scale bar: 20 μm. Bottom images represent cross-sections of the region of interest. The cross-section reveals that only under MMC single adipocytes are fully wrapped in a collagen IV cocoon. This situation would be normally seen only in cells seeded into hydrogels. Thus, MMC creates an ultra-flat 3D setting with regard to ECM distribution in a monolayer. (Reprinted from Lee et al. 2016, under Creative Commons CC-BY-4.0 license. Available from: https://doi.org/10.1038/srep21173)

Full size image

The assembly of collagen has now been extensively studied under macromolecular crowding in cell-free systems and in monolayer cultures of collagen-producing cells. The primordial ECM molecule, however, is fibronectin. It is produced and laid down first by cells (and or grabbed from serum as discussed above) and serves as a template for collagen deposition (Kadler et al. 1996). While we have not systematically studied fibronectin ECM formation in vitro, we have shown that macromolecular crowding does promote fibronectin deposition in vitro in mesenchymal stem cells seeded in the absence of serum (Chen et al. 2011). However, more work needs to be done to study self-assembly of fibronectin under MMC.

We have shown that MMC promotes ECM formation at least on two levels: on the enzymatic level and on the supramolecular assembly level (see Figs. 1 and 2). Thus, the logic of applying macromolecular crowding to empower cells to build their own microenvironment appears quite compelling. While adding macromolecular crowders to culture medium is straightforward, the relevant biophysics, however, are highly complex and have been subject to in vitro and extensive in silico considerations. These considerations can be found in more detail in a number of pertinent publications, some of which are cited throughout this text. One of these considerations is that in any given volume, macromolecular crowders occupy space. If we assume that all molecules in such a given volume are spherical and that we know their number (molarity) and their size (hydrodynamic radius, RH), we can then calculate the volume they occupy and set this in relation to the given volume they are in. Thus, the fractional volume occupancy (FVO) with the mathematical symbol Ψ can be determined, allowing a first approximation of crowdedness in a given system. Biological systems show a Ψ in the range of 5–32% and are expressed as volume/volume (Fig. 4) (Chen et al. 2011).

Fig. 4
figure 4

A simplified representation of the generation of the excluded volume effects through the presence of macromolecular crowders in culture medium. Here, a test molecule (red) is about to enter a given volume (blue box). The given volume already contains three macromolecules (crowders, black). If we assume that these crowders occupy 30% of this volume, their fractional volume occupancy (FVO, Ψ) would be given as 30% (v/v). The FVO can be calculated when the volume is known, along with the number of crowders (molarity) and their volume. The volume can be calculated assuming the crowders to be spherical. In this case, the hydrodynamic radius will be used. The additional unavailable volume is a challenge to calculate, as several factors such as electrostatic repulsion and hydration shell need to be considered. Water molecules and Brownian motion are not considered in this model. (Reprinted from Chen et al. 2011, with permission from Elsevier)

Full size image

For all intents and purposes, we shall equate this FVO with the excluded volume, which means that the crowder-occupied space excludes all other molecules from that space. This, of course, only applies for truly globular molecules and is a generalization that will not characterize all macromolecules well. For example, a collagen triple helix is a rod-like trimer with a diameter of 1.5 nm and a length of 280 nm, as aspherical as a molecule can get (Shoulders and Raines 2009; Lodish et al. 2000). However, assumptions must be made to get to grips with any system, and this approach has a time-honored tradition in engineering. Basically, the power of macromolecular crowding resides in its ability to confine reactants to less space, thereby leading to a virtual increase in concentration of reactants. However, macromolecular crowding does much more than just making reaction volumes smaller or reducing the volume of culture media. We have seen a moderate increase of collagen deposition in greatly reduced culture medium (200 μl versus 500 μl) in 24-well plates (Chen et al. 2011). While we created a concentration effect in these cell cultures, we put them in danger of drying out or running out of nutrients fast. Moreover, just reducing the culture medium volume will not change the relationship between total volume and available volume. In contrast, with a crowded standard culture medium volume (500 μl), we saw much stronger deposition effects because the thermodynamic activity is much larger in a crowded culture medium, which has simply to do with the substantially reduced available volume versus a total given volume (Fig. 5).

Fig. 5
figure 5

Medium volume reduction effects are negligible compared with macromolecular crowding of the culture medium. (a) Fibroblast cultures were grown under non-crowded conditions in decreasing medium volumes. Their collagen matrix deposition was compared with cells in the highest volume under MMC with dextran sulfate (DxS). A 2.2-fold increase of collagen deposition was indeed seen in 200 μl but was clearly outperformed in DxS-crowded fibroblasts (17-fold increase) in 500 μl. In 100 μl and 50 μl, respectively, wells dried out, resulting in cell death. (b) The ratio between total volume and available volume will always stay 1 under non-crowded conditions, regardless of the actual volume in the cell culture as both volumina do not change in their relationship to each other. For 500 μl standard culture, the ratio is 1 (green text), as well as in 200 μl. However, it could be argued that in a smaller volume, cellular secretion of macromolecules might introduce some form of MMC, as well, so reducing the available volume slightly. However, thermodynamic activities in 500 μl of DxS-crowded medium increase dramatically, as the available volume is substantially reduced under MMC. (Copyright 2010 Peng Y, Raghunath M. Published in (Peng and Raghunath 2010) and reproduced with permission under CC-BY-NC-SA 3.0 license. Available from: https://doi.org/10.5772/8573)

Full size image

6 The Power of Charge and More: Mixed Macromolecular Crowding

Macromolecular crowding can be created using a single size species macromolecule, and most model systems work with one crowder; however, the combination of two or more crowders is more powerful than a system with only one crowder. A selection of commonly used crowders can be found in Table 1. We shall use the term “mono-crowding” for usage of only one size of one particular macromolecular crowder and “mixed macromolecular crowding” for usage of two or more crowders with different molecular weights. The molecular weight, Mw, of a given crowding molecule would cause steric hindrance by occupying space. At physiological pH, most proteins are negatively charged, depending on their amino acid composition and their respective cumulative isoelectric points. Consequently, a negatively charged crowder would have an even greater volume-excluding effect due to electrostatic repulsion of negatively charged molecules by negatively charged crowders (Harve et al. 2006). In fact, we have demonstrated that the hydrodynamic radius (RH) of negatively charged crowders is substantially larger in physiological salt solutions than a comparably sized neutrally charged crowder (Lareu et al. 2007b). The resulting larger volume-excluding effect that virtually increases reactant concentration could explain the extremely accelerated collagen deposition seen in presence of dextran sulfate 500 kDa (Lareu et al. 2007a, b). The size differences, as described by the hydrodynamic radius, also explained why we saw much weaker effects with mono-crowding with neutrally charged dextran or Ficoll molecules. Full collagen deposition under dextran sulfate was effectively enhanced in vitro within 48 h with fetal lung fibroblast lines, as more collagen was detected within the matrix, as compared to cells cultured for several weeks without macromolecular crowding (Lareu et al. 2007b; Chen et al. 2009).

Table 1 Commonly used macromolecular crowders for mono- and mixed crowding, cell types analyzed, and overview of relevant outcomes
Full size table

However, the deposited collagen (and fibronectin) matrix appeared granular; ultrastructural imaging revealed these granules to contain thin cross-striated collagen fibrils, but a contiguous collagen meshwork was not apparent under negative mono-crowding, (Lareu et al. 2007b). We also came to realize that not every cell line thrives under these crowding conditions. We suspected that dextran sulfate, and likewise other negatively charged crowders, might interfere with a variety of growth factors that have a positive net charge. We initially saw weaker results using single neutrally charged crowders (Lareu et al. 2007a) but were inspired by modeling work predicting that the combination of two differently sized crowders would generate a stronger volume occupancy/exclusion effect (Zhou 2008). We therefore combined the sucrose copolymers Ficoll 70 kDa with Ficoll 400 kDa and after a few titration rounds determined that mixed macromolecular crowding had indeed a strong, albeit slower than dextran sulfate, effect on collagen deposition (Lareu et al. 2007a; Chen et al. 2009). The current Ficoll mixture we have been using ever since has a theoretical FVO of 18% v/v, which we determined to be equivalent to the albumin fraction in serum (Chen et al. 2011b).

Using cell-free collagen assembly as a readout system, we could experimentally confirm previous predictions (Zhou 2008) that the combination of a minimum of two differently sized species of crowders indeed creates more excluded volume than theoretically expected and discussed the underlying theory (Fig. 6) (Dewavrin et al. 2015). We have used mixed macromolecular crowding successfully also in settings of enzymatic reactions, such as PCR (Lareu et al. 2007a) and in DNA hybridization experiments (Harve et al. 2010). The advantage of mixed macromolecular crowding using Ficoll is that two synthetic polymers of defined size give a controlled biphasic size distribution pattern. However, theoretical and experimental data would suggest that a third or fourth crowder in descending size might even close more spaces. Indeed we used triple crowding in PCR and hybridization experiments (Harve et al. 2010). Therefore, reproducible polydispersity would be an interesting approach, as it would represent a bandwidth of different sizes. This approach has been put into practice using carrageenan, a polydisperse macromolecule that has shown powerful ECM deposition in vitro (Satyam et al. 2014; Kumar et al. 2015a, b; Cigognini et al. 2016; Satyam et al. 2016; Kumar et al. 2018). Therefore, mixed macromolecular crowding should be attempted in vitro wherever possible (Fig. 6).

Fig. 6
figure 6

Modelling of mixed macromolecular crowding . Left: Comparison to the volume that is accessible to a probe in a volume that is occupied by one species of crowders (single) or two different species (dual). In the mixed condition, the small crowder does not occupy a significant volume but reduces the average distance between larger crowders to a value lower than the probe diameter: a void volume is created. The total excluded volume is therefore greater than the sum of the crowders’ volumes, as the small crowder is granted with extra volume occupancy (the void volume) due to its proximity to the big crowders. Right: molecular modeling reveals that the total void volume generated by mixed MMC is larger than the sum of void volumes generated by either crowding component. Thus, mixing crowders is a powerful approach to tune an aqueous system. (Reprinted with permission from Dewavrin et al. 2015. Copyright (2015) American Chemical Society)

Full size image

7 Cell Sheet Engineering and Macromolecular Crowding: A New Match

Cell sheet engineering is based on growing cell layers on a thermosensitive polymer at body temperature and then to lift the cells off, along with their deposited ECM, as one coherent layer when the culture vessel is placed in a cooler environment (Yang et al. 2005; Yamato and Okano 2004). This feat is achieved by coating tissue culture polystyrene with poly (N-isopropylacrylamide) (variously abbreviated PNIPAAm). At a certain temperature, such as 37 °C, PNIPAAm strands are coiled up, but at lower temperatures, like 20 °C, the coils unwind. Cell seeded onto PNIPAAm would find a smooth adhesive surface at 37 °C and then would find themselves pushed away from underneath as the polymer uncoils when the culture medium cools down outside of the incubator (Yamato et al. 2007). The chemistry behind this material is based on its glass transition temperature (Rollason et al. 1993; von Recum et al. 1998). In conditions of low ECM deposition, these thermosensitive polymers work well. However, under macromolecular crowding conditions, where abundant ECM is deposited, copolymers of PNIPAAm are necessary to detach intact and ECM-rich tissue equivalents (Satyam et al. 2014).

8 Impact of Macromolecular Crowding on Three-Dimensional Extracellular Matrix Networks and Associated Cell Signaling Events

2D and 3D in vitro systems are extensively used in tissue engineering and regenerative medicine (Shologu et al. 2016; Li and Kilian 2015). Macromolecular crowding contributes to a blurring of the boundaries between what is called “2D” and “3D” cell culture . Traditionally, the term 3D tissue engineering is used to describe thick, tissue-like constructs, often cells seeded onto a scaffold. We posit that there is a clear Z-dimension (height) in rich in ECM, scaffold-free, crowded cultures, which better imitates the in vivo-like cell microenvironment without the shortcomings that typical 3D systems present in terms of microscopic analysis and viability assessment.

Confocal imaging analysis suggested that adipocytes generated from mesenchymal stem cells under MMC were covered and embedded in ECM, resembling a layered system with cells embedded between layers of ECM (Lee et al. 2016; Bolevich et al. 2016) (Fig. 7). With more collagen available around them, due to macromolecular crowding, the cells can generate more interfaces with the ECM and especially more attachment points, even on the top of the cells. This was graphically demonstrated by the presence of focal adhesions signifying an increased cell-matrix engagement under crowding conditions (Lee et al. 2016). With more focal adhesions engaging a richer ECM, the integrin-mediated signaling is likely to increase and even change, as cells are integrative systems which constantly process and combine various extracellular signals to adapt their behavior in response to external stimuli. This three-dimensional architecture of the ECM technology has also enabled investigations into the intrinsic brown potential of adipogenic differentiation of bone marrow-derived stem cells and progenitors from the stromal vascular fraction of subcutaneous fat (Lee et al. 2016). A similar 3D ECM network surrounding adipocytes has been observed in vivo by electron microscopy of native adipose tissue (Flynn 2010; Sbarbati et al. 1987). In vitro, this situation is usually found only in cells embedded in artificially created 3D matrices (Polacheck et al. 2014). This means that a monolayer culture system can be turned around into an ultra-flat 3D system via macromolecular crowding, a consideration that is important for the design of cell-based assays for the development of drug screening tests.

Fig. 7
figure 7

3D rendering of an ECM of collagen IV produced by human primary fibroblasts upon macromolecular crowding (MMC). (1) Human primary fibroblasts (Fb) were seeded on glass coverslips and incubated in the presence of crowders (MMC) resulting in an enhanced deposition of collagen IV (Col IV). (2) The cells were removed with sodium deoxycholate. The samples were then washed and fixed with PFA. (3) Collagen IV was stained by immunocytochemistry (ICC). (4) The samples were then imaged in 2D using a red LASER. (5) The samples were then imaged plane by plane (Z-stack) using an inverted confocal microscope (Zeiss LSM510) with an interval of 0.48 μm between two successive optical planes. (6) All images were converted into gray scale preserving pixel intensity but removing the original colors. (7) Each image was given an artificial color according to its position in the Z (vertical) axis (Depth Color Coding, DCC). Images acquired at the bottom of the sample (near the glass coverslip) were colored in blue. Images acquired on the top of the sample were artificially colored in red. (8) The stack of color-coded pictures was then imported into ImageJ software (NIH) for 3D rendering. (For more information on the Material and Methods, refer to Benny et al. 2015, PMID: 25058150)

Full size image

As we have seen, MMC can induce a variety of cells to deposit much more ECM than under standard conditions and with higher complexity than artificial/engineered matrices. This offers the opportunity to develop these rich in ECM cell constructs as an optimal niche for in vitro expansion of various cell types, following decellularization. For example, human embryonic stem cells benefit from a human fibroblast matrix deposited under crowding conditions and can be passaged effectively on such cell-produced matrix (Peng et al. 2012). The most recent example of the successful application of MMC was the generation of (decellularized) bone marrow matrix for the cultivation of hematopoietic stem cells, applying mixed MMC with novel crowder combinations (Prewitz et al. 2015). The application of crowding to produce the full depositional cascade of collagen in complete living fibroblast monolayers has led to the construction of a high content screening tool, the scar in a jar, for antifibrotic compound screening (Chen et al. 2009; Chen and Raghunath 2009).

9 Conclusion

Cells in tissue culture media basically represent seawater with some additional ingredients including a fraction of fetal calf serum. This makes for a very dilute environment when compared with the crowded intra- and extracellular milieus. This ocean of dilution leads to a number of biochemical and biophysical processes to occur with less speed and efficiency than they would in vivo. It is often claimed that in vitro cell culture is artificial to start with and therefore is of limited physiological relevance. While this argument is often used to make a case for animal experimentation, it should not be an excuse for not improving cell culture conditions. We believe that much can be done to move in vitro cell culture systems closer to human physiology by correcting some deficiencies, such as too much oxygen, lack of appropriate surface signals, the absence of biomechanical stimulation and macromolecular crowding, and a deficient microenvironment . It is worthwhile considering that the need for 3D tissue-engineered systems, living organ equivalents, is increasing in industry. There is societal pressure to abandon animal experimentation which is increasingly translating into political pressure and, therefore, legislation to enforce the 3R of animal welfare. Within this framework, tissue-engineered tissue equivalents will see a continued demand as long as they are robust and can be validated. In this context, macromolecular crowding can be a very valuable tool for not only answering fundamental questions but also increasing the functionality of cells in vitro by facilitating faster and deeper differentiation of progenitor cells or greater phenotypic stabilization, all via the generation of a complex human ECM.