Dynamics of Macro- and Supramolecular Structures in the Bacterial Cytoplasm

1. Introduction

The cytoplasm of bacteria lacks organelles, yet functions as a multi-part machine that responds to external stimuli. Densely packed macromolecular structures (e.g., granules, the nucleoid, proteins, enzymatic megacomplexes, and ribosomes), atoms, ions, and microcompartments create a highly constrained, microscopic meshwork that behaves like a glass-forming liquid in cells metabolically inactive or dormant [1,2,3,4]. In Escherichia coli, approximately 70% of the cytoplasmic space (1 μm³) is occupied by the 4.6 × 10⁶ bps nucleoid [5]. The average mesh size of an E. coli nucleoid is 50nm [5], which implies that macromolecular assemblies exceeding 50nm are excluded from the nucleoid. However, since the density of the nucleoid varies, smaller macromolecules may penetrate at random positions. The more compact the nucleoid, the smaller its mesh size (due to “shrinking”), and the more difficult it would be for proteins and other macromolecules to navigate a highly compressed DNA mesh [6]. The tight spacing slows down the diffusion of macromolecules, which may delay critical cellular processes such as transcription, replication, and DNA repair. Ribosomal subunits with Stokes radii of 15 to 20 nm and enzymes with a radius of 5 nm diffuse freely into the nucleoid [7,8,9,10]. This implies that 30S and 50S subunits diffuse freely, whilst fully assembled 70S ribosomes and polyribosomes are excluded from the nucleoid-occupied region [11]. The cytoplasm is a poor solvent, which supports genome compaction [5]. It is thus not surprising that for the largest part of the cell’s life cycle, the nucleoid is localized at the poles of the cell and the periphery of the cytoplasm [5].

The behavior of macromolecular structures in the cytoplasm is explained by using the ‘spacers-and-stickers’ model, in which multivalent interactions between binding motifs (‘stickers’) are separated by flexible regions (‘spacers’) [12]. In the ‘associative polymer with percolation and phase separation’ model [13], the spatial orientation of the nucleoid is explained according to the formation of polymer networks (percolation) and the separation of phases. The movement of macromolecules such as ribonucleoproteins (BRs), RNA polymerases, and heat-unstable (HU) nucleoid-associated protein (NAP) (HUNP)–DNA condensates in the bacterial cytoplasm is best described by liquid-liquid phase separation (LLPS) [2,14,15,16].

In a dynamic (metabolically active) state, the cytoplasm becomes less glassy and more fluidic [2]. Large and small particles in the cytoplasm interact directly or group in vacuole-like structures called “cages” [1]. Larger particles that migrate in cages include protein condensates, aggregates, and agglomerates; ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO); DNA and DNA-binding proteins from starved cells; polyphosphate biopolymers; ribonucleoprotein bodies; and glycogen [17]. Toward the end of the cell’s life, HUNP relaxes the highly compacted nucleoid, particles and cages rearrange, and open areas (“gaps”) form in the cytoplasm [1]. At this stage, the nucleoid divides, the mesosome forms, and cell division starts [2,5].

Protein-bound cages contribute to cytoplasmic viscosity [18]. In a study using HeLa, VMKO HeLa, and MDCK cells, Ebata et al. [19] concluded that the cytoplasm of metabolically active cells exhibited a single-exponent power-law rheology, expressed as G = G₁(−iω)^0.5 (G = complex shear modulus, combining elastic storage and viscous loss moduli; G₁ = material-specific scale factor or amplitude coefficient; −i = imaginary unit rotation representing phase lag; ω = angular frequency of the applied oscillation; 0.5 = the power-law exponent, representing critical jamming behavior). This is also observed in sheared, concentrated, starlike micelles [20] and dense, motile bacterial suspensions [21].

Several studies have suggested that ribosomes [22,23], pH [24], and ATP [1] controls cytoplasmic rheology. However, more recently, Losa et al. [25] demonstrated that cytoplasmic rheology is also controlled by protein agglomerates. When highly expressed, protein agglomerates obstruct the diffusion of nanoparticles. Membrane-bound organelles are rare and are only found in planctomyces (anammoxosomes), cyanobacteria (gas vescicles), and magnetotactic bacteria (magnetosomes) [26,27,28]. The rheology of the cytoplasm may also be altered by proteinaceous membrane-less microcompartments, also known as metabolosomes. Examples include carboxysomes encapsulating CO₂, and Pdu, Etu, and Eut organelles containing enzymes that metabolize 1,2-propanediol, ethanol, and ethanolamine, respectively [29]. Crystals of magnetite (Fe₃O₄) and greigite (Fe₃S₄) found in magnetotactic bacteria (Magnetospirillum, Magnetovibrio and Magnetococcus) are enclosed in nano-magnetosomes [28].

This review discusses how particle dynamics and spatial compartmentalization influence the formation of microscopic forces in the cytoplasm. We identify the drivers of cytoplasmic fluidity responsible for the generation of an out-of-equilibrium force that changes the cytoplasm from a glass-like to a semi-liquid phase. The review does not cover interactions between proteins and the cell membrane. The behavior of proteins near cell structures is influenced by electrostatic and/or hydrophobic interactions, as well as repulsive volume-exclusion interactions, and warrants a separate discussion. We acknowledge that interfacial forces such as hydrophobic, electrostatic, and hydrogen interactions may act as physical catalysts by lowering the energy barrier required for molecules to self-assemble. This may cause conformational shifts in proteins and may expose intrinsically disordered regions (IDRs) that act as drivers of condensate formation. These interfacial forces are mostly associated with membranes and nucleic acid scaffolds, which are not discussed in this review.

2. Physicochemical Properties of the Cytoplasm

Densely packed macromolecules (referred to as crowding) lead to the formation of large open areas/gaps (Figure 1) [1,2,3,4] and give the cytoplasm a glass-like appearance [1,2]. Crowding increases the effective concentration of molecules and may enhance biochemical reactions as molecules associate (Figure 1). At the same time, the dissociation constant (K_D) decreases (Figure 1) [30,31]. Because K_D is inversely proportional to the association rate constant (k_a in equation 1), a higher k_a (faster turning “on” of reactions) or a lower dissociation rate k_d (slower turning “off” of reactions), or both, decreases molecular dissociation and increases binding affinity (Figure 1). A decrease in K_D does not inherently (automatically) increase the reaction rate constant (k_react in equation 2). In a very stable enzyme-substrate (ES) complex (extremely low K_D), the initial free energy in the reaction is significantly lower, and the reacting molecules are in a deep thermodynamic pit, often referred to as a “substrate sticky trap” (visualized as a thermodynamic valley). Extremely stable enzyme-substrate complexes raise the overall activation energy barrier, making it harder to reach the transition state. In other words, in a deep thermodynamic pit (overly stable state), molecules collide less frequently and the energy difference between the ground state and transition state increases. Since fewer molecules have sufficient kinetic energy to reach the transition state, k_react is reduced (the overall reaction rate decreases) (Figure 1). For reactions driven by a high k_a, e.g., covalent inhibitors or proximity-induced reactions, a low K_D does not directly increase the chemical reactivity rate constant (k_react), but it increases the overall rate of product formation by driving the system toward the reactive complex.

$$K_D={\displaystyle \frac{kd}{ka}}$$

(1)

$$k_{\mathrm{a}}\mathrm{\alpha }{\displaystyle \frac{K_D}{k_{\mathit{react}}}}$$

(2)

With an increase in k_a, the sum of the inverse values of the rate constant under diffusion control (k_D) and k_react decreases (equation 3). This means the combined resistance (1/k_D + 1/k_react) decreases.

$$k{\mathrm{a}}_{=}{\displaystyle \frac{k_{\mathrm{D}}{k}_{\mathrm{react}}}{k_{\mathrm{D}}+k_{\mathrm{react}}}}$$

(3)

In a reduced volume (crowded space) with high viscosity, fewer separate undissociated reactions form, and the activation energy barrier is much lower. Thus, molecular motion is slowed in a higher-viscosity medium, and fast reactions that depend on random interactions are hindered [32,33,34]. The reaction between ribonuclease (RNase) and RNA is an example of ES complexes regulated by a low K_D. (approximately 10^-15 M, as opposed to standard biological ES interactions at 10^-9 to 10^-6 M). At this low K_D, the ribonuclease inhibitor (RI) protein wraps around RNase and serves as a biological “kill-switch”, protecting RNA from destruction.

Macromolecular crowding affects gene expression by altering the diffusion constants of transcription factors (TFs), as illustrated in Figure 2 [35]. Nucleoid-associated proteins (NAPs) bind to the genome and alter its three-dimensional structure, forming a compact nucleoid. Transcription factors (TFs) bind either specifically or non-specifically (randomly) to the nucleoid. The strong association with DNA causes the TFs to rotate more slowly along the grooves of the DNA molecule, a process referred to as facilitated diffusion/sliding (Figure 2). The physical distortion of DNA by NAPs disrupts the continuous helical path required for the TF to glide smoothly from one base pair to the next, severely shortening its one-dimensional sliding length (Figure 2) and causing premature dissociation. The rotational motion is associated with friction, proportional to the volume of the protein, and slows down the movement of TFs [36]. One-dimensional diffusion coefficients for TFs are substantially (two orders of magnitude) smaller than three-dimensional diffusion constants, suggesting that sliding down the nucleoid occurs much more slowly than diffusion through the cytoplasm. Norred et al. [37] have shown that macromolecular crowding influences the rate at which genes are expressed and described how mRNA and proteins are produced in bursts, a phenomenon associated with cytoplasmic noise (increased protein levels). This occurs at modest changes in mRNA levels, attributed to the non-homogeneous spatial distribution of mRNA in a crowded cytoplasm [37]. Thus, mRNA spatial noise leads to large temporal fluctuations in protein levels. Our understanding of the interplay between macromolecular crowding, gene expression, and spatial distribution (diffusion) of mRNA is limited and requires in-depth research. A better understanding of these processes will facilitate research in cell-based and synthetic biology.

The cytoplasm of metabolically active cells is in a homeostatic state, i.e., the physical and chemical conditions, such as macromolecular crowding, pH, ionic strength, and turgor pressure, are carefully controlled to maintain balance (reviewed by Poolman [38]). Protons, which participate in biochemical reactions as reactants and/or regulators of enzyme activity, serve as a source of electrochemical energy (proton motive force, PMF) and may influence LLPS (Figure 1). The PMF is influenced by the membrane potential (ΔΨ), which is typically negative in the cytoplasm, and the difference in cellular and external pH (ΔpH) (equation 4).

PMF = ΔΨ-ZΔpH

(4)

The symbol “Z” denotes the relation between the gas constant (R), absolute temperature (T), and the Faraday constant (F), formulated as 2.3RT/F [38]. Thus, to maintain a constant PMF, cells need to maintain electron charge (ion levels), temperature, and regulate thermodynamic and kinetic forces. PMF is generated by electron transport, light-driven proton translocation, ATP-driven proton pumps, or coupling of electrogenic transport to a metabolic reaction [39]. Each of these mechanisms increases the internal pH, which is kept in balance by cation/H⁺ antiporters, anion/H⁺ antiporters, and metabolite decarboxylation [38]. Protons are pumped out by respiration or other mechanisms and pumped back into the cell by PMF-consuming processes such as ATP synthesis or nutrient uptake [38]. Ionic strength influences the binding of proteins to nucleic acids, protein aggregation, the activity of enzymes, the structure of intrinsically disordered proteins (IDPs), ion channels, transporters, and phase separations [40,41,42]. Changes in the physical interactions between macromolecules in a crowded cytoplasm, e.g., due to increased ionic strength or lower pH, may affect the translational diffusion of proteins [43] and compromise proteome stability [44]. Proteins or protein domains that unfold as a result of, e.g., heat stress may represent IDPs and increase the viscosity of that specific area in the cytoplasm, causing the proteins to become entangled [45]. Below a critical concentration, proteins group and form a single-phase state (Figure 3). Changes in temperature, ionic strength, and post-translational modifications (PTMs) lower the critical threshold for phase separation. This allows proteins to disperse and associate at much lower concentrations (Figure 3). Proteins in separate liquid-liquid phases will attract each other, nucleic acids and other macromolecules in a nonspecific manner, and form an associative LLPS. Nonspecific interactions between molecules in the two phases may lead to the formation of segregated LLPS [46]. In this case, protein associations are enhanced via volume exclusion, resulting in one phase enriched in protein and the other phase depleted of protein, relative to the total composition. Macromolecules may also be dispersed via substrate channeling, mediated by metabolons (transient, noncovalently assembled enzyme clusters) that enable direct metabolite transfer between active sites. Examples are intramolecular tunnels (e.g., tryptophan synthase), electrostatic channels (e.g., between malate dehydrogenase and citrate synthase), chemical swing arms (e.g., pyruvate dehydrogenase), cluster channeling (e.g., purinosomes), and glycolytic bodies (reviewed by Fukuhara et al. [47]).

The behavior of proteins in a crowded cytoplasm was originally thought to be dictated by diffusion according to the StokesEinstein (SE) relation (τ∝^η/_T), where τ is the diffusion time, η is the viscosity of the medium, and T is the experimental temperature [48,49]. Recent findings on cytoplasm dynamics proved the SE relation incorrect. The translational diffusion coefficient (D) of proteins (study performed on Escherichia coli) is determined by the mass of a protein (polypeptide), multiplied by the oligomeric state, and not abundance [8,50]. Furthermore, D = αMβ, where M is the complex mass and α and β are fitting parameters. In the ES equation, β would be -0.33 for globular proteins that do not interact with each other. Instead, β was determined to be -0.6 for the diffusion of proteins in the cytoplasm of E. coli. [8,51]. This implies that the dependence on molecular mass is much stronger than originally anticipated in the SE equation (due to macromolecular crowding). This also indicates that the cytoplasm is a dilatant, non-Newtonian fluid that increases in viscosity during stress. A modified version of the SE equation was proposed (equation 5):

$$D_{=}{\displaystyle \frac{k_{\mathrm{B}}T}{{6\pi \eta }_{\mathit{MW}}r}}$$

(5)

where η_MW represents the perceived viscosity as a function of the molecular weight. The perceived macromolecular viscosity varies from 9.9 cP to 18.1 cP (centipoise) for proteins of 26 kDa to 318.9 kDa [52]. Deviation from the SE equation was confirmed by using fluorescence recovery after photobleaching (FRAP) and nuclear magnetic resonance (NMR) [53]. If diffusion is related to viscosity, as shown in the fractional Debye-Stokes-Einstein (FDE) equation τ∝ (^η/_T)^α (τ = time for a molecule to structurally rearrange or rotate, η = the fluid’s internal resistance to flow, T = absolute temperature, and α = the fractional exponent/scaling parameter of typically 1.0), an α value of less than 1.0 represents inconsistent diffusion [54]. Proteins in complex fluids near the glass transition have α-values of 0.7 to 1.0, depending on the degree of decoupling [55]. Junker et al. [56] have also shown a consistent negative deviation from the SE relation as the molecular weight increases. The diffusion of proteins in the mass range of 26-319 kDa correlates with the average 50-nm mesh size of the E. coli genome [5]. This means proteins with a Stokes radius up to 5 nm may not be affected by the genome’s position in the cytoplasm. Proteins at the poles of E. coli diffused more slowly than those in the rest of the cytoplasm [8,57]. Computer simulations demonstrated that the decrease in protein diffusion exceeds the constraining effects of the cell membrane. The accumulation of proteins most likely begins with genetic mutations, environmental stress (e.g., heat or oxidation), or aging. Misfolded proteins are cleared by chaperones that refold proteins and the proteasome. In bacteria, damaged proteins accumulate at cell poles [57]. As expected, the rate at which cytoplasmic proteins diffuse varies among species. An increase in macromolecular crowders may alter protein conformations from a relaxed, unfolded state to compact, quasi-spherical forms [58]. Proteins with no stable three-dimensional form under physiological conditions (referred to as intrinsically disordered proteins, IDPs) may become compacted in a crowded cytoplasm. Examples are carboxyamidated RNase T1 in the presence of dextran [59] and flavodoxin, apoflavodoxin, and Borrelia burgdorferi VisE in the presence of ficoll-70 k (ficoll PM 70), a neutral, highly branched, hydrophilic synthetic polymer of approximately 70 kDa [60].

Macromolecular crowding helps maintain the three-dimensional structure of DNA and proteins, regulates the kinetic and thermodynamic behavior of enzymes [61,62], facilitates the binding of molecules to ligands, the formation of insoluble aggregates, and the compaction/folding of proteins [4,63,64]. The nucleoid is kept intact by the ParA/MinD ATPase system, as shown in Figure 1 (reviewed by Dicks [2]). ParA orchestrates the segregation of genomic DNA, and MinD positions the divisome (the protein complex responsible for bacterial cell division) at the center of the cell [65]. Protein ParB binds to parS, a centromere-like site on the genome, and forms a large partition complex. ParB stimulates the ATPase of ParA and the ParA-ADP complexes. The ParA-ADP complex dissociates from the genome and is replaced by the next available ParA-ATP complex. This continuous cycle of ParA-ADP and ParA-ATP “shifts” the genome toward areas of higher ParA-ATP density located at opposite poles of the cell [66]. Genome shifting forms a “wave” [67]. Plasmids (and large proteins) are anchored by the helical actin homolog, protein MreB, positioned beneath the inner cell membrane. MreB also maintains cell shape and directs peptidoglycan synthesis [68]. The tubulin protein FtsZ, positioned at the center of the cell, tethers genomic DNA to the Z-ring, where cell division occurs [69]. Other cell particles are displaced in an equally well-controlled manner.

3. Drivers of Cytoplasm Dynamics

Metabolic activities fluidize the cytoplasm and allow large “pockets” of particles to group in overcrowded regions (nanopools), indicated in Figure 1. Regions in the cell with nanopools are metabolically more active [70], which implies that the rearrangement (turnover) of nanopools would promote the displacement of “pockets” over longer distances in the cytoplasm. Meng et al. [70] argued that the positioning of the “pockets”, ranging from subnanometers to micrometers in size, and the difference in diffusion rates of these “pockets” in the cytoplasm, may hold the answer to movement (cytoplasmic fluidity). In a dynamic state (metabolically active cytoplasm), particles are energized by the random “moving” of ATP, amino acids, and other metabolites [71,72,73,74]. Parry et al. [1], however, argued that DNA-related processes (e.g., transcription), the displacement of the nucleoid, and the turnover of cytoplasmic content due to metabolic processes (by definition, also particle “pockets”) are not the driving force behind cytoplasmic fluidity. This coincides with the view formulated by Monti et al. [75]. Both scientific groups, Parry et al. [1] and Monti et al. [75], argued that collisions between particles, or particles in “pockets,” occur randomly and thus out of equilibrium. For the fluctuation-dissipation (FD) theorem to apply, particles in the cytoplasm must be in equilibrium [75]. In the absence of FD, the internal fluctuations (e.g., Brownian motion), are not identical to the dissipation of energy in a weak force or agitation (with reference to conditions in the cytoplasm). Put differently, the fluctuations of particles must correspond to fluctuations in their energy to generate a fluidic cytoplasmic force that would force interactions between particles. For further reading on the FD theorem, the reader is referred to Monti et al. [75].

In modeling a bacterial cell, Meng et al. [70] have shown that cytoplasmic dynamics arise from a combination of metabolic processes. The authors determined the magnitude of the active force (F) in the cytoplasm to be 0.57 pN (pico Newton). This, they argue, is generated by the combined collisions between small molecules, amino acids, proteins, and ions that are out of equilibrium. Based on the size of a typical protein (10 nm, as defined by the authors), the active force generated by such a protein is 0.4 pN. This was calculated using the equation F = k_BT/a, where F is the force expressed in N, k_B is the Boltzmann constant [≈ 1.38 × 10⁻²³ J/K (Joules per Kelvin)], T is the temperature in K, and a is the size (in this case 10 nm). The Boltzmann constant relates the average microscopic thermal energy of particles to the macroscopic thermodynamic temperature of a system [76]. According to Meng et al. [70], a force of 0.4 pN is sufficient to overcome thermal fluctuations, propel proteins in a specific direction when signaling in the cytoplasm changes, and support three-dimensional conformational changes in proteins.

4. Size Matters in Diffusion

The rate at which macromolecules diffuse across the cytoplasm determines the speed or frequency of collisions. This is true when reactions are diffusion-limited, i.e., when the association rate constant (k_a) depends only on the translational diffusion coefficient (D). However, the k_a of a protein diffusing in the cytoplasm with D = 10 μm²/s and reacting with another molecule is approximately 10⁸ M^-1/s (M = molar). Since most proteins have catalytic regions and are not reactive across their entire surface, a more realistic diffusion-limited k_a would be 10⁵ to 10⁶ M^-1/s [74]. The Barnase-Barstar protein pair of Bacillus amyloliquefaciens has a k_a of 10⁸ to 10¹⁰ M^-1/s, which is beyond the diffusion limit. The interaction between these proteins is mediated by electrostatic attraction [77,78,79,80], enabling ribonuclease binding to the inhibitor protein at orders of magnitude above the non-electrostatic diffusion limit. Other diffusion-limited protein reactions with high k_a include the ternary complex of aminoacyl-tRNA, EF-TU, and GTP that interacts with the ribosome, known as tRNA selection or decoding [81,82]; proteins in low copy numbers and with longer distances to cover; and proteins binding to membranes or other large structures (e.g., the Min oscillation system) [83].

Protein-protein interactions, such as those between cationic Barnase and anionic Barstar, are rare in bacteria. Only 35 proteins in E. coli have a net surface charge >+10. Of these, 18 are ribosomal, 14 are DNA/RNA-associated, and 3 have unknown functions [52]. It thus seems that the bacterial proteome is largely anionic, thereby supporting rapid diffusion through the cytoplasm. It also implies that endosymbionts with highly cationic proteomes have evolved specialized mechanisms to avoid slow diffusion and impair ribosomal function. As crowding increases, the D of macromolecules decreases by orders of magnitude, and many reaction-limited interactions between macromolecules become diffusion-limited. Small molecules, such as sugars trapped in supramolecular protein aggregates, are not severely affected by osmotic upshifts and diffuse throughout the cytoplasm, even at NaCl concentrations of 1 M or higher [49,84]. The cytoplasm thus acts as a molecular sieve under high and low osmotic stress, but with different mesh sizes. This allows the rapid diffusion of ions, small metabolites, and signaling molecules across a crowded cytoplasm.

5. Conclusions

It is unlikely that cytoplasmic flow is caused by displacement of the nucleoid or by metabolic processes, but more likely by particles in “pockets” (“cages”) that move randomly in a non-equilibrium state, localized in nanopools at the two poles of the cell. The propulsion (displacement or movement) of proteins causes cytoplasmic streaming (flow) and increases biochemical reactions, thereby increasing metabolic activity. Macromolecular crowding is a key contributor to intracellular complexity and plays a major role in regulating protein-protein, protein-nucleic acid, and protein-lipid interactions. The spatial orientation of biomolecular condensates (macrocompartments) serves the role normally fulfilled by organelles and intracellular membranes. These condensates/compartments are regulated by supramolecular structures such as DNA and RNA, as well as by PTMs and the cell membrane. The exact manner in which PTMs regulate phase separation must be confirmed. The size of the nucleoid-free space in the cytoplasm and the concentration of proteins are likely determining the size of a condensate. Further research is required on the ultrastructure of condensates and the arrangement of their components, especially for heterotypic condensates. We do not know to what extent the stoichiometry of complexes in a semi-solid (dense, highly viscous, glass-like) cytoplasm changes during turnover to a semi-liquid phase. Examining crowding effects on the structure, function, and regulation of biomolecular condensates is challenging. Not all biomolecular condensates are manipulated by macromolecular crowding. In some cases, condensation is triggered by ionic strength, pH, ATP, and membrane surfaces. Many questions remain unanswered or not fully understood, e.g., the effect of ribosomes on protein mobility, the effect of protein levels on reaction rates and diffusion coefficients, the influence of metabolic conditions on LLPS formation, differences in the mobility of old and new proteins at cell poles, conditions or reactions affecting the compaction of the nucleoid, factors determining the size of biomolecular condensates, and the role of nucleotides in the assembly of condensates.