Polyphosphate: The “Dark Matter” of Bacterial Chromatin Structure

ABSTRACT

Polyphosphate (polyP), broadly defined, consists of a chain of orthophosphate units connected by phosphoanhydride bonds. PolyP is the only universal inorganic biopolymer known to date and is present in all three domains of life. At a first approximation polyP appears to be a simple, featureless, and flexible polyanion. A growing body of evidence suggests that polyP is not as featureless as originally thought: it can form a wide variety of complexes and condensates through association with proteins, nucleic acids, and inorganic ions. It is becoming apparent that the emergent properties of the condensate superstructures it forms are both complex and dynamic. Importantly, growing evidence suggests that polyP can affect bacterial chromatin, both directly and by mediating interactions between DNA and proteins. In an increasing number of contexts, it is becoming apparent that polyP profoundly impacts both chromosomal structure and gene regulation in bacteria, thus serving as a rarely considered, but highly important, component in bacterial nucleoid biology.

Polyphosphate condensates and their contributions to nucleoid structure. Polyphosphate condensates form in the ribosome‐depleted nucleoid region of bacterial cells, particularly under stress (in this case Pseudomonas aeruginosa undergoing nitrogen starvation; adapted with permission from (Racki et al. 2017)). This Perspective explores the emerging view that these membraneless organelles contribute structurally and functionally to bacterial chromatin.

1. Polyphosphate: The Ground That Everyone Walks on but Nobody Sees

From the early days of electron microscopy, polyphosphate “storage granules” have been noted as a ubiquitous feature of microbial cells. These structures are dense enough to generate contrast in the absence of heavy metal stains. Electron diffraction studies consistently demonstrate that these structures are amorphous, rather than crystalline, and elemental analysis with energy dispersive X‐ray spectroscopy (EDX) demonstrates these structures are enriched in the divalent cations Mg²⁺ and Ca²⁺ (Ward et al. 2012; Tocheva et al. 2013; Su et al. 2023). An expanding body of work has demonstrated that polyP granules are condensates incorporating both polyvalent cations and proteins, including, importantly, DNA structural proteins such as Hfq (Beaufay et al. 2021) and AlgP (Chawla et al. 2022), as well as both DNA and RNA. Recent advances particularly in the study of biomolecular condensates, and the formation of polyP‐protein‐nucleic acid complexes in microbial systems, have prompted us to re‐examine the role of polyP in the structure of the bacterial nucleoid, and in driving bacterial gene regulation and chromatin biology. As we will argue below, the conservation of polyP synthesis, as well as its promiscuity in interacting with other biomolecules, formation of a wide variety of condensates, and substantial dynamics in quantity and properties under different cellular growth phases, imbue it with substantial leverage in terms of potential for regulating cellular behavior. We argue that the polyP state of the cell constitutes a fundamental state variable of the cell, one that is sensed and interacted with through many different paths, and which fundamentally reshapes the biophysical and regulatory state of the cell. One major distinction possessed by polyP is that it may not have emerged as a signaling molecule due to selective pressure for the evolution of some signaling system; rather, it may always have been there, and the cellular regulatory and sensing systems may themselves have evolved with the presence of polyP as a backdrop. Furthermore, the general neglect of consideration of polyP as a key biomolecular interaction partner, and state variable, has likely prevented the field from recognizing a highly influential factor contributing to gene regulation under many conditions. It is important to note that polyP is generally not essential in bacteria (one can delete all polyphosphate kinase genes from bacteria where this has been attempted), and under ideal growth conditions few phenotypes are observed in polyP‐deficient mutants, but polyP‐dependent cells are sensitized to a wide range of environmental insults including DNA damaging agents and starvation (Rao et al. 2009; Beaufay et al. 2020). Here we will review key properties and observations of interactions between polyP and key biomolecules and metabolites both in vitro and in vivo, with an emphasis on interactions relevant to bacterial chromatin structure and gene regulation. We then close with a set of suggestions that we believe will help researchers incorporate measurement and consideration of polyP into their broader investigations into the bacterial nucleoid.

2. Empirical Observations of polyP‐Biomacromolecule Interactions In Vitro and In Vivo

Polyphosphate has a strong propensity to form condensates in cells, and a growing body of evidence suggests the emergent biophysical properties of these structures are likely at the heart of this polymer's biological functions, including its chromatin‐associated functions. The biomolecular condensate revolution of the past two decades has transformed our understanding of mesoscale organization in living systems, particularly in the context of eukaryotic chromatin (Strom et al. 2017; Sanulli et al. 2019), bacterial cell biology (Azaldegui et al. 2021), and now bacterial chromatin (Joyeux 2023; Gupta et al. 2023). Concomitant with this revolution in our understanding of modern cells has been renewed interest in an old hypothesis that phase separation could be an important mechanism of generating protocells in the absence of membranes (Oparin 1924; Haldane 1929), and allowing the formation of specialized compartments in cells without requiring membrane boundaries. Nucleic acids and nucleic‐acid binding proteins have been a central focus of our understanding of extant condensates, and RNA in particular has understandably been the polymer of choice for studies of origin‐of‐life condensate research (e.g., (Roden and Gladfelter 2021)). From a materials science perspective, polyphosphates have long been appreciated for their capacity to form materials with highly variable physical properties, from amorphous glasses to liquid‐like condensates, depending on chain length, cation composition, solvent, and temperature (Van Wazer 1958; Omelon and Grynpas 2008; Donovan et al. 2014; Schröder et al. 2022; Wadsworth et al. 2023). But polyP is also ubiquitous in cells from all three domains of life, and the condensates that it forms in cells make important interactions with, and often contain, DNA and RNA. The very ubiquity of polyphosphate also argues for its antiquity, and it is highly notable in this context that polyP so closely resembles nucleic acid biopolymers in both chemistry and biophysical properties. We propose that polyphosphate, the only universal inorganic polymer in all three domains of life, is the “ur‐polymer” of living systems and that polyP condensates, enriched in divalent cations in cells, may also be the “ur‐condensates” of cells.

Understanding how polyP condensates affect chromatin architecture and function requires interaction between a top‐down approach of characterizing the composition and spatiotemporal dynamics of polyP condensates in vivo, and a bottom‐up approach of measuring how individual components of polyP condensates affect phase separation, as well as the organization, conformational dynamics, and intermolecular interactions between polyP, nucleic acids, and nucleic acid binding proteins. Here we summarize what is known about how polyP itself, alongside cationic mediators, the emergent properties of polyP condensates and their interactions with nucleic acids in vitro, as well as the composition and biophysical properties of polyP condensates in vivo (emphasizing its role in the bacterial nucleoid), and how its behavior in both of those contexts may be connected.

2.1. Biophysical Properties and Intermolecular Interactions of polyP

2.1.1. Essential Properties of Polyphosphate

At its simplest, polyphosphate as observed in bacterial cells consists primarily of a linear chain of phosphoryl (PO₃) monomers connected by ester linkages; effectively it is a condensation polymer produced from orthophosphate/phosphoric acids (H_xPO₄) through elimination of water. In any discussion of the condition‐dependent dynamics of polyphosphate, it is important to keep in mind that in addition to gross changes in polyphosphate concentration and average chain length, there can be alterations in the chemistry of chain ends or differences in cyclization and branching levels of polyphosphate; these are frequently unconsidered in analysis and discussion of polyP (Figure 1A). Indeed, while methods for quantifying total polyP and average length distribution have improved (Christ et al. 2020), methods currently used for assessing other aspects of polyphosphate's structure are typically low throughput (e.g., NMR), hampering systematic characterization of polyP states under different conditions for a given organism. Long‐held dogma has suggested that cellular polyphosphate exists almost exclusively as a linear, rather than branched or cyclized, polymer (Rao et al. 2009). More recent evidence, however, has suggested that branched polyphosphate may be more stable than previously thought under cellular conditions and that the enzyme alkaline phosphatase can use branched polyphosphate as a substrate (Dürr‐Mayer et al. 2021). Any such structural variations in the polyphosphates occurring in biological systems would likely have been missed in many historical studies, even if they were present, due to conditions during polyP purification that favor breakdown of branched and cyclized polyP (Dürr‐Mayer et al. 2021). Numerous reports of cyclic polyphosphate in cells do exist in the literature (Niemeyer 1976; Mandala et al. 2020), whereas we are unaware of any direct observations of branched polyP (despite the circumstantial evidence noted above of alkaline phosphatase acting upon it).

FIGURE 1.

Structure, state, and roles of polyphosphate in bacterial cells. (A) Chemical structure of the fundamental polyphosphate unit, along with known and hypothetical arrangements and modifications. (B) “Polyphosphate attributes,” as defined here, noting that these attributes can vary independently of each other. (C) Example cases of polyphosphate‐nucleic acid interactions that may affect nucleoid structure and gene regulation. (D) Possible structures for polyphosphate condensates, which may be (clockwise from top left) homogeneous, consist of a shell and inner phase (which may or may not itself be heterogeneous), consist of a shell and inner phase plus some components that span the phases or include purely surface‐associated components which do not form a separate shell phase. The nature and relative amounts of polyP, nucleic acids, proteins, metabolites, and metal cations may all vary between the indicated regions—each color represents an abstract possible topology for condensate components or phases to take on, rather than a specific molecular entity.

Even considering purely linear polyphosphate, tremendous variation in the amount and chain length of polyP is known to arise in cells during stress conditions; in foundational studies on the subject, variations in both polyP concentration and average chain length across at least two orders of magnitude (< 10¹ to > 10³ phosphate units) were observed under a variety of stress conditions (Rao et al. 1985; Ault‐Riché et al. 1998). The extent to which polyP chain ends are decorated by specificity determinants or other covalent chemical modifications in vivo remains unknown, but proteins, including nuclear proteins in eukaryotes, have been suggested to be polyphosphorylated at specific residues in a manner that governs their partitioning within the nucleus (Azevedo et al. 2015). It is probable that different bacterial species will exhibit diverse changes to polyphosphate concentration, chain length, and modifications in response to different stresses and under different growth conditions. In the discussion below, we will collectively refer to the concentration, chain length, and modification states of polyP in cells as the “polyphosphate attributes” (see also Figure 1B). We note that here we must remain agnostic to the relative differences in the characteristics outlined above; at this point, it is not even generally clear to what extent correlations exist between the different axes of polyphosphate attributes. A recent survey of analytical methods suitable for investigating polyphosphate attributes, as we have defined it here, is given in (Christ et al. 2020), and may serve as a catalog of methods that could be more broadly adopted in the field.

2.1.2. Interactions With Nucleic Acids

Many polyphosphate condensates are closely associated with the nucleoid; from the early days of imaging polyP condensates by electron microscopy, their close physical association with the chromosome has been documented (Voelz et al. 1966; Henry and Crosson 2013; Racki et al. 2017; Rosigkeit et al. 2021) (schematized in Figure 1C). Using fixation and thin sectioning, Voelz et al. (1966) first described polyP granules as “embedded” in the nucleoid in Myxococcus xanthus , and suggested that nucleoid fibers may be “adherent to the granules.” Moreover, when cells were partially lysed by sonication, the nucleoid fibers protruding from channels created by sonication appeared to remain connected with the granules. A challenge with these early studies was the use of fixatives, which have been noted to disrupt granules (McGlynn et al. 2018). The potential for fixative artifacts in the context of condensates more generally is now well appreciated (Miné‐Hattab 2023; Dupont et al. 2023). The cryo‐electron tomography revolution, enabling the near‐native state visualization of the ribosome‐depleted nucleoid region of bacterial cells, validated the close association of polyP condensates with the nucleoid region in diverse taxa, though the details of its organization within the nucleoid region are species‐ and condition‐specific (discussed in more detail below). The localization of polyP condensates to the nucleoid region strongly suggested that polyP condensates interact with DNA, though the nature of this interaction between two of the most negatively charged polyanions in the cell was not clear. Additionally, from the earliest attempts to isolate polyP from cells, a strong divalent cation‐mediated association of polyP and RNA was observed (Stahl and Ebel 1963). In the following sections, we consider the three broad classes of potential mediators of polyphosphate‐nucleic acid interactions in the context of condensates: metals, proteins, and metabolites. We highlight the dual role that many of these mediators appear to play as both the molecular bridge between polyP and nucleic acids and drivers of polyP phase separation.

2.1.3. Interactions With Metals

Due to the polyanionic nature of polyP, it should be unsurprising that it interacts frequently with positively charged metal ions; Christ et al. even go so far as to argue that the nature of polyP counterions should be considered alongside the other characteristics that we have defined under the heading of “polyphosphate attributes” (table 1 of (Christ et al. 2020)). Divalent and polyvalent cations, including Mg²⁺, Ca²⁺, and transition metals including Mn²⁺ can themselves drive polyP condensate formation in a purely inorganic system (Schröder et al. 2022; Chawla et al. 2024; Wadsworth et al. 2023; Dai et al. 2023). The participation of other cationic species like DNA binding proteins and polyamines in polyP phase separation is therefore not strictly required to explain the formation of polyP condensates in vivo. Recent in vitro studies indicate that condensates of polyP and the divalent cations Mg²⁺ and Ca²⁺ are sufficient to drive dramatic changes in DNA architecture: a simple 3‐component system of polyP, Mg²⁺, and DNA drives DNA to form thin shells on the surface of polyP‐Mg²⁺ condensates (Chawla et al. 2024). Cryo‐EM imaging of the condensate interface reveals that DNA forms brush‐like filaments protruding from their surface. These studies indicate that completely inorganic polyP‐Mg²⁺ condensates can create two distinct environments: a surface that can confine DNA from 3D to 2D diffusion, as well as the condensate interior, relatively depleted of DNA. How confinement of DNA to the condensate surface affects its topology (particularly for supercoiled DNA) and potential molecular interactions remains to be determined. Interestingly, polyphosphate condensates formed with Mn²⁺ have been shown to exclude dsDNA and ssDNA (Dai et al. 2023).

Just as with assemblies of nucleic acids, and consistent with the in vitro findings outlined above, metals are a ubiquitous component of polyP condensates in bacteria. Bacterial polyP condensates have long been known to be electron‐opaque enough to generate contrast by Transmission Electron Microscopy (TEM) in the absence of the heavy metal stains often required to observe other subcellular structures. Elemental analysis with energy dispersive X‐ray spectroscopy (EDX) demonstrate these structures are enriched in metal ions, particularly the divalent cations Mg²⁺ and Ca²⁺, in diverse species, and can also sequester transition metals like iron and manganese (Ward et al. 2012; Tocheva et al. 2013; Racki et al. 2017; Li et al. 2019; Caffrey et al. 2024). In addition to the purely structural role played by metals in the formation of condensates involving polyP (as discussed in the in vitro section above), as a “sponge” for metals, polyP may sequester metals both important for chromatin architecture, notably Mg²⁺, and also metals that can damage DNA and RNA, such as iron (Beaufay et al. 2020). It is also important to note that metals strongly impact the fluorescent signals arising from detection of polyP with the recently characterized dye JC‐D7 (Deitert et al. 2024), providing further evidence for the strong effects of metal binding on polyP's characteristics (and also raising an important cautionary note in the use of dyes such as JC‐D7 in the characterization of polyP in living systems).

2.1.4. Interactions With Proteins

Recent studies in E. coli and P. aeruginosa using complementary approaches, and previous work in Ralstonia eutropha , have demonstrated that polyP fractions are enriched in nucleic acid‐binding proteins (Tumlirsch et al. 2015; Beaufay et al. 2021; Chawla et al. 2022). However, only a handful of bona fide polyP‐binding proteins have been validated using fluorescence microscopy. We discuss two key examples, Hfq and AlgP below, both nucleic acid‐binding proteins with disordered domains. From a biophysical perspective, we know that Hfq can drive polyP phase separation in vitro, resulting in condensates that interact with DNA in a distinct manner from PolyP‐Mg²⁺ condensates (Beaufay et al. 2021), as DNA goes into the droplet interior in 3‐component PolyP‐Hfq‐DNA systems, rather than associating with the surface as is observed with PolyP‐Mg²⁺ condensates. These studies indicate that partitioning between these two distinct environments, the condensate surface and interior, is likely tunable, depending on the composition of the condensate as a whole.

2.1.5. Interactions With Metabolites

In addition to metals, small molecules likely play an important role as mediators between nucleic acids and polyP within condensates, and also potentially as drivers of polyP phase separation. The metabolome of polyP condensates remains largely unknown, but one class of likely suspects is polyamines such as spermine, spermidine, and putrescine. These multivalent cationic small molecules, present at millimolar concentrations in cells, have long been known to bind to nucleic acids and directly affect their structure (Tabor and Tabor 1985; Lightfoot and Hall 2014), and functional interactions with proteins (Lightfoot and Hall 2014; Plateau et al. 2019; Duprey and Groisman 2020). Polyamines promote DNA condensation (Sarkar et al. 2007; Sarkar et al. 2009), important in phage genome packaging (Hershey 1957; Ames et al. 1958), and can drive phase separation of RNA in vitro (Wadsworth et al. 2023). In both E. coli and P. aeruginosa , polyamines promote polyP synthesis (Motomura et al. 2006; Peng et al. 2017), suggesting a functional association.

3. Phenotypes Associated With Polyphosphate (and the Loss Thereof)

Polyphosphate mutants have pleiotropic defects, long thwarting easy interpretation of their molecular function. Henceforth we use the term “∆ppk _all ” to refer to cells in which all identifiable polyphosphate kinases have been deleted. One seemingly universal theme across bacterial species observed in ∆ppk _all mutants is a failure to adapt to and survive diverse environmental stresses, from oxidants to metals to antibiotics to diverse forms of nutrient and energy limitation (Crooke et al. 1994; Rao and Kornberg 1996; Kuroda et al. 1999). PolyP synthesis is important for virulence and mediates interactions with the host in both commensal and pathogenic species, as well as interactions between species (Bowlin and Gray 2021; Shah et al. 2024). Of both medical and ecological importance, ∆ppk _all mutants have decreased starvation survival and persistence, making polyP an attractive antimicrobial target in the context of chronic infections where cells are often in non‐growing or slow‐growing states for which conventional antibiotics targeting activities important during growth are often ineffective.

3.1. The Useful Comparison of (p)ppGpp

The scale of the effect of polyP on gene expression is on par with another virtually universal global regulator of gene expression in response to starvation, the small molecule (p)ppGpp. Deletions of polyphosphate kinases have been observed to impact the expression of hundreds of genes in Escherichia coli (Beaufay et al. 2021), Pseudomonas aeruginosa (Rao et al. 2009), and Mycobacterium tuberculosis (Chugh et al. 2024), although the effects of an equivalent deletion in Campylobacter jejuni were more contained (Chandrashekhar et al. 2015). In Burkholderia pseudomallei , proteomic profiling of a ∆ppk1 strain revealed significant changes in abundance of at least 70 proteins, including master regulators RpoS and the quorum sensing LuxI homolog BpsI (Srisanga et al. 2019). Indeed, polyP and (p)ppGpp were originally thought to act in a common pathway in the stringent response, with (p)ppGpp acting upstream of polyP synthesis. While more recent detailed genetic studies are inconsistent with this model, at least in E. coli and P. aeruginosa (Racki et al. 2017; Gray 2019), the question remains whether polyP, like (p)ppGpp, can act through direct molecular interactions as a master regulator of gene expression. Generally, nucleoside phosphates and their derivatives have long been recognized as cornerstone regulators of gene expression, serving as a direct “pulse” on the cell's metabolic state (Atkinson 1968; Potrykus and Cashel 2008; Carling et al. 2011). PolyP, as a fundamental state variable, may have been co‐opted early in evolution to serve a similar role. A longstanding challenge to testing this hypothesis has been the lack of apparent (or at least known) specificity determinants on polyP. Whereas direct interactions between (p)ppGpp and central dogma machinery have been established (Artsimovitch et al. 2004; Wang et al. 2007; Corrigan et al. 2016; Diez et al. 2020), definitively demonstrating that polyP interacts with protein complexes in vivo is challenging. However, while lacking specificity at the primary level of organization, the condensate superstructures formed by polyP provide an opportunity to validate bona fide interactions using fluorescence microscopy. Two key examples of molecular interactions between polyP and proteins that have been validated by orthogonal methods are discussed in the case study sections below.

3.2. Effects on Nucleoid Organization, Dynamics, and Stability

In a wide range of species including E. coli (Gross and Konieczny 2020), C. crescentus (Boutte et al. 2012), P. aeruginosa (Racki et al. 2017), and S. cerevisiae (Bru et al. 2016), ∆ppk _all mutants have cell cycle progression defects, suggesting a broad coupling between polyP synthesis and the cell cycle (although overexpression of E. coli polyphosphate kinase in human cells did not cause detectable effects on cell cycle regulation (Bondy‐Chorney et al. 2020)). In both C. crescentus under carbon starvation, and in E. coli , ∆ppk _all cells inappropriately initiate DNA replication (Boutte et al. 2012; Gross and Konieczny 2020), possibly due to mis‐regulation of degradation of cell cycle regulators (CtrA and DnaA, respectively). In P. aeruginosa , ∆ppk _all cells fail to complete open rounds of the cell cycle during nitrogen starvation, through unknown mechanism(s) (Racki et al. 2017). PolyP may contribute to chromosome stability, as the SOS DNA damage response is activated in ∆ppk _all cells under nitrogen starvation in P. aeruginosa (Racki et al. 2017). In E. coli , polyP has been shown to contribute to chromosome integrity by yet another mechanism, regulating the acquisition and activity of foreign DNA, including the insertion sites of group II introns (Zhao et al. 2008), and the expression of horizontally acquired genes including prophage genes (Beaufay et al. 2021). PolyP also has been shown to affect the global compaction and dynamics of the chromosome, with the nucleoid exhibiting decompaction in ∆polyP cells in both E. coli and P. aeruginosa (Magkiriadou et al. 2024; Beaufay et al. 2021). Furthermore, in P. aeruginosa , the chromosome becomes “hypermobile” in ∆ppk _all mutant cells during nitrogen starvation (Magkiriadou et al. 2024). It is important to note that in many of the cases noted above, polyP is likely acting in concert with, or helping to organize, nucleoid‐associated proteins; this has been most clearly and directly observed in the aforementioned case of prophage repression, where polyP appears to act in concert with the nucleoid‐associated protein Hfq in exerting its regulatory effects (Beaufay et al. 2021); similar principles may well apply in other cases.

4. The Role of Biomolecular Condensates in Organizing, Functionalizing, and Specializing the Bacterial Nucleoid

While bacterial cells typically lack internal membranes (with several notable exceptions, (Ferrara et al. 2024) for review), and thus are often not subject to the internal spatial divisions so ubiquitous in eukaryotic cell biology, subcellular localization and specialization of bacterial components have long been recognized to play a major role in bacterial biology. Indeed, the formation of membraneless organelles has recently emerged as a widespread organizing principle in bacterial cells (reviewed in (Azaldegui et al. 2021)). The presence of the nucleoid itself acts as a major driver of cellular organization, as the nucleoid occupies a large fraction of the interior volume of the cell (Gray et al. 2019), excludes ribosomes (Hobot et al. 1985; Wang et al. 2011; Bakshi et al. 2012) (which are themselves often the most abundant macromolecular component of cells (Bakshi et al. 2012)), and plays a central role in organizing cell division through regulation of Z‐ring formation (Woldringh et al. 1991; Tonthat et al. 2011; Cho et al. 2011). While lacking a membrane‐bound nucleus to clearly delineate chromatin association, bacterial biomolecular condensates fall broadly into three categories: nucleoid‐excluded, nucleoid‐associated, and nucleoid‐integrated. PolyP condensates associate with the nucleoid in many bacteria, but the association may be dynamic and condition‐specific even in the same species.

4.1. Nucleoid‐Excluded Condensates

Many bacterial membraneless compartments and condensates are excluded from the nucleoid region, including (for example) PopZ microdomains (Bowman et al. 2010; Lasker et al. 2022), single‐stranded binding protein condensates (which can also become nucleoid‐associated (Harami et al. 2020)), TmaR‐driven polar sequestration of sugar transport machinery (Szoke et al. 2021), inclusion bodies (Linnik et al. 2024), protein nanocompartments (Jones and Giessen 2021), and membrane‐associated assemblies such as the M. tuberculosis ABC transporter Rv1747 (Heinkel et al. 2019).

4.2. Nucleoid‐Associated Condensates

In contrast with nucleoid‐excluded condensates, many biomolecular condensates are observed to associate with the nucleoid in terms of subcellular localization. For the present purposes, we further subdivide that category into condensates that are nucleoid‐associated (i.e., those that occur on the nucleoid periphery, and may or may not include DNA) vs. nucleoid integrated (those that are observed to actually occur in the nucleoid interior). It should be noted that light microscopy alone often cannot be used to resolve the distinction between these categories of condensates, as many microscopy experiments do not allow resolution between bacterial microcompartments (BMCs) occurring above or below the nucleoid with respect to the view of the microscope, vs. those actually within the nucleoid. Biochemical experiments or higher‐resolution imaging are often required to distinguish between these possibilities.

Even with the above caveat in mind, it has been shown that many BMCs, including polyhydroxyalkanoate (PHA) granules (Galán et al. 2011) and carboxysomes (MacCready et al. 2018), are positioned in the cell through the activity of DNA‐binding proteins that interact with the nucleoid (Savage et al. 2010; Wahl et al. 2012; MacCready et al. 2018; Galán et al. 2011), typifying what we would refer to as the nucleoid‐associated class of BMCs. The phasin protein PhaF in Pseudomonads is thought to tether PHA granules to the nucleoid, thereby ensuring their equal segregation during cell division (Galán et al. 2011), and also acts as a transcriptional regulator of PHA biosynthesis genes (Prieto et al. 1999; Tarazona et al. 2020). PhaF interacts with DNA through its histone H1‐like disordered C‐terminal domain (Maestro et al. 2013), which intriguingly has significant homology to the C‐terminal domain of AlgP, a polyphosphate granule binding protein discussed in more detail below in the case study section. The positioning of carboxysomes along the nucleoid has been particularly well studied, and has been shown to be dependent upon ParA/MinD family ATPases (Savage et al. 2010; Vecchiarelli et al. 2012; MacCready et al. 2018). By riding the chromosome, these cargo or “passenger” membraneless compartments—which have important metabolic functions—are evenly inherited by daughter cells, which has been shown to contribute to fitness in the case of carboxysomes (Savage et al. 2010).

Association with the nucleoid has long been thought to affect polyP granule organization. Changes in chromosome organization and compaction during cell cycle progression and stress alter granule organization in diverse species. For example, in Caulobacter crescentus , polyP granules are positioned at the ¼ and ¾ positions on the cell's long axis, suggesting a possible connection between granule number and chromosome number (Henry and Crosson 2013). Indeed, chemical and genetic perturbations of chromosome segregation in C. crescentus disrupt granule organization (Henry and Crosson 2013). In Synechococcus elongatus , granules associate with the nucleoid, and form closely associated “pairs” when the nucleoid becomes more compacted, which has been proposed to be the result of granules themselves dividing as a result of changes in chromosome folding (Seki et al. 2014). Similarly, timelapse imaging in Agrobacterium tumefaciens demonstrated that polyP granules can migrate to the new cell pole in a coordinated manner with cell division (Frank et al. 2022).

4.3. Nucleoid‐Integrated Condensates

Some membraneless compartments clearly also associate closely with the DNA itself and act as “drivers” of bacterial chromatin function, organization, and dynamics. ParA/MinD family ATPases, which are broadly known for their prominent role in partitioning nucleic acid cargoes across the vast majority of bacteria (Lutkenhaus 2012), provide an important example. The ParA ATPase exerts its activity through interaction with the DNA‐binding protein ParB, which itself forms condensates along with its target parS sites and are maintained and separated by ParA activity (Guilhas et al. 2020). Other prominent examples of biomolecular condensates directly integrated into the nucleoid, and often driving gene regulation at either transcriptional or post‐transcriptional levels, include the clustering of RNA polymerase organizing strong transcription (Ladouceur et al. 2020), protection of DNA by Dps in stationary phase cells (Janissen et al. 2018), enhancement of transcriptional termination by Rho phase separation in B. thetaiotaomicron (Krypotou et al. 2023), and the formation of BR‐bodies organizing the RNA degradosome machinery (Al‐Husini et al. 2018; Muthunayake et al. 2020).

As discussed in the above phenotypes section, polyP has long been known to affect nucleoid function. In recent years, exciting evidence has emerged supporting a model that polyP affects chromatin organization and function by directly altering DNA topology (Figure 2A) and DNA‐protein interactions in the context of condensates (Figure 2B). PolyP plays an essential role in mediating the formation and properties of three‐component droplets consisting of the protein Hfq, DNA, and polyP in nitrogen‐starved E. coli , which plays a major role in Hfq‐dependent gene regulation (Beaufay et al. 2021) (discussed further below). Somewhat similarly, in Pseudomonas aeruginosa under nitrogen starvation, many small polyP granules form in the nucleoid region and then consolidate and become evenly spaced on the long axis of the cell during the period of cell cycle exit. A specific DNA‐binding protein required for granule spacing within the nucleoid region, AlgP, has been recently identified, as discussed further in the case study below.

FIGURE 2.

Hfq and AlgP: Case studies of polyP‐DNA mediators in bacteria. (A) Polyphosphate condensates enriched in divalent cations, particularly Mg²⁺, can organize DNA on their surface as a thin shell (left, orange DNA). DNA binding proteins may change the mesoscale structure by recruiting specific DNA sequences selectively into the condensate interior (right, green DNA). The nature and concentration of polyP, DNA binding proteins, and available cations may dynamically tune DNA organization and accessibility in the context of polyP condensates. (B) PolyP may alter DNA‐protein interactions. For DNA binding proteins (blue) capable of forming condensates without polyP (left, orange DNA recruited), polyP (pink) may change the properties of the condensates, as well as alter the DNA sequence preference (right, green DNA recruited). (C‐D) Domain architecture of P. aeruginosa AlgP (C) and E. coli Hfq (D). Both proteins have structured N‐terminal domains, intrinsically disordered C‐terminal domain, and a short “tip” region with atypical amino acid compositions. The N‐terminal domain of AlgP is predicted to form an anti–parallel coiled‐coil domain, raising the possibility that this domain may participate in oligomerization. Hfq is known to form hexamers (and these hexamers super‐oligomerize in the presence of long chain polyP). (E) Comparisons of what is known about AlgP and Hfq as mediators. Dark green: Supported by experimental evidence; light green with question mark: Predicted but not experimentally confirmed; red: Not present, according to existing experimental evidence; white: Unknown (and in many cases, presumably negative).

One key area for future consideration is the extent to which polyP may participate in the assembly and properties of other membraneless compartments in bacteria, particularly those associated with the nucleoid or with nucleic acids—given the ubiquity and biophysical properties of polyP, it seems quite likely that this molecule plays roles in other bacterial condensates that have been previously missed.

5. Case Study: The P. aeruginosa DNA Binding Protein AlgP Orchestrates the Spatiotemporal Dynamics and Nucleoid Interactions of polyP Condensates

A histone H1‐like protein, AlgP, was recently identified as an important mediator between polyP condensates and the nucleoid in Pseudomonas aeruginosa . AlgP, a highly abundant DNA‐binding protein that was previously identified as a potential NAP (Amemiya et al. 2021), localizes to polyP granules and facilitates their even spacing within the nucleoid region, suggesting that polyP granules act as “passengers” that ride the nucleoid to achieve even spacing. AlgP has an intrinsically disordered C‐terminal domain with 54 repeats of “KPAA” and variants of this sequence (AlgP_CTD, Figure 2C). The AlgP_CTD is similar in overall sequence to PhaF_CTD, which as mentioned previously facilitates even spacing of PHA granules, consisting of a hydrophobic carbon storage polymer polyhydroxyalkanoate (PHA) (Prieto et al. 1999). Both polyP and PHA are synthesized in response to nitrogen starvation in Pseudomonads (Tobin et al. 2007; Racki et al. 2017). It is thought that PhaF tethers PHA granules to the nucleoid by this C‐terminal domain, and indeed the AlgP_CTD is also thought to bind DNA (Medvedkin et al. 1995). PhaF may also play a role in tuning the surface properties of the PHA condensates via its N‐terminal domain which has homology to another so‐called “phasin” protein PhaI. In the case of a ∆phaF mutant, PHA granules clump together but don't fuse, potentially due to the presence of PhaI, other surface‐interacting proteins, and the material properties of these condensates (Galán et al. 2011; Maestro and Sanz 2017). In the ∆algP mutant, polyP forms one large spherical granule in the nucleoid region, strongly suggesting that polyP condensates may consolidate over time by coalescence. AlgP may alter the surface properties of polyP condensates to inhibit fusion, or non‐exclusively simply tether the nascent granules to the nucleoid, such that they can only fuse with nearest neighbors.

The DNA binding capacity of the AlgP_CTD, along with its similarity in sequence composition to eukaryotic linker histones, raises the possibility that AlgP might contribute to polyP condensate association directly with bacterial chromatin. AlgP was originally identified as a transcription factor involved in alginate biosynthesis, though this function has now been called into question by recent studies demonstrating that AlgP does not affect the alginate operon, or indeed have large effects on transcription generally (Cross et al. 2020). However, studies of the transcriptional effects of AlgP have not yet been performed under conditions where polyP granules form. As a highly expressed DNA binding protein, AlgP has been suggested as a potential NAP (Amemiya et al. 2021), and indeed by fluorescence microscopy, an AlgP‐mApple chimera diffusely co‐localizes with the nucleoid in a ∆ppk _all mutant. How polyP synthesis affects AlgP's interactions with the chromosome remains to be determined. It is not yet clear for AlgP if polyP binding competes with DNA binding, or changes its sequence preference, as observed with Hfq. Intriguingly, there is evidence that the AlgP_CTD may be a mutational hotspot in the context of infections, subject to changes in the number of repeats (Deretic and Konyecsni 1990; Onteniente et al. 2003; Chung et al. 2012). How such changes might affect AlgP's interactions with DNA and/or polyP remains to be determined.

6. Case Study: Polyphosphate Acts as Both a Scaffold and Modulator of Behavior for the E. coli Nucleic Acid Binding Protein Hfq

A recent series of findings in E. coli has implicated polyphosphate as an important modulator of both biomolecular condensate formation and gene expression in bacteria mediated by the hexameric nucleoid associated protein Hfq.

As a highly abundant nucleic acid binding protein, Hfq associates with both the nucleoid and with RNA in the cytoplasm, and participates in a broad range of processes including chaperoning small RNA‐mediated gene regulation, ribosome biogenesis, gene silencing, and chromatin structure regulation (Valentin‐Hansen et al. 2004; Schu et al. 2015; Updegrove et al. 2016; Cech et al. 2016; Andrade et al. 2018; Arluison et al. 2022). E. coli Hfq is known to have three nucleic acid binding interfaces, referred to as the “proximal”, “distal”, and “rim” sites. Of these, all three binding sites are known to interact with RNA in at least some contexts, whereas only the proximal and rim faces have been shown experimentally to interact with DNA (Orans et al. 2020). In addition, many bacterial species have N‐terminal or C‐terminal extensions to Hfq that vary widely across species. In E. coli , for example, Hfq has a long, disordered C‐terminal extension that is essential for phase separation of Hfq (Beaufay et al. 2021, Figure 2D) and compaction of DNA by Hfq (Malabirade et al. 2018). It is important to note that Hfq is capable of forming homotypic condensates, although in the context discussed below it also forms mixed condensates with polyP and nucleic acids (Figure 2E).

The recent discovery that Hfq binds to the polyanionic polyphosphate is perhaps unsurprising, given that Hfq has long been known to bind both DNA and RNA. In the best‐characterized case of polyP affecting E. coli Hfq behavior in vivo, the foci formed by Hfq under nitrogen starvation (McQuail et al. 2020; McQuail et al. 2024) was observed to be polyphosphate‐dependent (Beaufay et al. 2021). Furthermore, in the absence of polyphosphate, in addition to the loss of the large Hfq assemblies, the distribution of Hfq on the genome fundamentally shifts, switching from an AT‐rich to a GC‐rich motif and leading to de‐repression of many genes in AT‐rich regions of the genome (Beaufay et al. 2021), and gross changes in the mobility of Hfq become apparent in single particle tracking analysis. Intriguingly, polyphosphate appears to compete effectively with DNA for binding Hfq but not with RNA when polyphosphate concentrations increase, potentially suggesting that polyphosphate has a higher affinity for the DNA binding than RNA binding interfaces of Hfq, and/or that different nucleic acid binding interfaces are favored by polyP, DNA, and RNA (Beaufay et al. 2021).

In the context of E. coli , where Hfq‐polyphosphate‐DNA trivalent interactions have been observed, the polyphosphate chain length plays a substantial role in the complex's qualitative behavior, providing insights into possible regulatory mechanisms for gene expression in bacteria. Indeed, whereas low molecular weight polyP can bind Hfq hexamers in vitro, higher chain length polyP induces a qualitative change in the behavior of Hfq that triggers the formation of higher order oligomerization, occurring at approximately the chain length beyond which polyP could no longer be wrapped by a single Hfq hexamer. These findings reveal a mechanism through which changes in polyphosphate attributes (and not simply polyP concentration) can impact Hfq‐nucleic acid interactions and the formation of higher order Hfq‐containing structures.

7. Likely Roles of polyP in Bacterial Chromatin Structure

Considering that polyphosphate attributes fluctuate over time, both in terms of chain length and concentration, and are particularly sensitive to certain environmental stresses, it is reasonable to conjecture that polyphosphate could significantly influence the restructuring of gene expression in response to major environmental changes. Indeed, as we have documented above, this has been discovered to be the case in at least several instances in bacteria. However, there remains a vast landscape of polyphosphate‐dependent changes yet to be identified, primarily due to challenges surrounding the study of this molecule which is, first, often overlooked or insufficiently characterized during a stress response, and second, inherently difficult to examine. A comprehensive understanding of polyphosphate's full regulatory effects on gene expression requires a thorough characterization of the variety of conditions under which polyphosphate attributes change and the precise nature of distinct polyphosphate species produced by bacteria under various biological conditions. While considering these possibilities, it is useful to examine what we now know about the contributions of polyP to the structure of the bacterial nucleoid, both as a topic in its own right and comparison with the richer existing literature on eukaryotic chromatin.

While lacking nucleosomes, and largely lacking canonical histones, bacteria have a diverse repertoire of nucleoid‐associated structural proteins (NAPs) with many parallel functions, including tuning access to the transcriptional machinery. A longstanding puzzle in the bacterial chromatin field is how bacteria regulate chromatin structure, including modulation of the organization and dynamics of high‐affinity and high‐avidity NAPs, in the absence of the diversity of ATP‐dependent chromatin remodeling enzymes found in eukaryotes. One possibility, first raised in 1976 before eukaryotic ATP‐dependent remodelers had been discovered, is that polyphosphate acts as a chromatin remodeler (Weinstein and Li 1976). That study demonstrated that polyP can evict histones from chromatinized DNA and enhance transcription in vitro. The capacity for polyP to evict NAPs from DNA may itself be dynamic and depend on the cationic species present. Such a mechanism would be broadly analogous to the “competitive substitution” mechanism invoked for prothymosin alpha chaperone proteins (which are disordered and negatively charged) in evicting linker histones from nucleosomes (Heidarsson et al. 2022).

It seems plausible to us that polyphosphate constitutes an important and integrated component of bacterial chromatin, acting as a “glue” in upholding the condition‐dependent changes in bacterial chromosomes, chromatin, and gene regulatory programs. The “glue” analogy applies in two capacities: firstly, in a literal structural sense, aiding in the assembly of nuclear protein complexes in a condition‐dependent manner to establish appropriate physiological complexes, often mediated by phase separation; secondly, in a metaphorical sense, assisting bacterial cells in maintaining specific physiological and regulatory states via its impact on gene regulation. That is, we hypothesize that polyP functions both as a structural component of bacterial chromatin, and (through its many interactions) as a non‐conventional remodeler of bacterial chromatin across changing physiological states, drawing upon the substantial evidence of both structural changes and changes in transcriptional regulation when polyP is lost or over‐produced. PolyP is uniquely positioned to fulfill the role of a major modulator of cellular state in response to changing environments due to the sheer number of mechanisms, based on pure biophysics and evolved sensing, that allow it to impact gene regulatory programs and its direct sensitivity to the metabolic state of the cell.

Notably, there is an important contrast between the differentiation landscapes experienced in higher eukaryotes vs. bacteria. Differentiation in multicellular eukaryotes is often visualized via the Waddington landscape, with canalized development into increasingly specialized cell types in which terminal differentiation is the rule (Slack 2002). In contrast, in bacteria, different physiological and morphological states typically exist, but there also tends to be more flexibility in switching between these in response to environmental changes (although many well‐studied examples of terminal differentiation certainly exist, such as Bacillus subtilis sporulation or Caulobacter crescentus stalk cell formation). In light of the observations that have been made to date, it seems quite likely that polyphosphate plays a key role in structuring the changes allowing transient “differentiation” in bacteria, and that the sensitivity of polyP attributes to environmental conditions may indeed play a key role in the very flexibility of bacterial phenotypic states. However, its full involvement remains to be investigated in most instances, and the vast range of polyP‐interacting proteins, and indeed, dynamics of polyP attributes, across the bacterial kingdom, surely add to the complexity of the problem. Fully measuring the organism‐ and condition‐dependent dynamics of polyP, its intermolecular interactions, and ultimately its effects on bacterial chromatin (both local and at cellular scale) will be a vast but important enterprise in further understanding bacterial cell and systems biology.

8. Outlook and Open Questions

Arthur Kornberg, who first discovered and characterized both known families of bacterial polyP kinases, famously laid out a list of maxims that, in his view, constituted ten commandments of enzymology (Kornberg 2000) (which he subsequently updated and expanded (Kornberg 2003)). While we cannot offer such definitive statements yet for work on the biological roles of polyphosphate, either on bacterial nucleoid structures or in general, given the foundational roles of polyP outlined above we can at least offer a set of four questions that scientists ought to keep in mind in order to not neglect the potential role of this important molecule in a multitude of processes. To begin with, the scientist without prior knowledge might ask, “how much polyP are cells making under the condition(s) that I care about?” With such information, another simple question that ought to be asked would be, “what kind of polyphosphate are cells making under the condition(s) that I care about?”, referring to the notion of polyphosphate attributes that we have outlined above. Having learned the answers to those first two questions, the skeptical scientist might ask, “what (phenotypic) effect is polyP actually having in cells under the condition that I care about?”, rightly wondering whether there is in fact any relevant role being played at all. But if the answer to that question is affirmative, a wise follow‐up would be to ask, “how (mechanistically) is polyP interacting with other relevant biomolecules under the conditions that I care about?” Pursuing these questions, or at least considering them as a checklist until one of the answers rules out meaningful involvement of polyP, is in our view an important addition to consideration of any process involving the bacterial nucleoid, given the foundational nature of polyP, and its demonstrated importance for many aspects of nucleoid structure and function as described above.

We need to measure polyP attributes and condensates under physiologically relevant conditions: To directly address how polyP affects chromatin, we need to make measurements under conditions where the polymer is abundant, where polyP condensates form (and what fraction of the cellular polyP is in condensates vs. cytoplasmic), and when polyP is functionally relevant. Exponential growth in LB is not the optimal condition for such measurements, particularly when it comes to addressing polyP's role in antibiotic tolerance and virulence. Excitingly, recent work in human plasma demonstrated that polyP condensates form under these conditions and that polyP synthesis contributes to complement evasion (Janet‐Maitre et al. 2023). A key omission to date in genetically tractable model systems is to test how polyP affects anaerobic survival. While many environmentally and medically important microorganisms make polyP anaerobically (Shirai et al. 2000; Tocheva et al. 2013; McGlynn et al. 2018), and the polyP‐binding protein AlgP is implicated in regulating anaerobic survival in P. aeruginosa (Cross et al. 2020), measuring the effects of polyP on chromosome dynamics and transcription under oxygen limitation in genetically tractable pathogens like P. aeruginosa will be revealing.

We need to characterize the polyP condensate proteome, under diverse stress and starvation conditions, and over time, including in deep starvation states. Several studies have implicated nucleic acid binding proteins, including NAPs, as polyP condensate clients, but little is known about how the polyP condensate proteome changes in response to different stress states. Given that the abundance of different NAPs changes dramatically as a result of growth rate, these measurements are critical to understanding the function of polyP in chromatin and beyond. Validation of partitioning using orthogonal techniques is critical here, but also not trivial, for three key reasons: (1) In the case of chromatin structural proteins like H‐NS, the proteins often do not tolerate modification with fluorescent proteins for imaging, as some such fusions are known to be subject to clustering artifacts (Wang et al. 2014). (2) PolyP condensates do not tolerate fixatives, complicating the use of antibodies (McGlynn et al. 2018). (3) Tags can themselves influence partitioning, the case‐in‐point being charge variants of GFP which not only differentially localize to polyP condensates, but likely change the properties of these structures by driving condensate formation as in the case of supercharged +36GFP (Wang et al. 2020). More generally the structure–function landscape for bacterial chromatin states, particularly outside of steady‐state growth, remains very poorly defined.

8.1. Uncoupling

We should think of methods to uncouple polyP biosynthesis from condensate formation: to address the specific function of these condensates, or develop other mechanisms to perturb the condensates. The functional significance of the condensates, including their organization, nucleoid localization, and dynamics, remains unclear. The uncoupling problem is particularly challenging for polyP, where there may be significant redundancy in partner species capable of driving demixing (as discussed in the biophysical properties section above).

As a field, we need to consider the integration of the protein chaperone and chromatin functions of polyP: PolyP functions as a non‐specific protein chaperone, and as we have argued here, the condensates it forms are an important component of bacterial chromatin. What is not yet clear is how these functions are spatially and temporally related: for example, does the protein chaperone function occur within granules, or via soluble polyP, or both? There is precedent for condensates playing a role in protein chaperone function in bacteria, notably in the context of inclusion bodies. Moreover, the nucleolus, the archetypal nuclear condensate in eukaryotes, couples transcriptional responses to protein chaperone function (Frottin et al. 2019). Given the importance polyP synthesis in proteostasis during oxidative stress in E. coli and P. aeruginosa , a better understanding of the chromatin‐associated roles of polyP under oxidative stress, as well as the cell biology of polyP condensates in response to oxidative stress, will be informative. A role for polyP in the physical and temporal coupling of transcriptional responses and protein chaperone activity makes sense in the context of starvation, when polyP is made, ribosome stalling increases, and resources for DNA repair become scarce.

We should consider how polyP might contribute to Epigenetics: PolyP may contribute to epigenetics in two key (but speculative) ways: (i) Persistence of condensates after a stressful event has ended, and inheritance of condensates, may differentially affect gene expression in progeny cells, and (ii) Post‐translational modification of NAPs. Technical approaches like the Mother Machine microfluidic device need to be employed to address condensate segregation into daughter cells and effects on transcription at the single‐cell level. For example, given that spacing of polyP condensates is regulated in P. aeruginosa , the inheritance of granules by daughter cells during cell cycle re‐entry may be functionally important. If polyP condensates persist after the first round of cell division and do not themselves split into two in a coordinated manner, then polyP and all associated DNA binding proteins will be unequally inherited between daughter cells. Quantitative single‐cell measurements across multiple timescales are necessary here to capture distinct but important behaviors: from milliseconds to capture fusion and fission events, to minutes and hours to capture cell cycle‐coordinated events, to hours or potentially longer to capture granule consumption and survival effects in starved states. While P. aeruginosa cells can live for weeks in growth‐arrested states (Eschbach et al. 2004; Glasser et al. 2014), very little is known about chromatin architecture and polyP condensates on these timescales. Such “slow” timescales are difficult to access, but very important from a biophysical and medical standpoint: biophysically, we know that the cytoplasm's properties change generally in starved states (Parry et al. 2014), that polyP can influence cytoplasmic motion in some but not all growth‐arrested states (Magkiriadou et al. 2024), and that condensate “aging” can change the viscoelastic properties of condensates themselves dramatically, making them more dense and viscous with time (Jawerth et al. 2020; Alberti and Hyman 2021). From a medical perspective, it is these deeply starved states where polyP condensates in bacterial pathogens may matter most for survival and persistence in chronic infections. (ii) Phosphorylation of chromatin‐associated proteins is important across all domains of life. We do not yet know if protein kinases can use polyP directly in bacteria to phosphorylate NAPs and transcription factors, though other enzymes such as NAD kinase (Kawai et al. 2001; Garavaglia et al. 2003) and glucokinase (Szymona and Ostrowski 1964; Phillips et al. 1993) in some bacterial species are thought to be more promiscuous and even actually favor polyP over ATP as a substrate. Modification of lysines within poly‐acidic lysine‐containing (PASK) domains by polyphosphorylation is also thought to occur in yeast, though whether this is an ionic association or covalent modification is an active area of debate (Azevedo et al. 2015; Azevedo et al. 2024; Neville et al. 2024).

Author Contributions

Lisa R. Racki: conceptualization, writing – original draft, writing – review and editing, funding acquisition. Lydia Freddolino: conceptualization, writing – original draft, writing – review and editing, funding acquisition.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.