Two important polymers cross paths

In an intriguing article by Gómez-García and Kornberg in this issue of PNAS (1), two important polymers, inorganic polyphosphate (poly P) and F-actin, cross paths.

Poly P and Its Enzymology Are Important Players in Cell Biology

Poly P, a ubiquitous polymer found in the environment and in all bacterial, fungal, animal, and plant cells, is an obvious energy store with high-energy phospho-anhydride bonds. Poly P is involved in a wide variety of critical cellular functions. It is, for example, essential for bacterial responses to stresses and starvation, as well as for survival (2). As an anion, it chelates Mg²⁺, Mn²⁺, Zn²⁺, and Fe³⁺ and affects their roles in the cell, and it also sequesters toxic metals such as Hg⁺ and Cd²⁺. As a polyanion, the flexible and relatively stable poly P can be localized in the cell and used in response to a wide variety of metabolic needs (3).

The enzymology of poly P has been elucidated primarily in the laboratory of Kornberg at Stanford University (2). The first enzyme activity for poly P synthesis was described in 1956 (4). Now known as PPK1 (poly P kinase 1), the gene for this enzyme was knocked out in Dictyostelium (H. Zhang, personal communication), revealing a second activity, DdPPK2, with remarkably different properties (1). As described below, the enzymatic activity of DdPPK2 is highly unusual in that it depends on and is coincident with assembly of the enzyme into a polymer, a property that it shares with the ubiquitous protein actin.

Actin Assembles into Filaments with Associated Hydrolysis of ATP

Biochemists have been studying actin since the 1940s. Actin is a highly conserved protein with >90% identity between yeast and human and serves a multitude of cellular functions (5). In its filamentous form (F-actin), it provides tracks along which a variety of myosin motors move carrying diverse cellular components to be strategically placed to create the organized but dynamic “city plan” of the cell. Together with myosin II bipolar thick filaments, actin filaments provide the contractile machine that drives muscle contraction and the furrowing process in a dividing cell. Actin also provides the force for cell migration by rapid assembly of filaments at the cell's leading edge. A host of actin-associated proteins regulate the dynamics of F-actin assembly and disassembly in the cell (5).

The assembly of actin monomers (G-actin) into F-actin is well characterized (6). Each 43-kDa G-actin monomer binds one ATP but does not act as an ATPase in this monomeric state. As a G-actin adds onto the growing end of an existing actin filament, a conformational change occurs in the newly added monomer that induces ATP hydrolysis. Inorganic phosphate is liberated, and ADP remains tightly bound to each monomer within the F-actin. At high G-actin concentration, rapid filament assembly occurs with an associated rapid hydrolysis of ATP. The amount of ATP hydrolyzed is equal to the number of G-actin monomers that have polymerized onto filaments. As the G-actin becomes depleted, the rate of disassembly approaches the rate of assembly, and at steady state a low level of ATPase activity is observed, manifesting the slow turnover of the actin filaments. The turnover of filaments can be increased manyfold by the action of accessory proteins that actively disassemble F-actin. Considering that in most eukaryotic cells ≈10% of the protein is actin and that filaments turn over rapidly and continuously, the energy available from ATP hydrolysis is very large. The amount of chemical energy needed to carry out the mechanical work involved, for example, in pushing the leading edge of a cell forward has not been determined.

The Activity of a New PPK Is Coincident with Polymerization into an Actin-Like Filament

Gómez-García and Kornberg (1) report the purification of a Dictyostelium polyphosphate kinase, called DdPPK2. The purified enzyme is a mixture of three proteins of ≈40 kDa each, similar in size to actin. Amino acid sequence analysis shows that each of the proteins has ≈60% identity to that of actin. This percent identity corresponds to that of the actin-related proteins, or Arps, which have been prominent players in cytoskeletal biology (7). Indeed, sequence analysis of the purified DdPPK2 revealed the presence of Arp1, Arp2, and an as-yet-uncharacterized actin-related protein. Arp1 and Arp2 have been extensively characterized in other organisms. Arp1, which is part of the dynactin complex associated with dynein-driven movement of vesicles along microtubules, can assemble into filaments resembling F-actin (8). Importantly, this is also true of DdPPK2. Arp2 forms a complex with Arp3 and five other proteins and helps regulate the spatial and temporal assembly of G-actin into filaments at the leading edge of a cell (5, 9). In Dictyostelium there are >10 genes that have 40–80% identity with actin and may therefore be members of the Arp family. Functions have been described in other organisms for Arps 1, -2, and -3 and, recently, for Arp11 (10), but there is little information regarding possible functions of other Arps.

As noted above, the purified DdPPK2 self-assembles into a thin filament with a 36-nm repeat, like F-actin and Arp1. The assembly is different, however, from actin in a number of ways. First, a small oligomer, possibly a tetramer, is the unit that adds onto a growing filament. Second, like Arp1, the critical concentration for assembly is much lower than that of actin. These differences are perhaps not surprising because DdPPK2 is clearly not actin per se but is related to the family of Arps. What is most remarkable is that poly P generation depends on concomitant filament assembly. Thus, analogous to actin, where ATP hydrolysis is coincident with actin filament assembly, the transfer of γ-phosphate of ATP to a growing chain of poly P is coincident with DdPPK2 assembly. In the case of DdPPK2, unlike the case for actin polymerization, a highly reversible reaction occurs that is not associated with a large free energy change.

It will be fascinating to watch this story unfold. Knockouts of the relevant genes are easy to carry out in Dictyostelium and will provide evidence as to which Arps are required for DdPPK2 activity. Gene knockouts will also reveal the functions of poly P in Dictyostelium. Assuming there are only two Dictyostelium PPKs, DdPPK1 and DdPPK2, a double knockout will reveal the phenotype of a poly P-free eukaryotic cell.

A challenge will be to understand mechanistically how this enzyme works, given the dimensions of the two polymers involved. The spacing of two ≈40-kDa subunits is ≈55 Å, whereas the spacing of two phosphates in a poly P polymer is only ≈3 Å. Yet their polymerization is coupled. It is difficult to imagine how this unusual enzymatic activity is mechanistically achieved. Translocation must be key, so that the penultimate P is at the end of the actin-like polymer. It is interesting that when the DdPPK2 filament is examined, poly P is bound to DdPPK2 such that its ends are blocked to the action of yeast exopolyPase. The poly P is, however, accessible to a yeast endopolyPase and is hydrolyzed to smaller chains and eventually to P_i and some small poly P oligomers.

It is worth emphasizing that poly P is itself a store for high-energy phosphate that can be transferred to ADP to make ATP, and that the polymerization of DdPPK2 is energetically much less costly than normal actin assembly. Gómez-García and Kornberg (1) speculate that bona fide actin might also have the capability of making poly P rather than hydrolyzing the ATP to ADP and P_i, thereby significantly reducing the energy costs of the very high level of actin turnover within the cell. Certainly all purified actins assemble into filaments with concomitant conversion of ATP to ADP and P_i. But was PPK activity lost during purification? This speculation leads to the following important aspect of this very interesting work.

The biochemistry of the 1950s and 1960s, which defined many of the metabolic pathways of the cell as well as the molecular basis of DNA replication, transcription, translation, and other biosynthetic pathways, had an underlying principle that differs from many modern methods used to characterize newly discovered proteins. More often than not, current approaches identify a protein from a gene sequence, and its function is inferred from its primary sequence. This approach is often highly successful. The sequence of a gene coding for the dilute mutation in mouse was found to have the key signatures of a myosin (11). It was later both purified from cell extracts and expressed in baculovirus to give rise to an interesting myosin that moves processively along an actin filament. In other cases, however, a protein is found by binding some other partner in the cell. A whole host of actin binding proteins have been identified in this way. Do any of these have enzymatic functions that have gone undetected?

Poly P generation depends on concomitant DdPPK2 assembly.

The underlying principle of the earlier approach was to first create a quantitative in vitro assay for a function of interest, PPK activity in the case of Gómez-García and Kornberg (1). Once the assay is established, the enzyme activity in crude cell extracts is sought and purified to homogeneity. The sequence can then identify the encoding gene(s). Although further characterization of the DdPPK2 will reveal its relationship with cytoskeletal functions in cells, an additional important aspect of this work from Gómez-García and Kornberg is their reminder of the importance of this fundamental approach in biochemistry. Suffice it to say, no one would think of discovering a cytoskeletal protein and then assaying it for PPK activity.