Phage Transition: Spatial Organization of Phage Assembly

Christopher Bazinet

I came to MIT as a fresh grad student in the fall of 1979. Four years out of undergrad, I had spent a couple of years teaching, then as a technician for Gail Bruns at Childrens Hospital. There, I found that I enjoyed life at the bench, so when a fellow technician mentioned to me that the good grad programs were tuition free and paid a living stipend, this mill town boomer applied to PhD programs. Recombinant DNA was the big thing then, and I arrived expecting to clone and sequence, but I fell in love with genetics, and was moved by the elegance and deep insights that came from phage genetics. Jon seemed more like someone I could relate to than most other faculty members, so it felt like the right fit for me.

Initially, I was attracted to the phage tail length problem that Phil Youdarian had worked on, but Jon had decided to focus on P22 head assembly, and suggested early on that I think about how the unique DNA-translocating vertex of the phage was built. David Goldenberg and Donna Smith taught me the basics of handling phage and I was off and running. After an unhappy year or so studying the semi-essential gene 14, (best to avoid leaky mutants if at all possible…), I finally got after the portal vertex problem. The phenotype of the T4 gene 20 mutants suggested strongly that T4 gp20 was required for efficient initiation of prohead assembly, so our working hypothesis was that gp1 of P22 (the likely structural orthologue of T4 Gp20) might have a role in head assembly initiation. If the portal vertex proteins were assembled as part of the initiation of coat protein polymerization, then the polymerizations at other vertices could be qualitatively different: the distinction between initiation and “elongation” could provide a means for structural differentiation of the portal vertex. We therefore expected that when we looked at the rate of prohead assembly in 1 infections, we would see significant depression relative to WT. Instead, the pulse-chase data showed clearly that heads were assembled just as quickly in 1infections as in the wild-type controls.

This was puzzling: If head assembly didn’t “wait” for the minor proteins to be incorporated into growing heads, what mechanism ensured that they were incorporated when available? The stoichiometry of head protein synthesis was very similar to the stoichiometry in proheads and mature phage, implying that some mechanism ensured the inclusion of all the required protein subunits into each particle. It seemed to me that there must be a form of spatial regulation: the gp1 (and presumably the other minor proteins) must be sequestered in or somehow channeled to a specific location where assembly initiation took place. Sherwood Casjen’s and Jon’s earlier work had showed that gp8, the scaffolding protein, appeared to be an autoregulator of its own synthesis, presumably at the translational level. The best-known forms of translational autoregulation involved binding to and interfering with the translational initiation sites of the autoregulator. Since gene 8 was immediately downstream of the gene 1 portal protein on the late message, gp8/scaffolding protein bound to its own translational start site could be ideally located for “catching” gp1 being released from just upstream. The mRNA could provide an “organizing center” for the initiation of phage head assembly. If gp8 bound to the RNA was activated to capture the portal protein, then initiation of prohead assembly there could effectively channel gp1 into the process; this initiation complex could then leave the site, ensuring that subsequently assembled (“elongation stage”) vertices would not incorporate gp1. Sure enough, Sherwood and Jon’s earlier work published in Nature showed that the beginning of phage production coincided nicely with the downregulation of scaffolding protein synthesis.

The lab had an uncharacterized gene 8 temperature-sensitive mutant which hadn’t been characterized. When I worked up this mutant, I found that it beautifully confirmed one part of this hypothesis: proheads were produced efficiently by the tsU172 mutant, but they lacked gp1 and the other minor proteins of the portal vertex: Scaffolding protein provided a function for inclusion of the portal protein(s), distinct and separable from its role in the formation of coat protein shells.

At the same time, Myeong-Hee Yu was sequencing tailspike (gp9) folding mutants. One of them, tsRAF, had been isolated by Jon Jarvik years earlier as a suppressor of a 1 cold-sensitive mutant that failed to be assembled into the prohead. This made no sense to us since gp9 had no known role in prohead assembly, and was not attached to the phage head until after DNA packaging and stabilization. In light of this, I hypothesized that tsRAF might represent a secondary “organizing site”, in which case it might show some sequence similarity to the gp8 translational initiation region. When Sherwood Casjens sent us the sequence of the scaffolding protein gene, a simple eyeball alignment between the tsRAF site and the gene 8 translational initiation site showed a significant similarity, as predicted by the “RNA organizer” hypothesis. I was excited. At my defense, the committee was dubious—it was the early days of “sequenceology” and my crude calculation of an “E Value” was new to all of us.

Jon kept me on as a postdoc for a year following my defense, during which further sequence sleuthing with a crude program written in basic revealed a second site in the gene 9 sequence that was also quite similar to the gene 8 start site. Because the lab had pretty much saturated the gene with ts mutations, there were several that mapped at or very close to this second candidate organizer site. When I crossed these into the 1cs background, they also suppressed the assembly mutation! That did it for me: First, a suppressor mutation led us to a site with significant similarity to the hypothesized “organizing center” for prohead initiation. Now, identification of a second potential organizing center by sequence similarity showed the same suppression of the original 1prohead assembly mutation. Interestingly, the purine-rich sequences found in all three of these sites was echoed in the TMV and ribosome assembly initiation sites, so there seemed to be potential for a more general theme of protein-RNA interactions in biological assembly initiation.

It wasn’t clear how to I was gripped with the desire to further study this, but life was crazy. My wife and I had buried both of our fathers in the year preceding my defense, our son Oliver had been born, and we were being threatened with eviction as Cambridge real estate and rents skyrocketed. By in vitro transcription, I synthesized a small piece of RNA centered around the gene8 start site and sent it to Peter Previlege to try in his in vitro assembly reactions, but the result was negative. In my wildest imaginings (if this is not enough…) I visualized the whole late messenger RNA as a complex dynamic machine channeling proteins to assembly sites formed at least in part by the RNA itself. To his credit, Jon was receptive to my musings, but we had no idea how to approach or test the idea experimentally. We had a viable 1deletion mutant, del1, which made a shorter gp1 distinguishable from WT by SDS gel electrophoresis. I crossed del1 into several different amber mutation backgrounds on the hunch that nonsense termination somewhere on the late RNA might affect assembly: If the message was somehow participating in the channeling of its own protein products to assembly initiation site(s), a cis-effect on the recovery of gp1 might be observable if an amber mutation in a highly dynamic site on the messenger disrupted such a channeling. I don’t think I presented the results of these experiments in group meeting, but the first experiment (effect of an 8amber mutation, if I remember right) showed no effect. In a second experiment, 9amber mutation did seem to show a modest effect—but I didn’t manage to repeat the experiment. I must have showed it to Jon, however, as he later sent me the notebook with this work in it in case I ever wanted to follow up on it, but the exigencies of life intervened and I never got back to it.

In my retirement in Seattle, I hope to find a lab that will sponsor a “playpen” for me. A hot topic in cell biology these days is the meta-organizational cellular rearrangements commonly referred to as “phase transitions”. There’s a strong consensus that most, if not all cellular phase transitions require RNA synthesis, indicating the participation of RNA molecules in these processes. The centrosome, organizing/control center for microtubule assembly initiation, may be a “compartment” operating on the phase transition principle. The assembly of phage proheads in an RNA organized cloud/compartment could be a simple prokaryotic model for cellular morphogenesis associated with RNA-dependent phase changes.

I thank all my fellow journeyers from the King Lab: David Goldenberg, Donna Smith, Harlee Strauss, Ed Loechler, Erika Hartwieg, Myeong-Hee Yu, Ben Fane, Peter Prevelige, Cammi Haase. I hope you are all well. Life goes by fast!