When I started my graduate work at TSRI in John Tainer’s lab, one of the first projects I worked on involved determining the structure of an RNase HII (or Type II RNase H) enzyme. In general, a ribonuclease (abbreviated RNase) is an enzyme that degrades RNA. There are many different types or classes of RNase enzymes, and several of these enzymes are essential for life. RNase HII enzymes are not well-characterized, although RNase H (note the missing “II”) has been studied extensively, mostly because HIV reverse trascriptase (a key enzyme involved in HIV viral replication) has an RNase H domain.

RNase H enzymes only degrade ribonucleotides in a hybrid duplex consisting of RNA hybridized to complementary DNA. This RNA/DNA hybrid structure forms as an intermediate in some key biological processes. For example, “standard” DNA replication in nearly all cellular organisms starts with a short piece RNA synthesized based on the DNA template. The RNA provides a starting point for enzymes that synthesize DNA. Later in the replication process, the RNA is removed and replaced with DNA. As mentioned previously, viruses that have RNA genomes (such as HIV) also form RNA/DNA hybrid during their replication process. Thus, the function of RNase H to specifically degrade the RNA in an RNA/DNA hybrid is biologically important.

We were interested in RNase HII because it specifically targets the structure of the hybrid duplex, and degrades the RNA independent of base sequence. In the fall of 1999, there was reasonable evidence that RNase HII enzymes were involved in DNA replication in eukaryotes. Biochemical and genetic studies showed that RNase HII interacts with flap endonuclease (FEN-1), which removes RNA primers generated during DNA replication. In contrast to RNase H, which degrades all the RNA in an RNA/DNA hybrid, RNase HII has a different substrate specificity. Biochemical experiments indicated that in vitro, RNase HII prefers a subtrate consisting of DNA+RNA in one strand, hybridized with a complementary DNA strand. RNase HII cleaves mostly near the DNA-RNA junction in the hybrid strand. Combined with genetic evidence show that RNase HII and FEN-1 were in the same pathway, we thought the RNase HII might be an important player in DNA replication.

So there I was, a rotating graduate student in John’s lab, given this RNase HII project that a senior graduate student in the lab had started. When I inherited the project, it came with one native data set to 2.0 angstrom resolution, and a collaborator, who had provided the protein. In order to obtain phases, I wanted to do a multiwavelength anomalous dispersion (MAD) experiment, which required protein substituted with selenomethionines in place of the normal methionine residues. My collaborator offered to make this protein, so I waited to get the shipment from them. They shipped concentrated samples, flash-frozen in small aliquots, which is a fairly common way to store proteins for long-term storage. Unfortunately, when I thawed the first tube, a significant amount of precipitate formed. SDS page analysis showed that it indeed was RNase HII protein. At this point I decided to take matters into my own hands.

As I’ve come to find out, crystallography, like most scientific techniques, is heavily dependent on the quality of the sample. It’s impossible to control the quality of your sample if you’re not making it yourself, so I asked my collaborator to send me the DNA over-expression plasmid, and set about to over-express and purify RNase HII myself. As an undergrad I had plenty of experience trying molecular cloning and failing miserably. I had never done protein expression or purification, but my cloning failures taught me alot about the value of controls and and how to troubleshoot. As it turns out, my first attempt went very smoothly. This particular RNase HII from the archaeabacteria Archaeoglobus fulgidus over-expressed very well in E. coli. Since the recombinant protein is thermostable and also contained a histidine tag, purification was straight-forward.

Everything was proceeding smoothly until I tried to concentrate the protein using a centrifugation-based filtration concentrator. As the concentration approached 20 mg/mL, a precipitate formed in the bottom of the concentrator (I was using a centricon concentrator, for those of you who know what that is). Upon closer inspection, I realized that this was no ordinary precipitate. It had a slight sparkle, and reproducibly sunk to the bottom of the concentrator if gently disturbed. I pipetted a few microlitres onto a coverslip and looked at it under the microscope. I was shocked to see crystals (granted, they were ugly crystals, but was I to complain) sitting on the coverslip. Of course, I made this discovery around 8 or 9 pm. These things always tend to occur when there is nobody around to share the news with. Fortunately, Karl-Peter Hopfner, who was a postdoc at the time, was still around, so I grabbed him and made him look at my concentrator-grown crystals. He laughed and told me that the same thing had happened to him a few months ago when he purified P. furiosus Rad50.

It turns out the this particular RNase HII enzyme crystallized in several different conditions and finding conditions that produced high quality crystals was not entirely trivial. In the end, I ended with conditions that were different from those used to crystallize the native protein. With the help of others in the lab we collected a MAD dataset at the ALS and determined the structure. Approximately 6 months later, while I was attempting to grow crystals of RNase HII with metal cofactors and DNA, I realized that the DTT added to prevent oxidation of selenomethionine had a significant affect on the diffraction quality of the crystals. Omitting DTT from the sample buffer produced crystals similar to the original native crystals that diffracted to slightly higher resolution.

RNase HII has the same basic fold as RNase H and related nucleases. However, it also has an extra C-terminal that is unique to RNase HII enzymes. Several conserved residues that are required for enzymatic function are located on a loop that extends down from the C-terminal domain towards the active site. Based on the structure, we proposed that RNase HII might recognize the spacing of adjacent phosphates in the backbone of the RNA/DNA hybrid. This spacing is a potentially unique feature of the hybrid, since the helical structure of an RNA/DNA hybrid is thought to lie somewhere between double-stranded DNA (B-form helix) and double-stranded RNA (A-form helix).

It’s worth noting that we actually got “scooped” on this work, since another group published the structure of RNase HII from M. jannaschii approximately one month before we were ready to submit our paper. Luckily, I had an additional structure showing the location of the metal ion necessary for catalysis, and our collaborator provided mutational and biochemcial data to support our findings. I modified the manuscript to include their result and emphasize how our data expanded upon their findings. It’s a good trick to keep in mind for the next time you get scooped.

Overall, I learned quite a bit about each step in the process of determining a crystal structure, starting from expression, all the way through writing a paper. After this, I was “hooked” on structural biology. I tried several other projects that didn’t work, and a few that did. I hope to write about those here eventually. That’s all for now, thanks for reading. If you’ve made it this far, you might be interested in reading the paper.