Which types of hydrogen bonds are known to exist in RNA secondary structures?

Question

I'm programming an implementation of an algorithm for pattern matching in RNA structures. The algorithm assumes the following types of base pairs:

• Plain: No base pairs (just the primary structure of the RNA)
• Single: Each base can be connected to at most one other base
• Multiple: Each base can be maximally connected to a number of other bases (where 'number' can be constant or infinite)

Although there is no problem with the implementation of the above types of base pairs (apart from the probable high computation complexity), I wonder if all of them were observed in real RNAs?

In particular, I'm interested in the following cases:

• A base pair between two adjacent bases i,j.
• One base (i) which connected to two different bases (j,k) (i is not adjacent to j or/and k).
• One base (i) which connected to more than two different bases, all of them not adjacent to i.

Can these exist? Some of them? If they can, how common is their existence?

Answer 1

By far the most common type of base pair is the Watson-Crick base pair in an RNA helix. Those are comparably easy to predict, e.g. Mfold and the Vienna RNA package can do this.

Base triples, three nucleobases that form hydrogen bonds to each other are not uncommon in RNAs with a complex tertiary structure. There is even a database of RNA triples, though this one also contains triples where not all three bases make contact with each other.

There are also quadruplexes, but there each base only makes contact with two other bases. I don't know if there are any structures where one base makes contact with more than two other ones.

Answer 2

A base pair between two adjacent bases i,j.

It is not possible. To form an intramolecular hydrogen bond, the RNA has to bend. The persistence length i.e. the minimum length of the chain required for bending, is around 4 bases for ssRNA (I am not very sure about this number. At this moment I am not able to locate the exact reference. I remember this number from the back of my head which I heard somewhere).

One base (i) which connected to two different bases (j,k) (i is not adjacent to j or/and k).

It is definitely possible for one pair after considering the persistence length. If a single base is bound to more than two bases then atleast one interaction should have to form a non Watson-Crick type bonding, as indicated by Mad Scientist.

One base (i) which connected to more than two different bases, all of them not adjacent to i

Very less likely. There are no known reports. Even if it happens, the bond will be weak because of limitation in the number of H-bond donor/acceptor sites in the bases.