Crystallization of RNA molecules other than simple oligonucleotide dup
lexes remains a challenging step in structure determination by X-ray c
rystallography. Subjecting biochemically, covalently and conformationa
lly homogeneous target molecules to an exhaustive array of crystalliza
tion conditions is often insufficient to yield crystals large enough f
or X-ray data collection. Even when large RNA crystals are obtained, t
hey often do not diffract X-rays to resolutions that would lead to bio
chemically informative structures. We reasoned that a well-folded RNA
molecule would typically present a largely undifferentiated molecular
surface dominated by the phosphate backbone. During crystal nucleation
and growth, this might result in neighboring molecules packing subtly
out of register, leading to premature crystal growth cessation and di
sorder. To overcome this problem, we have developed a crystallization
module consisting of a normally intramolecular RNA-RNA interaction tha
t is recruited to make an intermolecular crystal contact. The target R
NA molecule is engineered to contain this module at sites that do not
affect biochemical activity. The presence of the crystallization modul
e appears to drive crystal growth, in the course of which other, non-d
esigned contacts are made. We have employed the GAAA tetraloop/tetralo
op receptor interaction successfully to crystallize numerous group II
intron domain 5-domain 6, and hepatitis delta virus (HDV) ribozyme RNA
constructs. The use of the module allows facile growth of large cryst
als, making it practical to screen a large number of crystal forms for
favorable diffraction properties. The method has led to group II intr
on domain crystals that diffract X-radiation to 3.5 Angstrom resolutio
n. (C) 1998 Academic Press Limited.