326

23

Regulatory Networks

23.7.2

Crosslinking

The principle of this approach is to instantaneously crosslink all associated part-

ners (protein–protein and protein–DNA) using formaldehyde while the cell is still

alive. It is then lysed to release the crosslinked products, which can be identified

by mass spectrometry. In the case of a protein–nucleic acid complex, the protein

can be degraded with a protease, and the DNA fragments to which the protein was

bound—which should correspond to transcription factor binding sites—can be iden-

tified by hybridizing to a DNA microarray.

The specific instantiation for proteins (especially transcription factors) bound to

DNA is called chromatin immunoprecipitation (ChIP). In order to identify the DNA,

after crosslinking and cell lysis the DNA is fragmented by sonication and selected

complexes are precipitated using an appropriate antibody for the protein of interest,

following which the DNA can be sequenced. In order to determine where the protein

binds on the chromosome, the fragmented DNA can be exposed to an appropriate

microarray (ChIP-on-chip technology).

23.7.3

Correlated Expression

The assumption behind this family of methods is that if the responses of two (or more)

proteins to some disturbance are correlated, then the proteins are associated. As an

example, mRNA expression is measured before and after some change in conditions;

proteins showing similar changes in transcriptional response (increase or decrease,

etc—the expression profile) are inferred to be associated. Another approach is to

simultaneously delete (knock out) two (or more) genes that individually are not lethal.

If the multiple knockout is lethal, then it is inferred that the encoded proteins are

associated.

Although these approaches, especially the first, are convenient for screening large

numbers of proteins, the assumption that co-expression or functional association

implies actual interaction is very unlikely to be generally warranted, and, indeed,

strong experimental evidence for it is lacking.

23.7.4

Other Methods

Many other ways to identify protein complexes are possible; for example, A could

be labelled with a fluorophore, and B labelled with a different fluorophore absorbing

and emitting at lower wavelengths. If the cell is illuminated such that A’s fluorophore

is excited but the emission of B’s fluorophore is observed, then it can be inferred

that A and B are in sufficiently close proximity that the excitation energy is being

transferred from one to other by Förster resonance. This approach has a number of