One concept that is sometimes used in developmental biology is the idea of the “master control gene” or “master switch” — a single gene whose expression is both necessary and sufficient to trigger activation of many other genes in a coordinated fashion, leading to the development of a specific tissue or organ. It’s a handy concept on which to hang a discussion of transcription factors, but it may actually be of rather limited utility in the real world of molecular genetics: there don’t seem to be a lot of examples of master control genes out there! Pax-6 is the obvious one, a gene that initiates development of the eye, and other genes may be mentioned in certain stem cell pathways, but even in the eyes of vertebrates, for example, eye development is more complicated than a single switch, and similarly, many other developmental processes seem to use multiple or redundant regulatory controls — the cases where we have a single gene bottleneck are either rare or poorly represented in the literature.

They’re still at least pedagogically useful, though — it’s a simple case of imposition of a specific developmental pathway on a patch of tissue.

The cartoon below is a greatly simplified diagram of a piece of DNA encoding a transcription factor. The thick bar is the actual coding region — the part of the DNA that will be transcribed into messenger RNA and translated into a protein, in this case a protein that binds to specific sequences of DNA.

An important part of the gene is the regulatory DNA. Diagrammed here is the cis regulatory region — all “cis” means is that it is on the same strand of DNA as the coding region — on which I’ve draw a couple of little squiggles that represent short nucleotide sequences to which other transcription factors can bind. These are important: most genes will be silent unless there is a bank of proteins bound to the regulatory region that promote the binding of the polymerase protein complex to the start of the coding region.

I’ve only drawn two squiggles to stand for two cis regulatory elements (CREs) here, but genes will typically have many CREs. The level of activity of the gene will depend in a quantitative and qualitative way on which and how many CREs have protein bound to them. Note also that some CREs will bind inhibitory binding proteins, which can turn the coding region of the gene off. Molecular and developmental genetics is the study of regulatory logic — puzzling out how the combination of genes active in a cell controls how other genes are switched on and off. The patterns of activity can get very complex. The proteins that bind to DNA and switch gene states are also encoded by DNA that is regulated by other genes, and there are all kinds of feedback and feedforward mechanisms going on here — genetic programs are the original nightmarish spaghetti code.

Which is why master control genes are attractive examples of regulatory logic. We can actually see a direction to the flow of control.

Here’s another cartoon illustrating the pattern. The master control gene is diagrammed in red (it has cis regulatory elements, too — I just haven’t drawn them). When it is switched on, it makes a red protein that binds specifically to a particular CRE that I’ve drawn as a simple rectangle. When the red master control gene product binds to the CREs for the green eye tissue gene and the blue eye tissue gene, they are switched on, and make a green and blue DNA binding protein in turn. I’ve also illustrated two common complications. The blue protein can bind to a CRE for the green protein, so we can have some cross communication; you can also have one of these gene products feedback and bind to the master control gene, increasing it’s level of activity. I’m not showing all of the possible feedback mechanisms to keep this picture relatively uncluttered.

Another important consideration is that directing a cell into a particular developmental pathway means, in addition to turning on tissue specific genes, you have to turn off inappropriate genes. I’ve drawn one example: a purple gene that encodes a transcription factor important in the development of the gonad. Building a testis in the eyeball is not an optimal outcome, so some gene activity works to silence alternative pathways. There are many of these inhibitory interactions going on at the same time.

My blue and green proteins are going to go on to bind the regulatory elements of other genes, which I’ve labeled eye1, eye2, eye3, and eye4 — there will be many more than four. These genes may be yet more transcription factors that regulate yet more genes; they can be structural genes specific for that tissue, such as eye-specific ion channels; or they can be various signaling elements that will play a role in interactions between cells in this tissue, instructing some to make the photoreceptors, others to make supporting cells, and so forth. The interactions can get very hairy, and I should have little gray arrows snaking all over the place to illustrate all the possible interactions — but then the cartoon really would look like a plate of spaghetti.

Another property of some master control genes is that they are relatively well conserved — the mouse version of Pax-6 can substitute for the fly Pax-6 and do much of its job (the reverse doesn’t work as well, though: the mouse has a messier, more complicated set of genes regulating eye initiation, so the experiments are harder to do and harder to interpret). Here’s a real example, a comparison of sequence similarity of Pax-6 in several different animals.

Another complication: regulatory proteins can have multiple DNA binding sites, and Pax-6 has both a paired domain and a homeodomain. The segments of sequence below show first the paired domain, then the homeodomain, and then the C-terminal domain. The two DNA binding domains show significant sequence similarity across whole phyla; note that the human paired and homeodomains have 96% and 95% identity, respectively, to the squid (Loligo) paired and homeodomains. The C-terminal is not quite so similar, only 47% identity, and the overall sequence identity between vertebrates and squid is 67%.

Comparison of
the amino acid sequences between paired-, homeo- and C-terminal domains
of vertebrate and invertebrate species. The squid sequence is shown in
full; for other sequences only differing amino acids are shown. A -indicates gaps that were introduced to maximize similarities in the
C-terminal domain; an * marks the end of the known nemertine sequence.
The percent identities are shown for the squid sequence.

We can demonstrate that a master control gene is expressed in an appropriate region with in situ staining, where colored probes are used to mark regions where Pax-6 RNA is found. The photos below illustrate Pax-6 expression in the embryonic squid; the earliest regions of expression demarcate the eye field, the subset of cells that will go on to make the various cell types of the eye.

Expression of Pax-6 during
L. opalescens embryonic development.
(A–C, G, and
I) Whole mount in situ hybridization of
squid embryos using Pax-6
(A–C and G) and
S-crystallin (I) antisenseriboprobes. Sense probes were
used as controls (data not shown).(D–F,
H, and J) Frontal, 10-µm plastic
sections of embryos hybridizedas whole mount with Pax-6
(D–F and H) and
S-crystallin (J) antisenseprobes (20). (A) Stage
17 embryo. Brackets indicate the eye primordia where first traces of
Pax-6 expression are detected.(B) Stage 20 embryo. Arrows point to the arms. (C) Stage 27 embryo.The
yolk sac was removed. Arrows point to the arms.
(D–F) Sectionsof stage 27 embryo.
(D) Section through the arms and suckers.(E)
Section through the brain adjacent to the posterior edge of the eyes.
Bracket indicates the cerebral ganglion. (F) Section through
the posterior end of the brain. (G) Stage 27 embryo. Bracket indicates eye and arrow indicates olfactory organ, where expression of
Pax-6 is detected. (H) Section through the
eye of an embryo as in G. Brackets indicate areas from
which the cornea will develop as a fold from the edge of forward
growing arms and arrows point to lentigenic cells which do not express
Pax-6. (I) Stage 27 embryo, hybridized with
S-crystallin probe. Bracket indicates eye where expression of
S-crystallins is detected. Note that there is no expression in the
olfactory organ (arrow). (J) Section through the eye of an
embryo as in I. Brackets and arrows are as in
H.Label at the inner edge of the eye chamber is
background staining also observed with the sense probe. (Bar in
J = 300 µm in A andB, 150 µm in C, 100 µm in
D–F, 200 µm in G and
I, and 50 µm inH and
J.) Abbreviations: a, arms; b, brain; e, eye; ep, eye
primordia;i, iris; l, lens; m, mantle; ol, optic lobe; ols, outer lens
segment;ov, optic vesicle; r, retina; ys, yolk sac.

We can also test whether Pax-6 is necessary for eye formation by mutating it. This can be done fairly easily in flies; reducing expression of the genes that trigger eye formation in Drosophila produces flies that are either completely eyeless or have very tiny eyes. The experiment is harder to do in vertebrates — knocking out mouse Pax-6 completely is lethal, causing eyes to fail to form but also causing other cranial and brain abnormalities. Heterozygous mutations cause subtler phenotypes, all tangled with abnormalities of the eye, such as aniridia.

Is Pax-6 sufficient to induce an eye? If Pax-6 actually is a master control gene, than just flipping it on ought to induce eye formation in any region of the animal. We can do that by taking the coding region for Pax-6 and splicing it to a different control region, one that is either always switched on or one that we can regulate. If an aberrantly active Drosophila Pax-6 gene is injected into the wing disc of a larva, it will go on to recruit cells at the base of the wing to form an eye (B, in the photo below). These ectopic eyes can be induced in all kinds of interesting places in the cuticle where the gene can be switched on by human intervention.

The A photo is even more interesting. That is an ectopic eye at the base of the wing that was induced by squid Pax-6, not fly Pax-6. It shows that not only can Pax-6 initiate formation of an eye, but that particular function of the protein has been conserved for over half a billion years of evolution. It also suggests that the last common ancestor of arthropods and molluscs (and chordates; this experiment also works with mouse Pax-6) formed some kind of localized sensory organ under the control of the ancestral Pax-6 gene.

The ectopic expression of L. opalescens
Pax-6 cDNA1 by means of the GAL4 system induces the
formation of ectopic eye in Drosophila.
(A and B) Scanning electron micrograph of
an ectopic eye induced on the wing by misexpression of L.
opalescens Pax-6cDNA1 (A) and Drosophila
eyeless (B).

Of course this experiment with a squid Pax-6 gene only produces a fly compound eye. This is because all that has been transplanted from one species to another is the master control gene, a gene that is way upstream from the actual eye construction process, and which serves only to initiate eye formation. Referring to the spaghetti diagram up top, the experiment takes the red master control gene from a squid and puts it in a fly, where it turns on fly eye tissue gene 1, fly eye tissue gene 2, and fly eye1, eye2, eye3, and eye4. Putting a squid eye in a fly is almost certainly impossible: it would require transplanting the whole plate of spaghetti rather than just one small strand, and since all of the genes involved have multiple interactions with other gene networks, would disrupt the whole of the organism in unpredictable and devastating ways. The secret of the success of this experiment with Pax-6 is that master control genes require only a small nudge to switch on a well-integrated cascade of genes further downstream.

Tomarev SI, Callaerts P, Kos L, Zinovieva R, Halder G, Gehring W, Piatigorsky J (1997) Squid Pax-6 and eye development. Proc. Natl. Acad. Sci. USA
94:2421-2426.