At the same time, the 20 words are different enough that combining them can
make bats, bees, birds and bacteria. If
the amino acids are LEGOs and the task
is to build things for 4 billion years, you
had better have a diverse set, with rectangular blocks, joints and wheels, says
astrobiologist Stephen Freeland of the
University of Hawaii at Manoa.
There are plenty of other ways to
arrange the code. By the calculations
of a team including Peter Clote, now of
Boston College but formerly of Ludwig-Maximilians-Universität München in
Germany, there are 1084 alternate codes
that assign at least one codon to each of
the 20 amino acids (and “stop”). So it
would be weird if, by complete chance,
the code had such clever traits.
Tlusty thinks that if the genetic code is
so good at what it does, it’s because it
adapted under evolutionary pressure,
the way the beaks of some of Darwin’s
finches developed to be hefty seed crackers or long flower probes depending on
Three pressures, Tlusty argues, could
have shaped the current genetic code.
First, typos can’t be disastrous — if a random mutation changes one of the letters,
the cell should still infer which spelling
was intended. Second, the language
must be able to spell words with diverse
Error-proof code early on, the genetic
code may have relied on just two letters in its
three-letter words. a doublet code proposed
to have gotten things going (green) suffered
less cost from errors than other possible
doublet codes (average error cost for all codes
is in purple on a roughly bell curve).
Cost of errors among doublet codes
Number of codes
Relative cost of errors
meanings. Third, the language shouldn’t
take a lot of resources to write, forcing
the cell to make tons of extra molecules.
Assuming an evolving code, and assuming four letters in the alphabet and three-letter words, Tlusty tried to figure out an
ideal number of amino acids. He imagined the code as mapping onto a many-dimensioned doughnut of sorts, with all
64 possible codons spaced out so that
those that could easily be confused with
each other are neighbors. Then he tried to
find how many colors, or amino acids, are
needed to make a map that obeys his three
demands. In this way, Tlusty rephrased
a biological question (How many amino
acids would a changeable code settle
on?) as a classic mathematical one: What
is the fewest number of colors necessary
to color geographical divisions on a map
without any colors touching themselves?
On a two-dimensional map, like a map
of the United States, the answer is four
colors. In the higher dimensions necessary to map the code, the range of colors
is between 20 and 25, Tlusty reported
in Physics of Life Reviews in September.
That’s spot on for how many amino acids
are in today’s code.
Tlusty’s findings support his idea
that a changing code could have settled
on an optimal number of amino acids.
Another team suggests that an earlier
code, one that preceded today’s, could
have been a superstar at one of Tlusty’s
If randomly generated codes are placed
on a landscape, with higher elevations
designated for codes that are better at
preventing errors in protein manufacture, life’s code would be found on the
side of an unassuming hill. But a previous code could have been atop one of the
highest peaks, Koonin argues.
Because changing the third letter of
a codon doesn’t drastically change the
amino acid, Koonin and others think
that an early genetic code relied on only
the first two of its three letters. Such
a code could have expressed at most
16 amino acids. Koonin’s team backtracked the code to its most plausible
two-letter origins, playing with different codon assignments. In Biology Direct
in 2009, the researchers reported that
several such codes were exceptionally
robust against translation errors, suggesting that minimizing error drove
the code’s development early on. When
having more than 16 amino acids was
advantageous (allowing for more types
of proteins), the code started to use the
third letter — getting a little worse at
Chemistry behind the code
While agreeing that there is nothing
accidental about the code, other scientists suspect chemistry was a more
important driver. They are turning to
experiments in modern labs to try to
determine which amino acids came first.
In a legendary spark-tube experiment
in 1953, Stanley Miller created a handful of life’s amino acids by electrically
zapping a chamber filled with hydrogen, water, methane and ammonia gases.
And similar follow-up experiments have
yielded even more amino acids. A meteorite that crashed in Australia in 1969
contained some of the same amino acids,
suggesting that they were forged somewhere in the solar system.
Five amino acids made in spark-tube
experiments and found within meteorites— glycine, alanine, aspartic acid,
glutamic acid and valine—appear to
be related. Each of their codons begins
with a G, suggesting that whatever word
coded for the first amino acid, G may have
grabbed the first position.
“It’s sort of a hand-waving argument,”
says Paul Higgs, a bioinformaticist at
McMaster University in Hamilton, Canada. “I don’t know if you can really prove
that.” But in the absence of a detailed
chronology, deep hunches have led Higgs
and others to create models that begin
with these five amino acids. In 2009 in
Biology Direct, Higgs set out a plan for
how the remaining amino acids would be
added after the first handful were fixed.
Others argue, though, that the first
amino acids weren’t the most abundant,
but instead were the ones with a natural
chemical attraction to RNA, which some
scientists think got life off the ground
(SN: 7/3/10, p. 22).
adapted by janel kiley