麻豆传媒

鈥淗ERE, sniff this.鈥� The smell wafting from the open ampoule being held out by
Luca Turin suggests an unpleasant mixture of rotten eggs and onion soup. It has
to be a sulphur compound. 鈥淲rong!鈥� says Turin gleefully, revealing a formula
devoid of sulphur atoms. It鈥檚 a trick he has played on many eminent chemists,
and they have all fallen for it.

The mystery compound is decaborane, and its molecular structure looks nothing
like any of the chemicals usually responsible for eggy odours. It does have one
thing in common with them though鈥攊ts bonds vibrate at the same frequency.
And this, says Turin, is why they smell so similar. Among scientists studying
our olfactory sense, it鈥檚 a claim that is kicking up a bit of a stink.

Most researchers believe that a chemical鈥檚 smell comes from properties such
as the shape, chemical structure or electrical charge of its molecules. The idea
is that molecules with different shapes or structures dock at different
receptors in the nose, just like a key fitting into a lock, triggering a signal
that is passed on to the brain.

The problem is that no one has managed to develop this principle into a set
of rules for predicting how molecules should smell. Fragrance companies that
supply the ingredients for perfumes, aftershaves, deodorants and household
cleaners spend a fortune searching for new ingredients, and would dearly love to
have a set of guidelines to help them. Turin, a biophysicist in the anatomy
department at University College London, believes that researchers looking at
the shape and structure of smelly molecules are in a blind alley. He is
convinced that what matters is the frequency at which a molecule鈥檚 bonds
vibrate.

This is controversial stuff, and some researchers refuse even to consider
Turin鈥檚 ideas. Yet he has amassed plenty of supporting evidence. It was the
odour of decaborane that convinced him there might be something in the
vibrational theory. All chemical bonds vibrate, just as two balls joined by a
spring do. The frequency of this vibration (conventionally expressed in units
called wave numbers) depends strongly on what atoms or groups of atoms are at
either end of the bond. Most compounds that smell of rotten eggs or onion soup
contain a sulphur atom (S) connected to a hydrogen atom (H). An SH bond vibrates
at a frequency of about 2500 wave numbers. The only other bond to share this
frequency is boron-hydrogen (BH), found in decaborane and other boranes. 鈥淭he
problem with boranes is that they are rocket fuels,鈥� says Turin, explaining why
people hadn鈥檛 made a habit of sniffing them. 鈥淚t would be . . . it smells of
`boom鈥� 鈥�.

As it happens, decaborane isn鈥檛 quite that dangerous, and when Turin took a
sniff he was amazed to find that it smelled like a sulphur compound. Yet
chemically they are a world apart: if nasal receptors recognise a molecule鈥檚
shape and chemical structure, those that recognise decaborane would not
recognise sulphur compounds and vice versa. Somehow, two separate sets of
sensors would have to be 鈥渨ired鈥� to give a sulphurous smell in the
brain鈥攁nd that鈥檚 impossible, says Turin.

Others disagree. 鈥淪imply because two different molecules may interact with
different receptors does not necessarily mean that they will smell different,鈥�
says Joel White, an olfaction researcher in the neurosciences department at
Tufts University in Massachusetts. Different sets of signals from the receptors
could be interpreted to give the same smell. 鈥淧erception of odour goes on in the
brain,鈥� agrees Charles Sell, head of the fragrances discovery group at Quest
Fragrances in Ashford, Kent. 鈥淭he brain puts together an odour based on the
pattern of signals.鈥�

But Turin has an answer to these objections. Even if two separate sets of
sensors could stimulate the same perception in the brain, surely it would be
impossible for a single set of sensors to give rise to the perception of two
different smells. Yet this is what would have to happen in some cases, if size
and chemical structure are the all-important features. Take the compounds
ferrocene and nickelocene. Both molecules have a metal atom sandwiched between a
pair of hydrocarbon rings: in the case of ferrocene, the metal is iron, while in
nickelocene it is nickel. Both metal atoms are roughly the same size.

鈥淔errocene has a rather spicy, lovely smell,鈥� says Turin. But nickelocene
鈥渟mells of the carbon and hydrogen rings rather than anything else鈥�. This must
mean that the two molecules cannot trigger the same receptors, he says. And one
crucial difference between the two is the vibration frequency of the bond
between the metal atom and the rings. In ferrocene, it vibrates at 478 wave
numbers, but in nickelocene the frequency is 355 wave numbers. Turin thinks that
any bonds vibrating below 400 wave numbers get lost in 鈥渂ackground noise鈥�, and
that nickelocene therefore fails to trigger the receptor that gives ferrocene
its spicy aroma.

Turin claims more support for his vibration theory from a another pair of
molecules that smell different despite being almost the same shape and
size鈥攁nd chemically virtually identical. These are acetophenone and
deuterated acetophenone, in which normal hydrogen atoms are replaced by the
heavy hydrogen isotope, deuterium. These heavier atoms make the bonds vibrate at
lower frequencies. This should have an effect on the smell of a molecule, Turin
reasons.

Bitter whiff

Sure enough, he has found that deuterated acetophenone is fruitier than
acetophenome, and has a whiff of bitter almonds. Carbon-hydrogen bonds vibrate
at about 3000 wave numbers, while carbon-deuterium bonds vibrate at 2200 wave
numbers. Turin points out that this is also the vibrational frequency of nitrile
(CN) groups, which are present in many compounds that smell of bitter
almonds.

Not everyone agrees with Turin鈥檚 conclusions. 鈥淚t could be possible to
account for the differences in smell of ferrocene and nickelocene on the basis
of electrostatic and solvent interactions within the receptor pocket,鈥� says Ian
Connerton of the Institute of Food Research in Reading, who is studying
olfactory receptors. Something similar could also be true for acetophenone and
deuterated acetophenone.

These disagreements highlight how little we know about the mechanism of
smelling. One thing everyone agrees on is the way signals get from the nose to
the brain. Our noses contain several million receptors, of 500 to 1000 different
types, in an area called the olfactory epithelium at the back of the nasal
cavity. The receptors are long, chain-like proteins that are folded to create
docking sites for odour molecules. Each receptor is coupled to a partner, called
a g-protein. The first step in sending a signal to the brain is for the receptor
to release the g-protein; this triggers a second messenger which stimulates a
neuron into sending a signal to the olfactory bulb, where signals are sorted and
sent to the brain for processing.

The main controversy is about the first step. The leading theory is that a
receptor changes shape when an odour molecule docks, and that this is what makes
it release the g-protein. But according to Turin鈥檚 theory, receptors work by an
electron-transfer process: an electron uses a docked odour molecule to 鈥渢unnel鈥�
from a donor site to an acceptor site on the receptor protein chain, changing
the molecule鈥檚 chemistry so that it can no longer hold on to the g-protein.

Electron tunnelling normally only happens when the donor and acceptor sites
are at the same energy. Turin thinks that the electron-acceptor site in receptor
proteins is at a lower energy level than the donor, preventing the electron from
tunnelling unless it can lose some of its energy
(see Diagram). The odour
molecule allows it to do just this. 鈥淭he electron leaves the donor site, hits
the molecule and causes it to vibrate,鈥� says Turin. 鈥淚t loses some of its energy
to that vibration, and can then cross to the acceptor site.鈥�

Higher or lower frequencies of vibration will cause an electron to lose more
or less energy. So to make receptors that respond to different frequencies, all
you need to change is the energy gap between donor and acceptor sites. Turin
reckons this could be done by changing a few amino acids in the protein chain.
He thinks that the 500 to 1000 different receptor proteins in the human nose
fall into about 10 to 15 classes, according to the frequencies they detect.FIG-mg21154803.JPG

Having come up with this model, Turin set out to find evidence to support it.
He says he has found a sequence of amino acids in receptor proteins that could
act as an electron donor site. More importantly, he thinks he has identified the
acceptor site鈥攁 sequence of five amino acids bound to a zinc atom. 鈥淶inc
is known to be intimately involved with olfaction,鈥� says Turin. 鈥淚f you are zinc
deficient, the first thing that goes is your sense of smell.鈥� He also points out
that zinc is involved in other electron-transfer processes in biology.

It is this part of Turin鈥檚 theory that has drawn his critics鈥� most hostile
fire. Says White: 鈥淭he effects of zinc on olfaction could be due to any number
of mechanisms, not necessarily at the receptor level.鈥� Randall Reed, who
specialises in the molecular genetics of olfaction at Johns Hopkins University
in Maryland, agrees. 鈥淔or whatever reason, olfaction requires zinc,鈥� he says,
鈥渂ut there are many levels at which that could be manifest.鈥�

Despite this criticism, Turin鈥檚 receptor model has had its triumphs, too. One
big challenge for vibrational theories has been the enantiomers R- and
S-carvone鈥攁 mirror-image pair of molecules that are otherwise identical
and yet smell different. R-carvone smells of mint, while its S-partner smells of
caraway. If receptors work by sensing shape, this difference is easily
explained. 鈥淓nantiomers can be viewed by biological receptors as completely
different molecules,鈥� says White鈥攖he left-handed molecule won鈥檛 fit into
the same receptors as the right-handed version. And this view has no problem
explaining why other pairs of enantiomers can smell the same: small parts of the
molecules that are the same in both enantiomers could be the shape trigger for
the receptors.

The problem for Turin was that vibrational properties of enantiomers are
identical. 鈥淚t turns out this has a trivial answer,鈥� he says鈥攖he
vibrations are only identical if the molecules are in solution, where they are
oriented at random. If they are fixed in one orientation, as they are in the
nasal receptors, and if some of the vibrating bonds lie perpendicular to the
direction of electron tunnelling, these bonds won鈥檛 be detected, says Turin. The
electron will not reach the acceptor.

Right direction

Most molecules are flexible鈥攖he bonds can twist until they point in the
right direction for tunnelling. But in the carvones, the carbonyl bond joining
an oxygen atom to a carbon atom is fixed rigidly in a different position on each
enantiomer. Turin thinks that in caraway carvone, the carbonyl lies in the path
of electron tunnelling, and is picked up. In mint carvone, however, the bond
lies at right angles to the tunnelling direction, and isn鈥檛 detected. 鈥淪o take
mint carvone and add some carbonyl vibrations back by adding another molecule,鈥�
says Turin, 鈥渁nd you should get a caraway character.鈥� In an experiment with
French perfumers, this is exactly what he did鈥攁nd sure enough, mint
carvone mixed with butanone, which contains a carbonyl group, gave a caraway
smell.

Turin is now concentrating on trying to predict how new molecules will smell.
鈥淚f he can predict the smell characteristics of new odorants, the theory is of
some value,鈥� says Connerton. Fragrance companies such as Quest and Givaudan,
which is part of the Roche group, have already shown interest in using Turin鈥檚
theory to design new ingredients. This is a tough job. If Turin鈥檚 theory is
right, it shouldn鈥檛 be too hard to predict what vibrational spectrum will
produce a given smell. The tricky part is to come up with a brand new chemical
that has the same spectrum, and hence the same pleasant smell, as an existing
perfume ingredient. 鈥淚t鈥檚 a matter of how different the spectra are,鈥� he says.
Even small differences can have large effects on the smell, depending on the
fragrance involved. 鈥淚magine the multidimensional space of smell as a map,鈥� says
Turin. 鈥淪ome fragrances, such as rose or lily of the valley, might be like
Switzerland鈥攜ou go 50 miles in one direction, you could end up in
础耻蝉迟谤颈补.鈥�

Any successful theory linking structure to smell could speed up the
production of new fragrance ingredients. At present, many different compounds
have to be made for every successful candidate. Turin鈥檚 idea isn鈥檛 the only one
being tested though. 鈥淭here are already a lot of theories about structure and
activity,鈥� says Ton van der Weerdt, head of research and development at
Quest.

The company has its own huge database of smells from the past 40 years that
could provide some evidence for which of the theories are feasible. 鈥淲e鈥檝e been
trying to find a definitive way of proving or disproving all theories,鈥� says
Sell. 鈥淏ut the answer will come from biology.鈥�

In the meantime, Turin will continue to try to make more fragrant ingredients
using his theory. 鈥淧erfumes are to smells what music is to sound,鈥� he says.
鈥淭hey are stronger and more beautiful.鈥� If his theory is right, fragrance
chemists in the future will be relying on vibration spectra to compose their
symphonies.