Download by David Holzman Unlike its twin

Ribonucleic acid is the middleman in
the process whereby deoxyribonucleic acid, the primary genetic
material, is translated into protein, the
structural and functional material of all
life. As is appropriate to a middleman,
the study of RNA was all but neglected
until the last decade. 'Tor years we've
been thinking of RNA as nothing but an
information tape or a way of stringing
proteins together/ says Norman Pace of
the National Jewish Hospital in Denver.
"Now we're finding it's a much more
powerful molecule than that/'
In both DNA and RNA the sequence of
nucleotides, or bases, which are the letters of the genetic alphabet, encodes the
recipe for protein. Each protein is encoded in a stretch of bases that makes up
a gene. In DNA, other base sequences
appear in a region located upstream
from each gene or group of genes. These
sequences are the signals that regulate
the production of protein by controlling
the transcription of DNA Into RNA. Scientists also are finding that RNA often can
provide its own instructions for the synthesis of protein, that is when to make
protein and how much to make. Rather
than the arrangement of bases, however,
It Is the structure of the RNA, the way it
folds Itself into helices, that embodies
much information directing a cell's functioning. It is as if kinks appeared in the
tape to start a tape recorder when it was
time for the music to play and again
when it was time to stop.
The difference between RNA and DNA
that makes structure important in RNA
and nucleotide sequence important in
DNA is that RNA is single-stranded and
DNA is double-stranded. The two strands
of the DNA double helix connect at every
link In the chain of nucleotides. The result is a stiff and unreactive molecule,
one that must communicate principally
through base-sequence codes. By contrast, because it is single-stranded, RNA
can use form to function as well. Singlestranded RNA folds upon Itself, becom-
34 MOSAIC
by David Holzman
Unlike its twin-stranded cousin,
RNA uses structure
to help it perforin its function.
ing double-stranded In places and forming three-dimensional structures called
helices that are able to convey their own
messages in the cell.
Base pairing produces the helices in
RNA. In the DNA base pairing, adenine (A)
binds to thymine (T), and guanine (G)
binds to cytosine (c). [In RNA, uracil (u)
replaces thymine.] The twin-stranded
DNA exhibits full complementarity; every
base in each strand is paired. Helix formation occurs in RNA when a moderate
amount of complementarity exists between two nearby pieces of the same
strand, which twists on itself as complementary bases pair.
These helices are called the secondary
structure of RNA. They are represented
as two-dimensional structures called
hairpins. In addition RNA forms tertiary
structures when single-stranded loops
on the ends of hairpins pair with other
loops on the strand, orienting the hairpins three dimensionally.
just as rivers descend from the heights
and lose potential energy as they drop,
the pairing arrangements in RNA may
shuffle spontaneously until the molecule
lands in the structure with the least free
energy, a thermodynamic sea level.
TInoco measured the strengths of helices having different base sequences.
This task was not easy, because the
strength of any pair, say A with u, is
influenced by what the adjacent bases
are, and even by the neighbors beyond
those. These pairing strengths can then
be integrated into a computer program
that can find the most stable pairing arrangement for an RNA molecule.
Although Tinoco's rules can give a reasonable first approximation of an RNA
structure, they do have several serious
shortcomings. First, Investigators have
not tested all possible short sequences,
so the rules are based on considerable
extrapolation. Second, the strength of
the pairing is not the sole determinant of
structure. The order in which an RNA
chain forms affects structure, because
bases form pairs as the RNA chain grows;
they do not wait until the chain is fully
formed. A particularly strong helix early
in the chain could prevent the molecule
from reshuffling its helices into a more
stable structure when fully formed. A
third shortcoming of Tinoco's rules is
that they do not analyze tertiary structure. Computer attempts to impose a
two-dimensional structure on a molecule, neglecting three-dimensional pairing, frequently yields false results.
Tinoco's rules
Thermodynamic analyses of this kind,
based on comparison of the strengths of
the bonds between molecules, are applied quite successfully to secondary
structure where they can explain many
observations of the regulation of genetic
functioning by RNA. This is because the
Ignacio TInoco of the University of California at Berkeley has devised a set of
rules for predicting the structural folding
of an RNA from the sequence of its bases.
Tinoco's rules are based on the idea that
Stop sign. Stem and loop at the end of an
RNA transcript is thought to signal
transcription termination. Asterisks indicate
sites of norma! mutations, which can disrupt
the structure and eliminate termination.
C~G
C-G
A~U
GUUCAG
regulatory end of an RNA strand appears
to be too short to fold on itself, and so it
retains its secondary structure. It is also
because ribosomes translating the RNA
genes can prevent tertiary interaction
with the rest of the chain, maintaining a
tractable two-dimensionality.
Although some biochemists speak
condescendingly of computer studies of
RNA folding, it is possible to piece together quite accurate secondary structures by supplementing computer studies with biochemical studies. If linoco's
rules say a helix occurs at a certain point
in the RNA, an enzymatic probe that cuts
RNA only at double-stranded sites will
show whether the helix is really there.
Genetic analysis, testing the effect of a
mutation in paired bases on the strength
of the helix they help to form, can
provide the same information, but with
less certainty.
Stop signs
In messenger RNA a region called the
leader controls gene expression. Once
the enzymes, called polymerases, transcribe the RNA leader from DNA, the leader folds itself into shapes that the enzymes recognize as signals. The most
basic signal is a small hairpin with a loop
at Its end. The enzyme, recognizing the
loop structure as a stop sign, pauses
briefly before being directed further In
the transcription process. If the products
of the genes are not needed, the process
ends and the RNA separates from Its template. If they are needed, transcription
continues Into adjacent genes.
36 MOSAIC
Researchers had started to come upon
these stop signs, called terminators, in
the mid-1960s. That was long before anyone suspected that RNA structure had
anything to do with the mechanics of
terminators. Charles Yanofsky of Stanford University had a typical experience.
When his student Ethel Jackson deleted
genetic material in the then-mysterious
region between the promoter, which
binds the enzyme to start transcription,
and the genes, it was as if she had pulled
a plug. Suddenly the group of genes following the promoter was making five
times as much protein as predicted by
what were then current theories of gene
regulation.
Nevertheless terminators attracted little notice for some years. In the early to
middle 1970s, recalls Martin Rosenberg,
now of Smith Kline and French Research
Laboratories in Philadelphia, "everyone
thought that most regulation occurred at
the promoter. There were about 100 scientists studying promoters, and Charlie
Yanofsky and I were studying terminators. Everyone thought terminators
were just going to be the period at the
end of the sentence/'
Rosenberg made the first definitive
connection between the structure and
function of terminators in 1974. He had
been studying termination in p h a g e
Lambda, a favorite laboratory virus. In the
phage one terminator controls a set of
genes that is necessary for a stage in the
phage's life cycle. Only a small percentage of phages enter this stage, and the
terminator helps control the relative
numbers by having what can be considered a slow leak.
Rosenberg had two sets of mutant
phages. One Increased termination,
leading to fewer phages entering this
second stage of development, and one
reduced it, leading to more phages entering. Using newly developed techniques, Rosenberg determined the sequence of nucleotides in the terminator,
hoping to find clues to its operation. He
discovered two strikingly complementary sets of bases in the terminator,
implying that a helix might form. More
importantly, he discovered that the termination-enhancing mutant changed a
genetic letter to strengthen the helix,
while the leaky mutants weakened it.
Over the next few years the evidence
grew stronger that hairpins cause termination. Biochemical experiments proved
beyond a doubt that the structures
formed in terminators. Terry Piatt, at Yale
University, inserted synthetic hairpins
into the middle of a gene and found that
the stronger the hairpin, the greater the
frequency of termination.
Pausing
These experiments demonstrated that
hairpins are involved in termination, but
not how they are involved. Two experiments in the middle to late 1970s led to
the current model of how termination
takes place. There are two kinds of terminators, one requiring a protein to catalyze the separation of nascent RNA from
the DNA template, and the other having
no such requirement; this protein,, discovered by Jeff Roberts, was called rho.
In one experiment, Rosenberg found
that in the absence of rho, a polymerase
would pause at a rho-dependent terminator for 30 to 40 seconds. Then It would
continue making RNA. In the second,
Berkeley's Mike Chamberlin showed that
this pausing is not restricted to termination sites, but occurs wherever messenger RNA can form helices.
This discovery led to the idea that hairpins cause pausing. Terry Piatt describes
the generally accepted speculation about
this process: Recent evidence strongly
suggests that at any one time about 12
nucleotides of a growing RNA chain are
always paired with the DNA template inside the RNA polymerase. Then, says
Piatt, hairpin formation rips the RNA
from its DNA template. The resulting helix jams the enzyme.
The pause is thought to provide the
time necessary for the RNA chain-release
reaction to take place. In the rho-dependent terminators, this amounts to giving
rho enough time to interact with the
polymerase complex and to catalyze
chain separation. In the rho-independent terminators, the m e c h a n i s m is
more obvious. These terminator hairpins are followed Immediately by a string
of u bases, which, as determined by
Francis H. Martin, now of Applied Molecular Genetics Incorporated, and Ignacio TInoco In 1981, provide an extraordinarily feeble link to the As of DNA
template. In r h o - i n d e p e n d e n t terminators, therefore, the pause is thought to
halt transcription once the string of us
has been transcribed and while it is still
paired Inside the enzyme with Its DNA
template. The A-u bonds rupture, causing the RNA to fall away from the DNA.
W h e n expression of d o w n s t r e a m
genes Is needed, transcription continues
past the terminator. In some cases this
can h a p p e n because antitermination
proteins override the terminator. In other
cases, where terminators function like
leaky plugs to moderate the number of
downstream genes to be expressed, terminators probably fail to cause pausing.
This may happen because either the terminators fail to fold during transcription
or they fold then spontaneously unfold.
Either case would be consistent with evidence that frequency of termination is
related to the strength of the helix.
Attenuation
In a special class of terminators called
attenuators, the RNA takes a much more
MOSAIC 37
active role in its own transcription. The
heart of the attenuator is still the basic
terminator, b u t instead of being acted
upon by antitermination proteins, the
terminator competes with an intrinsic
antiterminator for use of some of the nucleotides letters of its helix. When the
bacterium n e e d s the products of the
downstream genes—which in one thoroughly studied case are the enzymes that
make the amino acid tryptophan—the
antiterminator helix comes together.
This action masks some of the nucleotides that would normally fold into
the terminator helix and permits transcription to proceed.
Two types of attenuator have been
found. One regulates protein synthesis
by controlling transcription, and the
other regulates by controlling translation. The mechanics of the process are
intricate: In each case individual segments of RNA fold to produce either a
terminator or an antiterminator, but never both at the same time. Which pattern
occurs depends in part on the circumstances of the cellular environment.
One example of transcriptional attenuation involves a gene that makes the
enzymes that synthesize the amino acid
tryptophan. In this case the production
of the amino acid is tied to the level of
tryptophan in the cell. High levels of the
amino acid favor the formation of a terminator, and low levels favor the formation
of an antiterminator. When an RNA
strand assumes the terminator shape,
the polymerase falls off the leader instead of transcribing the downstream
genes necessary for continued production of tryptophan. When tryptophan
levels fall a n d the m e s s e n g e r RNA
assumes the antiterminator shape, then
downstream genes are free to be transcribed and tryptophan production can
increase once again. Even when tryptophan is a b u n d a n t , however, these
genes are not always blocked; tryptophan production will never shut down
completely.
An example of translational attenuation occurs in bacteria resistant to the
antibiotic erythromycin, a drug that poisons ribosomes. In the absence of the
antibiotic, a terminator forms in these
bacteria, blocking translation of RNA to a
protein that makes ribosomes immune to
erythromycin. As is the case with trypt o p h a n , however, p r o d u c t i o n never
Holzman is a Washington-based science writer who works most frequently in biology.
38 MOSAIC
message by surviving ribosomes may begin. These survivors then translate immunity protein in bulk, producing more
immune ribosomes and quickly building
up.the bacterium's resistance.
Kinetic analysis
shuts down completely. Some immunity-producing protein is always incorporated into a few ribosomes, making
them immune to erythromycin. The few
immune ribosomes produced in the absence of erythromycin become important later when the drug, floods the cell.
Then messenger RNA assumes its antiterminator shape, and translation of the
Despite the success scientists have
had in explaining attenuation and termination with t w o - d i m e n s i o n a l t h e r modynamic analysis, some recent observations in messenger RNA defy analysis.
Most scientists studying RNA now think
that a complete u n d e r s t a n d i n g of Its
properties will require a better knowledge of the relative importance of kinetics and t h e r m o d y n a m i c s In RNA
structure formation, a better knowledge
of which proteins interact with messenger RNA and how they Interact, and
an understanding of the three-dimensional, tertiary structure of RNA.
The rate, or kinetics, of the chemical
and biochemical reactions may influence
the outcome of regulation by messenger
RNA as much as or more than the predictions that could be made by applying
Ignacio Tinoco's thermodynamic rules.
Tinoco's rules predict the structure an
RNA would assume if it had all the time in
the world to descend to thermodynamic
sea level. Regulatory events such as termination and attenuation, though, take
place in fractions of a second on chains
that are growing at 50 or 100 nucleotides
per second and folding into helices three
orders of magnitude more quickly.
"We don't know exactly what structures are being formed as the RNA is
made," says Berkeley's Mike Chamberlin. The data certainly s u g g e s t ,
however, that thermodynamics is not the
whole story. Yale's Terry Piatt had been
puzzled by the observation that a single
nucleotide mutation In the stem of a hairpin that changed the strength of the hairpin only slightly could drastically reduce
the frequency of termination at that site.
"I didn't understand this until I realized
that hairpin formation is almost certainly
a kinetic phenomenon," he says. "If you
introduce a single base-pair mismatch,
It's like missing a tooth in a zipper."
Chamberlin thinks that is putting the
case too strongly. He has watched many
RNA chains grow In vitro, In cell-free,
artificial systems, in his attempts to correlate the strength of hairpins with the
percentage of RNAs that pause at each
and the duration of the pauses. "None of
the hairpins we see implicated in pauses
are perfect hairpins," Chamberlin says.
"If Piatt were right, we wouldn't expect
any of those to cause pausing. I think the
effect could be kinetic but there's obviously a thermodynamic component."
In recentf studies Chamberlin, as well
as Don Mills and Susan LaFlamme at
Columbia University's College of Physicians and Surgeons, have found strong
hairpins without associated p a u s e s .
These findings have led to more quandaries about what structures are significant
in a dynamic context. Chamberlin's data
come from his further studies of growing
messenger RNA chains, while Mills and
LaFlamme's data come from experiments
that LaFlamme did while preparing her
Ph.D. dissertation.
The Columbia researchers used recombinant DNA techniques to make a
DNA template for a small RNA that is normally copied from its own RNA template.
The molecule is not a messenger RNA,
which means that its structure is unencumbered by ribosomes translating protein. The entire chain is free to bunch
and tangle. In the experiments the growing chain paused at only one of the molecule's many strong hairpins.
Hairpins, Chamberlin p o i n t s out,
"may not form rapidly enough to be sig-
clear magnetic resonance will ultimately
determine what structures are present in
the growing RNA chains. This determination will make it possible to ascertain the
relative contributions of kinetics and
thermodynamics.
Cause for pause
nificant in terms of the elongating complex." His observation is particularly salient because without the ribosomes on
the molecule, stable structures could be
preventing new helices from forming
quickly at the growing end of the RNA.
Chamberlin says that studies with nu-
These future studies should clear up
the mystery of hairpins without pauses,
but scientists are unsure about where to
look for solutions to the mystery of
pauses w i t h o u t h a i r p i n s . Mills and
LaFlamme have found two consecutive
pause sites downstream from a hairpin,
although the sites were not related to that
or any other hairpin. They tested this
finding by removing the hairpin, inserting in its place 51 bases lacking two-dimensional structure, and removing most
of the molecule downstream from the
pauses. During these manipulations
they left intact only a small number of
bases on either side of the pauses. The
pauses persisted throughout these manipulations. The Columbia researchers
believe that local factors within a few
bases of the pause sites were responsible. The most specific explanation Mills
can offer, though, is the possibility that
Protein-binding sites in RNA
Some proteins have jobs in the cell that involve binding
DNA or RNA. Many of these proteins regulate their own
number in a unique way. When they become so numerous
in the cell that they have saturated their natural binding
sites, they begin to bind to their own messenger RNA in a
way that prevents their own synthesis by ribosomes. This
binding happens because, in their messenger RNA, repression sites have apparently evolved to imitate the structure
of normal cellular binding sites. This molecular mimicry is
imperfect, though, so the proteins do not begin binding
their messages until ail their normal binding sites are
saturated.
The proteins appear to find their RNA binding sites by
recognizing the secondary structure but not the order of
bases in helical regions. In experiments on a repressor site
in the messenger RNA of phage R17, Olke Uhlenbeck of the
University of Illinois changed just about every nucleotide
in turn to find how proteins recognize their binding sites.
When he changed any base in a way that destroyed pairing,
the protein failed to bind. Changing two bases at a time in a
way that preserved the pairing, however, allowed the protein to bind to the site with normal strength.
Only four of the single-stranded residues, or segments,
were necessary for binding, explains Uhlenbeck, so "the
protein recognizes the hairpin by looking at the singlestranded residues held in a precise orientation by the helix." He was able to test this model by making a molecule
with a very different sequence but the same structure. He
showed that it still b o u n d protein, and he called it
"wierdmer."
This finding is consistent with the structural analysis of
nucleic-acid helix shape performed by Norman Pace of the
National Jewish Hospital in Denver. In the RNA helix, Pace
explains, the chemically unique properties of the bases are
sequestered well beneath the backbone of the molecule.
If proteins recognize RNA secondary structure, Larry
Gold of the University of Colorado thinks they can also
recognize its conspicuous absence. Gold has studied regulation of a protein whose job in the phage T4 is to bind to
single-stranded DNA during the phage's DNA-copying process, presumably to protect it from DNA-cutting enzymes.
When the growing DNA becomes saturated, excess protein
binds to its own message at a site that, like the DNA, is
conspicuous for its lack of structure.
Gold believes the protein interacts with the nucleic-acid
backbone, which is quite similar in DNA and RNA. "The
reason the protein can see the backbone is that there is no
secondary structure to interfere," says Gold.
Peter Model of Rockefeller University claims that the
protein recognizes sequences of nucleotides in DNA and
RNA, not the absence of structure. Gold does not think
Model is right, but he plans to make sure. He is constructing a wierdmer for this protein just as structureless as the
real thing but with an entirely different sequence. •
MOSAIC 39
the particular sequence of bases "could
result in some kind of stacking situation/'
Chamberlin, responding to pauses
without hairpins that turned up in his
own messenger RNA studies as well as in
the Columbia experiments, thinks that
"the fairest thing to say is we know hairpins can cause pausing, but whether
other structures can also do that or not is
up in the air. We have no idea what the
other structures would be. At this point
we're not even sure there are other structures, but we can't rule them out."
To complicate matters there are proteins and other molecules in living cells
that are known both to enhance pausing
and to depress it. Chamberlin had noted
early in his pausing experiments that RNA
grew much more slowly in artificial, cellfree systems than in living cells, and that
the difference was entirely attributable to
longer pausing. Postulating a pause suppressor in the living cell, he added a
small protein called nusA to. his templating broth. This action resulted in still
longer pausing. Adding an extract from
the cell shortened the pauses. The suppressor obviously appears to be there,
but Chamberlin still has not identified it.
He speculates that other extraneous cellular molecules may be responsible for
pause sites dissociated from hairpins,
perhaps by altering the way the transcribing enzyme sees structure.
Polarity sites
As is the case for hairpinless pause
sites, certain termination sites that require catalysis by rho also lack associated
hairpins. This special class of rho-dependent sites plays a unique regulatory role
in the cell. Such sites halt transcription of
RNA just as ordinary terminators do, but
in the middle of genes and only when
the just-transcribed RNA is defective.
The mechanism for termination at
these sites, which are called polarity
sites, depends on the fact that proteinmaking ribosomes in bacterial cells follow closely b e h i n d the polymerase,
translating protein from the newly
formed RNA. If a ribosome encounters a
nonsense codon, it falls off the RNA.
Transcription then is terminated at the
next available polarity site. Scientists
suspect that termination can h a p p e n
only after the ribosomes drop off and
their presence no longer blocks access of
the rho protein to its target terminator.
Though all this in part explains termination at polarity sites, it fails to address
pausing there. Some researchers think
40 MOSAIC
the cause of pausing at polarity sites will
not become clear until they can learn to
determine three-dimensional structure.
The problem is that messenger RNAs are
too scarce and too large to be subjected to
conventional m e t h o d s for analyzing
three-dimensional structure.
The notion that three-dimensional
structure plays some role in polarity is
attractive, says Martin Rosenberg, partly
because the absence of ribosomes between the mutation and the polarity site
would leave a relatively long stretch of
messenger RNA free to take on three-dimensional structure. "As soon as you
stop translating," he says, "you have an
RNA that will assume some stable tertiary
structure. The formation of that structure
may have a lot to do with inducing polarity," he says.
Rosenberg adds that, compared to the
mechanism of attenuation, which is
unique to certain operons, or groups of
genes regulated in common, "polarity is
a more global mechanism. It operates
within many of the translated operons in
an entire cell. It's probably far more subtle, since it has to take into account many
different translated operons." But, he
adds, "we do not yet really understand
how polarity works."
"Tertiary structure is a way we get
around explaining w h y all the pause
sites don't have secondary structure," he
says. •
The National Science Foundation contributes
to the support of research discussed in this
article through its Biochemistry and Genetics
Programs.
* U. S. GOVERNMENT P R I N T I N G O F F I C E : 1984
461-633/20000