Molecular Basis of Inheritance: Class 12 Biology · Chapter 5

Scrape the inside of your cheek with a clean toothpick and you collect a few hundred cells. Each one carries a full copy of your instructions. Stretched end to end, the DNA in a single one of those cells would run about two metres, yet it sits inside a nucleus only a few micrometres wide, and it is copied, read and switched on and off with remarkable precision.

In the previous chapter you followed Mendel's 'factors' from parents to children without asking what they are made of. Here we open the factor up. You will see how DNA is built and packed, how scientists proved that it, and not protein, carries heredity, and why RNA is the likely ancestor of both. Then we follow the information as it is copied (replication), rewritten into RNA (transcription) and decoded into protein (translation), and see how a bacterium decides when a gene should work. The chapter ends with two big applications: reading the whole human genome, and DNA fingerprinting. In Hindi the chapter is वंशागति का आणविक आधार.

5.1 The DNA

How do you measure a molecule that can be a metre long? By counting its building blocks. DNA, a polymer of deoxyribonucleotides, is sized in nucleotides when it is single-stranded and in base pairs (bp) when it is double-stranded. Each species has a characteristic genome size, and the range is huge: a few thousand units in small viruses, billions in mammals (Table 5.1).

5.1.1 Structure of Polynucleotide Chain

You met the building block in Class XI. Every nucleotide combines a phosphate group, a five-carbon (pentose) sugar and a nitrogenous base. Two details tell DNA and RNA apart at this level. DNA's sugar is deoxyribose and RNA's is ribose; and while both use adenine (A), guanine (G) and cytosine (C), DNA uses thymine (T) where RNA uses uracil (U). By shape the bases fall into two families: purines (A and G) are built on a double ring, pyrimidines (C, T and U) on a single ring.

The parts join in a fixed way (Fig. 5.1). The base attaches to carbon 1′ of the sugar through an N-glycosidic bond, giving a nucleoside. A phosphate then links to the OH on carbon 5′ through a phosphoester bond, giving a nucleotide. To grow a chain, the phosphate of one nucleotide bonds to the 3′ OH of the sugar in the next, a 3′–5′ phosphodiester bond. Two nucleotides make a dinucleotide; hundreds or millions make a polynucleotide.

Because every link runs from a 3′ carbon to the next 5′ carbon, a chain has direction. One end carries a free phosphate on a 5′ carbon (the 5′ end); the other ends in a free OH on a 3′ carbon (the 3′ end). Sugars and phosphates form the backbone, and the bases stick out from it. RNA differs in two details: every ribose carries an extra OH at carbon 2′, and uracil stands where DNA has thymine (thymine is simply 5-methyluracil).

Base pairing is what made the model famous. Because A can face only T, and G only C, each strand carries the complete information of its partner: the two strands are complementary. That at once suggested how DNA could be inherited. Separate the strands, rebuild a partner on each by the same rules, and you get two molecules identical to the original (Section 5.4).

1. Two chains. Two polynucleotide chains wind around one axis, with the sugar–phosphate backbones outside and the bases inside.
2. Antiparallel. The chains run in opposite directions: if one runs 5′→3′, its partner runs 3′→5′.
3. Specific pairs. A pairs with T through two hydrogen bonds; G pairs with C through three. A purine always faces a pyrimidine, which keeps the width almost constant (about 2 nm).
4. Right-handed twist. One full turn (the pitch) is 3.4 nm long and holds about 10 base pairs, so neighbouring pairs are about 0.34 nm apart.
5. Stacking. The flat base pairs lie on top of one another like a pile of coins. Stacking, together with the hydrogen bonds, holds the helix firm.

Soon after the double helix, Crick summed up how genetic information flows in a cell: DNA is copied into RNA, and RNA is used to make protein. He called this the central dogma (Fig. 5.3). DNA also copies itself before a cell divides. Some RNA viruses, such as HIV, run part of the path backwards: an enzyme called reverse transcriptase makes DNA from their RNA, a step aptly called reverse transcription.

5.1.2 Packaging of DNA Helix

How long is the DNA in one of your cells? A diploid human nucleus holds about $6.6\times {10}^9$ bp. Multiplying by 0.34 nm per pair gives about 2.2 m of double helix, packed into a nucleus only a few micrometres (${10}^{-6}$ m) across. Cells solve this with tight, orderly folding.

A bacterium manages without a nucleus. Its chromosome is confined to one part of the cell, the nucleoid, where it is folded into large loops held in place by proteins; positive charges on those proteins grip the negatively charged DNA.

In eukaryotes the basic unit of packing is the nucleosome: roughly 200 bp of DNA coiled around a core of eight histone proteins, the histone octamer (Fig. 5.4). Histones are small proteins loaded with lysine and arginine, amino acids whose side chains carry positive charges, so the negatively charged DNA clings to them. Chromatin, the material of the nucleus that takes up stain, is essentially nucleosome after nucleosome along the DNA, and in electron micrographs it looks like beads on a string.

Further folding needs another group of proteins, the non-histone chromosomal (NHC) proteins. With their help the bead string coils into thicker chromatin fibres, and at metaphase these fold so tightly that each chromosome can be seen under a light microscope. Even in a non-dividing nucleus the packing varies from place to place, which gives two kinds of chromatin.

5.2 The Search for Genetic Material

Miescher found nuclein in 1869, only four years after Mendel announced his laws, yet nobody connected the two for decades. By the 1920s the work of Mendel, Walter Sutton, Thomas Hunt Morgan and many others had placed genes on chromosomes inside the nucleus. Chromosomes, however, contain both DNA and protein. Which one carries the instructions? Many biologists bet on protein: with twenty kinds of amino acids it seemed far richer than DNA with its four bases.

Transforming principle

The first clue came from a medical microbiologist. In 1928 Frederick Griffith was studying Streptococcus pneumoniae, a bacterium that causes pneumonia. On a culture plate it grows in two forms. The S strain wraps each cell in a slimy polysaccharide capsule and forms smooth, shiny colonies; it kills mice. The R strain has no capsule, forms rough colonies and is harmless.

Griffith injected mice with four different preparations (Fig. 5.5). Live S cells killed the mice and live R cells did not, as expected. S cells killed by heat were also harmless. But a mixture of heat-killed S cells and live R cells killed the mice, and live S cells could be isolated from their blood. Something from the dead S cells had transformed R cells into capsule-making S cells, and their offspring stayed S. He called it the 'transforming principle'. It had to be hereditary material, but its chemistry was unknown.

Biochemical characterisation of transforming principle

Until the 1940s most scientists still assumed genes were proteins. Between 1933 and 1944 Oswald Avery worked with Colin MacLeod and Maclyn McCarty to pin down the chemical identity of Griffith's principle. They separated the components of heat-killed S cells and tested which could transform live R cells. Only the DNA fraction worked. Enzymes then gave the decisive test (Table 5.4).

Destroying protein or RNA made no difference, but destroying DNA abolished transformation, so the principle had to be DNA. Not everyone was convinced: some suspected that a trace of protein hidden in the DNA did the work.

5.2.1 The Genetic Material is DNA

In 1952 Alfred Hershey and Martha Chase put a sharp question to a very simple system. A bacteriophage, a virus that attacks bacteria, is little more than DNA packed inside a protein shell. When it infects a cell, something passes into the bacterium and takes over, so that the cell starts turning out new phages. If that something is protein, protein is the genetic material; if it is DNA, the case for DNA is closed.

They exploited a chemical difference. DNA contains phosphorus but no sulfur, while the coat protein contains sulfur but no phosphorus. Phages grown with radioactive phosphorus (${}^{32}$P) therefore had labelled DNA, and phages grown with radioactive sulfur (${}^{35}$S) had labelled coats. Each batch infected E. coli. A kitchen blender then shook the empty coats off the cells, and a centrifuge spun the heavy bacteria into a pellet, leaving the light coats in the liquid above (Fig. 5.6).

Bacteria infected by ${}^{32}$P phages were radioactive; those infected by ${}^{35}$S phages were not, and their radioactivity stayed in the liquid. Protein never entered the cells, while DNA did, and the next generation of phages was made from it. DNA, not protein, is the genetic material that passes from virus to bacterium.

5.2.2 Properties of Genetic Material (DNA versus RNA)

After Hershey and Chase, protein was out of the race. Yet DNA is not the only answer: in some viruses, such as tobacco mosaic virus, the bacteriophage Qβ and the influenza virus, the genes are made of RNA. Why, then, did most life settle on DNA, while RNA works as messenger and adapter? The answer lies in what any genetic material has to do.

1. Replicate: make accurate copies of itself.
2. Stay stable: keep its chemistry and structure through the life of the organism.
3. Change slowly: allow occasional mutations, the raw material of evolution.
4. Be expressed: show itself as traits that are inherited in Mendelian fashion.

Replication. Base pairing lets both nucleic acids direct their own copying. Proteins have no such pairing rule, so they fail the very first test.

Stability. Here the two molecules really differ. Every ribose in RNA carries a reactive OH group on carbon 2′, which makes RNA chemically fragile and quick to break down; that some RNAs can catalyse reactions is itself a sign of reactivity. DNA lacks that group and has a second, complementary strand. Even heat, which pulls the strands apart, does not ruin it, because they pair up again on cooling; that is how the DNA in Griffith's heat-killed S cells could still transform. Using thymine instead of uracil makes DNA steadier still.11Cytosine slowly loses an amino group and becomes uracil. Because normal DNA contains no uracil, repair enzymes can spot and fix every U they find; in a U-containing DNA they could not tell damage from normal bases.

Mutation. Change is needed too, but slowly. Being less stable, RNA picks up mutations faster than DNA. Combine that with the short generation times of viruses and RNA viruses evolve at great speed; influenza changes so quickly that its vaccine has to be updated most years.

Expression. Here RNA has the edge. A ribosome translates RNA directly, while DNA's information must first be copied into RNA, and the protein-making apparatus is itself built from RNA. The balance sheet: DNA wins as an archive and RNA as a courier, which is why cells store their genes in DNA and send RNA copies out to work.

5.3 RNA World

Today DNA needs proteins (its polymerases) to be copied, and proteins need DNA's instructions to be made. Which came first? A molecule that could do both jobs would break the circle, and RNA can: it stores information, and some RNAs, the ribozymes, catalyse reactions like enzymes. Tellingly, some of the oldest core processes of the cell, such as translation, splicing and parts of metabolism, still run on RNA; the peptide bond in every ribosome is made by RNA. Hence the idea of an early RNA world, in which RNA was the first genetic material.

That double life had a price: a molecule reactive enough to catalyse is also reactive enough to fall apart. DNA, a chemically modified RNA with no 2′-OH and with thymine, proved far steadier, and its second strand gave a backup copy from which damage could be repaired. So storage passed to DNA and most catalysis to proteins, leaving RNA in the middle (Fig. 5.7). You will meet this idea again in the chapter on evolution.

5.4 Replication

Watson and Crick saw the copying method in their own model. Near the end of their 1953 paper they remarked, almost in passing, that the pairing rule pointed straight to a way of duplicating the molecule. Unzip the two strands, let each one serve as a template, and add new nucleotides according to the pairing rules. Each finished molecule would then hold one old (parental) strand and one new strand: semiconservative replication.

This was a hypothesis, and two others were possible. In conservative copying the parent helix would stay intact and an entirely new helix would be built beside it. In dispersive copying old and new pieces would be scattered through both strands of every daughter. An experiment had to tell them apart.

5.4.1 The Experimental Proof

In 1958 Matthew Meselson and Franklin Stahl found a way to tell old DNA from new by its weight. First they made all of a bacterium's DNA heavy. E. coli was grown for many generations on a medium whose only nitrogen source was ${\phantom{X}}_{\phantom{}}^{\phantom{15}}{\phantom{X}}_{\phantom{2}}^{\phantom{2}15}\mathrm{N}\mathrm{H}{\phantom{X}}_4\mathrm{C}\mathrm{l}$, so nitrogen-15, a heavy but stable isotope (it is not radioactive), ended up in every base. Then the cells were switched to ordinary ${\phantom{X}}_{\phantom{}}^{\phantom{14}}{\phantom{X}}_{\phantom{2}}^{\phantom{2}14}\mathrm{N}\mathrm{H}{\phantom{X}}_4\mathrm{C}\mathrm{l}$, and from then on every strand they made was light. Samples were taken as the cells went on dividing, about once every 20 minutes. Each was spun at very high speed in a caesium chloride (CsCl) density gradient, a salt solution that grows denser towards the bottom of the tube, where every DNA molecule, still double-stranded, settles at the level that matches its own density (Fig. 5.8).

• Start: a single heavy band.
• After one generation (20 min): a single band of hybrid density, halfway between heavy and light, and no heavy band at all.
• After two generations (40 min): two bands of equal size, one hybrid and one light.

The results rule out both rivals. Conservative copying predicts a heavy band and a light band after one generation, but the real sample showed only hybrid DNA. Dispersive copying predicts a single band at every generation, drifting steadily towards light, but at 40 minutes the hybrid and light DNA formed two separate bands. Only semiconservative copying fits every sample. Try all three models yourself in Fig. 5.9.

Bacteria were not a special case. At about the same time, J. Herbert Taylor and colleagues fed radioactive thymidine to dividing root-tip cells of the broad bean Vicia faba and followed where the label went. The chromosomes, too, replicated semiconservatively, and the same was later shown for other plants and for human cells.

5.4.2 The Machinery and the Enzymes

How fast must copying go? E. coli duplicates its $4.6\times {10}^6$ bp in about 18 minutes, starting at one point and working both ways round its circular chromosome. Each of the two copying sites must therefore add roughly 2000 nucleotides every second, and add the right one almost every time, because an uncorrected slip becomes a mutation. The enzyme that manages this is DNA-dependent DNA polymerase: it reads a DNA template and strings deoxynucleotides into a matching strand.

The fuel comes with the building blocks. Nucleotides arrive as deoxyribonucleoside triphosphates (dATP, dGTP, dCTP and dTTP). When one is added to a growing strand its two outer phosphates are released; these are high-energy bonds of the same kind as in ATP, and breaking them drives the reaction. Copying DNA is expensive, and the dNTPs pay for it themselves.

DNA polymerase obeys one strict rule: it can attach a nucleotide only to the free 3′ end of a growing strand, so every new strand lengthens 5′→3′. The cell does not unwind the whole helix before copying, which would cost far too much energy; it opens a short stretch at a time, a Y-shaped replication fork that travels along the molecule (Fig. 5.10). The two templates exposed at the fork point in opposite directions, so the rule plays out differently on each. Along the template that runs 3′→5′ in the direction of travel, the new strand grows the same way the fork moves and is made continuously. Along the other template it has to grow away from the fork, so it is made discontinuously, in short pieces started as fresh template is exposed, and DNA ligase later seals the pieces into one strand.

Where does copying begin? Not at random, but only at particular sequences, the origins of replication; E. coli has exactly one. This has a practical consequence. A gene cut out of one organism and put into a bacterium carries no origin of its own, so it would never be copied. Genetic engineers therefore join it to a vector, which brings an origin along (Chapter 9). Note also that DNA polymerase can lengthen a strand but cannot begin one from scratch; a short RNA primer provides its starting point.

In eukaryotic cells copying is confined to one part of the cell cycle, the S phase, and it starts at many origins along each chromosome. The cell must then divide exactly once for each round of copying; a cell that replicates its DNA but skips division is left with extra sets of chromosomes, a state called polyploidy. Many details of how eukaryotic origins work are still being worked out.

5.5 Transcription

Replication copies everything; transcription is selective. In any one cell only some genes are copied into RNA, each as often as needed, and from each gene only one of the two strands is used. Transcription copies the information of that strand into RNA by the familiar pairing rules, except that the RNA puts uracil (U) opposite adenine. Because the process is so selective, the DNA must carry signals that mark where a transcribed stretch begins and ends, and which strand is to be read.

Suppose a cell did copy both strands of a gene. It would get two RNAs whose sequences are complementary, not identical. Those RNAs would find each other and pair, forming double-stranded RNA that ribosomes cannot read, so the work would be wasted. Even if they stayed apart, they would spell two unrelated messages, and one stretch of DNA would specify two different proteins, a recipe for confusion. Reading a single strand avoids both problems.

5.5.1 Transcription Unit

A transcription unit is a stretch of DNA with three parts: a promoter, a structural gene and a terminator.

Which strand gets read? The enzyme that makes RNA, DNA-dependent RNA polymerase, can only build 5′→3′, and its product must run antiparallel to the DNA it copies. So within the structural gene it reads the strand oriented 3′→5′, the template strand. The partner strand is not copied, but its sequence matches the RNA's (with T for U), so it is the one written down whenever we quote a gene. It is called the coding strand. Do not let the name mislead you: the coding strand is the copy we read, not the one the enzyme reads (Fig. 5.11).

RNA polymerase finds a gene by docking at its promoter, which sits at the upstream end, towards the 5′ end as measured along the coding strand. The promoter's orientation also fixes the direction of reading, and with it which strand serves as template: a promoter placed at the other end of the gene would make the opposite strand the template. At the downstream end, towards the coding strand's 3′ end, the terminator usually signals where transcription stops. Other regulatory sequences can lie further upstream or downstream; you will meet them in Section 5.8.

5.5.2 Transcription Unit and the Gene

Where does one gene end and the next begin? The classical definition, a functional unit of inheritance, does not translate neatly into a stretch of bases. Think of the DNA that is transcribed into a tRNA or an rRNA: it never codes for protein, yet it is a gene too. For protein-coding DNA biologists use a sharper term, the cistron: the stretch that codes for one polypeptide. A transcription unit carrying one cistron is monocistronic; one carrying several is polycistronic.

Most eukaryotic genes are also split. Their coding sequence is broken into pieces, the exons, separated by stretches called introns (intervening sequences). Both are copied into the first RNA transcript, but only the exons survive into the finished, mature RNA; the introns are removed. A eukaryotic gene is therefore not one continuous coding stretch, which makes defining it as 'a piece of DNA' even trickier.

A mutation does not have to fall inside a coding sequence to change a trait: damage to a gene's promoter or other control sequences can silence it just as surely. That is why such sequences are sometimes loosely called regulatory genes, although strictly they make neither RNA nor protein.

5.5.3 Types of RNA and the process of Transcription

Making a protein takes teamwork among three kinds of RNA, all transcribed from DNA. mRNA (messenger RNA) is the message itself. tRNAs (transfer RNAs) are the adapters that bring amino acids and read the code. rRNAs (ribosomal RNAs) form the core of the ribosome and carry out part of its chemistry.

A bacterium gets by with a single RNA polymerase for all its RNA. Follow one molecule of it along a gene (Fig. 5.12). It first settles on the promoter: initiation. It then travels along the template, prising the helix open just ahead of itself and joining ribonucleoside triphosphates that pair correctly with the template, while the RNA trails behind with only its newest end still held by the enzyme: elongation. At the terminator, the finished RNA and then the enzyme let go of the DNA: termination.

The enzyme does not manage all this alone. The core polymerase is good only at lengthening RNA. To find a promoter and begin, it briefly carries a partner protein, the initiation factor σ (sigma); to stop at a terminator it is helped by the termination factor ρ (rho). Each partner changes what the polymerase recognises, and each leaves once its job is done.

Bacterial mRNA is ready to use the moment it is made, and it is made in the same compartment as the ribosomes. Ribosomes therefore climb onto the 5′ end of an mRNA and start translating while RNA polymerase is still busy further along the gene: in bacteria transcription and translation are coupled.

Eukaryotes add two complications. The first is a division of labour. Their nuclei use three different RNA polymerases, each responsible for its own set of RNAs, while mitochondria and chloroplasts have polymerases of their own (Table 5.7).

The second is that the first copy of a gene, the primary transcript or hnRNA (heterogeneous nuclear RNA), contains introns as well as exons and cannot be used as it is. Three kinds of processing turn it into mRNA (Fig. 5.13). Capping fixes a modified nucleotide, a methylated guanosine joined through a triphosphate link, onto the 5′ end. In tailing, 200–300 adenylate (A) residues are added to the 3′ end without any template. In splicing, the introns are cut out and the exons joined in the right order. The finished mRNA then leaves the nucleus to be translated.

Why would cells bother cutting and pasting their own messages? One view is historical. Introns seem to be very old, and the splicing machinery is built around small nuclear RNAs that do much of the cutting and joining themselves, a likely relic of the RNA world (Section 5.3). The more biologists look, the more jobs they find that RNA does in cells.

5.6 Genetic Code

Copying DNA into DNA, or DNA into RNA, needs no dictionary: each base simply pairs with its partner. Translation is a different kind of problem. A sequence of four kinds of nucleotide has to become a sequence of twenty kinds of amino acid, and nothing in an amino acid's shape fits a base. What biologists did know was that mutations in genes produced altered amino acids in proteins. So the cell must follow a set of rules relating base sequence to amino acid sequence: the genetic code.

The first clue was arithmetic, and it came from a physicist, George Gamow. If one base stood for one amino acid, four bases could name only four amino acids. Pairs of bases give $4^2=16$ combinations, still short of twenty. Triplets give $4^3=64$, comfortably enough even with many left over. So a codon was probably three bases long. Showing that this was really so needed chemists, biochemists and geneticists working together.

The decisive tool was Marshall Nirenberg's cell-free system, a test-tube mix of cell contents that makes protein from whatever RNA is added. In 1961 an artificial RNA containing only U gave a chain containing only phenylalanine, so UUU had to mean phenylalanine. Artificial RNAs for such tests could be made with Severo Ochoa's enzyme, polynucleotide phosphorylase, which strings nucleotides together without a template, while Har Gobind Khorana's chemical methods produced RNAs with precisely repeating sequences (homopolymers and copolymers) that decoded codon after codon. By 1966 all 64 codons were known (Table 5.8).

The table is arranged by amino acid so you can go from protein back to RNA. Reading in the other direction, from a codon to its amino acid, is easiest in Fig. 5.14. Either way, a few features of the code stand out.

1. It is a triplet code. Of the 64 codons, 61 stand for amino acids and 3 (UAA, UAG and UGA) code for none; they are stop codons that end a chain.
2. It is degenerate. Most amino acids have more than one codon; leucine, serine and arginine have six each. Only methionine and tryptophan have a single codon.
3. It is unambiguous. Each codon means one amino acid and nothing else.
4. It has no punctuation. Codons follow one another without gaps or commas, so the message is read three bases at a time from a fixed starting point.
5. It is nearly universal. UUU means phenylalanine in bacteria, plants and people alike; the few exceptions occur in mitochondria and in some protozoans.
6. AUG has two jobs. It codes for methionine and also serves as the start (initiator) codon.

5.6.1 Mutations and Genetic Code

Mutations are the natural experiments that reveal how the code works. A big deletion or rearrangement can remove or duplicate whole genes, with an obvious loss or gain of function, but it says little about codons. Changes of one or a few bases are far more revealing.

A point mutation changes a single base pair. The classic example is sickle-cell anaemia. In the gene for the β-globin chain of haemoglobin, one base change turns the codon GAG into GUG, so glutamic acid (glutamate), the sixth amino acid of the finished chain, is replaced by valine. That single swap makes haemoglobin molecules stick together when oxygen is low, distorting red blood cells into sickles.

Because codons are read without punctuation, inserting or deleting bases has a special effect. Adding or removing one or two bases shifts the reading frame from that point onward, so every codon after it is read wrongly: a frameshift insertion or deletion. Adding or removing three bases (or a multiple of three) instead adds or removes whole codons. One or more amino acids are gained or lost, and the frame after the change stays intact (Table 5.9).

5.6.2 tRNA – the Adapter Molecule

Something has to stand between codon and amino acid. Transfer RNA does: it reads a codon at one end and carries the matching amino acid at the other. Francis Crick predicted such an adapter on logical grounds, since amino acids have no shape that could recognise a codon. Curiously, tRNA had already been discovered, under the name soluble RNA (sRNA), before anyone knew this was its job.

Drawn flat, a tRNA resembles a clover leaf; its true folded shape is a compact, upside-down L (Fig. 5.15). Two places matter. The anticodon loop carries three bases that pair, antiparallel, with a codon. The amino acid acceptor end (the 3′ end) is where the amino acid is attached. Each tRNA is charged with one particular amino acid, a special initiator tRNA carries the methionine that starts every chain, and stop codons have no tRNA at all.

5.7 Translation

Translation strings amino acids together by peptide bonds, in exactly the order of an mRNA's codons. A peptide bond does not form on its own, so the cell pays in advance.

Before translation, an enzyme uses ATP to activate each amino acid and load it onto a tRNA that matches it (its cognate tRNA). This loading is called charging of tRNA, or more precisely aminoacylation. A charged tRNA holds its amino acid by a high-energy link, so when two charged tRNAs sit side by side, joining their amino acids is energetically easy, and a catalyst makes it fast.

The work happens on ribosomes. Each is assembled from a small subunit and a large subunit, which float apart when idle; translation begins when a small subunit binds an mRNA. Together the subunits contain structural RNAs and roughly 80 kinds of protein. The large subunit has two neighbouring sites for charged tRNAs (usually called the P site and the A site), which hold two amino acids next to each other. The bond between them is formed by RNA: in bacteria the catalyst is the 23S rRNA, a ribozyme.

An mRNA is longer than the part that is read. At each end lies an untranslated region (UTR): the 5′ UTR before the start codon and the 3′ UTR after the stop codon. They help the mRNA to be translated efficiently. The stretch between, from AUG to the stop codon, is the translational unit that actually codes for the polypeptide.

1. Initiation. The ribosome assembles on the mRNA at the start codon AUG, which only the initiator tRNA (carrying methionine) can recognise.
2. Elongation. Charged tRNAs arrive one at a time; each binds only if its anticodon pairs with the codon in the free site. A peptide bond joins the growing chain to the new amino acid, and the ribosome moves along the mRNA by one codon. The cycle repeats.
3. Termination. When a stop codon reaches the ribosome, no tRNA fits. A release factor binds instead, the finished polypeptide is freed, and the ribosome comes apart.

5.8 Regulation of Gene Expression

A gene is 'expressed' when its product, usually a polypeptide, is made, and a cell can turn that output up or down at several points along the way. In eukaryotes, regulation can act at four levels:

1. the transcriptional level, deciding whether a primary transcript is made;
2. the processing level, for example which way an RNA is spliced;
3. the transport of mRNA from the nucleus to the cytoplasm;
4. the translational level, deciding how much protein each mRNA yields.

Cells do not run every gene all the time; that would waste energy and materials. A liver cell and a nerve cell carry identical DNA but use different sets of genes, and turning an embryo into an adult means switching whole batteries of genes on and off in the right order. Even a single bacterium tunes its genes to its surroundings. E. coli makes the enzyme β-galactosidase, which splits lactose into glucose and galactose for energy, mainly when lactose is actually there. Gene expression, in short, responds to the cell's metabolic, physiological and environmental conditions.

Bacteria do most of their regulating at the very first step: whether RNA polymerase starts transcribing a gene at all. Proteins that bind DNA near the promoter decide this. A repressor blocks the polymerase (negative control); an activator helps it (positive control). Many bacterial repressors work by sitting on an operator, a short DNA sequence next to the promoter, and each operator is recognised by its own repressor: the lac repressor binds the lac operator and no other.

5.8.1 The Lac operon

Bacterial genes that serve the same pathway are often lined up side by side and transcribed together into one polycistronic mRNA, sharing one promoter and one set of control sequences. Such a unit is an operon; the trp, ara, his and val operons are examples. The best known is the lac operon, which handles the sugar lactose. Its control was worked out by two French scientists, the geneticist François Jacob and the biochemist Jacques Monod, and it was the first system of transcriptional regulation to be understood.

The operon has one regulatory gene and three structural genes (Table 5.11, Fig. 5.17). The regulatory gene is called i, short for inhibitor (not inducer). All three structural-gene products serve lactose use, and in most operons, likewise, the grouped genes work in one pathway.

With no lactose around, the operon is off. The i gene is expressed nonstop (constitutively), so repressor is always present, and it sits on the operator where it blocks RNA polymerase. Now give the cells lactose, and no glucose, their preferred food. A trickle of transcription always escapes the repressor, so each cell has a few permease molecules, enough to let some lactose in; without them the sugar could never enter. Inside, lactose (more exactly allolactose, made from it) binds the repressor and changes its shape, so it lets go of the operator. RNA polymerase can now run through z, y and a, and the enzymes for using lactose are made. Because lactose both feeds the enzyme and flips the switch, it is called the inducer.

It is a neat economy: the substrate itself calls for the enzymes that use it. The products, glucose and galactose, cannot switch the operon on. Explore every combination, including two broken switches, in Fig. 5.18.

Because this switch works by removing a block, it is called negative regulation. The lac operon also has a second, positive control, an activator that keeps it low while glucose is plentiful, but that part of the story is beyond this chapter.

5.9 Human Genome Project

In 1990 biologists set out to read the complete DNA sequence of human beings: the Human Genome Project (HGP). The goal followed directly from this chapter. If a person's genetic make-up is written in the order of bases, then knowing that order, and how it differs between people, should explain a great deal. Two advances made the attempt realistic: genetic engineering allowed any piece of DNA to be isolated and cloned, and sequencing had become quick and routine.

It was a mega project by any measure. The genome has about $3\times {10}^9$ bp. Early planners guessed it would cost about US$3 to read each base pair, which comes to about US$9 billion. The data are just as daunting. Stored as one letter per base, one copy of the genome fills a text file of about 3 gigabytes; read aloud at five letters a second without a break, it would take nearly twenty years. Storing, searching and analysing so much information needed fast computers, and the project grew up together with a new field, bioinformatics, which uses computers to handle biological data.

Goals of HGP

1. Find all the genes in human DNA, then estimated at roughly 20 000–25 000.
2. Work out the order of all the roughly 3 billion base pairs.
3. Keep the information in databases open to researchers.
4. Build better tools for analysing the data.
5. Share the new technologies with industry and other sectors.
6. Think through the ethical, legal and social issues (ELSI) the project might raise.

The HGP ran for 13 years under the US Department of Energy and the National Institutes of Health, with the UK's Wellcome Trust soon joining as a major partner and Japan, France, Germany, China and other countries contributing. The sequence was declared complete in 2003. Mapping how DNA differs between people could change how diseases are detected, treated and eventually prevented. The genomes of other species were read too, among them bacteria, yeast, the nematode Caenorhabditis elegans, the fruit fly Drosophila, rice and Arabidopsis; knowing them helps us use such organisms in medicine, farming, fuel production and the clean-up of polluted sites.

Methodologies

The project attacked the problem from two directions. Expressed sequence tags (ESTs) picked out genes by starting from the RNA that cells actually make. The whole-genome route read everything, genes and the long stretches between them, without choosing, and worked out afterwards which parts do what; that labelling is called sequence annotation.

The whole-genome route was a giant jigsaw that the scientists first cut up themselves (Fig. 5.19). Sequencing machines read only short pieces, so DNA from many cells was broken at random into fragments. Each fragment was multiplied by cloning it in bacteria or yeast, carried in vectors called BACs (bacterial artificial chromosomes) and YACs (yeast artificial chromosomes). Automated sequencers read the clones by a method devised by Frederick Sanger, the same Sanger who first worked out the amino acid sequence of a protein. Because the pieces were cut at random they overlapped, and the overlaps showed which pieces lay next to each other. Matching millions of overlaps is impossible by hand, so special computer programs did the assembly.

The assembled sequences were annotated and assigned to their chromosomes. The last of the 24 different human chromosomes (22 autosomes plus X and Y) to be finished was chromosome 1, in May 2006. A further task was to build genetic and physical maps of the genome. These used variation (polymorphism) in the sites cut by restriction endonucleases, and short repeated sequences called microsatellites; the next section shows how such repeat polymorphisms are used.

5.9.1 Salient Features of Human Genome

Two findings deserve a word. Repeats were once dismissed as 'junk', but their arrangement is a record of how chromosomes are organised, how they move and change, and how they evolved. Single nucleotide polymorphisms (SNPs, said 'snips') have become the workhorse markers of human genetics: by checking which SNP variants travel with a disease in families, researchers can home in on the gene responsible, and SNP patterns across populations trace the migrations of our ancestors.

5.9.2 Applications and Future Challenges

The sequence was a beginning, not an end. Knowing the letters is not the same as understanding them, and turning sequence into biology will occupy many thousands of researchers, in laboratories and companies, for decades. The biggest change is in how questions are asked. Instead of studying one gene at a time, biologists can now measure thousands at once: which genes are switched on in a tumour compared with the healthy tissue beside it, which variants make a medicine work well or badly in a patient, or how a crop variety copes with drought. Reading a person's genome now takes days and costs a tiny fraction of the first one, which is bringing genome tests into hospitals, including in India.

5.10 DNA Fingerprinting

Forensic scientists and courts often need to know whether two DNA samples come from the same person, or whether two people are related. Since any two of us differ at only about 0.1 % of our bases, comparing whole genomes for every case would be wasteful. DNA fingerprinting takes a shortcut: it compares just a few regions that differ wildly from person to person.

The chosen regions are repetitive DNA, in which a short sequence is repeated over and over. Some of it first showed up as an oddity of centrifugation. In a density gradient most of a cell's DNA forms one big peak, but small extra peaks appear beside it, made of repeats whose base composition gives them a slightly different density: satellite DNA. Satellites come in many types, depending on how rich they are in A and T or in G and C, how long the repeating unit is and how many times it repeats; microsatellites and minisatellites are two of them. They code for no protein, yet they make up a large part of the genome, and the number of repeats varies enormously between people.

Every tissue of a person, whether blood, hair roots, skin, bone, saliva or semen, carries the same pattern of repeats. That makes it a powerful identification tool in forensic work. And because the repeats are inherited from parents, the same test can settle disputes about parentage.

Since the method rests on DNA polymorphism, we should be clear about the term.

Polymorphisms are inherited, so they begin as mutations in germ cells; a mutation in a body (somatic) cell dies with the person. If a germ-line change does not stop its carriers having children, it can spread through the population over many generations. Most polymorphisms lie in non-coding DNA, where a change rarely affects health or fertility, so it is not weeded out and keeps accumulating. They range from a single changed base to large rearrangements, and they supply the variation on which evolution and the formation of new species depend.

The method was invented by Alec Jeffreys in 1984–85. His probe was a highly variable satellite sequence, a variable number of tandem repeats (VNTR), and the original procedure detected it by Southern blot hybridisation with the probe made radioactive.

Why does this give a personal pattern (Fig. 5.20)? A VNTR is a minisatellite: a short sequence repeated head to tail. People differ in how many repeats they carry at each VNTR site, and even the two copies of a chromosome in one person usually differ, so cutting beside the repeats gives fragments anywhere from about 0.1 to 20 kb long. The probe lights up every one of them, and the film shows a ladder of bands unique to each person. Only identical (monozygotic) twins, who come from one zygote, share it.

Today the test is far more sensitive. The polymerase chain reaction (PCR, Chapter 9) multiplies chosen stretches of DNA millions of times, so the DNA of a single cell can be enough, and many different probes and markers are in use. Beyond crime and paternity cases, DNA fingerprints help measure genetic diversity, for example within an endangered species or among crop varieties.