Gene Regulation and Biotechnology

Every cell in your body carries the same genome, yet a neuron, a liver cell and a red blood cell precursor look and behave nothing alike. The difference is not which genes they have but which genes they use, when, and how much. This chapter explains how bacteria and eukaryotes switch genes on and off, how controlled gene expression builds an embryo, and how the tools that grew out of that knowledge (PCR, cloning, sequencing and CRISPR) now let us read and rewrite DNA in medicine, agriculture and forensics.

Section 1Gene regulation in bacteria: operons

Why regulate genes at all?

A bacterium such as Escherichia coli has roughly 4,300 protein coding genes, but it never needs all of their products at once. A cell that keeps making the enzymes to digest a sugar that is not present, or the enzymes to build an amino acid that is already plentiful, wastes resources and grows more slowly than a competitor that does not. Natural selection has therefore favored cells that match gene expression to their environment.

Bacteria can adjust metabolism in two broad ways. They can change the activity of enzymes that already exist, for example through feedback inhibition, in which the end product of a pathway binds an allosteric site on the first enzyme and slows it within seconds. Or they can change the amount of enzyme by controlling transcription of the genes that encode it. The second response is slower (minutes) but saves far more resources. In bacteria the main control point is the start of transcription, because transcription and translation are coupled in the cytoplasm and most bacterial mRNAs last only a few minutes.

The operon model

In 1961 François Jacob and Jacques Monod proposed that bacterial genes with related functions are often clustered and transcribed together under a single on and off switch. They called the unit an operon. An operon has three essential parts:

• A promoter, the DNA sequence where RNA polymerase (guided by its sigma factor) binds and begins transcription.
• An operator, a short DNA segment that usually overlaps or lies just downstream of the promoter. It is the switch: when a repressor protein binds the operator, it blocks RNA polymerase.
• The structural genes, the coding sequences for the enzymes or other proteins. They are transcribed as one long polycistronic mRNA that carries several coding regions, each with its own start and stop codon and ribosome binding site.

The repressor is encoded by a separate regulatory gene, which has its own promoter and is usually expressed continuously at a low level. Because the repressor is a diffusible protein, it can act on any copy of the operator in the cell. The operator, by contrast, is a DNA sequence that controls only the genes physically attached to it. This difference between a trans-acting factor (a diffusible product) and a cis-acting element (a DNA site) is one of the most useful ideas in all of gene regulation.

The lac operon: an inducible operon

E. coli prefers glucose, but it can live on lactose, the disaccharide of milk, by splitting it into glucose and galactose. The lac operon contains three structural genes. lacZ encodes β-galactosidase, which hydrolyzes lactose and also converts a little of it into the isomer allolactose. lacY encodes lactose permease, a membrane transporter that brings lactose into the cell. lacA encodes a transacetylase whose role in lactose use is minor. Upstream lies the regulatory gene lacI, which encodes the lac repressor.

When lactose is absent, the lac repressor is active and binds tightly to the operator. RNA polymerase can still sit at the promoter but cannot move into the genes, so almost no lac mRNA is made. When lactose enters the cell, the few molecules of β-galactosidase always present (a small leaky level of expression) convert some of it to allolactose. Allolactose is the inducer: it binds the repressor at an allosteric site, changes its shape and drops its affinity for the operator. The repressor falls off, and transcription begins. Because a small molecule switches the operon on, the lac operon is called an inducible operon. Inducible operons usually encode catabolic enzymes that break down a nutrient, and it makes sense to make them only when the nutrient is present.

Researchers often use IPTG, a synthetic look alike of allolactose, as an inducer. IPTG binds the repressor but is not broken down by β-galactosidase, so it keeps the operon switched on at a constant level.

Positive control: catabolite repression and CAP

Removing the repressor is necessary but not sufficient for strong expression. The lac promoter is weak; RNA polymerase binds it poorly on its own. Expression becomes high only when an activator helps polymerase bind. That activator is the catabolite activator protein (CAP, also called CRP). CAP can bind DNA only when it carries the small signaling molecule cyclic AMP (cAMP). When glucose is scarce, the enzyme adenylyl cyclase is active and cAMP levels rise; cAMP binds CAP, and the cAMP-CAP complex binds a site just upstream of the lac promoter, bends the DNA and contacts RNA polymerase, increasing transcription roughly fiftyfold. When glucose is plentiful, glucose transport keeps adenylyl cyclase inactive, cAMP falls, CAP leaves the DNA and lac expression stays low even if lactose is present. Glucose also reduces lactose uptake directly by inhibiting the permease, a process called inducer exclusion.

The result is a two input logic circuit. The repressor reports whether lactose is present; CAP reports whether glucose is absent. The cell commits to lactose only when it has lactose and lacks its preferred sugar. Because the lac operon and many other operons for alternative sugars all depend on cAMP-CAP, the cell can coordinate them as a group, an arrangement called catabolite repression. When E. coli is grown in a mixture of glucose and lactose, it uses up the glucose first, pauses briefly, then switches to lactose, giving the two phase growth curve (diauxie) that Monod first described.

Mutations that revealed the logic

Jacob and Monod worked out the system with mutants. A lacI^- mutant makes a nonfunctional repressor, so the operon is constitutive (always on). An operator constitutive mutant (lacO^c) has an altered operator that the repressor cannot bind, so it too is constitutive. A lacI^S (super repressor) mutant makes a repressor that cannot bind allolactose, so the operon can never be induced. To tell these apart they built partial diploid bacteria carrying a second copy of the lac region on an F plasmid. A normal lacI gene on the plasmid restored control to a lacI^- chromosome, because the repressor diffuses (trans acting, and the mutation is recessive). But a normal plasmid could not rescue a lacO^c chromosome, because an operator controls only the genes on its own DNA molecule (cis acting). The super repressor was dominant: its noninducible protein sat on both operators.

The trp operon: a repressible operon

The trp operon holds five genes (trpE, D, C, B and A) that encode the enzymes for making the amino acid tryptophan. Here the logic is reversed. The operon is normally on, because the cell usually needs tryptophan. The regulatory gene trpR makes a repressor in an inactive form (an aporepressor) that cannot bind the operator by itself. When tryptophan accumulates, tryptophan binds the repressor as a corepressor, activating it; the complex binds the operator and shuts transcription down. Because a small molecule turns the operon off, it is a repressible operon. Repressible operons usually encode anabolic enzymes that build a product, and it makes sense to stop building it when the product is abundant.

Attenuation: a second dial on the trp operon

The trp operon has a second control layer that depends on the coupling of transcription and translation. Before the first structural gene lies a leader sequence of about 160 nucleotides with four regions (1, 2, 3 and 4) that can pair in different ways. Region 1 encodes a short leader peptide with two tryptophan codons in a row. As soon as the 5' end of the transcript appears, a ribosome begins translating it.

• When tryptophan is plentiful, charged tryptophan tRNA is abundant, and the ribosome moves quickly through region 1 and covers region 2. Regions 3 and 4 then pair into a hairpin followed by a run of U's, which acts as a transcription terminator. RNA polymerase stops before reaching the structural genes.
• When tryptophan is scarce, the ribosome stalls at the two tryptophan codons in region 1. Region 2 is free to pair with region 3, forming an antiterminator that prevents the 3 to 4 terminator from forming, and RNA polymerase continues into the genes.

This attenuation fine tunes expression. Repression alone changes trp operon output about seventyfold; attenuation adds roughly another tenfold, so the combined range is several hundredfold.

Beyond operons

Operons are only part of bacterial control. Bacteria use alternative sigma factors to redirect RNA polymerase to whole sets of genes at once, for example the heat shock genes or the genes for spore formation in Bacillus. Some mRNAs contain riboswitches, folded regions that bind a metabolite directly and change their own shape to stop transcription or block the ribosome binding site, with no protein involved. Many bacteria also sense their own population density by releasing and detecting small signal molecules, a process called quorum sensing, and switch on genes for light production, biofilm formation or toxins only when enough cells are present.

Section 2Gene regulation in eukaryotes and epigenetics

Many control points, one goal

Eukaryotic cells face a harder problem than bacteria. A human has about 20,000 protein coding genes, and a multicellular body contains hundreds of cell types, each of which must express its own set of genes stably for years. A typical human cell expresses perhaps a third to a half of its genes, and the particular set defines what the cell is. Eukaryotic genes are also separated from ribosomes by the nuclear envelope, wrapped in chromatin, interrupted by introns, and transcribed one gene at a time rather than in operons. Each of these features creates an extra point where expression can be adjusted. Control is possible at every step from DNA to finished protein:

1. Chromatin: how tightly the DNA is packed and chemically marked.
2. Transcription: whether and how often RNA polymerase II starts at a promoter. This is still the most important step for most genes.
3. RNA processing: capping, polyadenylation and especially alternative splicing.
4. mRNA transport and stability: export from the nucleus and how long an mRNA survives in the cytoplasm.
5. Translation: whether ribosomes start on an mRNA.
6. Protein processing and degradation: cleavage, chemical modification, targeting and destruction.

Chromatin: packing as a control step

In the nucleus DNA wraps around octamers of histone proteins (two each of H2A, H2B, H3 and H4) to form nucleosomes, about 147 base pairs per core, like beads on a string. Nucleosomes fold further into fibers, loops and territories. Tightly packed heterochromatin is largely inaccessible to the transcription machinery and its genes are usually silent; looser euchromatin is where most active genes lie. Whether a gene sits in open or closed chromatin depends heavily on chemical marks on the histones and on the DNA itself.

Each histone has a flexible tail that sticks out of the nucleosome. Enzymes add and remove chemical groups on these tails:

• Histone acetylation: histone acetyltransferases (HATs) add acetyl groups to lysines. Acetylation neutralizes the positive charge of lysine, weakening the tail's grip on the negatively charged DNA and on neighboring nucleosomes. Chromatin opens, and acetylated lysines also serve as docking sites for proteins that promote transcription. Histone deacetylases (HDACs) remove the acetyl groups and help close chromatin.
• Histone methylation: methyl groups on lysines can either activate or repress, depending on which residue is marked. Trimethylation of lysine 4 on histone H3 (H3K4me3) marks active promoters; trimethylation of lysine 9 or lysine 27 (H3K9me3, H3K27me3) marks silenced regions.

DNA methylation works directly on the DNA. Enzymes called DNA methyltransferases add a methyl group to carbon 5 of cytosine, almost always where a C is followed by a G (a CpG site). Heavily methylated promoters are usually silent: methylated DNA attracts proteins that recruit histone deacetylases and compact the chromatin. Many promoters sit in CpG rich stretches called CpG islands that stay unmethylated in cells where the gene is active. Methylation patterns are copied when DNA replicates, because a maintenance methyltransferase recognizes a half methylated CpG on the new strand and methylates its partner. This copying lets a cell pass its pattern of silent genes to its daughters.

Transcription factors, promoters and enhancers

RNA polymerase II cannot recognize a promoter by itself. It needs general transcription factors (GTFs), a set of proteins that assemble at the core promoter. In many genes the core promoter contains a TATA box about 25 to 30 nucleotides upstream of the start site; the TATA binding protein, part of the factor TFIID, binds it and bends the DNA, and other factors and the polymerase join to form the transcription initiation complex. GTFs alone give only a low, basal rate of transcription. High rates, and the differences between cell types, come from specific transcription factors that bind control elements farther away.

Proximal control elements sit just upstream of the promoter. Distal control elements, grouped into enhancers, can lie thousands or even hundreds of thousands of base pairs upstream, downstream or inside an intron of the gene they control. Proteins that bind enhancers and stimulate transcription are activators; proteins that bind silencers and reduce it are repressors. A typical activator has a DNA binding domain that recognizes a specific sequence and an activation domain that contacts other proteins. DNA bending brings enhancer bound activators next to the promoter, where they contact the large mediator complex and the general factors, speeding assembly of the initiation complex. Activators also recruit histone acetyltransferases and chromatin remodeling complexes, opening the local chromatin. Insulators are boundary elements that keep an enhancer from acting on the wrong gene.

How can a few thousand transcription factors control 20,000 genes in hundreds of cell types? The answer is combinatorial control. Each enhancer contains binding sites for several different factors, and full activation requires a particular combination. A liver cell and a lens cell may both contain the same albumin and crystallin genes, but the liver cell has the right set of activators for the albumin enhancer and the lens cell has the set for the crystallin enhancer. A small number of factors, used in different combinations, can produce an enormous number of distinct patterns. Genes that must be expressed together, such as the genes a cell turns on in response to a steroid hormone, are coordinated not by being in an operon but by sharing the same control element in their enhancers, so that one activated factor binds all of them.

Control after transcription

Alternative splicing lets one gene make several proteins. By including or skipping particular exons, a cell can produce different isoforms from the same pre-mRNA. More than nine in ten human genes with multiple exons are alternatively spliced, which helps explain how a modest number of genes yields a far larger set of proteins.

Once in the cytoplasm, an mRNA's lifespan strongly affects how much protein it makes. Some mRNAs last minutes, others days.

Noncoding RNAs are a major regulatory force. MicroRNAs (miRNAs) are about 22 nucleotides long. A miRNA gene is transcribed into a precursor that folds into a hairpin; the enzyme Dicer trims the hairpin, and one strand is loaded into a protein complex (built around an Argonaute protein) called the RNA induced silencing complex. The miRNA then base pairs with matching sequences, usually in the 3' untranslated region of target mRNAs. Partial pairing blocks translation and speeds degradation; near perfect pairing triggers cleavage. One miRNA can regulate hundreds of mRNAs, and well over half of human protein coding genes are thought to be miRNA targets. Small interfering RNAs (siRNAs) are similar in size and use the same machinery but come from longer double stranded RNA, for example from viruses or experimentally introduced RNA. The silencing of genes by these small RNAs is called RNA interference (RNAi), and it is now used both as a research tool and in approved drugs. Longer long noncoding RNAs help organize chromatin, as XIST does in X inactivation.

Finally, cells control translation and protein lifetime. Initiation can be blocked globally, for instance by phosphorylating an initiation factor during stress, or for single mRNAs by proteins that bind regulatory sequences. Many proteins are inactive until cut or chemically modified. Proteins destined for destruction are tagged with chains of the small protein ubiquitin and fed into a proteasome, a barrel shaped protease complex. Timed destruction of regulatory proteins, such as cyclins, drives events like the cell cycle.

Epigenetics

Epigenetic inheritance refers to changes in gene expression that are passed through cell divisions (and sometimes across generations) without any change in the DNA sequence. DNA methylation and some histone marks are the best understood carriers of this cellular memory. When a skin stem cell divides, its daughters remain skin cells, partly because the pattern of silenced and open genes is copied along with the DNA.

Several classic phenomena are epigenetic:

• X chromosome inactivation. In female mammals one X chromosome in each cell of the early embryo is randomly chosen to be silenced. The chosen X expresses the long noncoding RNA XIST, which coats it and recruits repressive histone marks and DNA methylation. It condenses into a Barr body, and all descendants of that cell keep the same X silent. Females are therefore mosaics, as seen in the orange and black patches of calico cats.
• Genomic imprinting. For a small number of genes (on the order of 100 to 200 in humans), only the copy from one parent is expressed, because methylation marks laid down in egg or sperm silence the other. Loss of the same chromosome region on chromosome 15 causes Prader-Willi syndrome if the paternal copy is missing and Angelman syndrome if the maternal copy is missing.
• Environmental effects. Diet, stress, toxins and aging can shift methylation. In mice carrying a particular agouti allele, the mother's diet during pregnancy alters methylation of the gene and changes the coat color and obesity risk of her pups.

Section 3Development and differential gene expression

Three processes build a body

A human begins as a single fertilized egg, the zygote, and becomes a body of tens of trillions of cells. Three processes overlap to make this happen. Cell division increases the number of cells. Cell differentiation makes cells specialized in structure and function. Morphogenesis ("creation of form") shapes tissues and organs through cell movement, changes in cell shape, selective growth and programmed cell death. All three depend on cells expressing different genes from the same genome, a principle called differential gene expression.

The evidence that differentiated cells keep a complete genome came from nuclear transplantation. In frogs, nuclei taken from tadpole intestinal cells and placed into enucleated eggs sometimes produced swimming tadpoles. In 1996 the sheep Dolly was cloned from the nucleus of an adult mammary cell. These experiments showed that development does not usually remove genes; it changes which genes are used, and an egg's cytoplasm can, at least sometimes, reset a nucleus to an embryonic state.

Determination and differentiation

Before a cell looks specialized, its fate is often already fixed. Determination is the commitment of a cell to a particular fate; differentiation is the visible result, when the cell makes tissue specific proteins. Muscle offers the clearest example. Embryonic cells become myoblasts when they begin expressing MyoD, a master regulatory gene that encodes a transcription factor. MyoD activates more MyoD (a positive feedback loop that locks the decision in) and turns on genes for other muscle transcription factors, which in turn activate genes for myosin, actin and other muscle proteins. Myoblasts fuse into multinucleate muscle fibers.

How cells learn where they are

Cells acquire different fates through two kinds of information.

• Cytoplasmic determinants are maternal mRNAs and proteins placed unevenly in the egg. When the zygote divides, each daughter inherits a different share, so the nuclei in different cells are exposed to different regulators from the start.
• Induction is signaling between cells. Neighboring cells release or display signal molecules that change gene expression in target cells through signal transduction pathways. Induction is how the developing eye cup tells overlying ectoderm to form a lens, and how the notochord tells the ectoderm above it to form the neural plate.

Many signals form gradients. A substance that spreads from a source and tells cells their position by its concentration is a morphogen. Cells exposed to high, medium and low concentrations turn on different genes, so a single gradient can divide a field of cells into several zones. The overall process by which cells acquire a spatial organization, with a head here and a tail there and limbs in the right places, is called pattern formation, and the molecular cues involved are called positional information.

The fruit fly: a cascade from gradient to segment

Much of what we know about pattern formation comes from Drosophila melanogaster, whose larva and adult are built of repeated segments: head segments, three thoracic segments and eight or so abdominal segments. Researchers screened thousands of mutant embryos for defects in the body plan and found that a hierarchy of genes lays out the pattern step by step.

1. Maternal effect genes are expressed by the mother and their products are placed in the egg. The best known is bicoid. Its mRNA is anchored at the anterior end of the egg; after fertilization it is translated, and the protein diffuses backward, forming a concentration gradient. Embryos from mothers lacking bicoid have no head or thorax and instead develop tail structures at both ends. Injecting bicoid mRNA into the middle of an egg produces head structures there. Bicoid was the first morphogen to be identified, and it confirmed that a gradient of a single protein could set up an axis. Another maternal product, nanos, is concentrated at the posterior.
2. Gap genes, such as hunchback and Krüppel, are switched on in broad bands depending on the Bicoid concentration. A mutation in a gap gene removes a block of several adjacent segments, leaving a "gap".
3. Pair-rule genes, such as even-skipped and fushi tarazu, are expressed in seven stripes. Mutants lack every other segment.
4. Segment polarity genes, such as engrailed and wingless, are expressed in fourteen stripes and set the front and back of each segment.
5. Homeotic genes, also called Hox genes, specify what each segment becomes: antenna bearing, leg bearing, wing bearing and so on. Mutations transform one body part into another. The dominant Antennapedia mutation puts legs where antennae should be; loss of Ultrabithorax function turns the third thoracic segment into a copy of the second, giving a fly with four wings.

Each class of genes encodes transcription factors (or signaling proteins) that regulate the next, so a crude gradient is refined into a precise pattern. This is a pure example of differential gene expression: every nucleus has the same genes, but each reads a different combination.

Hox genes are shared by animals

Every Hox gene contains a 180 nucleotide stretch called the homeobox, which encodes a 60 amino acid DNA binding homeodomain. Homeobox genes are found in nearly all animals, as well as in plants and fungi. Flies have one Hox cluster; mammals have four clusters with 39 Hox genes. In both, the genes lie on the chromosome in the same order as the body regions they pattern, from head to tail, a striking property called colinearity. The deep conservation of this genetic toolkit is why much of evolution of body form reflects changes in where and when shared regulatory genes are expressed, rather than the invention of new genes, a theme taken up in later chapters.

Apoptosis and the worm

The nematode Caenorhabditis elegans has a fixed, fully mapped cell lineage: the adult hermaphrodite has 959 somatic cells, and exactly 131 other cells die by apoptosis (programmed cell death) during development. Genetic studies of these deaths identified the core death machinery (proteins called Ced-9, Ced-4 and Ced-3) whose human relatives include the Bcl-2 family and the caspases.

Stem cells and reprogramming

A stem cell can divide indefinitely and produce both more stem cells and specialized cells. Cells differ in how many fates remain open:

• A totipotent cell, such as the zygote and the earliest blastomeres, can form an entire organism, including the placenta and other extraembryonic tissues.
• A pluripotent cell can form every cell type of the body but not a whole organism. Embryonic stem cells, taken from the inner cell mass of a blastocyst, are pluripotent.
• A multipotent cell can form a limited set of related types. Adult hematopoietic stem cells in bone marrow make all blood cell types but not neurons.

In 2006 researchers showed that adding just four transcription factor genes (Oct4, Sox2, Klf4 and c-Myc) to adult mouse skin cells converted them into induced pluripotent stem cells (iPS cells) that behave much like embryonic stem cells. Human iPS cells followed a year later. Because iPS cells can be made from a patient's own tissue, they avoid immune rejection and many of the ethical objections to using embryos, and they are used to model diseases in a dish, screen drugs and develop cell replacement therapies.

Section 4DNA technology: PCR, cloning, sequencing and CRISPR

Cutting and joining DNA

Modern biotechnology began with enzymes borrowed from bacteria. Restriction enzymes (restriction endonucleases) protect bacteria by cutting foreign DNA, such as phage DNA, at specific short sequences called restriction sites, usually 4 to 8 base pairs long. The bacterium's own DNA is protected because its sites are methylated. Most restriction sites are palindromic: the sequence reads the same 5' to 3' on both strands. The enzyme EcoRI, for example, recognizes GAATTC and cuts between G and A on each strand. The staggered cuts leave short single stranded overhangs called sticky ends, which can base pair with any other fragment cut by the same enzyme. Other enzymes cut straight across, leaving blunt ends. DNA ligase then seals the sugar phosphate backbones, making a permanent recombinant DNA molecule.

Gene cloning in bacteria

To make many copies of a gene, researchers insert it into a cloning vector, a DNA molecule that can replicate in a host cell. The most common vectors are bacterial plasmids, small circular DNAs engineered to carry an origin of replication, an antibiotic resistance gene for selection and a multiple cloning site with several unique restriction sites. A typical workflow:

1. Cut the plasmid and the source DNA with the same restriction enzyme.
2. Mix them so sticky ends pair, and seal with DNA ligase.
3. Transform bacteria with the products, using heat shock or an electric pulse to let DNA enter.
4. Grow the cells on medium containing the antibiotic, so only cells that took up a plasmid survive.
5. Screen for colonies whose plasmid carries the insert. In blue-white screening, the cloning site lies inside part of the lacZ gene; plasmids without an insert make β-galactosidase and turn colonies blue on the indicator X-gal, while plasmids with an insert disrupt lacZ and give white colonies.

Each colony grows from one cell, so all its plasmids carry the same insert: a clone. Eukaryotic genes expressed in bacteria are usually cloned as cDNA (complementary DNA), made by copying mature mRNA with reverse transcriptase, because bacteria cannot remove introns. Collections of clones representing a whole genome or all the mRNAs of a tissue are called genomic libraries and cDNA libraries.

Gel electrophoresis

Gel electrophoresis separates DNA fragments by size. Samples are loaded into wells in a slab of agarose gel and an electric field is applied. Because DNA's phosphate backbone is negatively charged, fragments move toward the positive electrode. The gel acts as a sieve, so shorter fragments move farther than longer ones. A ladder of fragments of known size run beside the samples allows sizes to be estimated, and a fluorescent dye makes the bands visible.

The polymerase chain reaction

The polymerase chain reaction (PCR) copies a chosen stretch of DNA billions of times in a few hours without cells. A reaction contains the template DNA, two short single stranded DNA primers (each about 18 to 25 nucleotides) that flank the target and match opposite strands, the four dNTPs, and a heat stable DNA polymerase, classically Taq polymerase from the hot spring bacterium Thermus aquaticus. A thermal cycler repeats three steps:

1. Denaturation at about 94 to 98 °C separates the two strands.
2. Annealing at about 50 to 65 °C lets the primers base pair with their complementary sequences.
3. Extension at about 72 °C lets the polymerase add nucleotides to the 3' end of each primer.

Each cycle doubles the number of target molecules, so n cycles give up to 2ⁿ copies: 30 cycles produce about a billion. After the first few cycles, most products are exactly the length defined by the two primers. Heat stability matters because an ordinary polymerase would be destroyed in the first denaturation step.

Variations extend PCR's reach. RT-PCR first converts RNA to cDNA with reverse transcriptase, so it detects RNA viruses or measures gene expression. Quantitative (real time) PCR follows a fluorescent signal cycle by cycle; the earlier the signal crosses a threshold, the more template was present at the start. Diagnostic tests for many infections, including the respiratory virus tests used during the COVID-19 pandemic, are RT-PCR assays.

DNA sequencing

Sanger sequencing (the dideoxy chain termination method) copies a template with DNA polymerase in the presence of normal dNTPs and a small amount of dideoxynucleotides (ddNTPs). A ddNTP lacks the 3' hydroxyl group, so once it is added no further nucleotide can attach and the chain stops. Each of the four ddNTPs carries a different fluorescent dye. The result is a set of fragments ending at every position, each labeled by its final base. Capillary electrophoresis sorts them by length, and a detector reads the colors in order, giving the sequence. Sanger reads are about 700 to 1,000 bases and very accurate, and the method was the workhorse of the original Human Genome Project.

Next generation sequencing reads millions to billions of short fragments in parallel, usually by detecting each nucleotide as it is added in a cycle of synthesis, then reassembles them by computer. Together these methods have cut the cost of a human genome by roughly a millionfold since the first draft, making whole genome sequencing routine in research and increasingly in medicine. Sequencing RNA (as cDNA) in bulk, called RNA-seq, measures the expression of every gene in a tissue at once.

CRISPR-Cas9 genome editing

CRISPR (clustered regularly interspaced short palindromic repeats) began as a bacterial immune system. Bacteria store short pieces of viral DNA as "spacers" between repeats in their own genome, transcribe them into small guide RNAs, and use Cas nucleases guided by those RNAs to cut matching viral DNA in a later infection. In 2012 researchers showed that the Cas9 protein from Streptococcus pyogenes, combined with a single engineered guide RNA, could be directed to cut almost any chosen DNA sequence, work recognized with the 2020 Nobel Prize in Chemistry.

The guide RNA contains a 20 nucleotide spacer that matches the target, joined to a scaffold that binds Cas9. Cas9 scans DNA for a short protospacer adjacent motif (PAM), which for this Cas9 is NGG (any base followed by two G's). At a PAM, Cas9 unwinds the adjacent DNA; if the spacer base pairs with it, Cas9 makes a blunt double strand cut about 3 base pairs upstream of the PAM. The cell then repairs the break in one of two ways:

• Nonhomologous end joining (NHEJ) glues the ends back together but often adds or removes a few nucleotides. These small indels usually shift the reading frame and knock out the gene.
• Homology directed repair (HDR) copies a template with matching sequence on each side. If researchers supply a template carrying a desired change, the cell writes that change into the genome.

Newer variants avoid double strand breaks. Base editors fuse a disabled Cas9 to an enzyme that converts one base to another (for example C to T) directly, and prime editors carry their own short RNA template for small insertions, deletions or substitutions. A Cas9 with both cutting sites disabled can be fused to activators or repressors to switch genes on or off without changing the sequence at all.

Section 5Applications and ethics of biotechnology

Medicine

The first product of recombinant DNA approved as a drug was human insulin, made by bacteria carrying the human insulin gene, in 1982. It replaced insulin purified from pig and cattle pancreas and could be produced in unlimited amounts. Human growth hormone, clotting factors for hemophilia, erythropoietin for anemia, and many monoclonal antibodies are now made in engineered bacteria, yeast or cultured mammalian cells. The hepatitis B vaccine is a viral surface protein produced in yeast, and mRNA vaccines deliver an mRNA that the body's own cells translate into a viral protein.

Gene therapy introduces a working gene, or edits a faulty one, in a patient's cells. Common carriers are disabled viruses: adeno-associated virus vectors, which mostly stay outside the chromosomes, and lentivirus vectors, which integrate. Approved gene therapies now treat an inherited form of blindness, spinal muscular atrophy in infants and some blood disorders. In CAR T cell therapy, a patient's T cells are engineered to carry a receptor that recognizes a cancer cell protein, then returned to attack the cancer. In 2023 the first CRISPR based therapy was approved for sickle cell disease: blood stem cells are removed, an enhancer of the BCL11A gene is disrupted so that the cells resume making fetal hemoglobin, and the cells are returned.

Forensics and identification

Forensic DNA profiling compares short tandem repeats (STRs), sites where a 2 to 5 base unit is repeated a variable number of times. Each person has two alleles at each STR locus, one from each parent, and the number of repeats varies widely among people. PCR with fluorescent primers amplifies a standard panel of loci (20 in the United States database), and the products are sized by capillary electrophoresis. Because the loci are inherited independently, the chance that two unrelated people match at all of them is vanishingly small.

Agriculture and the environment

A genetically modified organism (GMO) carries DNA altered by genetic engineering; if the DNA came from another species it is transgenic. Plants are often transformed with the Ti plasmid of the soil bacterium Agrobacterium tumefaciens, which naturally transfers a segment of DNA into plant chromosomes. Important examples include Bt crops, which make a Cry protein from Bacillus thuringiensis that is toxic to certain insect larvae but not to mammals and so reduce insecticide spraying; herbicide tolerant crops; papaya engineered to resist ringspot virus, which rescued an island papaya industry; and Golden Rice, which makes β-carotene to combat vitamin A deficiency. CRISPR now allows small edits, such as knocking out a browning gene, without adding foreign DNA. Engineered microbes help with bioremediation (cleaning up oil and toxic metals) and with producing biofuels and industrial enzymes.

Ethical questions

Biotechnology raises questions that science alone cannot settle, and informed citizens need to weigh them:

A useful habit is to separate three questions: is it safe (an empirical question), is it effective (also empirical), and is it right (a question of values about consent, fairness and the kind of world we want). In 2018 the birth of twins whose embryos had been edited with CRISPR was widely condemned on all three grounds: the edit was medically unnecessary, its effects were uncertain, and the families and future children could not give truly informed consent.

Words worth knowing