Welcome to chapter four, foundations of biology. Every chapter after this one describes tissues, organs and systems, and every one of those is built from cells, run by cells, repaired by cells, and, when something goes wrong, usually broken at the level of cells. Chapter one introduced the cell's membrane and chapter three its molecules. This chapter puts them together: what a cell is, how a gene becomes a working protein, how cells divide, specialise and age, what bacteria and viruses are, and, the part that matters most for you, how biologists actually work and with what equipment.
Three practical reasons to know this. First, a great deal of modern medicine is applied cell biology. Insulin is made by bacteria, cancer is diagnosed by sequencing tumour DNA, vaccines are designed from viral genomes, and some leukaemias are treated by re-engineering the patient's own immune cells. Second, the laboratory is a technology-dense workplace, full of engineered instruments that are often expensive, closed and poorly suited to teaching. Optics, electronics, automation and software are all places a technologist is useful. Third, cells set the physical limits on what you can build: diffusion distances, growth rates, contamination and biofilm.
Biologists do not agree on one crisp definition of life, but living things share a set of properties, and each has an engineering reading. They are organised into cells. They run a metabolism, an open system kept far from equilibrium by a constant energy supply. They hold internal conditions steady by feedback. They grow by an internal programme, reproduce with an error rate, sense and respond, and evolve. Viruses sit on the boundary: they carry instructions and evolve, but have no metabolism and can only replicate inside a host cell. In medicine, what matters is that they are not cells, so the drugs that kill bacteria do nothing to them.
Two ideas make biology hang together. Cell theory says all living things are made of cells, the cell is the basic unit, and cells arise only from existing cells. That last point is why a sterile object stays sterile and why breaking the chain of infection works. Evolution by natural selection says that heritable variants that survive and reproduce better become more common. Three consequences. Bodies are historical compromises, not optimised machines: the airway crosses the food passage, the retina is wired backwards. Variation is the normal state, not noise. And microbes evolve fast. A billion bacteria dividing every half hour make antibiotic resistance evolution you can watch.
All cellular life falls into three domains. Bacteria and archaea are prokaryotes: small cells with no nucleus. Bacteria include most pathogens, the gut microbiome, and the strains used to make insulin. Archaea are biochemically distinct, live in extreme environments and in our gut, and cause no known human disease. Eukaryotes have a nucleus and internal compartments, and include animals, fungi, plants and single-celled protists such as the malaria parasite. Viruses infect members of every domain.
A bacterium is one to two micrometres long; a human cell ten to thirty across. Bacterial DNA is one circular chromosome plus small plasmids; ours is linear chromosomes inside a nucleus. Bacterial ribosomes are smaller, and most bacteria have a peptidoglycan cell wall we lack. These differences are exactly what antibiotics target. Penicillin attacks the cell wall; several families block the bacterial ribosome. A drug is selective only to the extent that its target differs between pathogen and patient, which is why antibacterials are plentiful, antifungals fewer, because fungi are eukaryotes like us, and antivirals hardest of all.
Here is a generalised animal cell. The nucleus holds about two metres of DNA packed into a sphere five to ten micrometres across. Ribosomes translate messenger RNA into protein; on the rough endoplasmic reticulum they make proteins for export, like antibodies and collagen, which the Golgi then modifies and sorts. Smooth ER makes lipids and, in the liver, detoxifies drugs. Mitochondria make most of the ATP and carry their own DNA, evidence that they were once free-living bacteria. Lysosomes digest. And the cytoskeleton gives shape, carries cargo, and lets cells sense mechanical load, which is the basis of bone remodelling.
Scale matters. A DNA helix is two nanometres wide, a membrane seven or eight, a ribosome about twenty-five, a virus twenty to three hundred. The light microscope's limit is about two tenths of a micrometre, so all of those are invisible to it, which is why so much of molecular biology is inferred from indirect measurement rather than seen. Above the limit sit bacteria at one or two micrometres, red cells at seven or eight, typical human cells at ten to thirty, an egg just visible at a tenth of a millimetre, and a nerve axon that can be a metre long.
Here is the machinery between gene and function. In the nucleus, DNA is the stored source, about three point one billion base pairs per set. A gene is transcribed into pre-messenger RNA, which is spliced and processed into mature mRNA. That is exported to the cytoplasm, where a ribosome reads it and links amino acids into a chain. The chain folds and is modified in the ER and Golgi into a working protein. And regulation acts at every step: which genes are read, how often, how long the mRNA lasts, and whether the protein is switched on.
DNA is two strands of four bases, A pairing with T and G with C, so either strand is a template for the other. The human genome is about three point one billion base pairs per set, with roughly twenty thousand protein-coding genes, which make up only one or two percent of the total. Copying it is a feat of process engineering. Polymerase works at tens of bases per second, so replication starts at tens of thousands of origins at once. Proofreading and mismatch repair bring the error rate to about one per billion bases. Chromosome ends carry protective telomeres that shorten with division. And DNA is damaged and repaired constantly; radiotherapy works partly by overwhelming that repair.
To use a gene, RNA polymerase copies it into messenger RNA. In our cells the first transcript contains non-coding stretches, introns, which are cut out and the exons joined. Alternative splicing lets one gene make several protein variants. A ribosome then reads the mRNA three bases at a time. There are sixty-four codons: sixty-one specify the twenty amino acids and three are stop signals. The code is redundant, so many single-base changes are silent. And it is nearly universal across life, which is what lets a bacterium make human insulin from a human gene. That is the foundation of biotechnology.
If you think in software terms, DNA is source code in long-term storage, messenger RNA a working copy checked out for execution, the ribosome an interpreter, the protein the running process, and regulation the configuration. It is a useful picture, and it breaks in three places. Nothing runs sequentially on one processor; thousands of copies run in parallel at rates set by chemistry. Bugs are often partial: a mutation may cut function by thirty percent rather than crash anything. And the configuration, which genes are on in which cell, is itself heritable within the body. That is epigenetics: chemical marks on DNA and its packaging that keep a liver cell's daughters liver cells.
Cells spend most of their time in interphase: G1 for growth and normal work, S for copying DNA, G2 for checking. Mitosis is short: a spindle pulls the chromosome copies apart and the cell splits in two. For cultured human cells the whole cycle is about a day. Some cells leave the cycle for good; most neurons and heart muscle cells never divide again, which is why damage there is largely permanent. Checkpoints, guarded by genes such as p53, stop a damaged cell dividing, and cancer is the failure of that control. That is also why chemotherapy hurts fast-dividing hair, gut lining and bone marrow. Meiosis makes eggs and sperm, halving and shuffling the chromosomes.
A fertilised egg becomes thirty to forty trillion cells of a few hundred recognised types, by progressive and largely stable changes in which genes are used. Stem cells can both renew themselves and give rise to specialised cells. Totipotent cells can make everything including placenta; pluripotent cells any body cell, including induced pluripotent cells reprogrammed from adult skin; multipotent cells a family, such as the blood-forming stem cells behind bone marrow transplants, the longest-established stem cell therapy. A caution: stem cell therapy is also the name of a large unregulated market. The question to ask is whether it is an approved therapy or a registered trial.
Bacteria are split by the Gram stain: Gram-positive with a thick wall, Gram-negative with an outer membrane that keeps drugs out. Some form spores that survive heat and disinfectants, and plasmids pass resistance genes between species. In fresh medium they show a lag, then exponential doubling, then a stationary phase. Viruses attach to specific receptors, hijack the cell to copy themselves, assemble and leave. And then biofilms: bacteria attached to a surface in a self-made matrix, tolerating antibiotics at hundreds of times the normal concentration. Biofilms on catheters, implants and endoscope channels are a major cause of device infection. A surface that can be colonised but not cleaned is the design problem.
The immune system is a distributed sensing and response network built from cells. Neutrophils are the first responders. Macrophages clear debris, coordinate repair, and wall off a foreign body, which is the foreign body response to an implant. Dendritic cells capture fragments of invaders and present them to T cells, linking the fast innate response to the slower, targeted adaptive one. T cells coordinate and kill infected cells; CAR-T therapy re-engineers them to attack cancer. B cells make antibodies, the basis of immunoassays and antibody drugs. And memory cells give a faster second response, which is why vaccines work.
Two pieces of arithmetic explain a great deal. First, surface to volume: for a sphere it is three over the radius, so bigger cells have less surface per unit volume, which is why cells are small, flat or folded. Second, diffusion time scales with the square of distance. For oxygen in water, ten micrometres takes about twenty-five milliseconds, a hundred micrometres about two and a half seconds, but a millimetre about four minutes and a centimetre seven hours. So most cells sit within a hundred or two hundred micrometres of a capillary. That is why thick tissue-engineered constructs die in the middle, why cartilage heals poorly, and why a tight cast can starve tissue.
Exponential growth is the second theme. One bacterium doubling every twenty minutes becomes a billion in ten hours, which is why a warm contaminated solution left overnight is a different object in the morning. Human cells doubling daily go from a million to a billion in about ten days, the timescale of making a CAR-T dose. PCR ideally doubles its target each cycle, but at ninety percent efficiency thirty cycles give five times less, so quantitative PCR must measure efficiency, not assume it. And the genome as data: two bits per base is about seven hundred and seventy-five megabytes, and two people differ at roughly one base in a thousand.
Light microscopes are limited by diffraction to about two hundred nanometres, so bacteria are visible and viruses are not. Super-resolution techniques reach tens of nanometres, and electron microscopes, especially cryo-electron microscopy, reach near-atomic detail. Fluorescence imaging counts photons, so the square-root law applies, but more excitation light bleaches dyes and damages living cells; every live-imaging protocol is a negotiation, like dose in radiography. And centrifuges are specified in multiples of gravity, not rpm, because force depends on radius. Three thousand rpm at ten centimetres is about a thousand g. A protocol written in rpm does not transfer between rotors.
Genetics and genomics run newborn screening, diagnose rare diseases by genome sequencing, and sequence tumours to choose targeted drugs. Microbiology and virology run hospital labs, infection control and sterility testing of devices. Immunology gives vaccines, monoclonal antibody drugs and immunoassays. Biomanufacturing uses living cells as factories: recombinant insulin since the nineteen-eighties, antibodies in bioreactors, and mRNA vaccines. Cell and gene therapy uses cells or genes as the medicine, including CAR-T and, since twenty twenty-three, the first approved CRISPR therapy. Tissue engineering and organ-on-chip build tissue and test drugs. And bioinformatics is the most direct entry point for a software-literate engineer.
Almost every experiment, research or diagnostic, follows the same shape. A sample, whether blood, a swab, tissue or cultured cells, is sometimes grown, then separated and prepared, then measured, then analysed in software, and finally recorded and stored. The tools of the lab are easiest to understand by where they sit in that chain, so the next few slides go through it in order.
For seeing: brightfield microscopy for stained tissue and blood films, phase contrast for living unstained cells, fluorescence and confocal for labelled proteins and three-dimensional images, light-sheet for gentle imaging of whole embryos, and electron and cryo-electron microscopy for nanometre detail. For growing: a biological safety cabinet with filtered airflow protects worker, sample and room, and it is not a fume hood. A carbon dioxide incubator holds thirty-seven degrees, five percent carbon dioxide and high humidity, an embedded-control product. And bioreactors scale up to thousands of litres, where transferring oxygen without shearing the cells is the central engineering problem.
For preparing: the micropipette is the most-used instrument in biology, and its accuracy depends on calibration and technique. Liquid-handling robots pipette across plates of ninety-six or more wells. Centrifuges pellet cells and separate blood. Extraction purifies DNA or RNA, the first step in every PCR test, and gel electrophoresis sorts fragments by size, at voltages that are a real electrical hazard. For measuring: plate readers read every well of a microplate, flow cytometers classify thousands of cells per second with lasers, Western blots and ELISAs detect specific proteins, mass spectrometry identifies bacteria in minutes, and patch-clamp rigs measure currents through single ion channels.
PCR and quantitative PCR amplify a chosen sequence by repeated heating and cooling; thermal uniformity and ramp rate are the engineering problems. Sanger sequencing reads one fragment very accurately. Short-read sequencing reads billions of short fragments in parallel, and a human genome now costs hundreds of dollars in reagents at scale. Long-read sequencing reads tens of thousands of bases, and nanopore devices, which measure current changes as DNA threads through a pore, can be pocket-sized. And CRISPR uses a short guide RNA to direct a cut at a chosen sequence, used in research, in approved therapy, and in rapid diagnostics.
Samples and cells are stored in freezers down to minus eighty and in liquid nitrogen at minus a hundred and ninety-six, and freezer failure alarms are a classic monitoring problem. Biobanks curate samples with consent and linked data; laboratory information systems and electronic notebooks keep the records. On the software side, NCBI, Ensembl and UniProt are the free reference databases, and BLAST answers "what is this sequence?" in seconds. Biopython, Bioconductor and Galaxy run analysis. Fiji, CellProfiler and napari count and measure cells in images. And AlphaFold, recognised by a Nobel Prize, predicts protein structures from sequence.
Labs that work with living organisms are classified by biosafety level. Level one is for organisms not known to cause disease in healthy adults, on an open bench with good practice. Level two covers moderate-risk organisms and most clinical samples, with safety cabinets, protective equipment and restricted access; hospital labs work largely here. Level three is for serious airborne organisms such as tuberculosis, and level four for Ebola and Marburg. The rule for you is simple. Do not culture microorganisms or human cells outside a properly supervised lab. A plate grown from a swab is a culture of unknown organisms at numbers far higher than the original sample.
Biology labs are full of instruments that are expensive, proprietary, or poorly suited to teaching, and open hardware has become a serious movement in response. Open-source microscopes combine flexure stages, off-the-shelf optics, a camera and software. Temperature control for incubators and heat blocks is a classic embedded problem. Custom labware holders are a constant need. Teaching bioreactors and microfluidic chips make excellent integrated projects, and image analysis needs no lab at all. Two boundaries. Centrifuges store dangerous energy and gel power supplies use hazardous voltages, so build accessories, not rotors. And an instrument used for patient results must be validated.
A teaching lab has one commercial gel imager for forty-eight students and asks whether you could build twelve. Understand the measurement first: the stain fluoresces green under blue light, so the imager needs even blue excitation, an amber filter to block it, and no room light. Then ask questions. Use blue-light stains only, to avoid a UV hazard. Make the stage wipeable with the lab's own disinfectant, which decides between acrylic and powder-coated sheet metal. Decide phones or a fixed camera. And keep it entirely separate from the gel tank's power supply. Define acceptance with the lab: a five nanogram band visible, even light across ten centimetres. And price twelve commercial units before building anything.
Exercises. Look at onion skin and baker's yeast under any microscope and measure the cells. Extract DNA from strawberries with washing-up liquid, salt and cold alcohol, and explain what each ingredient did. Watch food colouring diffuse through gelatin and plot spread against the square root of time. Watch yeast grow exponentially with balloons or a carbon dioxide sensor. Find the insulin gene in NCBI and translate it. Count nuclei in a public image and see how the count changes with the threshold. And do the centrifuge arithmetic. Keep to kitchen materials and food-grade yeast.
The limits. Knowing what a flow cytometer measures is not knowing how to run, quality-control or interpret one; biomedical scientists are trained and often registered professionals. You may not handle clinical samples or culture organisms except within a lab's training and biosafety framework. A home-built instrument is not a validated one. Understanding how a CAR-T product is made says nothing about who should receive it. And the field moves quickly, so treat specific numbers as orientation. Next, tissues, skin, and contact with an object.