Life, considered as a chemistry problem. Proteins, sugars, fats, and nucleic acids; metabolism, signalling, and the molecular logic of the living cell.
Cells are wet bags of organic chemistry. Biochemistry is the study of those reactions — what molecules, organised how, doing what. Reductive in method; vast in subject matter.
Biochemistry sits between molecular biology (which works with DNA, RNA, and proteins) and chemistry (which works with smaller molecules). Its concerns are metabolism, the macromolecules that catalyse and structure life, the regulatory networks, and increasingly the systems-scale dynamics.
This deck covers the four families of biological macromolecules; the central catabolic and anabolic pathways; enzymes; signalling; and the modern toolkit (structural biology, omics, AI for protein structure). It assumes you remember high-school chemistry.
The chemical understanding of life is recent. Until 1828, "organic" molecules were thought to require a "vital force" only living things possessed. Friedrich Wöhler's synthesis of urea (NH₂)₂CO from inorganic ammonium cyanate broke that wall. Organic chemistry became chemistry.
Eduard Buchner's 1897 demonstration that cell-free yeast extract still ferments sugar to alcohol — that fermentation does not require living cells, just enzymes — won the 1907 Nobel and effectively founded biochemistry as a separate discipline.
The early 20th century filled in the basics. Vitamins (Funk, 1912). Amino acid composition of proteins (Fischer's century). The Lipmann/Krebs/Cori metabolic pathway era (1930s-50s). Pauling's protein structures (1951). Watson-Crick-Franklin DNA (1953). Sanger sequencing (1955 insulin, 1977 DNA). The recombinant DNA revolution (Cohen and Boyer, 1973).
Cells are 70% water. Water's properties — high specific heat, high latent heat of vaporisation, high dielectric constant, ability to form hydrogen bonds — make it the universal biological solvent. Most biochemistry is what happens in water.
Hydrophobic effect. Non-polar molecules don't dissolve well in water; they cluster together to minimise the disruption of water's hydrogen-bond network. This drives lipid bilayer formation, protein folding (hydrophobic residues bury inside, hydrophilic on the surface), and most molecular recognition.
pH. Cells maintain pH near 7. Many enzymes work only in narrow pH bands. Buffer systems (bicarbonate-CO₂ in blood, phosphate in cells) keep pH stable against metabolic acid/base loads.
Without water's hydrogen bonding and dielectric properties, none of this works. Some astrobiologists speculate about ammonia or hydrocarbon-based life; the chemical case for water is hard to beat.
The workhorses. Polymers of 20 standard amino acids, linked by peptide bonds. A typical cell has 10⁴ different protein species; the human proteome has roughly 20,000 protein-coding genes producing perhaps 100,000 protein variants after splicing and modification.
Four levels of structure:
Primary. The amino acid sequence, encoded by DNA.
Secondary. Local structures stabilised by backbone hydrogen bonds — α-helices and β-sheets, identified by Linus Pauling (1951).
Tertiary. The 3D fold of a single polypeptide. Determined by the sequence (Anfinsen's rule, 1961). Predicted with high accuracy by AlphaFold (DeepMind, 2020-21) — one of the genuinely transformative scientific results of the AI era.
Quaternary. Assemblies of multiple polypeptide chains — hemoglobin (4 chains), the ribosome (~80 chains), the proteasome.
Function follows structure follows sequence. Mutations that destabilise structure produce disease (sickle-cell anaemia, cystic fibrosis, prion diseases).
Enzymes are catalysts — they speed up reactions without being consumed. Most enzymes are proteins (some are RNA — ribozymes — including the catalytic core of the ribosome). Enzymes accelerate reactions by factors of 10⁶ to 10²² over uncatalysed rates.
v = Vmax · [S] / (Km + [S]) Michaelis-Menten
The Michaelis-Menten equation (1913) describes enzyme kinetics. Vmax is the maximal rate at saturating substrate; Km is the substrate concentration giving half-maximal rate (a measure of affinity).
Enzymes work by stabilising transition states, positioning reactants in geometries impossible in solution, and excluding water from active sites where it would interfere. The active site is typically a few amino acids that do the chemistry.
Enzymes are regulated — by allosteric binding (cooperative, hemoglobin-style), covalent modification (phosphorylation), proteolytic activation (zymogen → active enzyme), and synthesis/degradation rates. Most metabolic regulation works by enzyme regulation.
DNA and RNA are polymers of nucleotides — each nucleotide a sugar (deoxyribose or ribose), a phosphate, and one of four bases (A, T or U, G, C). Phosphodiester bonds link the sugars; the bases pair across strands (A-T/U, G-C) via hydrogen bonds.
DNA is the genetic information storage molecule. Double-stranded, helical (Watson-Crick-Franklin, 1953). The base sequence encodes proteins (in coding regions, ~1.5% of human DNA) and regulatory information (the other 98.5% is mostly not "junk" — it's regulatory, structural, or evolutionary residue).
RNA is the working molecule. mRNA carries protein-coding information from DNA to ribosome. tRNA delivers amino acids during translation. rRNA forms the catalytic core of the ribosome. Other RNAs (microRNA, lncRNA, circRNA) regulate gene expression.
The "RNA world" hypothesis proposes that early life used RNA both as information storage and catalyst — DNA and proteins came later. Ribozymes (catalytic RNAs) and the ribosome (an RNA machine) support this picture.
Carbohydrates. Sugars and their polymers. Glucose (energy currency, blood sugar). Glycogen (animal energy storage). Starch (plant energy storage). Cellulose (plant structural). Chitin (insect and fungal structural). Glycoproteins and glycolipids (cell-surface recognition, blood types). The "sugar coding" of cell surfaces — glycomics — is a growing area; sugars determine much of cell-cell recognition.
Lipids. Hydrophobic biomolecules. Fatty acids (energy storage and signalling). Triglycerides (long-term energy storage in adipose tissue). Phospholipids (membrane components — lipid bilayers form the boundary of every cell and most organelles). Steroids (cholesterol, hormones — testosterone, estrogen, cortisol).
The cell membrane is a fluid bilayer of phospholipids, ~5 nanometres thick, with embedded proteins (channels, receptors, pumps, structural). Singer and Nicolson's "fluid mosaic" model (1972) is still the working picture, refined by lipid-raft and membrane-microdomain concepts.
Cells run on ATP — adenosine triphosphate. Hydrolyzing the terminal phosphate (ATP → ADP + Pᵢ) releases ~30 kJ/mol that drives most cellular work — biosynthesis, ion pumping, mechanical motion, signalling.
ATP + H₂O → ADP + Pᵢ ΔG ≈ -30 kJ/mol
An average human turns over their body weight in ATP every day. The pool is small (~250 g of ATP at any moment); it cycles dozens of times per minute.
ATP is regenerated from ADP by:
Substrate-level phosphorylation (in glycolysis and the TCA cycle).
Oxidative phosphorylation (in mitochondria — the major source for most cells).
Photophosphorylation (in chloroplasts, using light energy).
The other major energy currencies are NADH and FADH₂ (reducing power, mostly used in oxidative phosphorylation) and NADPH (reducing power for biosynthesis).
Glycolysis — the breakdown of glucose to pyruvate — is the universal core of metabolism. Every living thing uses some version of it. Ten enzymatic steps in the cytoplasm, no oxygen required.
Net: 1 glucose → 2 pyruvate + 2 ATP + 2 NADH
The pathway was worked out in the 1930s and 1940s — Embden, Meyerhof, Parnas, Cori. The whole sequence is encoded in 10 highly conserved enzymes; the same chemistry runs in E. coli, yeast, and human muscle.
Glycolysis is regulated at three control points. Hexokinase (entry), phosphofructokinase (commitment step), pyruvate kinase (exit). Each is responsive to cellular energy state — ATP, AMP, citrate, fructose-2,6-bisphosphate as allosteric effectors.
Pyruvate's fate depends on conditions. With oxygen: enters the TCA cycle. Without oxygen: fermentation to lactate (in muscle and cancer cells) or ethanol (in yeast).
The tricarboxylic acid cycle (also Krebs cycle, citric acid cycle) is metabolism's central hub. Hans Krebs worked it out in 1937 (Nobel 1953). Eight steps in the mitochondrial matrix; net oxidation of an acetyl group to two CO₂.
Per acetyl-CoA: 3 NADH + 1 FADH₂ + 1 ATP (or GTP) + 2 CO₂.
The TCA cycle is amphibolic — both catabolic (oxidation for energy) and anabolic (its intermediates are starting points for biosynthesis of amino acids, heme, fatty acids). Many disease states (cancer, hereditary metabolic disorders, oncometabolites like 2-hydroxyglutarate) involve TCA cycle dysregulation.
The full path of glucose oxidation: glucose → glycolysis → pyruvate → pyruvate dehydrogenase → acetyl-CoA → TCA cycle → NADH/FADH₂ → electron transport chain → ATP. Net per glucose: ~30-32 ATP.
The 30+ ATP per glucose come not from substrate-level phosphorylation but from the electron transport chain in the mitochondrial inner membrane. NADH and FADH₂ donate electrons; the chain pumps protons across the inner membrane, building a proton gradient.
Peter Mitchell's chemiosmotic theory (1961) — initially considered absurd — proposed that this proton gradient itself, not a chemical intermediate, is the energy form. Protons flow back through ATP synthase, a rotary molecular machine that uses the gradient to phosphorylate ADP. Mitchell got the 1978 Nobel after years of skepticism.
ATP synthase is one of the most beautiful machines in biology. A rotor in the membrane spins driven by proton flow; the rotation drives conformational changes in three catalytic subunits, each of which makes one ATP per third-revolution. ~100 ATP per second per ATP synthase.
Chloroplasts use the same machinery, driven by photosynthetic electron transport. Bacteria do it on the inner membrane. The mechanism is universal.
The reverse direction. Plants, algae, and cyanobacteria use light energy to fix CO₂ into glucose, releasing O₂.
6 CO₂ + 6 H₂O + light → C₆H₁₂O₆ + 6 O₂
Two stages:
Light reactions (thylakoid membranes). Chlorophyll absorbs photons, water is split (releasing O₂), electrons travel through Photosystem II → Photosystem I → NADPH. ATP is made by chemiosmosis.
Calvin cycle (stroma). The enzyme RuBisCO fixes CO₂ onto a 5-carbon sugar, generating two 3-carbon molecules. ATP and NADPH from the light reactions reduce these into glucose precursors.
RuBisCO is the most abundant protein on Earth — roughly 700 million tons. It's also slow (3-10 turnovers/second vs. 1000+ for typical enzymes) and confuses CO₂ with O₂ (photorespiration). Improving RuBisCO is a long-running goal of crop science.
Cyanobacteria invented oxygenic photosynthesis ~2.5 billion years ago, oxygenating Earth's atmosphere — the Great Oxidation Event.
Watson and Crick's 1953 paper ended with a famously understated sentence: "It has not escaped our notice that the specific pairing we have postulated immediately suggests a possible copying mechanism for the genetic material."
The mechanism: DNA replication is semi-conservative. Each strand templates a new complementary strand. Meselson and Stahl (1958) confirmed this with elegant density-gradient centrifugation.
The machinery: helicases unwind the double helix. Primase lays down RNA primers. DNA polymerases extend, reading the template, picking the complementary base, catalysing the phosphodiester bond. Ligase seals nicks. Topoisomerases manage twisting. The replisome is one of the most complex machines in cells — 10+ proteins working together.
Polymerase fidelity: ~1 mistake in 10⁹ bases. This is achieved by base-pairing (10⁻⁴-ish), 3'→5' proofreading exonuclease (10⁻⁶ish), and post-replication mismatch repair (down to 10⁻⁹). Defects in mismatch repair cause Lynch syndrome (hereditary colon cancer).
Proteins are synthesised on ribosomes. The ribosome reads mRNA in groups of three nucleotides (codons), each codon specifying one amino acid via tRNA matching. The genetic code (Nirenberg, Khorana, 1961-66) translates 64 codons into 20 amino acids plus stop.
The ribosome is itself an RNA-protein machine — bacterial ribosomes are ~65% RNA, ~35% protein. The catalytic site (peptidyl transferase) is RNA. The ribosome is a ribozyme. Ada Yonath, Tom Steitz, Venki Ramakrishnan got the 2009 Nobel for solving its atomic structure.
Translation rate: bacteria ~20 amino acids/second; eukaryotes ~5/second. A typical protein takes seconds to minutes to make. Multiple ribosomes can translate one mRNA simultaneously (polysomes), increasing throughput.
Translation is the principal target of most clinically important antibiotics — chloramphenicol, erythromycin, tetracycline, streptomycin all target bacterial ribosomes. Antibiotic resistance has been a slow-moving public-health crisis since the 1990s.
Cells respond to external chemical signals via receptor proteins on the plasma membrane (or, for steroid hormones, intracellular). Major receptor families:
G-protein coupled receptors (GPCRs). ~800 in humans. Targets of ~30% of all FDA-approved drugs. Bind hormone, neurotransmitter, light, smell, taste; activate intracellular G proteins; trigger secondary messenger cascades. 2012 Nobel to Lefkowitz and Kobilka for structural work.
Receptor tyrosine kinases. ~60 in humans. Bind growth factors; dimerise; phosphorylate themselves and downstream substrates. EGFR mutations drive lung cancer. HER2 amplification drives some breast cancers.
Ion channel receptors. The nicotinic acetylcholine receptor opens an ion channel directly upon ligand binding. Underlies neuromuscular transmission.
Nuclear receptors. Steroid and thyroid hormones bind cytoplasmic receptors that translocate to nucleus and act as transcription factors directly.
Downstream: kinase cascades (MAPK, PI3K-AKT-mTOR), second messengers (cAMP, Ca²⁺, IP3, DAG), and ultimately transcription factor activation.
Christian Anfinsen's 1961 ribonuclease experiment showed that the sequence determines the fold — denatured RNase reassembles its native structure. Anfinsen got the 1972 Nobel.
But how? Levinthal's paradox (1969): a 100-residue protein has 10⁹⁵ possible conformations. Random search would take longer than the universe's age. Yet proteins fold in milliseconds.
The answer is hierarchical, funneled folding. Local secondary structures form fast. Hydrophobic collapse drives non-polar residues inward. The native fold is the bottom of an energy funnel; intermediate states channel toward it.
Some proteins need help — chaperones (Hsp70, Hsp90, GroEL) prevent misfolding and aggregation. Misfolded proteins cause Alzheimer's (amyloid β, tau), Parkinson's (α-synuclein), prion diseases (PrP), and some cancers (mutant p53).
The protein folding problem — predicting structure from sequence — was the central open challenge in biochemistry for sixty years. AlphaFold 2 (DeepMind, 2020) substantially solved it — Demis Hassabis and John Jumper share the 2024 Chemistry Nobel for this work.
Knowing what molecules look like is foundational. Three main techniques:
X-ray crystallography. Hen egg-white lysozyme was the first enzyme structure (Phillips, 1965). Today the Protein Data Bank holds 200,000+ structures, mostly from crystallography. The technique requires growing protein crystals — sometimes the hardest step. Synchrotron sources (the LCLS, ESRF, Diamond, APS) provide bright X-rays for diffraction.
NMR spectroscopy. For smaller proteins (≤30 kDa) in solution. Gives dynamics in addition to structure. Wüthrich Nobel 2002.
Cryo-electron microscopy. Flash-freezes proteins in glassy ice; computer-reconstructs structure from thousands of 2D projections. The "resolution revolution" (Bai et al. 2015, advances by Henderson, Frank, Dubochet — Nobel 2017) brought cryo-EM to atomic resolution. Now competes with crystallography for many problems and excels at large complexes (ribosomes, viruses, membrane proteins).
And AlphaFold prediction. 2024 Nobel for Hassabis, Jumper (and David Baker for protein design). AlphaFold has now predicted structures for ~200 million proteins — essentially every known protein.
Sanger sequencing (1977) read DNA one base at a time using chain-terminating ddNTPs. The Human Genome Project (2003) used industrialised Sanger sequencing — $3 billion, 13 years.
Next-generation sequencing (Illumina, ~2007) cut cost by orders of magnitude. The first human genome at $1000 (2014, NIH milestone). Today: a complete human genome at $200, an exome at $50, in 24-48 hours.
Long-read sequencing (Pacific Biosciences, Oxford Nanopore) reads single molecules over kilobases — solving the assembly problem for repetitive regions. The first complete human genome (T2T-CHM13, 2022) was made possible by long reads.
The "omics" suffix proliferated: genomics (DNA), transcriptomics (RNA, especially via RNA-seq), proteomics (mass spectrometry), metabolomics (small molecules), epigenomics (DNA modification), microbiome (the bacterial communities on/in us).
Single-cell omics (10x Genomics, ~2015) profile expression in individual cells; the Human Cell Atlas project aims to catalog all cell types.
CRISPR-Cas9 — the bacterial adaptive immune system that became the universal genome editor. Discovered as bacterial sequences in 1987 (Ishino), recognised as immune system in 2007 (Barrangou et al.), repurposed for editing in 2012 (Jennifer Doudna and Emmanuelle Charpentier — Nobel 2020).
The mechanism: a guide RNA targets Cas9 (a DNA endonuclease) to a specific genomic site. Cas9 cuts both strands. The cell repairs — either by error-prone end-joining (knocking the gene out) or, in the presence of a template, by homology-directed repair (precise edits).
Development since 2012 has been rapid. Base editing (Liu lab, 2016): single-base changes without cutting. Prime editing (Liu lab, 2019): arbitrary edits with reduced collateral damage. Casgevy (Vertex/CRISPR Therapeutics, 2023): the first FDA-approved CRISPR therapy, for sickle-cell disease.
The 2018 He Jiankui controversy — illegal CRISPR-edited human embryos — set off a global ethics conversation that has only partially settled.
Biochemistry and pharmaceutical development are inseparable. Drugs work by binding biomolecular targets — usually proteins (enzymes, receptors, ion channels, transporters), occasionally nucleic acids.
Small-molecule drugs. Aspirin (Bayer 1899). Sulfonamides (Domagk 1932). Penicillin (Fleming 1928, Florey-Chain 1940s). Statins (Endo 1976). The kinase inhibitor revolution (imatinib for CML, 2001).
Biologics. Recombinant insulin (1982, the first recombinant drug). Erythropoietin. Monoclonal antibodies (CD20: rituximab; HER2: trastuzumab; PD-1: pembrolizumab — the cancer immunotherapy revolution).
Vaccines. Smallpox (Jenner 1796 — eradicated 1980). The mRNA platform (Karikó and Weissman, with Pfizer/BioNTech and Moderna 2020-21) — Nobel 2023.
GLP-1 agonists. Semaglutide (Ozempic, Wegovy) and tirzepatide (Mounjaro, Zepbound) — the most consequential pharmaceutical class of the 2020s. Originally developed for type 2 diabetes; the obesity indication has reshaped the industry.
Cancer is fundamentally a disease of disregulated cellular biochemistry. Hanahan and Weinberg's "Hallmarks of Cancer" (2000, expanded 2011, 2022) catalog ten signatures:
Sustained proliferative signalling. Evasion of growth suppressors. Resistance to cell death. Replicative immortality. Sustained angiogenesis. Invasion and metastasis. Reprogrammed energy metabolism (the Warburg effect — cancer cells often prefer glycolysis even in oxygen). Immune evasion. Tumor-promoting inflammation. Genome instability.
The molecular logic: most cancers begin with mutations in proto-oncogenes (gain-of-function in growth-promoting genes — KRAS, MYC, EGFR) or tumor suppressors (loss-of-function in growth-restraining genes — TP53, RB, BRCA1/2).
Therapy follows molecular logic. Targeted therapies (kinase inhibitors against specific oncogenic drivers — imatinib, gefitinib, vemurafenib). Immunotherapy (checkpoint inhibitors against PD-1/PD-L1 and CTLA-4 — Allison and Honjo, Nobel 2018; CAR-T cells; bispecific T-cell engagers). Antibody-drug conjugates (trastuzumab deruxtecan).
You contain ~10⁴ species of bacteria, archaea, fungi, and viruses, mostly in your gut, totalling roughly the same number of cells as your "own" body cells — a few kilograms of microbial biomass.
The microbiome makes vitamins (K, B12, biotin), digests fibre humans can't, modulates immune system development, influences neurotransmitter production. Disturbances correlate with inflammatory bowel disease, obesity, depression, autism, cardiovascular disease.
Modern microbiome research relies on 16S rRNA sequencing (taxonomy) and shotgun metagenomics (function). The Human Microbiome Project (2007-2016) was the inventory phase. Causation studies are now the focus.
Therapeutic implications: fecal microbiota transplant for C. difficile infection (FDA-approved Rebyota, 2022). Live biotherapeutics. Diet- and prebiotic-based interventions. The clinical translation has lagged the basic science.
The cell is genuinely complicated. A short list of open problems:
Aging. Why do cells (and we) get old? Hallmarks proposed (López-Otín et al. 2013, 2023): genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis, mitochondrial dysfunction, cellular senescence, stem cell exhaustion, altered intercellular communication, dysbiosis, etc. Synthesis still pending.
Origin of life. How did the first self-replicating system arise? RNA world is the leading hypothesis but the path from prebiotic chemistry to functional ribozymes remains speculative.
Conditions for protein folding rules. AlphaFold predicts structures but doesn't fully explain folding pathways. Misfolding diseases remain hard to treat.
Membraneless organelles. Phase-separated condensates (P-granules, stress granules, the nucleolus) are a discovered organising principle of cells, only explored in the last decade.
The gut-brain axis. Real but mechanistically unclear.
↑ The Krebs Cycle and metabolism
Watch · DNA replication (updated)
Watch · The web of complex biology
Three paths.
For non-scientists. Mukherjee's The Gene and The Song of the Cell are the lyrical introductions. Lane's Power, Sex, Suicide is the best book on mitochondria. Goodsell's The Machinery of Life for visualisation.
For undergraduate biology and chemistry. Nelson and Cox's Lehninger Principles of Biochemistry is the universal undergraduate textbook. Stryer's Biochemistry is the classic alternative. Alberts et al. Molecular Biology of the Cell for cell-biology focus.
For depth and current research. The reviews in Annual Review of Biochemistry, Cell, Nature Reviews Molecular Cell Biology. Specific monographs by topic: Cooper's cell, Pollard's cell biology by the numbers (data-rich), Sigee's Drug Discovery for pharmacology.
For tools. Beginning bioinformatics: Bourne's The Bioinformatics Workshop. Beginning structural biology: Drenth's Principles of Protein X-ray Crystallography. AlphaFold, ColabFold, and ESMFold are the practical AI structural-biology tools.
Three claims.
It is the bridge science. Biology asks what living things do. Chemistry asks what molecules do. Biochemistry asks what living things are made of and how those things do what they do. Without biochemistry the bridge is missing — biology becomes natural history; chemistry becomes synthesis without function.
It is the foundation of medicine. Most drugs target proteins. Most diagnostics measure metabolites or proteins. Most genetic diseases are biochemistry diseases (a missing or malfunctioning enzyme). Most therapies of the last 40 years (recombinant biologics, monoclonal antibodies, kinase inhibitors, mRNA vaccines, GLP-1 agonists, CRISPR) come from biochemistry.
It is in a productive boom. The 2010s and 2020s have been one of biology's best decades. AlphaFold solved the folding problem. CRISPR became a clinical tool. mRNA vaccines worked at planetary scale. GLP-1 drugs are reshaping public health. Single-cell sequencing changed our basic taxonomy of cell types. The pace is faster than at any time in the field's history.
Four directions worth watching.
Designed proteins. David Baker's lab has shown that machine learning can design entirely new proteins with arbitrary specified functions. RFdiffusion and ProteinMPNN are the public tools. Therapeutic candidates designed this way (for cancer, autoimmunity) are entering clinical trials.
In silico biology. AlphaFold, ESMFold, RoseTTAFold — predicting structure. Then predicting interactions, dynamics, and ultimately whole-cell models. Karr et al.'s 2012 M. genitalium whole-cell model was the proof of concept; the field is now scaling.
Editing without cutting. Base editing and prime editing reduce the off-target risk of standard CRISPR. The Casgevy approval (2023) is the first; many more in trials.
Aging therapeutics. Senolytics, partial reprogramming (Yamanaka factors with off-switches), rapamycin-class mTOR modulators. Altos Labs is the most-watched private effort. Whether biological aging will be tractably modifiable in this decade is unclear; the field is well-funded enough to find out.
Biochemistry — Volume III, Deck 14 of The Deck Catalog. Set in Source Serif Pro with monospace metadata. Cool botanical paper #f3f7f2; leaf-green and berry accents.
Twenty-eight leaves on the molecular logic of life. The wettest of sciences.
↑ Vol. III · Sci. · Deck 14