Carbohydrate chemistry: monosaccharides, glycosidic bonds, polysaccharides, and glycoconjugates

Intuition Beginner

Carbohydrates are sugars and their polymers. The simplest sugars are monosaccharides — single-molecule units with the general formula ${\text{C}}_n({\text{H}}_2\text{O}{)}_m$. Glucose is the most important: a six-carbon sugar that serves as the primary fuel for cellular metabolism. Fructose, found in fruit, is another six-carbon sugar with the same atoms arranged differently.

Two monosaccharides can join together through a glycosidic bond, formed when a hydroxyl group ($\text{-OH}$) on one sugar reacts with the hydroxyl group on another, releasing a water molecule. The resulting two-unit molecule is a disaccharide. Lactose (milk sugar) is glucose joined to galactose. Sucrose (table sugar) is glucose joined to fructose.

When hundreds or thousands of monosaccharides link together, the result is a polysaccharide. Two polysaccharides made entirely of glucose illustrate how structure determines function. Starch stores energy in plant cells; its bonds are easily broken by enzymes to release glucose. Cellulose forms the rigid structural framework of plant cell walls; its bonds are arranged differently, and most animals cannot digest them.

Carbohydrates also attach to proteins and lipids on cell surfaces, forming glycoproteins and glycolipids. These sugar-decorated molecules act as molecular name tags, allowing cells to recognise each other. Your blood type (A, B, AB, or O) is determined entirely by the specific sugar chains on the surface of your red blood cells.

Visual Beginner

The major carbohydrate classes and their relationships:

Class	Size	Example	Function
Monosaccharide	1 unit	Glucose (${\text{C}}_6{\text{H}}_{12}{\text{O}}_6$)	Cellular fuel
Disaccharide	2 units	Sucrose (glucose + fructose)	Transport sugar in plants
Polysaccharide (storage)	100s-1000s of units	Starch, glycogen	Energy reserve
Polysaccharide (structural)	1000s of units	Cellulose, chitin	Structural scaffold
Glycoconjugate	Variable	Glycoproteins, glycolipids	Cell recognition, signalling

Worked example Beginner

Consider lactose, the sugar in milk. It is a disaccharide of galactose bonded to glucose.

Step 1. Identify the monomers. Both galactose and glucose are six-carbon sugars with the formula ${\text{C}}_6{\text{H}}_{12}{\text{O}}_6$. They differ in the orientation of the hydroxyl group on carbon-4: in glucose it points one way, in galactose the other.

Step 2. Identify the bond. A glycosidic bond connects carbon-1 of galactose to carbon-4 of glucose. Because this bond locks the anomeric carbon of galactose, but leaves the anomeric carbon of glucose free, lactose is a reducing sugar — the free anomeric carbon can undergo the open-chain reaction.

Step 3. Digestion. The enzyme lactase (beta-galactosidase) hydrolyses the beta-1,4-glycosidic bond, releasing free glucose and galactose. People who produce insufficient lactase cannot break down lactose efficiently, leading to lactose intolerance.

Step 4. What this illustrates. Two sugars with identical molecular formula differ in biological behaviour because of a single stereochemical difference. A single bond type (glycosidic) creates an enormous diversity of structures depending on which carbons are linked and whether the linkage is alpha or beta.

Check your understanding Beginner

Formal definition Intermediate+

Monosaccharides are polyhydroxy aldehydes (aldoses) or polyhydroxy ketones (ketoses) with three to seven carbon atoms. The simplest aldose is glyceraldehyde (three carbons, one chiral centre). The simplest ketose is dihydroxyacetone (three carbons, no chiral centre). Each additional carbon introduces a new chiral centre, so an aldose with $n$ carbons has $2^{n-2}$ stereoisomers. Glucose, a six-carbon aldose (aldohexose), is one of $2^4=16$ possible aldohexoses.

D/L designation. In the Fischer projection (a 2D representation of a 3D molecule), the hydroxyl group on the highest-numbered chiral carbon (the penultimate carbon) determines the D/L designation. If this hydroxyl is on the right, the sugar is the D-isomer; if on the left, the L-isomer. Biologically active sugars are almost exclusively D-sugars. This convention derives from the reference standard D-glyceraldehyde, as established by Emil Fischer (1891) ^{[Fischer 1891]}.

Ring formation. In aqueous solution, monosaccharides with five or more carbons cyclise. An aldose cyclises when the hydroxyl on carbon-5 (or carbon-4 in pentoses) attacks the aldehyde carbonyl on carbon-1, forming a hemiacetal. A ketose cyclises by analogous attack on carbon-2, forming a hemiketal. The resulting ring is a six-membered pyranose (five carbons + one oxygen) or a five-membered furanose (four carbons + one oxygen). Glucose predominantly forms a pyranose ring in solution.

Anomeric carbon and mutarotation. Cyclisation creates a new chiral centre at carbon-1 (the anomeric carbon). Two stereoisomers at this position are called anomers: the alpha-anomer has the C1-hydroxyl trans to the CH${}_2$OH group at carbon-5 in the Haworth projection, and the beta-anomer has it cis. In solution, the two anomers interconvert through the open-chain form, a process called mutarotation, until an equilibrium mixture is reached (for D-glucose, approximately 36% alpha and 64% beta at equilibrium).

Glycosidic bonds form when the anomeric carbon of one sugar reacts with a hydroxyl group on another molecule, creating an acetal linkage with the loss of water. The bond is designated by: (i) the configuration at the anomeric carbon (alpha or beta), (ii) the carbon number of the donor, and (iii) the carbon number of the acceptor. For example, maltose has an alpha-1,4-glycosidic bond between carbon-1 of one glucose and carbon-4 of the next.

Polysaccharides are classified by their degree of polymerisation, linkage type, and branching pattern. Homopolysaccharides consist of a single sugar type; heteropolysaccharides contain two or more types. The major homopolysaccharides of glucose are:

• Amylose: unbranched alpha-1,4-glucan, forming a left-handed helix with 6 residues per turn
• Amylopectin: alpha-1,4 backbone with alpha-1,6 branches every 24-30 residues
• Glycogen: alpha-1,4 backbone with alpha-1,6 branches every 8-12 residues
• Cellulose: unbranched beta-1,4-glucan, forming extended flat chains that pack into microfibrils

Glycoconjugates are molecules in which carbohydrate chains are covalently attached to proteins or lipids:

• N-linked glycoproteins: oligosaccharide attached to the amide nitrogen of asparagine, in the consensus sequence Asn-X-Ser/Thr (where X is any amino acid except proline)
• O-linked glycoproteins: oligosaccharide attached to the hydroxyl oxygen of serine or threonine
• Proteoglycans: proteins with long, unbranched glycosaminoglycan (GAG) chains, forming the extracellular matrix
• Glycolipids: oligosaccharides attached to sphingosine (glycosphingolipids) embedded in the outer leaflet of the plasma membrane

Key mechanism Intermediate+

Mechanism (Glycosidic bond formation and hydrolysis). Glycosidic bond formation is a condensation reaction in which the anomeric hydroxyl of one monosaccharide is replaced by the oxygen of a hydroxyl group from another molecule. The reaction proceeds through an oxocarbenium-ion-like transition state at the anomeric centre. In biological systems, glycosyltransferases catalyse formation and glycosidases catalyse hydrolysis, both proceeding with either retention or inversion of configuration at the anomeric carbon depending on the enzyme mechanism.

Formation. Glycosyltransferases transfer a sugar from a nucleotide-sugar donor (UDP-glucose, GDP-mannose, and others) to an acceptor molecule. The enzyme positions the acceptor hydroxyl for nucleophilic attack on the anomeric carbon while displacing the nucleotide diphosphate leaving group. The stereochemistry of the product (alpha or beta) is determined by the enzyme's active site geometry, not by the thermodynamic anomeric ratio of the free sugar.

Hydrolysis. Glycosidases cleave glycosidic bonds using one of two mechanisms. Retaining glycosidases use a double-displacement mechanism: a catalytic nucleophile (typically a glutamate or aspartate side chain) attacks the anomeric carbon, forming a covalent glycosyl-enzyme intermediate with inversion, then a water molecule attacks with a second inversion, netting retention of configuration. Inverting glycosidases use a single-displacement mechanism: a general base activates a water molecule for direct nucleophilic attack on the anomeric carbon, producing inversion in a single step.

Application to lactose hydrolysis. Lactase (beta-galactosidase in humans) cleaves the beta-1,4-glycosidic bond of lactose by a retaining mechanism. Glu1749 acts as the nucleophile, forming a covalent galactosyl-enzyme intermediate. Glu2011 acts as the acid/base catalyst, protonating the glycosidic oxygen to facilitate departure of glucose, then deprotonating water in the second step to hydrolyse the covalent intermediate. Mutations that reduce expression or stability of lactase cause lactose intolerance, which affects approximately 65-70% of the global adult population.

Worked example: classifying a glycosidic bond

Isomaltose is a disaccharide of two glucose units linked by an alpha-1,6-glycosidic bond. Compare this to maltose (alpha-1,4).

The alpha-1,6 linkage in isomaltose involves the anomeric carbon-1 of one glucose and the primary hydroxyl on carbon-6 of the other. Because carbon-6 is outside the ring, the alpha-1,6 linkage creates a branch point rather than a linear extension. In glycogen and amylopectin, alpha-1,6 branches are introduced by branching enzyme, which cleaves a segment of the alpha-1,4 chain and reattaches it via an alpha-1,6 linkage to a glucose residue in the same or a neighbouring chain. The alpha-1,6 linkage is hydrolysed by debranching enzyme, a two-step process involving transfer of a trisaccharide stub to a nearby chain end followed by alpha-1,6-glucosidase activity.

Exercises Intermediate+

Conformational analysis of sugar rings and the anomeric effect Master

The Haworth projection, introduced in the Formal definition as a flat ring representation, is a simplification. Real sugar rings adopt three-dimensional conformations governed by the same principles that govern cyclohexane conformational chemistry.

Pyranose chair and boat conformations

The six-membered pyranose ring exists in two chair conformations (${}^4{\text{C}}_1$ and ${}^1{\text{C}}_4$, using the IUPAC notation where the superscript is the "upper" carbon and the subscript is the "lower" carbon in the chair) and several boat and twist-boat intermediates. For D-glucopyranose, the ${}^4{\text{C}}_1$ chair is strongly favoured because all large substituents (hydroxyl groups and the CH${}_2$OH at C5) occupy equatorial positions, minimising 1,3-diaxial steric interactions. The alternate ${}^1{\text{C}}_4$ chair places all substituents axial and is energetically disfavoured by approximately 25 kJ/mol ^{[Berg Tymoczko Stryer 2019 Ch. 7]}.

In the ${}^4{\text{C}}_1$ chair of beta-D-glucopyranose, the hydroxyl groups at C1, C2, C3, and C4 are all equatorial, and the CH${}_2$OH at C5 is also equatorial. This all-equatorial arrangement is the basis for glucose's exceptional thermodynamic stability among the aldohexoses. In alpha-D-glucopyranose, the C1 hydroxyl is axial, introducing one 1,3-diaxial interaction that makes the alpha anomer less stable by approximately 1.8 kJ/mol at equilibrium. This energy difference corresponds to the 36:64 alpha ratio at room temperature through the Boltzmann distribution: $N_{\beta }/N_{\alpha }=\exp (-\Delta G/RT)=\exp (1800/(8.315\times 298))\approx 2.1$, giving approximately 68% beta, in reasonable agreement with the observed 64%.

The anomeric effect

The anomeric effect is a stereoelectronic phenomenon that favours the axial orientation of an electronegative substituent at the anomeric carbon, counter to the predictions of pure steric analysis. In the absence of solvent effects, the axial (alpha) anomer is stabilised relative to the equatorial (beta) anomer by approximately 5 kJ/mol for methoxy-substituted tetrahydropyrans. This stabilisation arises from the interaction between the lone pair on the ring oxygen and the antibonding orbital (${\sigma }^{\ast }$) of the C1-O bond, which is maximised when the lone pair is antiperiplanar to the C1-O bond — a geometry achieved in the axial orientation.

In aqueous solution, the anomeric effect is largely overridden by solvation and steric effects. The equatorial hydroxyl of beta-D-glucose is better solvated by water, and the steric preference for the equatorial orientation dominates. The result is that the anomeric effect is a measurable but secondary contributor to anomeric equilibria in water, becoming more significant in nonpolar environments or in glycosidic bonds where the hydroxyl is replaced by an alkoxy group.

Furanose ring conformations

Five-membered furanose rings are more flexible than pyranose rings. They adopt envelope (E) and twist (T) conformations in which one or two atoms deviate from the plane defined by the remaining atoms. The pseudorotation cycle describes the continuous interconversion between these conformers. Ribose in RNA adopts a C2'-endo or C3'-endo sugar pucker, and the preference for one over the other has consequences for the backbone geometry of A-form versus B-form DNA and RNA helices. This conformational switching is central to the structure and function of nucleic acids and will be treated in detail in the nucleic acids unit.

Carbohydrate-protein recognition and the glycome Master

Carbohydrates serve as information carriers on cell surfaces, and proteins that recognise specific glycan structures are central to cell-cell communication, immune function, and pathogen entry.

Lectins

Lectins are proteins that bind specific carbohydrate structures with high selectivity but without catalysing a chemical reaction (unlike glycosidases). The binding is mediated by a network of hydrogen bonds between the protein's amino acid side chains and the hydroxyl groups of the sugar, plus van der Waals contacts and, in some cases, coordination of a calcium ion that bridges the protein and the sugar. The specificity arises from the precise spatial arrangement of these interactions: a lectin that binds mannose will not bind glucose, even though the two sugars differ only in the configuration of a single hydroxyl group (at C2).

Selectins are a family of lectins that mediate the initial adhesion of white blood cells to endothelial cells during the inflammatory response. L-selectin on lymphocytes binds to specific sulfated glycans on endothelial cells in lymph nodes. P-selectin on activated platelets and endothelial cells binds to PSGL-1 (P-selectin glycoprotein ligand-1), a mucin-like glycoprotein on leukocytes whose binding activity depends on specific sialylated, fucosylated, and sulfated oligosaccharide structures. The selectin-carbohydrate interaction is transient (on the order of seconds), allowing leukocytes to roll along the endothelium before firm adhesion and extravasation.

The glycome as an information carrier

The glycome is the complete set of glycan structures produced by a cell, tissue, or organism. Unlike the genome (a fixed sequence) or the proteome (determined by translation), the glycome is not directly template-driven. Glycan structures are assembled stepwise by glycosyltransferases and glycosidases in the ER and Golgi, and the repertoire depends on enzyme expression levels, substrate availability, and compartmental pH. This means that the glycome is dynamic, cell-type-specific, and responsive to environmental cues.

The information-encoding capacity of glycans is enormous. Consider a simple disaccharide of two hexoses: the linkage can be alpha or beta, can involve any of several hydroxyl positions on each sugar, and the component sugars can be any of 10-15 common monosaccharides. Even a modest trisaccharide can encode more structural diversity than a tripeptide built from 20 amino acids ($20^3=8000$ for peptides, but far more when glycosidic linkage position and anomeric configuration are included). This diversity is why the glycome is sometimes described as the "third language of life" after the genome and the proteome.

The ABO blood group system

The ABO blood group system is a direct medical application of glycan diversity. The H antigen, present on type O erythrocytes, is an oligosaccharide chain terminating in fucose linked alpha-1,2 to galactose. The A gene encodes a glycosyltransferase that adds N-acetylgalactosamine (GalNAc) in alpha-1,3 linkage to the terminal galactose of the H antigen. The B gene encodes a different glycosyltransferase that adds galactose (Gal) in alpha-1,3 linkage to the same position. The two enzymes differ by only four amino acid substitutions, which alter the active site to accommodate either GalNAc or Gal. The O allele has a single-nucleotide deletion that produces a frameshift and a nonfunctional enzyme, leaving the H antigen unmodified.

Type AB individuals express both enzymes and carry both A and B antigens. Type O individuals carry neither modification. Antibodies against the missing antigens (anti-A in type B individuals, anti-B in type A individuals, both in type O individuals) are produced by exposure to environmental bacteria carrying similar glycan structures. Transfusion of incompatible blood triggers antibody-mediated destruction of the transfused cells (hemolysis), which is why blood typing is essential before transfusion.

Glycosylation site prediction

Predicting which asparagine residues in a protein will be glycosylated requires more than just identifying the Asn-X-Ser/Thr consensus sequence. Approximately one-third of consensus sites are not glycosylated in practice, because glycosylation depends on: (i) accessibility of the site to oligosaccharyltransferase during translocation into the ER; (ii) the local folding rate of the protein (if the consensus site folds into the protein interior before OST can act, glycosylation is blocked); (iii) the spacing of the site from the transmembrane domain (sites too close to the membrane are inaccessible). Computational predictors such as NetNGlyc combine sequence features with structural accessibility estimates to predict glycosylation sites from sequence alone, achieving accuracies of approximately 70-80%.

Bacterial cell wall peptidoglycan

The bacterial cell wall is a glycoconjugate structure of immense medical importance. Peptidoglycan (murein) consists of long glycan chains of alternating N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) linked by beta-1,4-glycosidic bonds — the same beta-1,4 linkage found in cellulose, but with the lactyl group of MurNAc serving as the attachment point for a short peptide chain. These peptide chains cross-link adjacent glycan strands, forming a covalent mesh that gives the cell wall its mechanical strength. In Gram-positive bacteria, the peptidoglycan layer is thick (20-80 nm) and may be cross-linked through a pentaglycine bridge (as in Staphylococcus aureus). In Gram-negative bacteria, the peptidoglycan layer is thin (2-7 nm) and sits in the periplasm between the inner and outer membranes.

Penicillin and related beta-lactam antibiotics inhibit the transpeptidase enzymes (penicillin-binding proteins) that catalyse the cross-linking reaction. Without cross-links, the peptidoglycan mesh weakens and the cell lyses under its own osmotic pressure. Lysozyme, an enzyme found in tears and mucus, cleaves the beta-1,4-glycosidic bond between GlcNAc and MurNAc, directly degrading the glycan chains. Together, these two vulnerabilities — the cross-linking reaction and the glycosidic bond — are the targets of the innate immune defence against bacterial infection.

Connections Master

To glycolysis and energy metabolism (17.04.04). Glucose, the central monosaccharide introduced here, is the substrate of glycolysis. The alpha and beta anomeric equilibrium, the ring-to-open-chain interconversion, and the phosphorylation of hydroxyl groups are all directly relevant to the ten enzymatic steps that convert glucose to pyruvate. Glycogen phosphorylase cleaves alpha-1,4 linkages from the non-reducing ends of glycogen, releasing glucose-1-phosphate that enters glycolysis after conversion to glucose-6-phosphate. The branching structure of glycogen (every 8-12 residues) directly determines how many glucose units can be mobilised simultaneously.

To membrane structure (17.02.01). Glycolipids (glycosphingolipids) are carbohydrate-bearing lipids embedded in the outer leaflet of the plasma membrane. Their sugar chains project into the extracellular space, where they serve as recognition markers. Gangliosides (sialic acid-containing glycosphingolipids) are concentrated in neuronal membranes and participate in cell signalling and pathogen binding (cholera toxin binds to GM1 ganglioside). The asymmetric distribution of glycolipids (exclusively in the outer leaflet) is maintained by the same biosynthetic pathway that determines lipid asymmetry.

To cell signalling (17.07.01). Many signalling receptors are glycoproteins whose glycan chains affect ligand binding, receptor dimerisation, and signal transduction. The epidermal growth factor receptor (EGFR) is N-glycosylated at multiple sites, and the glycan structures modulate its interaction with EGF. Heparan sulfate proteoglycans in the extracellular matrix bind and concentrate growth factors such as FGF, creating a reservoir that can be released by heparanase cleavage.

To immunology (17.10.01). The major histocompatibility complex (MHC) presents peptide antigens to T cells, but MHC molecules are themselves glycoproteins whose glycan shields affect antigen loading and T-cell receptor interaction. The ABO blood group system, described in this unit, is one of the simplest and best-characterised examples of glycan-based immune recognition. Natural killer cell receptors (KIR family) recognise specific HLA glycoforms on target cells.

To nucleic acid structure. The ribose sugar in RNA and the deoxyribose in DNA are five-carbon furanose rings whose conformational preferences (C2'-endo vs C3'-endo sugar pucker) determine the global geometry of the nucleic acid helix. The 2'-hydroxyl that distinguishes ribose from deoxyribose is the chemical basis for RNA's greater chemical lability (base-catalysed hydrolysis) and its capacity for catalysis (ribozymes).

Historical notes Master

Emil Fischer established the stereochemistry of glucose in a landmark series of papers beginning in 1888, culminating in the 1891 publication on the configuration of glucose and its isomers ^{[Fischer 1891]}. Fischer introduced the projection convention that bears his name and used it to assign the relative configurations of all 16 aldohexoses. His work earned him the 1902 Nobel Prize in Chemistry. Fischer's assignment was based on chemical reasoning and degradation experiments; the absolute configuration was not confirmed until 1951, when Bijvoet used X-ray crystallography to show that Fischer's arbitrary choice of D-glyceraldehyde as the reference standard happened to match the true absolute configuration.

Walter Haworth proposed the ring structure of glucose in 1926, replacing the previously accepted open-chain model. The Haworth projection, which represents the sugar ring as a flat hexagon with substituents drawn above or below the plane, became the standard teaching representation. Hassel and Ottar's conformational analysis of pyranose rings (1947) later showed that the ring is not flat but adopts a chair conformation, but the Haworth projection remains widely used for its simplicity.

The glycosidic bond was named and characterised in the late 19th century. The distinction between alpha and beta glycosidic linkages was established by comparison of the optical rotation of methyl glycosides. The structural difference between starch (alpha-1,4) and cellulose (beta-1,4) was determined by Haworth and colleagues in the 1920s-1930s, explaining why humans can digest one but not the other.

The field of glycobiology emerged as a distinct discipline in the 1980s, driven by the realisation that glycoconjugates play essential roles in cell-cell recognition, immune function, and development. The term "glycobiology" was coined by Raymond Dwek in 1988. Ajit Varki and colleagues established the first major glycobiology textbook (Essentials of Glycobiology, first edition 1999, now in its 4th edition, 2022) ^{[Varki 2022]}, which remains the definitive reference.

Karl Landsteiner discovered the ABO blood group system in 1901, observing that sera from some individuals agglutinated red blood cells from others. The chemical basis — differences in terminal sugar residues on surface glycolipids — was not elucidated until the 1950s by Morgan and Watkins. Landsteiner received the 1930 Nobel Prize in Physiology or Medicine.

Alexander Fleming's discovery of lysozyme in 1922 and Howard Florey's development of penicillin in the 1940s (building on Fleming's 1928 observation) both target the carbohydrate backbone of the bacterial cell wall: lysozyme cleaves the beta-1,4-glycosidic bond, and penicillin inhibits the transpeptidase that cross-links the peptide chains. These discoveries inaugurated the antibiotic era and remain among the most consequential applications of carbohydrate chemistry to medicine.

Bibliography Master

• Alberts, B. et al. Molecular Biology of the Cell, 7th ed. Garland Science, 2022. Ch. 2: The Chemical Components of the Cell — Carbohydrates. The standard cell-biology textbook treatment of monosaccharide structure, glycosidic bonds, polysaccharide diversity, and glycoconjugates.

• Berg, J. M., Tymoczko, J. L. & Stryer, L. Biochemistry, 9th ed. W. H. Freeman, 2019. Ch. 7: Carbohydrates and Glycobiology. A thorough treatment of sugar stereochemistry, ring conformations, glycosidic bond chemistry, and glycoconjugate structure with quantitative detail.

• Varki, A. et al. Essentials of Glycobiology, 4th ed. Cold Spring Harbor Laboratory Press, 2022. Ch. 1-6. The definitive reference for monosaccharide structure, glycosidic bond formation, glycoprotein classification, and the glycome as an information carrier. Freely available online through the NCBI Bookshelf.

• Fischer, E. Ueber die Configuration des Traubenzuckers und seiner Isomeren. Ber. Dtsch. Chem. Ges. 24 (1891) 1836-1845. The foundational paper establishing glucose stereochemistry and the Fischer projection convention.

• Haworth, W. N. The Constitution of Sugars. Edward Arnold, London, 1929. Haworth's synthesis of his work on ring structures and the projection that bears his name.

• Dwek, R. A. Glycobiology: toward understanding the function of sugars. Chem. Rev. 96 (1996) 683-720. A comprehensive review that helped establish glycobiology as a unified discipline.

• Sharon, N. & Lis, H. Lectins, 2nd ed. Springer, 2003. The standard reference on carbohydrate-binding proteins and their biological roles.

• Landsteiner, K. Zur Kenntnis der antifermentativen, lytischen und agglutinierenden Wirkungen des Blutserums und der Lymphe. Zentralbl. Bakteriol. 27 (1900) 357-362. The discovery of the ABO blood group system.