Post-Translational Modification of Proteins

A newly synthesized polypeptide chain is chemically unfinished. It may still be tethered to the ribosome. Or it may be freshly released into the cytosol. It may fold correctly. But it cannot yet respond to a hormone. It cannot cut itself free of an inhibitory propeptide. It cannot be tagged for demolition. It cannot be routed with the correct sugar barcode to the correct organelle. All of that chemistry happens after translation ends. This layer of covalent editing is called post-translational modification (PTM). PTMs convert the twenty-letter alphabet written by the ribosome into a vastly larger functional vocabulary. A single gene product can be phosphorylated, glycosylated, cleaved, and ubiquitinated. These combinatorial patterns each mean something different to the cell. The stakes are not abstract. A single misplaced phosphate on cyclin-dependent kinase controls whether a cell divides or arrests. A single failure to glycosylate a clotting factor causes a bleeding disorder. A single proteolytic nick converts inert trypsinogen into digestive trypsin. That trypsin can dissolve the pancreas that made it, if the nick happens in the wrong place. A single chain of ubiquitin molecules marks a misfolded protein for destruction. This destruction occurs via the proteasome, the cell's principal quality-control mechanism. Diseases as diverse as cystic fibrosis, cancer, and Creutzfeldt–Jakob disease trace to failures in these modification systems. Roughly two-thirds of the human proteome carries at least one PTM. Many proteins carry dozens, layered in time and space to build molecular logic circuits. The unifying chemical idea is simple. PTMs exploit a small set of nucleophilic side chains as reactive handles. These include the hydroxyls of Ser/Thr/Tyr, the ε-amine of Lys, the thiol of Cys, and the N-terminal amine. They use activated donor molecules, such as ATP, UDP-sugars, and ubiquitin-adenylate, to install modifications. These modifications are thermodynamically uphill from the modified protein's point of view. But they are downstream of a phosphoanhydride or thioester bond cleavage that pays for the reaction. Removal is equally enzymatic and equally chemical. Phosphatases hydrolyze phosphoesters. Glycosidases trim sugars. Deubiquitinases cleave isopeptide bonds. The reversibility of each modification is itself functionally meaningful. This is true for phosphorylation and ubiquitination. So is the irreversibility of most proteolysis. This article works through four major PTM classes mechanistically. These are: phosphorylation by protein kinases and its reversal by phosphatases; N- and O-linked glycosylation in the secretory pathway; proteolytic maturation of zymogens and precursor proteins; and the ubiquitin-proteasome system. Each section builds the chemistry from first principles. It covers the electrophile, the nucleophile, the leaving group, and the energetic driving force. It then connects this chemistry to physiological logic. It builds on The Ribosome and the Elongation Cycle and Protein Targeting and the Secretory Pathway. These are the delivery mechanisms that put substrates in the right compartment to be modified.

Key takeaways

Protein kinases transfer the γ-phosphate of ATP to Ser/Thr/Tyr hydroxyls via in-line nucleophilic attack on phosphorus; phosphatases hydrolyze it back off.
Phosphorylation is reversible and fast, making it the currency of second-to-second signal transduction (e.g., insulin, MAPK cascades, cell cycle checkpoints).
N-linked glycosylation begins co-translationally in the ER: a 14-sugar dolichol-linked oligosaccharide is transferred en bloc to Asn-X-Ser/Thr sequons, then trimmed and remodeled in the Golgi.
Proteolytic processing (zymogen activation, signal peptide removal, proprotein convertase cleavage) is irreversible and often the last step that switches a protein from inactive to active.
Ubiquitination attaches a 76-residue protein to lysines via an E1–E2–E3 cascade ending in isopeptide bond formation; polyubiquitin chains of different linkage types mean different things (degradation vs. signaling).
Key idea: PTMs let a fixed genome encode a combinatorially larger, faster, and more reversible functional proteome by using small, chemically distinct, enzymatically installed and removed covalent tags as molecular switches and addresses.

Where this lives today

Kinase inhibitors are now a mainstay of oncology: imatinib (Gleevec) inhibits the BCR-ABL tyrosine kinase in chronic myeloid leukemia by occupying the ATP-binding cleft, directly blocking the in-line phosphoryl-transfer chemistry described above. Dozens of kinase inhibitors targeting EGFR, ALK, CDK4/6, and JAK kinases now treat cancers by exploiting exactly the mechanistic vulnerability of the kinase active site. On the opposite side, phosphatase modulators (and the recognition that phosphatases are not simply "always on" housekeeping enzymes but are themselves tightly regulated) are an active area of drug discovery, particularly for PP2A reactivation in cancer and PTEN pathway restoration. Glycosylation engineering underlies biologics manufacturing: therapeutic antibodies are produced in CHO (Chinese hamster ovary) cell lines specifically because their glycosylation machinery closely mimics human patterns, and afucosylated antibody variants (engineered to lack a core fucose on the N-glycan) show enhanced antibody-dependent cellular cytotoxicity, a property now deliberately engineered into next-generation cancer antibodies like obinutuzumab. Congenital disorders of glycosylation (CDG), caused by mutations in the dolichol-pathway or OST enzymes, cause severe developmental disease, directly validating the essential nature of the N-glycosylation pathway traced in Example 2. Proteolytic maturation is exploited pharmacologically and industrially: proprotein convertases (furin and relatives) that activate viral fusion proteins (including the SARS-CoV-2 spike protein's furin cleavage site) are targets for antiviral drug design, while PCSK9 (a proprotein convertase that degrades LDL receptors) is the target of a major class of cholesterol-lowering biologic drugs (evolocumab, alirocumab). The ubiquitin-proteasome system is now drugged directly: proteasome inhibitors (bortezomib, carfilzomib) treat multiple myeloma by blocking degradation of pro-apoptotic factors, while an entirely new drug modality — PROTACs (proteolysis-targeting chimeras) — hijacks E3 ligases to force ubiquitination and destruction of previously "undruggable" disease proteins, directly exploiting the E1–E2–E3 chemistry derived above as a targeted-degradation therapeutic platform. These modification systems, once understood chemically, become levers for Regulation of Gene Expression: Chemical Logic at the protein level, complementing the transcriptional and translational control layers.

Ask the tutor: How do phosphorylation and ubiquitination get wired together in series to build a single fast, irreversible degradation switch — using IκB or cyclin as a worked example?

The physical picture — chemistry on a finished chain

Translation gives you a linear sequence of amino acids folded (with help from chaperones) into a three-dimensional structure. But the genetic code specifies only twenty side chains, and most cellular decisions — divide or don't, secrete or degrade, active or latent — need to be made and unmade on timescales the ribosome cannot provide, since making a new protein from scratch takes minutes and cannot be instantly reversed. Post-translational modification solves this by attaching or removing small chemical groups on an existing, already-folded polypeptide, using enzymes that recognize very specific sequence and structural contexts. The four systems treated here differ enormously in chemistry but share a common architecture: an activated donor (ATP, a UDP-sugar, ubiquitin-adenylate), a nucleophilic acceptor site on the substrate protein, and a dedicated transferase enzyme that couples donor activation to acceptor modification with high specificity. Removal machinery (phosphatases, glycosidases, deubiquitinases) is equally specific and, for the reversible modifications, keeps the system poised at a steady state that can be shifted rapidly in either direction — this is the chemical basis of switch-like signaling. Glycosylation is the outlier: it is essentially irreversible under physiological conditions (no glycosidase runs the reaction backward to re-form the same linkage against the concentration gradient) and instead serves as a permanent structural and address-label modification, built and edited compartment by compartment as the protein moves through the secretory pathway. Proteolysis is even more starkly irreversible — peptide bond hydrolysis has no cellular "ligase" waiting to re-form the exact same bond — so proteolytic maturation is used for one-way, often life-or-death switches: activating a protease, releasing a hormone, triggering apoptosis. ```svg ``` Phosphorylation works by nucleophilic substitution at phosphorus. A protein kinase active site positions ATP·Mg²⁺ so that its γ-phosphate lies in line with the incoming Ser/Thr/Tyr hydroxyl oxygen; a catalytic base (often an Asp in the kinase's catalytic loop) deprotonates the hydroxyl, and the resulting alkoxide attacks the electrophilic γ-phosphorus in a single associative in-line displacement, expelling ADP as the leaving group. The mechanism is essentially SN2 at phosphorus, proceeding through a trigonal-bipyramidal transition state with partial negative charge on the axial oxygens, stabilized by Mg²⁺ and conserved active-site lysines/arginines. ```svg ```

The core ideas — governing relationships

Phosphorylation transfer potential. The reaction catalyzed by a protein kinase,

\text{ATP} + \text{protein–OH} \longrightarrow \text{ADP} + \text{protein–O–PO}_3^{2-}

is thermodynamically favorable because it couples phosphoester bond formation to hydrolysis of the ATP phosphoanhydride bond, whose standard free energy of hydrolysis is large and negative:

\Delta G^{\circ\prime}_{\text{ATP}\to\text{ADP}+P_i} \approx -30.5\ \text{kJ/mol}

where: $\Delta G^{\circ\prime}$ is the standard transformed free energy at pH 7, 1 M standard states (in practice cellular ΔG is more negative, roughly −50 kJ/mol, because cellular [ATP]/[ADP][Pi] ratios are far from 1 M standard state).

Reading: the kinase reaction is essentially irreversible under cellular conditions (equilibrium lies overwhelmingly toward the phosphoprotein), so the reverse reaction — dephosphorylation — must be catalyzed by a chemically distinct enzyme (a phosphatase performing simple hydrolysis) rather than the kinase running backward. This chemical asymmetry between "writer" and "eraser" enzymes is a recurring PTM motif.

Phosphatase hydrolysis. Protein phosphatases (Ser/Thr phosphatases like PP1, PP2A; Tyr phosphatases like PTP1B) catalyze:

\text{protein–O–PO}_3^{2-} + \text{H}_2\text{O} \longrightarrow \text{protein–OH} + P_i

\Delta G^{\circ\prime}_{\text{hydrolysis}} \approx -10\ \text{to}\ -13\ \text{kJ/mol (phosphoester)}

where: this value is much smaller in magnitude than the phosphoanhydride hydrolysis of ATP, which is exactly why the forward (kinase) and reverse (phosphatase) reactions are each essentially irreversible in their own direction — the system behaves as a futile-cycle-avoiding one-way switch pair rather than a near-equilibrium interconversion.

Fraction phosphorylated at steady state. For a substrate cycled between kinase (rate constant $k_1$ , effectively pseudo-first-order given saturating ATP) and phosphatase ( $k_2$ ), the steady-state fraction phosphorylated is

f_P = \frac{k_1}{k_1+k_2}

where: $k_1$ is the pseudo-first-order phosphorylation rate (kinase activity × its concentration), $k_2$ the dephosphorylation rate.

Reading: because both $k_1$ and $k_2$ can be independently modulated (by upstream signals activating the kinase or inhibiting the phosphatase), $f_P$ can be driven from near 0 to near 1 rapidly — this zero-order or near-linear sensitivity, especially when kinase/phosphatase operate near saturation, is the molecular basis of ultrasensitive signaling switches (Goldbeter–Koshland kinetics).

Ubiquitin conjugation cascade. The overall reaction, spanning three enzyme classes, is

\text{E1-SH} + \text{Ub} + \text{ATP} \rightarrow \text{E1-S-Ub} + \text{AMP} + \text{PP}_i

\text{E1-S-Ub} + \text{E2-SH} \rightarrow \text{E1-SH} + \text{E2-S-Ub}

\text{E2-S-Ub} + \text{Lys-NH}_2\text{-substrate} \xrightarrow{\text{E3}} \text{E2-SH} + \text{Lys-NH-CO-Ub (isopeptide)}

where: E1 is the ubiquitin-activating enzyme (forms a high-energy thioester using ATP, releasing AMP + PPi — an energetic investment equivalent to two ATP hydrolyses since PPi is subsequently hydrolyzed), E2 is a ubiquitin-conjugating enzyme (thioester-linked intermediate carrier), E3 is a ubiquitin ligase that confers substrate specificity and catalyzes (or, for RING E3s, merely scaffolds) isopeptide bond formation between ubiquitin's C-terminal Gly76 carboxylate and a substrate lysine ε-amine.

Reading: the two-thioester relay (E1→E2) followed by amide bond formation is mechanistically a transesterification/aminolysis chain — each step conserves the high-energy acyl bond until the final, committed step forms a stable isopeptide (amide) bond, paying the ATP energy forward into a covalent mark that is read by downstream effectors (the proteasome, endocytic machinery, or signaling scaffolds depending on chain topology).

Derivation AUltrasensitivity of a phosphorylation-dephosphorylation cycle: Goldbeter–Koshland zero-order switch

Goal: show that a kinase/phosphatase cycle operating near enzyme saturation converts a graded input signal into a sharp, switch-like output in the fraction of substrate phosphorylated — the chemical basis for why phosphorylation cascades (e.g., MAPK cascades) behave digitally despite being built from smooth Michaelis–Menten enzymes.

Step 1 — Set up the cycle. Let $W$ (total substrate) exist in unphosphorylated ( $W$ ) and phosphorylated ( $W^*$ ) forms, with $W + W^* = W_T$ fixed. Kinase converts $W \to W^*$ with Michaelis constant $K_1$ and maximal rate $V_1$ (set by kinase activity); phosphatase converts $W^* \to W$ with constant $K_2$ and maximal rate $V_2$ . At steady state the two fluxes are equal:

V_1 \frac{W}{K_1+W} = V_2 \frac{W^*}{K_2+W^*}

Step 2 — Define the saturation regime. If $K_1, K_2 \ll W_T$ (both enzymes are saturated by their substrate — the "zero-order regime," as opposed to first-order/linear kinetics when $K \gg W_T$ ), each Michaelis–Menten term approaches a step-like dependence on the fraction phosphorylated $f = W^*/W_T$ .

Step 3 — Nondimensionalize. Let $v = V_1/V_2$ (ratio of maximal kinase to phosphatase rate — the "signal"), and $K_1' = K_1/W_T$ , $K_2' = K_2/W_T$ (small parameters in the zero-order limit). The steady-state condition becomes

v\,\frac{1-f}{K_1' + (1-f)} = \frac{f}{K_2' + f}

Step 4 — Solve the quadratic. Cross-multiplying and collecting terms in $f$ gives the Goldbeter–Koshland quadratic:

a f^2 - \big[a + v(1-K_2') + K_1'\big] f + v(1-K_2')= 0,\qquad a \equiv v - 1

with the physically meaningful root

f = \frac{ (v-1) + K_1' + v K_2' - \sqrt{\big[(v-1)+K_1'+vK_2'\big]^2 - 4(v-1)v(1-K_2')} }{2(v-1)}

Step 5 — Take the zero-order limit. As $K_1', K_2' \to 0$ (deep saturation), the response sharpens dramatically: $f$ jumps from near 0 to near 1 over a very narrow range of $v$ near $v=1$ , in contrast to the graded, Michaelis-like response obtained when $K_1', K_2' \gg 1$ (substrate far below $K_m$ , first-order/linear regime), where $f \approx v/(1+v)$ — a smooth hyperbola.

Step 6 — Interpret. Zero-order ultrasensitivity means that when both the kinase and the phosphatase are working near saturation with respect to their shared substrate pool, a small percentage change in the kinase/phosphatase activity ratio $v$ can flip the system almost entirely from off ( $f\approx0$ ) to on ( $f\approx1$ ). This is the chemical-kinetic reason phosphorylation switches (cell-cycle checkpoints, MAPK cascades) can behave as near-digital decisions built from analog Michaelis–Menten chemistry, without requiring cooperativity in the classical allosteric sense.

Go deeper · Derivation B · 3 minFree-energy accounting for the E1–E2–E3 ubiquitination cascade

Derivation BFree-energy accounting for the E1–E2–E3 ubiquitination cascade

Goal: quantify why ubiquitin conjugation, which forms a thermodynamically uphill isopeptide (amide) bond between ubiquitin Gly76 and a substrate lysine, is nonetheless efficiently driven forward — and why the process is effectively irreversible without a dedicated deubiquitinase.

Step 1 — Identify the target bond. The terminal, committed step forms an isopeptide bond:

\text{Ub-Gly76-COOH} + \text{H}_2\text{N-Lys-substrate} \rightarrow \text{Ub-Gly76-CO-NH-Lys-substrate} + \text{H}_2\text{O}

Amide bond formation from a free carboxylic acid and a free amine is thermodynamically unfavorable in water, with $\Delta G^{\circ\prime}$ typically +10 to +20 kJ/mol for simple amide synthesis (peptide-bond-like), because the reaction releases water into an already water-rich solvent and forms a bond with modest intrinsic stabilization relative to the separated, solvated acid and amine.

Step 2 — Activate the carboxylate: E1. Rather than reacting the free Gly76 carboxylate directly, the cell first activates it. E1 uses ATP to form ubiquitin-adenylate:

\text{Ub-COOH} + \text{ATP} \rightarrow \text{Ub-CO-AMP} + PP_i,\qquad \Delta G^{\circ\prime}\approx -30\ \text{kJ/mol (net, after PPi hydrolysis)}

The mixed anhydride Ub-CO-AMP is a high-energy acyl-adenylate, chemically analogous to the aminoacyl-AMP intermediate used in The Ribosome and the Elongation Cycle's charging of tRNAs — the same "activate a carboxylate as a mixed anhydride with AMP, paid for by ATP→AMP + PPi" strategy recurs across biochemistry precisely because it converts a poor leaving group (OH⁻ implicit in a free acid) into an excellent one (AMP²⁻).

Step 3 — Transfer to E1's active-site cysteine. The adenylate is attacked by a nucleophilic active-site Cys-SH on E1 itself:

\text{Ub-CO-AMP} + \text{E1-SH} \rightarrow \text{Ub-CO-S-E1} + \text{AMP}

This forms a thioester — an acyl group on sulfur — which retains substantial group-transfer potential (thioester hydrolysis $\Delta G^{\circ\prime}\approx-30$ to $-35$ kJ/mol, comparable to ATP's phosphoanhydride bonds, because the poor π-overlap between sulfur 3p and carbonyl π* orbitals means little resonance stabilization of the thioester relative to a normal amide/ester, leaving it high in energy and reactive — the identical chemical logic that makes acetyl-CoA a "high-energy" metabolic currency).

Step 4 — Relay to E2 by transthioesterification. A second Cys-SH, on the E2 enzyme, attacks the E1-bound thioester carbonyl in a direct nucleophilic acyl substitution, displacing E1-SH and forming a new E2-bound thioester:

\text{Ub-CO-S-E1} + \text{E2-SH} \rightarrow \text{Ub-CO-S-E2} + \text{E1-SH}

Because both species are thioesters, this step is close to thermoneutral — no additional energy is spent, and the acyl-transfer potential is conserved intact from E1 to E2.

Step 5 — E3-directed transfer to substrate lysine. E3 ubiquitin ligases fall into two mechanistic classes. RING-type E3s (the majority) do not form a covalent intermediate; they simply bind both E2-Ub and the substrate simultaneously, orienting the substrate lysine ε-amine for direct, in-line nucleophilic attack on the E2-bound thioester carbonyl:

\text{Ub-CO-S-E2} + \text{H}_2\text{N-Lys-substrate} \xrightarrow{\text{RING-E3 (scaffold)}} \text{Ub-CO-NH-Lys-substrate} + \text{E2-SH}

HECT-type E3s instead form their own transient thioester intermediate (Ub-CO-S-E3) before transferring to the substrate lysine, adding one more relay step but the same net chemistry.

Step 6 — Tally the energetics. The overall reaction from free Ub-COOH and ATP to the isopeptide-linked conjugate:

\Delta G^{\circ\prime}_{\text{overall}} = \Delta G^{\circ\prime}_{\text{ATP}\to\text{AMP}+2P_i} + \Delta G^{\circ\prime}_{\text{isopeptide bond formation}}

\Delta G^{\circ\prime}_{\text{overall}} \approx (-45.6\ \text{kJ/mol}) + (+15\ \text{kJ/mol}) \approx -30\ \text{kJ/mol}

where the ATP→AMP + 2Pi hydrolysis (via PPi hydrolysis by pyrophosphatase) delivers roughly twice the free energy of a single ATP→ADP hydrolysis, comfortably overpaying the endergonic amide bond formation and rendering ubiquitin conjugation effectively irreversible on cellular timescales — exactly analogous to how tRNA charging (also an ATP→AMP+PPi-driven acyl-adenylate mechanism) drives an otherwise unfavorable ester bond formation.

Step 7 — Why this must be enzymatically reversed, not chemically. Because the forward reaction is strongly exergonic overall, the reverse (deubiquitination) cannot occur by simple mass-action reversal; it requires dedicated deubiquitinating enzymes (DUBs), cysteine or metalloproteases that hydrolyze the isopeptide bond in a chemically independent reaction, releasing free ubiquitin and substrate. This "irreversible-forward, enzyme-catalyzed-reverse" pattern mirrors the kinase/phosphatase asymmetry exactly, and is a general design principle for regulatable covalent switches.

Worked examples

Example 1 — Foundational: energetics of a single phosphorylation event by protein kinase A. Protein kinase A (PKA) phosphorylates a serine on glycogen phosphorylase kinase using ATP·Mg²⁺. Given $\Delta G^{\circ\prime}_{\text{ATP hydrolysis}} = -30.5$ kJ/mol and cellular concentrations [ATP] = 3 mM, [ADP] = 0.03 mM, [Pi] = 1 mM at 37 °C (310 K), compute the actual cellular free energy released, and confirm the reaction is far from equilibrium (driving it essentially to completion).

\Delta G = \Delta G^{\circ\prime} + RT\ln\frac{[\text{ADP}][P_i]}{[\text{ATP}]}

Using $R = 8.314\times10^{-3}$ kJ/(mol·K), $RT = 8.314\times10^{-3}\times310 = 2.577$ kJ/mol:

\frac{[\text{ADP}][P_i]}{[\text{ATP}]} = \frac{(0.03\times10^{-3})(1\times10^{-3})}{3\times10^{-3}} = 1.0\times10^{-5}\ \text{M}

\ln(1.0\times10^{-5}) = -11.51

\Delta G = -30.5 + (2.577)(-11.51) = -30.5 - 29.66 = -60.2\ \text{kJ/mol}

This is nearly double the standard-state value because cellular [ATP] is kept high relative to [ADP][Pi], displacing the reaction far from equilibrium. Since phosphoryl transfer to the Ser-OH is coupled directly to this hydrolysis in a single concerted in-line step (no free Pi intermediate is released — the γ-phosphate goes directly onto the serine oxygen), essentially the entire −60.2 kJ/mol drives the phosphorylation forward, making the kinase reaction, as noted above, essentially irreversible under physiological ATP/ADP/Pi ratios. As a limiting-case check: if [ATP] and [ADP][Pi] were at the (nonphysiological) standard 1 M reference state, $\Delta G$ would collapse exactly to $\Delta G^{\circ\prime}=-30.5$ kJ/mol — confirming the extra driving force comes entirely from the cell's nonequilibrium ATP/ADP/Pi ratio, not from any change in the chemistry.

Example 2 — Applied: N-glycosylation sequon recognition and the fate of a secreted glycoprotein. A newly synthesized polypeptide destined for secretion contains the sequence …Leu-Asn-Lys-Thr-Val-Gly… at residue 142. Determine whether this Asn will be N-glycosylated, and trace the oligosaccharide's fate. The consensus sequon for N-linked glycosylation is Asn-X-Ser/Thr, where X is any amino acid except proline. Here, X = Lys (not Pro) and the +2 residue is Thr — the sequon Asn-Lys-Thr matches Asn-X-Ser/Thr exactly, so Asn142 is a valid acceptor. Mechanistically, while the nascent chain is still being translocated into the ER lumen (co-translationally, via the machinery built up in Protein Targeting and the Secretory Pathway), the oligosaccharyltransferase (OST) complex transfers a pre-assembled tetradecasaccharide,

\text{Glc}_3\text{Man}_9\text{GlcNAc}_2

en bloc from a dolichol-phosphate lipid carrier in the ER membrane to the Asn side-chain amide nitrogen, forming an N-glycosidic bond:

\text{Dolichol-PP-Glc}_3\text{Man}_9\text{GlcNAc}_2 + \text{Asn-NH-CO-} \rightarrow \text{Dolichol-PP} + \text{Asn-NH-GlcNAc}_2\text{Man}_9\text{Glc}_3

The nucleophile here is the amide nitrogen of the asparagine side chain (activated for nucleophilic attack on the anomeric carbon of the innermost GlcNAc by the OST catalytic Asp, which deprotonates it), attacking the anomeric carbon of the reducing-end GlcNAc in an SN1-like (oxocarbenium-ion-stabilized) or SN2-like glycosyl transfer, depending on the specific glycosyltransferase mechanism. Downstream in the ER, glucosidases I and II trim the three terminal glucoses — this trimming is the signal read by the calnexin/calreticulin chaperone cycle, which checks folding status before allowing ER exit. In the Golgi, mannosidases remove mannose residues and Golgi glycosyltransferases add GlcNAc, galactose, and sialic acid, converting the original high-mannose structure into a mature complex-type glycan. Because no cellular enzyme hydrolyzes this specific glycosidic bond to Asn under normal conditions (only lysosomal degradation removes it, destructively), the modification is functionally permanent for the life of the mature protein — a limiting case confirming the earlier claim that glycosylation, unlike phosphorylation, is not a reversible switch.

Example 3 — Exam/competition level: reconstructing a proteolytic activation cascade and computing its amplification. Chymotrypsinogen (a 245-residue zymogen) is activated by a proteolytic cascade: trypsin cleaves the Arg15–Ile16 peptide bond of chymotrypsinogen to generate active π-chymotrypsin, which then autocatalytically removes two dipeptides (Ser14-Arg15 and Thr147-Asn148) to yield mature α-chymotrypsin (three chains held together by disulfides). Given that one molecule of active trypsin can cleave $k_{cat}=100\ \text{s}^{-1}$ per active site (order-of-magnitude typical serine protease turnover) and that each chymotrypsin molecule, once active, itself turns over dietary protein substrates at a similar rate, calculate the fold-amplification of proteolytic capacity 10 seconds after a single trypsin molecule is introduced into a pool of $10^4$ chymotrypsinogen molecules, assuming pseudo-first-order kinetics with $k_{cat}/K_M = 10^5\ \text{M}^{-1}\text{s}^{-1}$ and substrate far below $K_M$ so the reaction is effectively first order in substrate. First, treat trypsin acting on chymotrypsinogen as a pseudo-first-order process (single trypsin molecule at fixed low concentration, substrate in excess but below $K_M$ , so rate $= (k_{cat}/K_M)[\text{trypsin}]\times[\text{substrate}]$ ). Let $[\text{trypsin}] = 1$ molecule in a small effective volume such that its concentration is, say, $1$ nM (a representative intracellular/digestive value); then the pseudo-first-order rate constant for conversion of chymotrypsinogen is

k_{obs} = (k_{cat}/K_M)\,[\text{trypsin}] = (10^5\ \text{M}^{-1}\text{s}^{-1})(10^{-9}\ \text{M}) = 10^{-4}\ \text{s}^{-1}

After $t=10$ s, the fraction of chymotrypsinogen converted to active enzyme is

f = 1 - e^{-k_{obs}t} = 1 - e^{-(10^{-4})(10)} = 1 - e^{-10^{-3}} \approx 1.0\times10^{-3}

so of the $10^4$ zymogen molecules, approximately

N_{\text{active}} = 10^4 \times 1.0\times10^{-3} = 10\ \text{molecules}

have been converted to active chymotrypsin in the first 10 seconds — this looks small, but each of those 10 active chymotrypsin molecules is itself a full-turnover protease. If each newly active chymotrypsin molecule also has $k_{cat}/K_M \sim 10^5\ \text{M}^{-1}\text{s}^{-1}$ against dietary protein substrate (present at, say, millimolar concentration, i.e. $10^{-3}$ M, well above enzyme concentration so pseudo-first-order in enzyme), each chymotrypsin molecule cleaves peptide bonds at

\text{rate per molecule} = (k_{cat}/K_M)[\text{substrate}] = (10^5)(10^{-3}) = 100\ \text{s}^{-1}

matching the stated $k_{cat}$ ceiling (substrate saturates the enzyme at this concentration), so 10 active chymotrypsin molecules hydrolyze dietary peptide bonds at a combined rate of $10\times100 = 1000$ bonds/s — compare this to the single trypsin molecule's own direct proteolytic output of $k_{cat}=100$ bonds/s on its own substrates. The one-trypsin-in, ten-chymotrypsin-out cascade has amplified total proteolytic flux roughly tenfold within 10 seconds, and the amplification factor grows exponentially with time as more zymogen is converted and as chymotrypsin's own autocatalytic self-activation (which was neglected here for simplicity) kicks in — illustrating why proteolytic cascades (also seen in blood clotting and complement activation) are so effective at converting a trace initiating signal into an overwhelming, rapid, irreversible response. As a limiting-case check: if $t\to0$ , $f\to0$ and no amplification has yet occurred, consistent with the cascade requiring finite time to build up active enzyme.

Limiting cases, scope & common traps

Situation	What actually happens	Common misconception
Kinase and phosphatase both far below $K_M$ ( $K \gg W_T$ )	Response is graded/hyperbolic, $f\approx v/(1+v)$	Assuming all phospho-switches are digital; only the zero-order-saturated regime gives ultrasensitivity
Ubiquitin chain linked through Lys48	Canonical signal for 26S proteasomal degradation	Assuming any polyubiquitination means degradation — Lys63 chains signal endocytosis/DNA repair/signaling, not degradation
Monoubiquitination (single Ub, no chain)	Often a non-degradative signal (endocytosis, histone regulation, DNA repair recruitment)	Assuming ubiquitination always means "destroy this protein"
N-glycosylation sequon Asn-Pro-Ser/Thr	NOT glycosylated — proline's ring geometry blocks OST recognition even though the sequon "matches" the X = Pro slot	Assuming any Asn-X-Ser/Thr is automatically modified regardless of X
Zymogen with the activation peptide re-added synthetically	Does NOT regain full activity — the peptide bond, once hydrolyzed, cannot re-ligate spontaneously; conformational relaxation of the new N-terminus is required for active-site formation	Thinking proteolytic activation is "just removing a plug"; it is bond cleavage plus a conformational reorganization (e.g., new N-terminal Ile16 ion pair with Asp194 in chymotrypsin)
O-GlcNAcylation vs. O-glycosylation in Golgi	Nuclear/cytosolic O-GlcNAc is added by a single enzyme (OGT) directly with UDP-GlcNAc, dynamically reversible, and often competes directly with phosphorylation at the same Ser/Thr	Confusing this with the elaborate, irreversible, multi-enzyme Golgi O-glycosylation of secreted mucins

The chemist behind it

The modern understanding of reversible protein phosphorylation as a regulatory mechanism owes its foundation to Edwin G. Krebs and Edmond H. Fischer, who in the late 1950s worked out how glycogen phosphorylase is activated by phosphorylation and inactivated by dephosphorylation — establishing, using the specific case of glycogen metabolism in rabbit skeletal muscle, that a covalent chemical modification (not just allosteric ligand binding) could switch an enzyme's activity on and off. Their painstaking biochemistry — purifying phosphorylase kinase, demonstrating the phosphate was covalently attached to a specific serine, and showing a distinct phosphatase removed it — took more than two decades to be fully appreciated as a general regulatory principle rather than a metabolic curiosity specific to glycogen. It is now recognized as the prototype for essentially all kinase-based cell signaling, including the tyrosine kinase cascades central to growth factor and hormone signaling and to cancer.

Origins · 1 minEdwin G. Krebs and how this was discovered

Edwin G. Krebs

1918–2009 · University of Washington, discoverer with Edmond Fischer of reversible protein phosphorylation

Krebs trained as a physician before turning to enzymology, and his collaboration with Fischer began almost by accident when they set out simply to purify the enzyme that activates glycogen phosphorylase. What they found — a specific protein kinase installing a phosphate group that could later be removed by a specific phosphatase — turned out to be a universal chemical mechanism operating in essentially every signal transduction pathway in eukaryotic cells, from insulin action to the cell cycle to memory formation in neurons. Krebs and Fischer shared the 1992 Nobel Prize in Physiology or Medicine for this discovery.

History

1955

— Fischer and Krebs demonstrate that glycogen phosphorylase is interconverted between active and inactive forms by a covalent modification, phosphorylation.

1963

— Discovery that cyclic AMP activates protein kinase A, linking hormone signaling to phosphorylation cascades.

1975

— Structural and chemical characterization of proteolytic zymogen activation mechanisms (chymotrypsinogen, trypsinogen) clarifies the conformational basis of proteolytic switches.

1980

— Aaron Ciechanover, Avram Hershko, and Irwin Rose identify ubiquitin-mediated proteolysis as an ATP-dependent process, founding the ubiquitin-proteasome field (2004 Nobel Prize).

2001

— Genome-wide phosphoproteomic and glycoproteomic mass spectrometry methods begin to map PTMs systematically across entire proteomes.

Is phosphorylation always activating, and dephosphorylation always inactivating?

No. Phosphorylation can activate or inhibit a given protein depending entirely on where the phosphate is placed and what conformational change it induces. Glycogen synthase, for example, is inactivated by phosphorylation (the opposite of glycogen phosphorylase, which is activated by phosphorylation) — both respond to the same hormonal signal (insulin vs. glucagon/epinephrine) but in opposite directions, which is precisely how the cell coordinates glycogen synthesis and breakdown reciprocally from a single signaling input. The functional consequence is determined by the specific site and the conformational or steric effect of the added phosphate group, not by any universal rule about phosphorylation's meaning.

Why is N-linked glycosylation added co-translationally but O-linked glycosylation added post-translationally?

N-linked glycosylation requires the oligosaccharyltransferase complex embedded in the ER membrane to access the Asn-X-Ser/Thr sequon as the nascent chain threads through the translocon, before the protein folds into a structure that might bury the sequon — this coupling to translocation is why it is co-translational. O-linked glycosylation (on Ser/Thr hydroxyls) instead occurs in the Golgi, after the protein has folded and moved out of the ER, and does not depend on a fixed consensus sequence, instead depending on local, folding-dependent accessibility of Ser/Thr side chains to Golgi-resident polypeptide GalNAc-transferases. This is why O-glycosylation sites are harder to predict from sequence alone than N-glycosylation sites.

How does the cell prevent trypsin, once activated, from digesting the pancreas that made it?

Multiple layers of proteolytic control operate simultaneously. Trypsinogen is stored in zymogen granules segregated from active cytoplasmic proteins; a small-molecule trypsin inhibitor (pancreatic trypsin inhibitor / SPINK1) is co-secreted to neutralize any trypsin activated prematurely; and activation itself is restricted to the duodenal lumen, where enteropeptidase (enterokinase), a membrane-anchored protease expressed only on duodenal brush-border cells, cleaves trypsinogen's specific activation peptide. Mutations that impair any of these safeguards (e.g., PRSS1 mutations that make trypsinogen resistant to autolytic inactivation, or SPINK1 mutations) cause hereditary pancreatitis, directly demonstrating the physiological necessity of each layer.

What is the chemical reason ubiquitin, specifically, rather than some other small protein, is used as the degradation tag?

Ubiquitin's chemistry is well suited to its role: it is small (76 residues, ~8.5 kDa) and extremely stable (a tight β-grasp fold), so it survives being conjugated, chained, and later removed intact by deubiquitinases without itself being degraded or unfolding. Its C-terminal Gly-Gly motif (ending at Gly76) provides a flexible, unhindered carboxylate for E1 activation and subsequent isopeptide bond formation without steric interference from a bulky side chain. Crucially, ubiquitin itself carries seven lysines (K6, K11, K27, K29, K33, K48, K63) plus its own N-terminal Met1, each of which can be the acceptor for another ubiquitin's Gly76 — this is what allows the combinatorial "chain code" (different linkages meaning different things) that no simpler tag could support.

Can the same protein be both phosphorylated and ubiquitinated, and does this matter functionally?

Yes, extensively, and the interplay is often the actual regulatory point. A classic example is the transcription factor IκB: phosphorylation by IKK kinase on two specific serines creates a phosphodegron — a phospho-dependent binding site recognized by the SCF^β-TrCP E3 ubiquitin ligase, which then polyubiquitinates IκB for proteasomal degradation, freeing NF-κB to enter the nucleus. Here phosphorylation is not itself the functional signal; it is an upstream chemical prerequisite that creates the ligase-recognition motif for ubiquitination, illustrating how PTM systems are frequently wired in series rather than acting independently.

Why can't glycosylation be "undone" the way phosphorylation is, given that both are just enzymatic bond formation and hydrolysis?

Glycosidic bonds absolutely can be hydrolyzed enzymatically (lysosomal glycosidases do this routinely during glycoprotein turnover), so the chemistry of reversal exists. What glycosylation lacks is a regulated, cytosolic/ER/Golgi-resident "eraser" system analogous to phosphatases that removes a specific glycan mark on a specific residue in response to a signal, on a fast timescale, while leaving the rest of the protein intact and functional. Glycan remodeling in the secretory pathway is unidirectional and maturational (trimming and building, compartment by compartment) rather than a fast toggle; the only "reversal" available is wholesale degradation of the glycoprotein in the lysosome, which destroys the protein rather than resetting it. The one documented exception, O-GlcNAcylation in the cytosol/nucleus, does have both a dedicated transferase (OGT) and a dedicated hydrolase (OGA) and behaves as a genuinely reversible PTM.

How do phosphoproteomics and glycoproteomics experimentally detect these modifications across an entire proteome?

Mass spectrometry is the primary tool: proteins are digested into peptides, and modified peptides are enriched (using titanium dioxide or immobilized-metal affinity chromatography for phosphopeptides, or lectin affinity/hydrazide chemistry for glycopeptides) before tandem MS/MS fragmentation. A phosphorylated residue shifts peptide mass by +79.97 Da (HPO₃) relative to the unmodified peptide, and characteristic neutral losses (e.g., loss of 98 Da, H₃PO₄, from phospho-Ser/Thr during fragmentation) help localize the modified residue precisely. Glycopeptides show diagnostic oxocarbenium fragment ions (e.g., m/z 204 for HexNAc) that flag glycosylation and, combined with genetic sequon prediction, allow mapping of site occupancy across the entire secreted proteome.