# Water, Weak Bonds, and the Generation of Order

> The second law seems to say everything should fall apart. So how does a protein fold itself into its working shape, again and again? Water does much of the work. It pushes oily groups together so that they are buried, and it lets many weak bonds, each easy to break and re-form, add up to a specific structure. Water's own hydrogen bonds last only trillionths of a second, and the weak bonds work at the few-Ångström scale where atoms touch. How strongly charges attract depends on whether water is between them, and pH sets which groups carry charge at all.

Course: Biochemistry (Foundations). Level: university elective. About 27 minutes.
Lesson page (link this one when citing): https://stemhop.com/lessons/water-weak-bonds/

Builds on: Water and Life · What Am I Made Of?

## What you'll learn

- Explain water's polarity and its three jobs in the cell: cohesion, dissolving charged things, and rejecting nonpolar things.
- Rank the weak interactions (ionic, hydrogen, van der Waals) by strength in vacuum and in water, explain why water evens them out, and explain why hydrogen bonds and van der Waals contacts work only within a few Ångströms.
- Explain the hydrophobic effect as water pushing nonpolar groups together, mainly an entropy story about water that is clearest for small groups near room temperature, and connect it, with packing and weak bonds, to protein folding and membrane assembly.
- Use pH = pKa + log([A⁻]/[HA]) to predict whether a group is protonated, and explain how the bicarbonate buffer keeps blood between 7.35 and 7.45 although its pKa is 6.1.

**A protein folds itself into its working shape, often in milliseconds, with no instructions but its own sequence — and at first glance the second law of thermodynamics seems to forbid it.** Disorder is supposed to win: a sandcastle never builds itself. Yet inside every cell, polypeptide chains snap into precise structures, membranes seal themselves shut, and DNA strands find their exact partners.

The film above gives the resolution, the idea most students get backwards: entropy can *build* structure, with water’s help, once you count *whose* entropy. At constant temperature and pressure, the total entropy (protein, water and surroundings) rises exactly when ΔG = ΔH − TΔS for protein-plus-water is negative (ΔH is the change in enthalpy, or heat content; T the absolute temperature; ΔS the change in entropy, the number of ways a system can be arranged, which this lesson calls freedom). Why the same thing? Heat given off warms the surroundings and raises their entropy by −ΔH/T, so the entropy of everything changes by ΔS − ΔH/T, which is −ΔG/T.

> **The one idea to carry through this lesson.** Count the water, not just the molecule. A change goes forward when the free energy of the molecule *plus its water* falls (ΔG < 0). In water, each weak interaction is worth only about one to a few RT net (RT, the gas constant times the absolute temperature, ≈ 2.6 kJ/mol at 37 °C), so no single contact decides anything. The sum decides. That sum includes the freedom water gains when nonpolar surface is buried, and the charges that pH allows: one pH unit changes the free energy of a group’s protonation by 2.303 RT, about 5.9 kJ/mol, the size of one weak bond.

The rule is for dilute watery solution near 25–37 °C. It says which state is favored, not how fast a chain gets there, or whether it gets there without help.

## Water is the engine, not the stage

Water is polar. The oxygen pulls electrons far harder than the hydrogens do, so the oxygen carries a partial negative charge and each hydrogen a partial positive. The molecule bends at about 104.5 degrees, so those charges don’t cancel: water is an electric dipole, with a minus end and a plus end.

That polarity gives water three jobs. It sticks to itself, hydrogen end to oxygen end, in a flickering network where each bond forms and breaks in a few trillionths of a second. It dissolves many charged things, wrapping a sodium ion in a shell of oxygens and a chloride in a shell of hydrogens. And it rejects anything nonpolar, refusing to make room for an oil droplet the way a crowded room refuses to absorb someone who won’t talk to anyone. Hold onto that third job.

*Figure (AI-generated illustration).* One polar molecule, three jobs. Water sticks to itself in a network whose bonds flicker on and off in trillionths of a second; it wraps ions in shells, negative end toward a cation and positive end toward an anion; and it offers nothing to a nonpolar surface. That third job is the one that builds proteins and membranes. Image: a three-panel schematic of water molecules drawn as dipoles: a flickering hydrogen-bonded network, hydration shells with the negative end facing a sodium ion and the positive end facing a chloride ion, and water lining but not bonding to an oil droplet.

## Weak on purpose, measured in RT

Covalent bonds *build* the cell — the backbone of a protein, the sugar-phosphate rails of DNA — and at 100 to 1,000 kilojoules per mole they are made to last. But the cell’s work is done by bonds you can break: hydrogen bonds, ionic attractions, van der Waals contacts. When you smell a cat, odorant molecules bind your receptors, trigger a signal, and fall off. If that binding were covalent, you would smell that cat for the rest of your life. Reversibility is the point.

What breaks a weak bond is heat. A particle in water jitters as water molecules hit it, the random walk that, summed over billions of molecules, we call diffusion. Warm the water or shrink the tracer and watch it speed up.

*Interactive on the lesson page:* Brownian motion — water never sits still

**RT** is the yardstick for that jostling: covalent bonds are about 40 to 400 RT, so heat almost never breaks them, while a bond worth one or two RT breaks and re-forms all the time.

### Hydrogen bonds: who can donate

A hydrogen bond forms when a hydrogen on a strongly electronegative atom, O or N (the **donor**), is drawn to a lone pair on another O or N (the **acceptor**).

**Predict.** Which can donate a hydrogen bond: serine’s –OH, lysine’s –NH₃⁺, or alanine’s –CH₃?

> **Check your prediction.** The –OH and –NH₃⁺ can. The –CH₃ cannot, in any useful way: carbon’s electronegativity (2.55) is too close to hydrogen’s (2.20) to leave the hydrogen much positive charge, and the methyl carbon has no lone pair to accept one. That is why methyl groups are nonpolar.

In a vacuum a hydrogen bond is worth about 10 to 40 kJ/mol. In water the donor and acceptor were already hydrogen-bonded to water, so pairing them mostly trades one bond for another, and the net gain is a few kJ/mol, one to a few RT.

### Charges: why water evens things out

Coulomb’s law gives the energy between two charges q₁ and q₂ a distance r apart: E = kq₁q₂/(Dr). Here k is a constant and D, the dielectric constant, measures how well the medium screens charge: 1 for a vacuum, about 80 for water, about 2 to 4 inside a protein. Two opposite charges 4 Å apart attract with about 350 kJ/mol in a vacuum, about 90 inside a protein (D ≈ 4), and about 4 in water. So in a vacuum, ionic beats hydrogen bond beats van der Waals, but in water an exposed salt bridge (an ion pair between side chains) gets at most about 4 to 6 kJ/mol from Coulomb’s law alone, measured values are usually smaller, and it is no stronger than a hydrogen bond’s net share.

**Predict.** Move a glutamate–lysine salt bridge from the surface (D ≈ 80) into the core (D ≈ 3) at the same distance. How much stronger is the attraction, and does the protein gain that much stability?

> **Check your prediction.** By Coulomb’s law alone, about 27-fold stronger (80 / 3). But to reach the core both charges had to leave their water, and that **desolvation penalty** can cost as much as the attraction gains. The net effect ranges from destabilizing to several kcal/mol.

Ethanol precipitation of DNA runs the same law the other way: change the water instead. Ethanol lowers D, the attraction between the negative phosphates and counterions such as sodium strengthens, and the DNA clumps into a white thread.

### The four-Ångström window

Hydrogen bonds and van der Waals contacts work at the same length scale: roughly three to four Ångströms (a few ten-billionths of a meter) between atoms. Closer, atoms repel; a little farther, the attraction fades fast. Ionic attraction reaches farther, though water and dissolved salt damp it down. Life happens in that narrow window. On the ruler below, slide the distance past 5 Å and watch the energy all but vanish.

*Interactive on the lesson page:* Weak-bond ruler pick two groups, set the distance Group A Group B Medium around the groups: Vacuum (D = 1) Protein core 

One van der Waals contact is worth under about 4 kJ/mol (1 kcal/mol), about 1.5 RT. But weak bonds add, and since an equilibrium constant grows exponentially with energy, their effects multiply: one 4 kJ/mol contact gives a constant of about 5, ten acting together about 5 million (an upper bound: in water the net gain per contact is smaller, because water was touching those atoms too, and a real interface pays some entropy back). A gecko hangs from a ceiling mainly on van der Waals forces, made over and over by the branching hairs on its feet (figure). Drag the “How many bonds” slider above and watch the total grow.

*Figure (AI-generated illustration).* One weak bond is almost nothing, but weak bonds add. A gecko hangs from a ceiling mainly on van der Waals forces: millions of tiny hairs on its feet each branch into many finer tips, making countless trivial contacts, and the sum can hold many times its body weight. Image: a gecko hanging upside down from a ceiling by its feet, with two zoom insets: a toe pad covered in a dense brush of tiny hairs, and one hair whose end splits into many fine branches, each tip touching the ceiling. Adhesion is mainly due to van der Waals forces.

## What pays for order: the hydrophobic effect

The film above animates this section: the stiffer layer of water around one methyl group, the release when oily groups gather, and the folding ledger. Here are the details it skips. The water next to a –CH₃ keeps its hydrogen bonds by pointing them toward other water.

*Figure (AI-generated illustration).* Nonpolar groups end up together mainly because water gains freedom, not because they attract each other. Each exposed group is lined by a layer of water only modestly more ordered than the bulk (not an ice cage); gathering the groups shrinks that surface and frees the water. Near room temperature this gain is almost all entropy. Image: a schematic in two panels. Left, scattered nonpolar groups each lined by a thin, slightly ordered layer of water. Right, the same groups gathered into one cluster with less surface to line, and the released water moving freely.

**Predict.** When a protein chain folds and buries its nonpolar side chains, what is the sign of ΔS for the chain? For the water?

> **Check your prediction.** The chain’s entropy falls: a floppy chain that could take an astronomical number of shapes is held in roughly one. The water’s entropy rises: water that lined the buried surface is released into the bulk. Which is bigger? The ledger below answers it.

That release is the **hydrophobic effect**. The groups’ own van der Waals attraction is weak. At a large, flat surface such as an oil droplet’s, enthalpy’s larger share comes from water simply losing hydrogen bonds it cannot replace.

Here is the film’s ledger for a protein of about 100 residues near 25 °C, in round numbers (positive opposes folding, negative favors it):

| Term | Rough size (kJ/mol) |
| --- | --- |
| The chain loses its freedom of shape | about +550 |
| Water freed from about 4,800 Å² of nonpolar surface buried in the core | about −500 |
| Hydrogen bonds, ion pairs and tight van der Waals packing inside the fold, counted net of the bonds those groups used to make with water | about −90 |
| **ΔG of folding** | **about −40** |

The first two rows come from standard rules of thumb: about 25 cal/mol for each Å² of nonpolar surface buried (4,800 × 25 cal is about 500 kJ), and 4 to 7 kJ/mol of lost freedom per residue. The total is what is measured: small proteins typically fold with ΔG between about −20 and −60 kJ/mol. So the third row is simply what is left over (Going deeper).

The film reads the ledger for the second law and for the small difference between big numbers. Two numbers it leaves out. Near 25 °C nearly all of the −500 is the water’s entropy, about −480: a little less than the chain’s +550. And a 10% change in the burial row is 50 kJ/mol. The same thin margin keeps folded proteins only modestly stable, which also leaves room for the flexing many proteins need to work.

The same logic builds membranes. A phospholipid with two greasy tails can’t dissolve, and it is too bulky to pack into the small sphere, a micelle, that single-tailed soaps form. So it lines up tail-to-tail into a two-layer sheet, a bilayer, that closes on itself so no tails are left exposed at an edge. Enzymes make the lipids and proteins insert them into existing membranes, but no enzyme holds a bilayer together. Water does: the lipids lose freedom when they line up, and the water freed from their tails pays for it.

*Figure (AI-generated illustration).* Single-tailed soaps pack into a small sphere, a micelle. Two-tailed phospholipids are too bulky for that, so they line up tail to tail in a bilayer, and the sheet closes on itself because an open edge would leave tails touching water. No enzyme holds the bilayer together; the water freed from the tails pays for the lipids' lost freedom. Image: a schematic of a closed lipid bilayer vesicle with two-tailed lipids arranged tail to tail, beside a small micelle of single-tailed soap molecules and a flat bilayer fragment whose open edge exposes tails to water.

## pH: the ledger for charge

Ionic attractions, and many hydrogen bonds, depend on whether a group is charged, which depends on pH. Back to the one idea: a proton coming on or off is one more term in the ledger, worth about one weak bond per pH unit.

Below a group’s pKa, the group mostly holds its proton; above it, it mostly lets go; at the pKa, half and half. Call the protonated form HA and the form that has lost its proton A⁻. The Henderson-Hasselbalch equation writes the rule out: pH = pKa + log([A⁻]/[HA]).

| pH − pKa | [A⁻] : [HA] | Share deprotonated |
| --- | --- | --- |
| −2 | 1 : 100 | about 1% |
| −1 | 1 : 10 | about 9% |
| 0 | 1 : 1 | 50% |
| +1 | 10 : 1 | about 91% |
| +2 | 100 : 1 | about 99% |

**Predict.** Glycine has a carboxyl group (pKa 2.3) and an amino group (pKa 9.6). At pH 7.4, what form does each take, and what is the net charge?

> **Check your prediction.** The carboxyl’s pKa is 5.1 units below the pH, far beyond the table’s last row: it is –COO⁻, more than 99.99%. The amino group’s pKa is 2.2 units above the pH, so it stays –NH₃⁺, about 99.4%. One negative, one positive: a **zwitterion**, net charge 0.

The table also explains buffers. Far from the pKa, one form has nearly run out, leaving little to absorb added acid or base, so a simple buffer in a sealed flask works best within about one unit of its pKa.

Blood pH lives between 7.35 and 7.45, a window one tenth of a pH unit wide, guarded by the bicarbonate buffer (CO₂ + H₂O ⇌ H₂CO₃ ⇌ H⁺ + HCO₃⁻), effective pKa 6.1. At 7.4 that is 1.3 units off, about 20 bicarbonate ions per dissolved CO₂ (101.3 ≈ 20): a poor buffer by the sealed-flask rule. It works because blood is not sealed: breathing carries the extra CO₂ away. Model it with bicarbonate alone, 24 mM HCO₃⁻ and 1.2 mM dissolved CO₂, and add 10 mM of strong acid; each mM of acid turns one mM of bicarbonate into CO₂. In a sealed flask the pH falls to about 6.2 (14 mM HCO₃⁻ against 11.2 mM CO₂). If breathing keeps CO₂ at 1.2 mM, it falls only to about 7.17. Real blood does better still: the lungs lower CO₂ in response, and hemoglobin and other buffers take up part of the acid. The kidneys then restore the bicarbonate.

**Predict.** A patient hyperventilates until blood CO₂ halves, bicarbonate unchanged. Which way does blood pH move?

> **Check your prediction.** Up, to about 7.7: a respiratory alkalosis. Removing CO₂ pulls the equilibrium left, so [H⁺] falls (in this bicarbonate-only model, the bicarbonate used up is negligible against 24 mM). CO₂ is the acid side of the ratio, so halving it doubles [HCO₃⁻]/[CO₂], and pH rises by log 2, about 0.3.

Cross much past the window and proteins change their charges and work less well, the heart rhythm can falter, and a patient can slide toward coma. Less than one pH unit separates ordinary chemistry from a medical emergency. Life is organized chaos, held together by bonds weak enough to let go, paid for by water’s demand for its own freedom, and tuned by a single number, pH.

## Going deeper

### How temperature changes the hydrophobic effect

The hydrophobic contribution to stability tends to *strengthen* with temperature over the physiological range, so some proteins are less stable in the cold and can even cold-denature. The cause is a positive heat-capacity change, ΔCp, on exposing nonpolar surface: as temperature rises the entropy term shrinks and the enthalpy term grows. In a standard (convergence) model, entropy’s share of the cost is about 96% at 25 °C, 82% at 37 °C and 56% at 60 °C, while the total grows only about 4% from 25 to 37 °C. The same ΔCp bends protein stability into a two-armed curve: heat unfolds a protein because the chain’s entropy wins, and cold because the hydrophobic effect weakens.

### Which favorable terms matter most

The ledger’s third row is a remainder, not a measurement, so how the favorable side splits is still argued. Burying oily groups is widely considered the largest single contribution to folding, but some calorimetric and mutational analyses argue that hydrogen bonds contribute comparably.

### Favored is not the same as fast

Anfinsen proposed that the native shape is the chain’s lowest-free-energy state in its environment. Many proteins fit, but some need help to fold, and amyloid can be even more stable than the native form. Folding avoids a random search of every shape (Levinthal’s paradox) by running downhill on a funnel-shaped energy landscape.

---

From STEMhop (https://stemhop.com), free lessons by Christopher Deppmann. Lesson text © the authors. Practice questions: https://stemhop.com/lessons/water-weak-bonds/
