Epithelia and Tissue Organization

The lining of your small intestine is a single layer of cells, yet it moves glucose one way, from the gut into the blood, while keeping the gut’s contents from seeping in between the cells.

A sheet one cell thick can do that only if each cell knows which way is up, and only if the gaps between cells are sealed. That sheet is an epithelium: cells packed side by side that cover a surface or line a tube. This lesson takes it apart in three steps. First the junctions that hold the cells together and seal them. Then the polarity that makes the top of each cell different from the bottom. Then the transport that uses both. A last section asks how plants, whose cells have walls instead of junctions, connect their cells at all.

The idea: an epithelium has a top and a bottom

Each epithelial cell has two faces. The apical membrane faces the lumen, the space inside a tube such as the gut, or the outside world. The basolateral membrane faces the neighbouring cells and, below them, the basement membrane, a thin mat of extracellular matrix covered in the lesson on adhesion and the extracellular matrix. Apicobasal polarity is the name for that difference: different proteins and lipids along the cell’s apical-to-basal axis.

Think of a row of shops on a street. Each has a front door for customers and a loading bay at the back, and the stock on each side is different. If the walls between shops had gaps, customers could wander straight into the loading bays and the difference would be lost.

In the standard scheme, which comes largely from fruit-fly genetics and mammalian cell lines, three groups of proteins mark the sides. The Crumbs complex sits apically. The Par complex, which includes the kinase aPKC, sits at the boundary where apical meets lateral. The Scribble complex sits basolaterally. The groups are thought to restrain one another, so each holds its own territory. Polarized membrane traffic then delivers the right proteins to each face, as described in the lesson on vesicle traffic.

Four junctions, four jobs

Along the lateral membrane of a vertebrate epithelium, junctions are stacked in a fixed order. Each has a transmembrane protein that reaches across to the neighbour and, inside, a link to part of the cytoskeleton.

• Tight junction, the most apical. Transmembrane proteins: claudins, with occludin among others. Scaffold proteins called ZO proteins link it to actin. Job: seal the space between cells and decide which ions may pass.
• Adherens junction, just below. Transmembrane protein: E-cadherin, a calcium-dependent adhesion protein. Inside, catenins link it to a belt of actin. Job: couple the actin of neighbours mechanically.
• Desmosomes, spot welds further down. Transmembrane proteins: desmogleins and desmocollins, cadherins of their own family. Desmoplakin in the plaque links them to keratin intermediate filaments. Job: resist stretching and tearing.
• Gap junctions, scattered along the lateral membrane. Each channel is two hemichannels, one from each cell, and in the crystal structure of human connexin 26 each hemichannel is six connexins. In the standard scheme there is no major cytoskeletal anchor. Job: let ions and small molecules pass from one cytosol to the next.

A fifth anchor, the hemidesmosome, joins the cell’s base to the basement membrane through integrin alpha6 beta4 and plectin, linked to keratin.

Order matters, and it is not universal. Tight junctions lie above adherens junctions in vertebrates. Insect epithelia seal with septate junctions that lie below the adherens junction, so the vertebrate stack is not the rule for all animals.

Gap junctions are not alike either. In pairs of cultured cells made to express one connexin, the permeability of the channel to cAMP relative to potassium was 0.180 for connexin 43, 0.027 for connexin 26 and 0.018 for connexin 40. Two cells can be equally coupled electrically and still share a messenger very differently.

Tight junctions: seal, pore, and what the knockouts showed

In 1963 Marilyn Farquhar and George Palade surveyed epithelia from rat and guinea-pig organs by electron microscopy (see the lesson on microscopy and images). In rat kidney tubules filled with hemoglobin, the tracer stopped at the zonula occludens, the junction we now call the tight junction. That placed a barrier there for a protein-sized tracer. It did not show that the junction is closed to ions.

Freeze-fracture, which splits membranes down the middle and makes a metal replica of the inside face, shows tight junctions as branching strands. In 1998 Mikio Furuse, Shoichiro Tsukita and colleagues showed that claudin-1 or claudin-2 alone built such strands in mouse L fibroblasts, cells that normally have none.

Then a surprise. In 2001 Furuse’s group put dog claudin-2 into MDCK I cells, a tight strain of a dog kidney line. Resistance fell more than twentyfold, and strand number did not change significantly. Claudin-2 makes the junction leakier to ions, in keeping with the idea, which this experiment alone does not test, that it forms a paracellular pore, a channel through the strands between cells that favours cations. This is the gate function, and it is set by which claudins a tissue makes.

The other textbook job is the fence: the junction is said to stop membrane proteins and lipids from drifting between the apical and basolateral faces. The case is weaker than the slogan. In 2019 Tetsuhisa Otani, Furuse and colleagues disrupted five major claudins in MDCK II cells. Strands vanished, and resistance fell from 52.5 to 14.5 ohm·cm² in one clone, yet polarity was not disorganized. They did not rule out other claudins, and membrane contacts containing JAM-A remained, so a fence made by other junction proteins was not excluded. Still, in these cells polarity did not need claudin strands.

Worked problem: reading a resistance measurement

TEER, transepithelial electrical resistance, is measured on cells grown as a sheet on a porous filter insert. Here are made-up numbers. MDCK monolayers grow on inserts of area 1.12 cm². A blank insert with no cells reads 120 ohm. A control monolayer reads 250 ohm, and a monolayer expressing claudin-2 reads 126 ohm.

1. Subtract the blank, because the filter and fluid add resistance of their own. Control: 250 − 120 = 130 ohm. Claudin-2: 126 − 120 = 6 ohm.
2. Multiply by area. A larger sheet has more paths side by side, so its raw resistance is lower; multiplying corrects for that. Control: 130 × 1.12 = 145.6 ohm·cm². Claudin-2: 6 × 1.12 = 6.72 ohm·cm².
3. Compare: 145.6 / 6.72 = 21.7-fold drop, the size of effect Furuse and colleagues reported.
4. Know the limit. TEER measures how easily ions carry current. It does not say which ion, or whether sugars or proteins leak too.

TEER (ohm·cm²) = (R monolayer − R blank) × area of insert

Never divide by area, and never quote one TEER for “MDCK cells”: MDCK I and MDCK II differ greatly.

Moving glucose across a cell sheet

There are two routes across an epithelium. The paracellular route runs between cells, through the tight junction. The transcellular route runs through a cell, in at one face and out at the other. Glucose absorption in the small intestine is the classic transcellular case (its place in nutrition is in the lesson on digestion).

• Apical entry. SGLT1, a sodium-glucose symporter, carries glucose into the cell with sodium. For human SGLT1 expressed in frog oocytes, coupled transport moves two sodium ions per sugar. The sodium runs down its gradient into the cell and drags glucose in against its own.
• Basolateral exit. A sodium-independent carrier, GLUT2 in the standard account, lets glucose leave down its concentration gradient by facilitated diffusion.
• The power supply. The sodium-potassium pump on the basolateral membrane uses ATP to push sodium out, keeping cytosolic sodium low. It does not touch glucose, but without it the gradient that drives SGLT1 runs down. The driving forces are worked out in the lesson on membrane transport.

The polarity matters. Put SGLT1 on both faces and the basolateral copies would pull glucose back in from the blood, working against absorption.

The 2:1 ratio holds for SGLT1, not the whole family: human SGLT2 coupled one sodium per glucose in HEK293T cells. And the paracellular route works alongside this one. In leaky epithelia, claudin pores let sodium move between cells, so the same tissue can use both routes at once.

Plants: walls and plasmodesmata

Plant cells have no tight or adherens junctions. Each is boxed in a cell wall, an extracellular matrix rich in cellulose, and neighbouring walls are glued together. The route between plant cells is the plasmodesma: a channel through the wall lined by plasma membrane that is continuous from one cell to the next. Down its middle runs the desmotubule, a narrow strand of endoplasmic reticulum continuous with the ER of both cells. Cytosol fills the sleeve between the two.

That makes a plasmodesma different in kind from a gap junction. A gap junction is a protein pore joining two separate membranes. A plasmodesma makes the two cells’ membranes one.

Appearance can mislead here. In 2019 Dawei Yan and colleagues studied an Arabidopsis root mutant, plm-2. At one interface, between the phloem-pole pericycle and the endodermis, they saw no plasmodesmata of the type with a visible cytoplasmic sleeve. Yet GFP moved across that interface faster than in wild-type roots, by fluorescence recovery after photobleaching. A pore with no visible sleeve still passed protein, at least at this interface.

Design problem

Claudin-2 lowered MDCK I resistance more than twentyfold without changing strand number. Design experiments to decide whether it makes a selective cation pore or a general leak.