Cell Biology · Unit 5: Cells in tissues
Stem Cells and Tissue Renewal
A stem cell is defined by what it does over time, not by how it looks or which marker it carries. This lesson pairs each part of the definition (self-renewal, potency, the niche) with the experiment that tests it, using the mouse gut crypt as the working example, and then asks what resetting a specialized cell to pluripotency does and does not show. Afterwards you can read a lineage-tracing result, predict clone numbers under neutral drift, and design a test of whether a niche cell is truly required.
Watch first Chance Decides Which Stem Cell Wins · 3:02
The epithelium lining a mouse’s small intestine is replaced every few days, yet the animal keeps it for life. Something at the bottom of each crypt never runs out.
That something is a stem cell. The word is used loosely, so this lesson pins it down the way cell biologists do: by tests. You will see what counts as evidence that a cell self-renews, how many fates it can produce, and what its neighbours contribute. Then you will meet the experiments that showed a specialized cell’s nucleus can be reset.
What would count as a stem cell: tests, not looks
A stem cell has two properties, and each has its own test.
Self-renewal means that dividing cells maintain a stem population over long periods. One division that makes a copy proves little, because fast-dividing transit-amplifying cells (short-lived progenitors that divide a few times and then differentiate) also make copies. The test is persistence: transplant the cells and then transplant their descendants into a second host, or grow them from one cell and replate them again and again.
Potency is the range of cell types a cell can produce. The scale runs from totipotent (one cell can make an entire conceptus, including placenta and other extra-embryonic tissues), through pluripotent (all the lineages of the embryo proper), multipotent (the several cell types of one tissue) and unipotent (one cell type). The test is clonal: start from one cell and count the kinds of descendant.
The third idea is the niche, the local surroundings that keep a stem pool going. Its test is loss and rescue: remove a candidate signal or cell, see the stem cells fail, then restore the signal and see them recover.
The pattern was set in 1961 by James Till and Ernest McCulloch. They injected marrow cells into irradiated mice and counted nodules in the spleen; the count rose in proportion to the cells injected. In 1963, with Andrew Becker, they used chromosome markers to show that each colony was a clone. One cell, its descendants, and a count of lineages: every later definition follows that template. Notice what is missing from it. Stem cells need not be dormant, and nothing in the definition says each division must be asymmetric. A marker is not a definition either. A marker is useful only after the functional tests have been passed.
The gut crypt: a model renewing tissue
The small intestine is a sheet of epithelium folded into finger-like villi and pocket-like crypts (see epithelia). Cells are born in the crypts, move up the villus as they differentiate into absorptive cells, mucus-secreting goblet cells and hormone-secreting enteroendocrine cells, and are shed at the tip. At the very bottom of each crypt sit Paneth cells, which secrete antibacterial proteins and stay put. Squeezed between them are slender crypt-base columnar cells.
In 2007 Nick Barker, Hans Clevers and colleagues found that a Wnt target gene, Lgr5, is expressed in these cycling columnar cells in mouse small intestine and colon. Expression alone shows nothing about fate, so they traced the cells. The method relies on Cre, a recombinase that cuts DNA between two loxP sites. Fused to a mutant estrogen-receptor domain (CreERT2), Cre enters the nucleus only after the animal receives tamoxifen, and it then removes a stop sequence from a Rosa26 reporter gene, leaving a mark every daughter inherits (more in tools for development).
After one pulse in Lgr5-CreERT2 mice, label began at the crypt base and, over 60 days, filled ribbons of epithelium containing every epithelial lineage. That is long-term multilineage output from cells in their own tissue, which the authors took as evidence that these cells are the stem cells. It shows multipotency within the gut, not pluripotency, and it does not show that Lgr5 cells are the only cells able to restore the crypt after injury.
Lineage tracing and neutral drift
How do the stem cells share the crypt? One classic cartoon has each stem cell divide asymmetrically, every time, into one stem cell and one transit-amplifying cell. In 2010 Hugo Snippert, Clevers and colleagues tested it with the Confetti reporter, which after Cre acts turns on one of four fluorescent proteins at random, so that neighbouring clones differ in colour.
They found that mouse small-intestinal Lgr5-high cells divide about once a day, so these stem cells are busy, not dormant. Most divisions were symmetric, giving two stem cells or two non-stem cells. Over 1 to 6 months, crypts drifted toward a single colour. Surviving clones grew while others vanished. The data fit neutral drift: equivalent stem cells crowd one another out at random, and chance, not a fixed hierarchy, decides which lineage wins. Balance is kept by the population, not by each division.
Here is the logic in numbers. The model below is illustrative, with round numbers chosen to show the arithmetic.
Suppose a low dose of tamoxifen labels 100 of 1,000 crypts, each with a single labelled stem cell at day 2. Assume each crypt holds N = 5 equivalent, competing stem cells. This is an effective number for the model; it is smaller than the roughly 14 Lgr5-high cells per crypt reported for the mouse small intestine (a figure from the clonal-tracing literature), possibly because not every marked cell competes equally.
probability a single clone takes over its crypt = 1 / N
- Fixation. With N = 5, each labelled clone takes over its crypt with probability 1/5 = 0.2. Expect 100 x 0.2 = 20 fully labelled crypts. The other 80 clones are lost.
- Conservation check. Labelled stem cells at the start: 100 clones x 1 cell = 100. At the end: 20 crypts x 5 cells = 100. Under neutral drift the total is conserved on average.
- An intermediate time. Suppose 50 labelled clones survive. Their mean size must be about 100 / 50 = 2 stem cells each. Surviving clones grow precisely because others vanish, the pattern Snippert and colleagues reported.
Change N to 14 and the expected count drops to 100 / 14, about 7 crypts. Counting monoclonal crypts therefore estimates how many cells really compete.
labelled stem cell unlabelled stem cell clone lost clone survives, crypt mixed crypt fully labelled
| Of the 100 labelled crypts | Observed now | Expected |
|---|---|---|
| Clones surviving | ||
| Mean size of a surviving clone | ||
| Fully labelled crypts | ||
| Labelled stem cells in total |
Rule to keep: with N equivalent stem cells, a labelled clone takes over its crypt with probability 1/N, so 100 labelled crypts give about 100/N fully labelled ones, and labelled stem cells are conserved on average (100). In this sketch each event replaces one randomly chosen stem cell with a copy of a randomly chosen one, and the time axis is in events, not days. N here is an effective number for the model, which the lesson puts below the roughly 14 Lgr5-high cells counted per crypt (the slider runs up to 14). The numbers are round and illustrative, as in the lesson.
Niches, and why the Paneth cartoon was too simple
In 2009 Toshiro Sato and colleagues grew single sorted mouse Lgr5 cells into organoids, small crypt-villus structures, embedded in Matrigel (a basement-membrane matrix) with EGF, R-spondin 1 and Noggin (a BMP inhibitor). Organoids could be broken up and replated repeatedly. They contained no stromal cells, which made them a clean test of propagation, though matrix and growth factors still stood in for a niche (see organoids).
In 2011 Sato and colleagues reported that Paneth cells express Wnt3, EGF, TGF-alpha and the Notch ligand Dll4. Sorted stem cells formed organoids far better with Paneth cells beside them, and a pulse of Wnt could substitute for the Paneth cells. That is a clean sufficiency result in culture. The same paper also reported that genetic removal of Paneth cells in vivo coincided with loss of Lgr5 cells, a result that later Paneth-specific ablation (below) did not reproduce (for the pathways, see Wnt signaling and Notch signaling).
Design problem
The culture data make Paneth cells look like the niche. Are they indispensable in the living mouse gut? Design an in vivo test, state what “Paneth cells only” and “redundant sources” each predict, and list your controls.
One way to do it
Ablate Paneth cells genetically. Express the human diphtheria-toxin receptor from a Paneth-cell gene such as a lysozyme gene, so toxin kills Paneth cells and spares other cells. Score Lgr5-positive cell number, EdU incorporation (S-phase entry; see cell-cycle control) and lineage output from a separate Lgr5 tracing pulse over days to weeks.
Separately, block Wnt secretion at its source. Delete Porcn, which encodes Porcupine, the enzyme that adds the lipid Wnt needs to be secreted, either in the epithelium or in stromal FOXL1-positive telocytes, flat cells with long extensions lying just beneath the epithelium.
Predictions: if Paneth cells are the only niche, ablation should lose Lgr5 cells and stall renewal, and telocyte Porcn deletion should do nothing. If sources are redundant, stem cells persist after ablation and telocyte Wnt matters.
Controls: toxin given to mice without the receptor; receptor mice without toxin; staining to confirm how completely Paneth cells were removed; tamoxifen alone for the Cre lines.
The published mouse results favour redundancy. Michal Shoshkes-Carmel and colleagues (2018) found that telocyte Porcn deletion cut crypt Wnt signalling and proliferation within 24 hours. Johan van Es and colleagues (2019) found that Lgr5 cells survived Paneth ablation, with enteroendocrine and tuft cells supplying Notch ligand instead.
The niche is a set of signals with more than one source. Fates are also less fixed than the cartoon implied: in 2020 Murata and colleagues found that after lethal injury to Lgr5 cells in mouse gut, absorptive and secretory progenitors dedifferentiated to rebuild the stem pool, switching on Ascl2 days before Lgr5. That is an injury result, and it does not mean mature cells act as stem cells during normal renewal. Neural stem cells raise similar questions (see adult neurogenesis and injury and regeneration).
Resetting a cell: Gurdon to Yamanaka
Does a specialized cell lose the genes it no longer uses? In 1962 John Gurdon transplanted nuclei from intestinal cells of feeding Xenopus tadpoles into eggs whose own nuclei had been destroyed by ultraviolet light. Most failed, but a small fraction became normal feeding tadpoles. At least some specialized nuclei still held the instructions; the egg could read them out again.
In 2006 Kazutoshi Takahashi and Shinya Yamanaka asked whether defined proteins could do what egg cytoplasm does. Four transcription factors, Oct3/4, Sox2, Klf4 and c-Myc, turned mouse fibroblasts into induced pluripotent stem (iPS) cells under embryonic-stem-cell culture conditions (for how such factors reshape chromatin, see eukaryotic regulation). The cells formed tumours with tissues of all three germ layers and contributed to mouse embryos. Those first clones, selected with an Fbx15 reporter, did not make adult chimaeras; Nanog-selected clones in 2007 did, and one passed through the germline. Also in 2007, Yamanaka’s laboratory reprogrammed adult human skin fibroblasts with the same four factors, and James Thomson’s laboratory reported human iPS cells with a different set (OCT4, SOX2, NANOG, LIN28).
Read the result for what it is. It shows that nuclear state is resettable. It does not show that every iPS line matches embryonic stem cells in every test, it does not make them totipotent, and passing developmental tests is a separate question from clinical safety.
How we measure it
Inducible Cre lineage tracing
A gene expressed in the cells of interest drives CreERT2, a Cre recombinase that enters the nucleus only after tamoxifen. A pulse of tamoxifen removes a stop sequence from a reporter gene, and the mark is inherited by every descendant. Looking weeks later shows what the labelled cells made.
Multicolour clonal analysis (Confetti)
A reporter that, after Cre acts, expresses one of four fluorescent proteins at random gives neighbouring clones different colours. Counting how many clones survive and how large they grow over time tests models of how stem cells compete.
Single-cell organoid culture and replating
One sorted cell is embedded in basement-membrane matrix with defined growth factors and grows into a crypt-villus organoid. Breaking the organoid up and replating it again and again tests sustained propagation under those stated conditions.
Spleen colony transplantation assay
Marrow cells are injected into irradiated mice, and the colonies that appear in the spleen are counted and analysed. Colony number per cell injected, colony clonality and the lineages in each colony measure blood-forming stem and progenitor cells.
Finish line
Test yourself
8 questions
Practice and MCAT-style, with explained answers.
Next · Lesson 25
Cell Identity, Cell State and Disease
Most of the lining of an older person's healthy oesophagus can carry a mutation found in cancer. Why is it still healthy?
Continue →