Research Article

Single-cell Wnt signaling niches maintain stemness of alveolar type 2 cells

See allHide authors and affiliations

Science  09 Mar 2018:
Vol. 359, Issue 6380, pp. 1118-1123
DOI: 10.1126/science.aam6603

Fibroblasts as lung stem cell niche

Each breath that we take provides oxygen to the bloodstream via tiny sacs in the lung called alveoli. AT1 cells line the alveoli and mediate gas exchange, whereas AT2 cells secrete lung surfactant. A subset of AT2s also serve as stem cells that slowly generate new alveolar cells throughout adult life. Nabhan et al. show that the rare AT2 stem cells have a special niche next to a fibroblast secreting Wnts. This Wnt activity is needed to select and maintain the stem cells. Injury expands the stem cell pool by transiently inducing autocrine Wnts in other surfactant-secreting alveolar cells. This simple but expandable niche sustains oxygen delivery, and it is co-opted in lung cancer.

Science, this issue p. 1118


Alveoli, the lung’s respiratory units, are tiny sacs where oxygen enters the bloodstream. They are lined by flat alveolar type 1 (AT1) cells, which mediate gas exchange, and AT2 cells, which secrete surfactant. Rare AT2s also function as alveolar stem cells. We show that AT2 lung stem cells display active Wnt signaling, and many of them are near single, Wnt-expressing fibroblasts. Blocking Wnt secretion depletes these stem cells. Daughter cells leaving the Wnt niche transdifferentiate into AT1s: Maintaining Wnt signaling prevents transdifferentiation, whereas abrogating Wnt signaling promotes it. Injury induces AT2 autocrine Wnts, recruiting “bulk” AT2s as progenitors. Thus, individual AT2 stem cells reside in single-cell fibroblast niches providing juxtacrine Wnts that maintain them, whereas injury induces autocrine Wnts that transiently expand the progenitor pool. This simple niche maintains the gas exchange surface and is coopted in cancer.

Although there has been great progress identifying tissue stem cells, much less is known about their niches and how niche signals control stem cell function and influence daughter cell fate (1, 2). The best understood examples come from genetic systems (3) such as the Drosophila testis niche, where 10 to 15 cells (“the hub”) provide three short-range signals to the 5 to 10 stem cells they contact (4). These signals promote stem cell adhesion to the niche and inhibit differentiation, but after polarized division, a daughter cell leaves the niche, escaping the inhibitory signals and initiating sperm differentiation. In mammalian systems, stem cells and their niches are typically more complex, with more cells and more complex cell dynamics. Even in the best-studied tissues (58), there is incomplete understanding of niche cells, signals, and the specific aspects of stem cell behavior each signal controls. Here, we describe an exquisitely simple stem cell niche and control program that maintains the lung’s gas exchange surface.

Mouse genetic studies have identified a hierarchy of stem cells that replenish the alveolar surface (9), some of which are active only after massive injury (10, 11). Normally, the epithelium is maintained by rare “bifunctional” alveolar type 2 (AT2) cells, cuboidal epithelial cells that retain the surfactant biosynthetic function of standard (“bulk”) AT2 cells (12) but also serve as stem cells (13, 14). Their intermittent activation gives rise to AT1 cells—exquisitely thin epithelial cells that mediate gas exchange—and generates slowly expanding clonal “renewal foci” that together create ~7% new alveoli per year (13). Dying cells are proposed to provide a mitogenic signal transduced by the epidermal growth factor receptor (EGFR)–KRAS pathway that triggers stem cell division (13). However, it is unclear how stem cells are selected from bulk AT2 cells, how they are maintained, and how the fate of daughter cells—stem cell renewal versus reprogramming to AT1 identity—is controlled.

Here, we molecularly identify alveolar stem cells as a rare subpopulation of AT2 cells with constitutive Wnt pathway activity and show that a single fibroblast near each stem cell comprises a Wnt signaling niche that maintains the stem cell and controls daughter cell fate. Severe injury recruits ancillary stem cells by transiently inducing autocrine Wnt signaling in “bulk” AT2 cells.


Wnt pathway gene Axin2 is expressed in a rare subpopulation of AT2 cells

Canonical Wnt signaling activity marks stem cells in various tissues (8), and the Wnt pathway is active in developing alveolar progenitors (1517). To determine whether AT2 cells in adult mice show Wnt activity, we examined expression of Wnt target Axin2 (18) using an Axin2-Cre-ERT2 knock-in allele crossed to Cre reporter Rosa26mTmG. After three daily tamoxifen injections to induce Cre-ERT2 at age 2 months, fluorescence-activated cell sorting (FACS) showed 1% of purified AT2 cells expressed the green fluorescent protein (GFP) reporter (Fig. 1D). Immunostaining for GFP and canonical AT2 marker SftpC showed labeled AT2 cells distributed sporadically throughout the alveolar region (Fig. 1, A to C). Multiplexed single-molecule fluorescence in situ hybridization [proximity ligation in situ hybridization (PLISH) (19)] confirmed a distributed population of Axin2-expressing AT2 cells (fig. S1). Axin2+ AT2 cells represent a stable subpopulation because the percentage of labeled AT2 cells did not increase when tamoxifen injections were repeated 1 and 2 weeks after the initial induction, and the percentage was similar among animals induced at different ages (Fig. 1G). This subpopulation expressed all AT2 markers, including surfactant proteins and lipids (fig. S2), suggesting that the cells are physiologically functional. No AT1 or airway epithelial cells were marked under these “pulse-labeling” conditions (>1000 AT1 cells scored in each of three mice), although other (nonepithelial) alveolar cells were. Thus, Axin2+ AT2 cells represent a rare, stable subpopulation of mature AT2 cells.

Fig. 1 Axin2 marks rare AT2 cells with stem cell activity.

(A to C) Alveoli of adult (2 months) Axin2-CreERT2;Rosa26mTmG mouse lung immunostained for Cre reporter mGFP (membrane-bound form of GFP), AT2 marker pro-surfactant protein C (SftpC), and 4′,6-diamidino-2-phenylindole (DAPI) 5 days after three daily injections of 3 mg tamoxifen (3-day pulse) to lineage-label Axin2-expressing (Axin2+) cells (Axin2>GFP). Close-ups show (B) Axin2 bulk AT2 cell and (C) rare Axin2+ AT2, indicating Wnt pathway activation. (D) FACS of AT2 cells in (A) to (C) shows 1.0 ± 0.5% (n = 3 biological replicates) express Axin2>GFP. (E) Alveoli labeled as in (A) to (C), harvested 1 year later (1-year chase). There are increased Axin2-lineage AT2 cells (arrowheads) and labeled AT1 cells and fibroblasts (arrows), the latter from another Axin2+ lineage. (F) Close-up of lung as in (E), showing Axin2+ lineage-labeled AT1 cells (arrows). RAGE, AT1 membrane marker; arrowhead, AT2 cell. (G) Quantification of Axin2-lineage labeled AT2 cells immediately after 3-day or 3-week pulse (3-day pulse each week) at age 2 months, 3-day pulse at 4 months, or after 3-day pulse at 2 months plus 1-year chase. Mean ± SD (n = 3500 AT2 cells scored in two to four biological replicates). (H and I) Alveoli of Axin2-CreERT2;Rosa26Rainbow mice given limiting dose of tamoxifen (2 mg) at 2 months to sparsely label Axin2+ cells (H) with different fluorescent clone markers [mOrange in (H) and (I)], and immunostained for SftpC 1 week (H) or 6 months (I) later to detect AT2 clones. (J) Quantification of AT2 clone sizes 1 week and 6 months after labeling. ***P < 0.001 (Student’s t test). Scale bars, 25 μm (A) and (E); 5μm (C) and (F); and 20 μm (I).

Axin2+ AT2 cells have alveolar stem cell activity

The fate of the labeled AT2 cells was examined a half or 1 year later (half or full year “chase”) (Fig. 1, E to I). Labeled cells exhibited three features of stem cells. First, unlike most AT2 cells, which are quiescent (20), 79% of Axin2+ AT2 cells generated small clones of labeled cells (Fig. 1, H to J). Daughter cells remained local (Fig. 1I), with some found as doublets (fig. S3B), indicating recent division; on occasion, an Axin2+ AT2 cell was seen dividing (fig. S3A), an intermediate that we never observed for bulk AT2 cells in normal lungs. Second, lineage-labeled AT2 cells expanded sixfold relative to unlabeled cells during a 1-year chase (Fig. 1G). Third, labeled cells gave rise to another alveolar cell type, shown by appearance of AT1 cells expressing the lineage label (Fig. 1F). Like AT2 daughter cells, daughter AT1 cells were typically found in close association with the presumed founder Axin2+ AT2 cell. Thus, Axin2+ cells constitute a rare AT2 subpopulation with stem cell activity, which slowly (about once every 4 months) self-renew and produce new AT2 and AT1 cells.

Fibroblasts provide short-range Wnt signals to neighboring AT2 stem cells

Wnts are local signals with a typical range of just one or two cells (21). Fibroblasts were an excellent candidate for the Wnt source because some contact AT2 cells (22), such as Pdgfrα-expressing fibroblasts that support surfactant production and formation of alveolospheres in culture (14, 23, 24). Transmembrane protein Porcupine, which acylates and promotes secretion of Wnts (25) and marks Wnt signaling centers (26), was expressed in rare alveolar stromal cells (fig. S4A), most of which were Pdgfrα-expressing fibroblasts (fig. S4C) and some were closely associated with AT2 cells (fig. S4B). Serial dosing of Porcupine Porcn inhibitor C59 reduced the pool of Axin2+ AT2 cells by 68% (Fig. 2A). Targeted deletion in lung mesenchyme (by using Tbx4LME-Cre) or fibroblasts (Pdgfrα-Cre-ER) of Wntless, another transmembrane protein required for Wnt secretion (27), also reduced the pool (Fig. 2, B and C). The remaining Axin2+ AT2 cells could be due to incomplete deletion or perdurance of Wntless in PDGFRα+ fibroblasts or to another Wnt source.

Fig. 2 Single, Wnt-secreting fibroblasts comprise the stem cell niche.

(A) Effect of five daily doses of Wnt secretion [Porcupine (PORCN)] inhibitor C59 (+) or vehicle control (–) on Axin2 expression in AT2 cells at age 2 months, measured with PLISH for Axin2 and Sftpc. Mean ± SD (n = 900 AT2 cells scored in three biological replicates). ***P = 0.001 (Student’s t test). (B and C) Effect on Axin2-expresssion in AT2 cells of inhibiting fibroblast Wnt secretion by deleting Wntless with (B) Tbx4LME-Cre or (C) Pdgfrα-CreERT2 induced with 3 mg tamoxifen 3 days before analysis. *P = 0.02; ***P = 0.003 (Student’s t test). (D) Expression of Wnts, fibroblast markers (Pdgfrα and Col1a2), Axin2, and ubiquitous controls (Ubc, Ppla, and Actb) in 47 alveolar fibroblasts (columns) from B6 adult lungs analyzed by means of scRNA-seq. (E) Wnt5a, Axin2, and SftpC mRNA detected with PLISH of adult (2 months) alveoli. Blue, DAPI. Shown is a rare Wnt5a-expressing cell (dotted box “g”). (F and G) Close-ups showing AT2 cells (F) far from or (G) near a Wnt5a-expressing cell. (Insets) Axin2 channel of AT2 cell. AT2 near Wnt5a source expresses Axin2. (H) Quantification showing percent (mean ± SD) of AT2 cells, located far (>15 μm, n = 132 cells from three biological replicates) or near (<15 μm, n = 150 cells) a Wnt5a source, that express Axin2. ***P < 0.0001 (Student’s t test). (I) Axin2 expression in AT2 cells isolated from adult Axin2-lacZ mice and cultured (5 days) with indicated Wnts at 0.1 μg/mL or antagonist Dickkopf 3 at 1 μg/mL, then assayed (Spider-gal) for LacZ. ***P < 0.0001 (Student’s t test). Scale bars, 10 μm (E) and 5 μm (G).

Single-cell RNA sequencing (scRNA-seq) of alveolar fibroblasts revealed a subset expressing Wnt5a, most of which (74%) also expressed Pdgfrα (Fig. 2D). Many Wnt5a+ fibroblasts also expressed low levels of one or two other Wnts—including Wnt2, Wnt2b, Wnt4, and Wnt9a—as did other smaller subpopulations of fibroblasts (Fig. 2D). AT2 cells did not express Porcupine (fig. S4) or any Wnt genes (fig. S5) under normal conditions. Wnt5a-expressing fibroblasts (fig. S6) were scattered throughout the alveolar region, most near an Axin2+ AT2 cell (Fig. 2, E to H). Although Wnt5a is sufficient to induce Axin2 in AT2 cells (Fig. 2H), it is not the only Wnt operative in vivo because others can also induce Axin2 (Fig. 2I), and deletion of Wnt5a with Tbx4LME-Cre reduced Axin2+ AT2 cells in vivo by 15%, and the effect did not reach significance (P = 0.12). We conclude that Wnt5a and other Wnts expressed by the fibroblasts activate the canonical Wnt pathway in neighboring AT2 cells. This signal is short range because AT1 cells derived from Axin2+ AT2 cells do not express Axin2 (n > 1000 cells scored in three lungs at age 4 months), implying that they do not maintain Axin2 expression once they move away from the Wnt source. Some Wnt-expressing fibroblasts themselves expressed Axin2 (Fig. 2D and fig. S6), suggesting that they can also provide an autocrine signal.

Wnt signaling prevents reprogramming of alveolar stem cells into AT1 cells

To investigate the function of Wnt signaling, we deleted β-catenin, a transducer of canonical Wnt pathway activity, in mature AT2 cells by using Lyz2-Cre or SftpC-CreERT2 while simultaneously marking recombined cells by using Rosa26mTmG. We reasoned that only cells with active Wnt signaling (Axin2+ AT2 cells) would be affected. The number of AT1 cells expressing the AT2 lineage mark tripled, while preserving the percentage of lineage-labeled AT2 cells (85 ± 3% of AT2 cells versus 82 ± 3% in wild-type β-catenin controls, n = 500 AT2 scored in three biological replicates) and alveolar structure (Fig. 3, A, B, and D; and fig. S7, A, B, and D). Of the AT2-lineage-marked AT1 cells in this experiment (48 of 172 scored cells in three animals), 27% were not physically associated with a marked founder AT2 cell (Fig. 3, E and F), implying that the stem cell had directly converted into an AT1 cell, which was rare in control lungs (4%; n = 145 scored cells in three biological replicates). Thus, abrogation of constitutive Wnt signaling promotes transdifferentiation of Axin2+ AT2 cells into AT1 cells.

Fig. 3 Wnt signaling prevents reprogramming to AT1 fate.

(A to C) Alveoli of adult (8 months) (A) Lyz2-Cre;Rosa26mTmG, (B) Lyz2-Cre;Rosa26mTmG β-cateninfl/fl and (C) Lyz2-Cre;Rosa26mTmG;β-cateninEx3/+ mice immunostained for SftpC and Lyz2-Cre (AT2) lineage trace (mGFP). Dashed circles indicate alveolar renewal foci identified through squamous AT1 expressing AT2 lineage trace. Scale bar, 50 μm. (D) Quantification showing percent (mean ± SD) alveoli with AT2 lineage-labeled AT1 cells (n = 25 100-μm z-stacks scored in two or three biological replicates). ***P = 0.002 (Kruskal-Wallis test). (E) Close up of renewal foci as above in control (Lyz2-Cre;β-catenin+/+, top) or β-catenin conditional deletion (Lyz2-Cre;β-cateninfl/fl, bottom). Shown are AT1 (arrows) and its AT2 parent (arrowheads) in control, but absence of AT2 parent (asterisk) in β-catenin deletion, implying loss of stem cell by reprogramming to AT1 fate. (F) Quantification shows percent (mean ± SD) AT1 cells from AT2 lineage (GFP+) lacking AT2 parent. ***P = 0.0004 (Student’s t test). (G to I) AT2 cells from B6 adult mouse lungs cultured in Matrigel [AT2-maintaining conditions, (G) and (H)] or on poly-lysine–coated glass [AT1-promoting conditions (I)] without (control) or with indicated Wnt pathway antagonist (1 μg/ml Dkk3) or agonists (100 ng/ml Wnt5a or 10nM CHIR99201). After 4 days, cells were immunostained for SftpC and Podoplanin (Pdpn) (G), and percent (mean ± SD) of AT2 (cuboidal SftpC+; arrowheads) and AT1 cells (large, squamous, Podoplanin+; arrow) were quantified (n = 500 cells from three biological replicates) [(H) and (I)] **P = 0.002; ***P < 0.001 (Student’s t test). (J) AT2 cells isolated from adult (2 months) Sftpc-CreER;Rosa26mTmG mice were cultured in Matrigel as above without (control) or with indicated Wnts (100 ng/ml) and EGF (50 ng/ml), then proliferation was assayed by means of EdU incorporation. (K) Quantification (n = 400 cells scored, four biological replicates). *P = 0.007; ***P < 0.001 (Student’s t test). n.s., not significant. Scale bars, 50 μm (C) and (G), 10 μm (E), and 5 μm (J).

We also prevented AT2 cells from down-regulating Wnt signaling by expressing a stabilized β-catenin (β-cateninEx3). This did not induce proliferation (fig. S8) or other obvious effects on AT2 cells but reduced lineage-marked AT1 cells 63% (Fig. 3, C and D, and fig. S7, C and D).

Under culture conditions that maintain AT2 identity (28), Wnt antagonist Dickkopf 3 (29) increased the percentage of cells that reprogrammed to AT1 fate 3.8-fold (2.5 ± 1.5% versus 9.5 ± 0.1%) (Fig. 3, G and H). Conversely, under conditions that promote differentiation to AT1 fate (28), Wnt5a inhibited this transdifferentiation 2.6-fold (21 ± 1% versus 8 ± 1%) (Fig. 3I). CHIR99021, a pharmacological activator of canonical Wnt signaling, had a similar effect (Fig. 3I).

Thus, canonical Wnt signaling maintains the AT2 stem cell pool by preventing their reprogramming to AT1 identity, both in vivo and in vitro. Although Wnt signaling alone had little effect on AT2 proliferation (Fig. 3, J and K, and fig. S8), it enhanced EGF’s mitogenic activity (Fig. 3, J and K).

Wnt signaling is induced in “ancillary” AT2 stem cells after epithelial injury

To investigate stem cell activity after injury, we established a genetic system (30) to ablate alveolar epithelial cells. Diphtheria toxin receptor was expressed throughout the lung epithelium of adult mice by using Shh-Cre. Diphtheria toxin (DT) (150 ng) triggered apoptosis in ~40% of alveolar epithelial cells (fig. S9, A and B) but spared enough for repair (fig. S9C) and survival. Nearly all remaining AT2 cells (85%) began proliferating after injury (Fig. 4, A and C), indicating that bulk AT2 cells are recruited as ancillary progenitors during repair; similar recruitment of bulk AT2 cells was observed after hyperoxic injury (see below) (31, 32). Most AT2 cells (73%) expressed Axin2 after DT-triggered injury, indicating that canonical Wnt signaling is broadly induced in ancillary stem cells (Fig. 4, B and C, and fig. S10C). Inhibition of Wnt signaling with C59 abrogated AT2 proliferation and blocked repair (Fig. 4C and fig. S9D). Thus, Wnt signaling recruits ancillary AT2 cells with progenitor capacity after severe injury.

Fig. 4 Genetically targeted epithelial injury induces Wnt signaling and proliferation of bulk AT2 cells.

(A and B) Alveoli of Shh-Cre;Rosa26LSL-DTR (Diphtheria toxin receptor) animals injected with vehicle (control, top) or limiting dose (150 ng) of Diphtheria toxin (DT) to induce sporadic epithelial cell ablation (bottom) then (A) immunostained 5 days later for SftpC and Ki67 or (B) probed with PLISH for SftpC and Axin2 expression. (C) Quantification (n = 250 cells in four animals) of percent AT2 cells expressing Ki67 (mean ± SD) or Axin2 (mean ± SD). Ablation induces proliferation and Wnt signaling in most AT2 cells, both abrogated by Wnt secretion inhibitor C59. ***P < 0.001 (Student’s t test). (D) Alveoli as (A) and (B) probed with PLISH for Wnt7b and SftpC mRNA 1 day after vehicle (control) or DT injection (ablation). (E) Kinetics of Wnt7b induction after ablation (n = 300 AT2 cells scored per animal, four biological replicates per time point, mean ± SD). **P = 0.005 (Kruskal-Wallis). Scale bars, 10 μm (A), 5 μm (B) and (D).

Injury induces autocrine signaling in AT2 cells

There was no change in Wnt5a expression (fig. S10C) or stromal expression of Porcupine (fig. S11, A and B) after DT-triggered injury. By contrast, Porcupine was broadly induced in AT2 cells (figs. S10C and S11, A to C), suggesting that injury activates autocrine Wnt secretion. We found Wnt7b, expressed in alveolar progenitors during development (16) but not healthy adult AT2 cells (fig. S5), was broadly induced in AT2 cells after DT-triggered injury (Fig. 4, D and E). AT2 expression of Wnt7b and Porcupine and activation of canonical Wnt signaling were induced within 24 hours of injury (Fig. 4, D and E, and fig. S10C). AT2 proliferation initiated over the next 2 days, peaking at day 5 as epithelial integrity was restored, after which AT2 proliferation and gene expression returned toward baseline and new AT1 cells appeared (fig. S10, B and C).

To explore the generality of injury-induced autocrine Wnt signaling, we used hyperoxic injury (75% oxygen) to induce alveolar repair (fig. S12A) (31). This allowed us to mark and genetically manipulate alveolar cells through endotracheal delivery of an adeno-associated virus encoding Cre (AAV9-Cre) into lungs of mice carrying a Cre-dependent reporter and conditional Wnt pathway alleles. Quantitative polymerase chain reaction (PCR) analysis of FACS-purified, lineage-labeled AT2 cells (fig. S12, B and C) showed hyperoxic injury induced AT2 expression of Wnt7b and six other Wnt genes by 3- to 12-fold and similarly induced Axin2 (fivefold) and Lef1 (sevenfold), indicating autocrine activation of the canonical Wnt pathway (Fig. 5A). The suite of induced Wnts did not include most Wnts expressed by the fibroblast niche, including Wnt5a (Fig. 5A). AAV9-Cre–mediated mosaic deletion of Wntless in ~50% of alveolar epithelial cells (fig. S12, D to F) decreased AT2 proliferation after injury (Fig. 5, B and C). The effect was cell-autonomous because AT2 cells expressing Cre-GFP, but not neighboring AT2 cells, showed diminished proliferation (Fig. 5, B and C). This autocrine effect is mediated by multiple Wnts because deletion of just one induced Wnt (Wnt7b) did not diminish proliferation. Thus, epithelial injury induces AT2 expression of a suite of autocrine Wnts, which transiently endow bulk AT2 cells with progenitor function and proliferative capacity.

Fig. 5 Hyperoxic injury induces autocrine Wnt signaling in bulk AT2 cells.

(A) Expression of Wnt genes and targets (Axin2 and Lef1) measured with quantitative reverse transcription PCR of FACS-sorted, lineage-labeled bulk AT2 cells from mice 2 days after hyperoxic alveolar injury (5 days at 75% O2) (fig. S12B), normalized to values before hyperoxic injury (mean ± SD, n = 3 mice per condition). Red, Wnts expressed by fibroblast niche (Fig. 2D). *P < 0.05, ***P < 0.001 (Student’s t test). (B) Alveoli immunostained as indicated from adult (age 2 months) Wntlessfl/fl mouse with alveolar epithelium infected with AAV9-GFP-Cre virus (fig. S12, D to F) to mosaically delete Wntless 1 week before hyperoxic injury. Shown is injury-induced proliferation (Ki67 staining) of AT2 cells (Muc1 apical marker) (fig. S12A), except AT2 infected with AAV-Cre-GFP to delete Wntless. (C) Quantification showing percent AT2 cells Ki67+ after hyperoxic injury of AAV9-Cre-GFP–infected control (Wntless+/+) and Wntlessfl/fl mice, scored for all or just GFP+ (AAV9-Cre-GFP–infected) AT2 cells. Mean ± SD (n = 500 total AT2 cells scored per mouse, three mice per condition, including 52 GFP+ cells). Scale bar, 10 μm.


We molecularly identified a rare subset of AT2 cells with stem cell function (AT2stem) scattered throughout the mouse lung in specialized niches that renew the alveolar epithelium throughout adult life. AT2stem cells express Wnt target Axin2, and many lie near single fibroblasts expressing Wnt5a and other Wnt genes that serve as a signaling niche (Fig. 6). AT2stem cells divide intermittently, self-renewing and giving rise to daughter AT1 cells that lose Wnt activity when they exit the niche. Maintaining canonical Wnt signaling blocked transdifferentiation to AT1 identity, whereas loss of Wnt signaling promoted it.

Fig. 6 Model of alveolar stem cells and their niches.

(Top) (Left) During homeostasis, the niche is a single fibroblast constitutively expressing Wnt5a and/or other Wnts that provide “juxtacrine” signal (arrow) to the neighboring AT2 cell (green cytoplasm, lineage trace; black nucleus, Wnt pathway active), selecting/maintaining it as a stem cell. (Middle) Upon receiving a mitogenic signal from the dying AT1 cell, the activated stem cell proliferates. Daughter cells (green) compete for the niche. (Right) One remains in the niche as a stem cell; the other leaves the niche, losing Wnt signal and transdifferentiating into a new AT1 cell. (Bottom) After severe injury, bulk AT2 are recruited as “ancillary” stem cells through induction of autocrine Wnts, allowing unlimited proliferation in response to mitogens. Autocrine Wnts diminish as injury resolves.

Our results support a model in which each Wnt-expressing fibroblast defines a niche accommodating one AT2stem cell, and this short-range (“juxtacrine”) signal selects and maintains AT2stem identity (Fig. 6). When an AT2stem cell divides, daughter cells compete for the niche. The one that remains in the niche retains AT2stem identity; the other leaves the Wnt niche, escaping the signal and reprogramming to AT1 fate. Cells leaving the niche can also become standard AT2 cells, presumably if they land near a signaling center that selects bulk AT2 fate. This streamlined niche, comprising just a single Wnt-expressing fibroblast and stem cell, minimizes niche impact on alveolar gas exchange. It also explains why each expanding focus of new alveoli is clonal, derived from a single AT2stem that typically remains associated with the growing focus (13). Although our model posits that the niche cell selects the stem cell, it remains uncertain how the scattered niche cells are selected. Some niche cells themselves are Axin2+ (Fig. 2D and fig. S6) (24), suggesting that autocrine Wnt signaling might maintain the niche as juxtacrine signaling maintains the stem cell within it.

Our results also reveal a transient expansion of the alveolar progenitor population after epithelial injury, when the rare AT2stem are insufficient. Many normally quiescent bulk AT2 cells turn on Axin2 and serve as ancillary progenitors that rapidly proliferate and regenerate lost alveolar cells (Fig. 6). This widespread recruitment of AT2 cells to AT2stem function is not achieved through expansion of the fibroblast Wnt niche. Rather, injury induces Porcupine and another suite of Wnts, including Wnt7b, in AT2 cells. This switch to autocrine control of stem cell identity obviates dependence on the stromal niche. As the epithelium is restored, Wnt expression subsides in ancillary AT2stem, and they cease proliferating and begin differentiating into AT1 cells or returning to bulk AT2 identity. The Wnt pathway is broadly active during alveolar development (16, 17) and cancer, where most cells proliferate, so it may have a general role in maintaining alveolar stem or progenitor states.

Our data suggest that Wnt signaling, whether juxtacrine Wnts from a fibroblast or autocrine Wnts induced by injury, endows AT2 cells with two stem cell properties. One is AT2stem gene expression and identity, preventing reprogramming to AT1 (and presumably bulk AT2) fate (Fig. 3). The other is an ability to proliferate extensively, as observed after injury when ancillary AT2stem divide rapidly to restore the epithelium (Fig. 6). Thus, Wnt signaling confers stem cell identity (“stemness”) on AT2 cells but does not itself activate the stem cells (Fig. 3, J and K, and fig. S8). Proliferation is controlled by EGFR/KRAS signaling (Fig. 3, J and K) (13), presumably activated by EGF ligand(s) from dying cells (33). We propose that Wnt and EGFR/KRAS pathways function in parallel to select and activate alveolar stem cells, respectively, explaining their synergy (Fig. 3, J and K).

The above findings have implications for lung adenocarcinoma, the leading cancer killer (34) initiated by oncogenic mutations that activate EGFR/KRAS signaling in AT2 cells (13, 35, 36). Although most AT2 cells show a limited proliferative response, a rare subset proliferates indefinitely, forming deadly tumors (13): The tumor-initiating cells could be AT2stem. Indeed, a subpopulation of adenocarcinoma cells was recently found to have Wnt pathway activity and function as tumor stem cells, with associated cells serving as their Wnt signaling niche (26). As oncogenic EGFR/KRAS drives stem cell proliferation, Wnt signaling would maintain their identity; this explains why Wnt signaling has little proliferative effect on its own but potentiates KRASG12D and BRAFV600E mouse models of lung carcinogenesis and why Wnt inhibition induces tumor senescence (26, 37, 38). Wnt antagonists might thus be powerful adjuvants in adenocarcinoma therapy, attacking stem cell identity, while EGFR antagonists target stem cell activity. One reason stem cell identity may have restricted during evolution to rare AT2 cells is that it minimizes cells susceptible to transformation.

There is growing appreciation that some mature cells in other tissues can also provide stem cell function (8, 9, 39). Like AT2stem, their clinical utility has been overlooked as more classical “undifferentiated,” and pluripotent stem cells have been sought. Our study shows that stem cells and their niche cells can each represent minor, solitary subsets of mature cell types. By molecularly identifying such rare subpopulations and niche signals, it should be possible to isolate and expand them for regenerative medicine.

Supplementary Materials

References and Notes

Acknowledgments: We thank A. Andalon for technical assistance; D. Riordan and M. Nagendran for advice on PLISH; B. Treutlein and S. Quake for help with scRNA-seq; R. Nusse and colleagues for generously sharing mouse lines and reagents; members of the Krasnow, Desai, and Nusse laboratories for discussion; and M. Peterson for help preparing the manuscript and figures. Funding: This work was supported by a National Heart, Lung, and Blood Institute (NHLBI) U01HL099995 Progenitor Cell Biology Consortium grant (M.A.K., T.J.D., and P.B.H.), NHLBI grant 1R56HL1274701 (T.J.D.), and Stanford BIO-X grant IIP-130 (T.J.D. and P.B.H.). A.N.N. was supported by NIH Comparative Medicine Branch training grant fellowship 2T32GM007276. M.A.K. is an investigator of the Howard Hughes Medical Institute. Data and materials availability: Expression-profiling data sets were deposited in Gene Expression Omnibus ( GEO accession numbers are GSE109444 for the adult mesenchyme and GSE52583 for the adult AT2 cells. Competing interests: T.J.D. and P.B.H. are co-inventors of a patent application (#62475090) submitted by Stanford University that covers the technology used for the multiplexed in situ hybridization experiments. Author contributions: A.N.N., T.D., and M.A.K. conceived, designed, and analyzed experiments and wrote the manuscript. All experiments except fibroblast and AT2 scRNA-seq were performed by A.N.N. D.G.B. performed and analyzed RNA-seq of alveolar fibroblasts and AT2 cells. P.B.H. conceived and advised on PLISH (19).

Stay Connected to Science

Navigate This Article