Control of Mitotic Spindle Angle by the RAS-Regulated ERK1/2 Pathway Determines Lung Tube Shape

Many organs, including lung, develop from epithelial tubes that branch to form complex networks. Each branch grows to attain the final size and shape characteristic of its position in the network (1). The genetic programs and cellular mechanisms that determine branch architecture in mammalian organs are poorly understood. Here we characterized how airway branch shape, defined as the ratio (ρ) of average tube circumference to length (Fig. 1A), changes during early mouse lung development (2). In the left primary branch, ρ decreased significantly over ~24 hours (Fig. 1, B to D). This change in airway shape (ρ) was not accompanied by alterations in cell size or shape (fig. S1), and the airway epithelium remained a monolayer. These data demonstrate that allometric growth (3), in this case a greater increase in length than circumference, is a characteristic of early airway development.

• Download high-res image
• Open in new tab
• Download Powerpoint

Fibroblast growth factor (FGF)–mediated receptor tyrosine kinase (RTK) signaling is required for lung development (4). To determine whether such signaling plays a role in regulating airway shape, we used Cre-mediated recombination to express a mutationally activated form of RAS, a critical effector of RTK signaling, throughout the airway epithelium. We employed Shh^cre (5) to recombine a Kras conditional allele, Kras^LSL-G12D (6) (hereafter referred to as Kras^LSL) (fig. S2A), to obtain expression of KRAS^G12D, which is predicted to activate signaling by way of RAF→MEK→ERK and other pathways (MEK: mitogen-activated or extracellular signal–regulated protein kinase kinase) (7). In control lungs at the 47 somite stage [som; embryonic day 11.75 (E11.75)], phosphorylated ERK1/2 (pERK1/2) is highly localized (8) (fig. S2B), whereas in Shh^cre/+;Kras^LSL/+ lungs, pERK1/2 was detected throughout the epithelium (fig. S2C). In the mutant lungs, the normal change in the ratio of airway circumference to length (ρ) did not occur (Fig. 1, E to G), i.e., airway growth was isometric rather than allometric, suggesting that increasing ERK1/2 signaling in lung epithelium prevents the normal airway shape change.

To confirm this hypothesis, we generated Shh^cre/+;Braf^CA/+ embryos, in which a mutationally activated RAF family member, BRAF^V600E (9), is expressed in developing lung epithelium (fig. S3A). BRAF^V600E is predicted to increase ERK1/2 but not PI 3-kinase→PDK→AKT signaling (PI 3-kinase: phosphatidylinositol 3-kinase; PDK: phosphoinositide-dependent protein kinase) (10). Shh^cre/+;Braf^CA/+ lungs displayed the same abnormal shape as Shh^cre/+;Kras^LSL/+ lungs (compare fig. S3, B and C, to Fig. 1, D and G). Next, we treated Shh^cre/+;Kras^LSL/+ embryos in utero with PD0325901, a highly specific inhibitor of MEK1/2 (11, 12). Whereas vehicle-treated Shh^cre/+;Kras^LSL/+ embryos displayed the same abnormal airway shape as untreated mutant embryos, lungs from PD0325901-treated Shh^cre/+;Kras^LSL/+ embryos appeared normal (fig. S4). Together, these data provide compelling evidence that the lack of normal change in airway shape (ρ) in Shh^cre/+;Kras^LSL/+ lungs is mediated by RAS-regulated RAF→ MEK→ERK signaling.

We next investigated what cell-biological processes might be responsible for the mutant airway phenotype. We detected no increase in epithelial cell proliferation in Shh^cre/+;Kras^LSL/+ lungs (fig. S8A), consistent with previous reports (13), and observed little or no epithelial cell death in Shh^cre/+;Kras^LSL/+ or control lungs. Moreover, epithelial cell size and shape were similar in mutant and control lungs at 38 or 47 som (fig. S1). Furthermore, no effects were detected on apical-basal polarity or on primary cilia or airway smooth muscle formation. We then assayed for effects on mitotic spindle orientation because it can influence tube shape (14–16) (Fig. 2A).

• Download high-res image
• Open in new tab
• Download Powerpoint

Mitotic spindle orientation in dividing airway epithelial cells can be deconstructed into two component angles, determined with respect to tube longitudinal axis (θ; Fig. 2B) and the apical-basal axis of the epithelium (ϕ; fig. S5). In control left primary airways at 38 som, spindle angle θ distribution was highly nonuniform and skewed toward low angles: In 42% of mitotic cells, the spindle was essentially parallel to the airway longitudinal axis (θ = 0° to 10°), whereas in 58% spindle orientation distribution over the remaining angular range (θ = 11° to 90°) was not different from uniform (P = 0.27; Kolmogorov-Smirnov test). However, in Shh^cre/+;Kras^LSL/+ lungs the bias toward low angles was lost, and spindle orientation distribution across the entire angular range (0° to 90°) was not different from uniform (P = 0.32, Kolmogorov-Smirnov test; Fig. 2C). In contrast, spindle angle ϕ distribution was not affected by KRAS^G12D expression (P = 0.11; fig. S5). Moreover, these component angles can be independently controlled: In individual mitotic cells in control lungs, there was no correlation between the values of θ and ϕ (r = –0.13; P = 0.1, two-tailed t test).

Together, our data underscore the importance of regulating ERK1/2 signaling during airway development and suggest that genes encoding endogenous regulators of ERK1/2 signaling are required for normal airway shape change and mitotic spindle angle distribution. Because members of the sprouty family negatively regulate ERK1/2 signaling (17–20), and Spry1 and Spry2 RNAs are detected in lung epithelium from at least 38 som (fig. S6, A and B), we tested the effect of sprouty loss-of-function (LOF). In Spry1 and Spry2 double-knockout (Spry1/2 DKO) embryos, but not in their double-heterozygous (Spry1/2 Dhet) littermates, we observed both an expansion of the ERK1/2 activation domain (compare fig. S6, C and D, with fig. S2, B and C) and a lack of the normal change in airway shape (ρ) between 38 and 47 som (Fig. 3, A to F), similar to the effect observed when ERK1/2 activity was increased following KRAS^G12D expression (Fig. 1, B to G). Moreover, spindle angle distribution was altered in Spry1/2 DKO lungs in the same manner as in Shh^cre/+;Kras^LSL/+ lungs, with a loss of bias toward low angles; the distribution of spindle orientations across the entire angular range (0° to 90°) was not different from uniform (P = 0.28 for Spry1/2 DKO lungs; Kolmogorov-Smirnov test; Fig. 3J). Furthermore, airway shape change and mitotic spindle angle distribution were normal when Spry1/2 DKO embryos were treated with MEK1/2 inhibitor PD0325901 (fig. S7).

• Download high-res image
• Open in new tab
• Download Powerpoint

However, there were differences between the Shh^cre/+;Kras^LSL/+ and Spry1/2 DKO lungs. In the latter, the percentage of mitotic epithelial cells increased ~twofold (fig. S8A), secondary branching was impaired (Fig. 3, C and F), and infoldings of the epithelium and underlying basement membrane into the airway lumen were observed (fig. S8, B to E). Because increased cell proliferation was observed in Spry1/2 DKO but not Shh^cre/+;Kras^LSL/+ airways, it is unlikely to cause the lack of normal shape change seen in both, but presumably underlies the abnormal infoldings and suppressed branching observed in Spry1/2 DKO lungs.

Sprouty genes encode negative-feedback antagonists of signaling pathways downstream of FGF receptor (FGFR) and other RTKs (20). We hypothesized that Spry1 and Spry2 LOF removes a brake on signaling via Fgf10 and Fgfr2, genes that are required for lung development (21–25). If so, then reducing FGF10-mediated FGFR2 signaling by reducing Fgf10 or Fgfr2 gene dosage in Spry1/2 DKO embryos might rescue the mutant phenotype. This approach has been used to identify ligand and receptor genes involved in other sprouty LOF phenotypes caused by excess RTK signaling (26–29). When Fgf10 dosage was reduced in Spry1/2 DKO embryos, airway shape, branching, the domain of ERK1/2 phosphorylation, and mitotic spindle angle distribution appeared normal (Fig. 3, G to J, and fig. S6E). Reducing the dosage of Fgfr2 or Fgf9, another FGF gene required for normal lung development (30), did not rescue the Spry1/2 DKO phenotype (fig. S9). These data demonstrate that Spry1 and Spry2 LOF results in an increase in FGF10-mediated ERK1/2 signaling that causes abnormal spindle angle distribution and a lack of normal change in airway shape.

To address whether the distribution of mitotic spindle angles in the epithelium at 38 som (initial time, t_i) can account for the changes in airway shape (ρ) observed between 38 and 47 som (final time, t_f), we devised a mathematical model to predict epithelial tube shape from spindle angle distribution [see supporting online material (SOM) text I to IV] and tested whether predicted and experimentally measured values were in close agreement. Spindle angle distribution can be described as comprising two underlying distributions and characterized by a single parameter, the fraction (f) of mitotic cells in which spindle angle is uniformly distributed across the angular range (Fig. 4A; SOM text V). f can be estimated with a maximum likelihood approach (SOM text VI). Because the fraction of mitotic cells in which θ = 0° to 10° is lower in the mutant lungs, we assume that ERK1/2 signaling influences f such that its value increases as the proportion of cells with high-level ERK1/2 signaling increases. We tested this assumption by measuring pMEK1/2 level and determining spindle angle θ in individual mitotic cells. pMEK1/2 level was significantly higher in cells in which mitotic spindle angle θ = 11° to 90° than in cells in which θ = 0° to 10° (P = 0.003; one-tailed t test; Fig. 4B), supporting the notion that a high level of ERK1/2 signaling increases the probability that a cell will not divide parallel or near parallel to the airway longitudinal axis.

• Download high-res image
• Open in new tab
• Download Powerpoint

Our model predicts that there is a scaling relationship between changes over time in epithelial tube shape (ρ, average tube circumference/length) and tube surface area (A, average tube circumference × length) that is determined solely by the value of f. Specifically, a plot of the fold-change in ρ versus fold-change in A on a log-log scale gives a straight line, the slope of which is a function of f (SOM text V). From such a plot, log fold-change ρ (i.e., the log of the ratio of ρ at t_f to ρ at t_i) can be predicted for any given value of log fold-change A (Fig. 4C). On the basis of the spindle angle distribution for each genotype, we calculated f and used this value to predict fold-change ρ given the measured fold-change A between t_i and t_f. As expected, the calculated value of f was greater in mutant lungs with increased ERK1/2 signaling than in controls, and the predicted values of fold-change ρ for both mutant and control embryos closely matched the measured values (Fig. 4D and fig. S10). Such agreement suggests that there is no need to invoke effects of ERK1/2 on cellular processes other than spindle orientation to explain the observed shape changes in the primary airway.

To investigate whether a common mechanism controls the change in airway shape in primary and later branch generations during normal lung growth, we examined L.L2, a left lateral secondary branch (31) (fig. S11, A to C). As in the control primary branches, spindle orientation in wild-type L.L2 was strongly biased toward low angles (fig. S11D). Notably, as for the primary branch, the fold-change ρ for L.L2 between t_i (55 som) and t_f (60 som) predicted on the basis of the observed spindle angle distribution closely matched the measured value (fig. S11E).

Our analysis suggests that cell division in airway epithelium may by default be oriented longitudinally and ERK1/2 activity functions as a switch to override the default orientation, with the level of ERK1/2 activity determining the probability that override will occur. Presumably, ERK1/2 activity does not itself provide any orientation cue but instead determines whether or not a cell responds to such cues. The planar cell polarity (PCP) pathway, which plays an important role in regulating mitotic spindle angle distribution in other tissues (14, 32–34), may provide these cues. However, as yet, there is no evidence that PCP pathway genes function to regulate airway shape or spindle orientation in the lung. Indeed, these parameters were normal in lungs from Vangl2^Lp homozygotes (fig. S12).

The mechanism by which ERK1/2 signaling influences spindle orientation remains to be determined. Possibilities include effects on the centrosomal components of the mitotic spindle, on the cytoskeletal elements that control centrosome position and spindle orientation, or on some aspect of cell behavior. Consistent with observations that cell movement in presomitic mesoderm and limb bud is affected by ERK1/2 signaling (35, 36), it is conceivable that high levels of signaling cause mitotic cells to rotate within the epithelium such that spindle angle with respect to the airway longitudinal axis is altered.

The hypothesis that RAS-regulated ERK1/2 signaling influences airway shape change by regulating mitotic spindle angle distribution not only explains the adverse effects on airway shape of increasing ERK1/2 activity, but also posits a role for ERK1/2 signaling in controlling normal airway shape change during development. It is tempting to speculate that such a mechanism also regulates epithelial tube shape changes in other branched organs. Moreover, our data identify sprouty genes as key regulators that function to ensure that the shape changes required for normal development occur.