Modulation of integrin activity is vital for morphogenesis

Abstract

Cells can vary their adhesive properties by modulating the affinity of integrin receptors. The activation and inactivation of integrins by inside-out mechanisms acting on the cytoplasmic domains of the integrin subunits has been demonstrated in platelets, lymphocytes, and keratinocytes. We show that in the embryo, normal morphogenesis requires the alpha subunit cytoplasmic domain to control integrin adhesion at the right times and places. PS2 integrin (alphaPS2betaPS) adhesion is normally restricted to the muscle termini, where it is required for attaching the muscles to the ends of other muscles and to specialized epidermal cells. Replacing the wild-type alphaPS2 with mutant forms containing cytoplasmic domain deletions results in the rescue of the majority of defects associated with the absence of the alphaPS2 subunit, however, the mutant PS2 integrins are excessively active. Muscles containing these mutant integrins make extra muscle attachments at aberrant positions on the muscle surface, disrupting the muscle pattern and causing embryonic lethality. A gain- of-function phenotype is not observed in the visceral mesoderm, showing that regulation of integrin activity is tissue-specific. These results suggest that the alphaPS2 subunit cytoplasmic domain is required for inside-out regulation of integrin affinity, as has been seen with the integrin alphaIIbbeta3.

Figure 1

Amino acid sequence of the two mutated α_PS2 cytoplasmic domains. The cytoplasmic tail and part of the transmembrane region of α_PS2 subunit are shown at the top. Beneath this, the two mutants generated are shown: αPS2Δcyt and αPS2ΔGFFNR.

Figure 2

The α_PS2 cytoplasmic domain is not required for localization of the PS2 integrin to the end of the muscles. An antibody against the β_PS subunit was used to visualize expression of the PS2 integrin. (a) In a wild-type embryo, the PS2 integrin is localized at the muscle attachment sites (m.a.). (b) In mutant embryos that lack the α_PS2 subunit, the β_PS staining at the end of the muscles is lost. (c) Expression of the UAS-α_PS2Δcyt construct in an α_PS2 mutant embryo produces a mutant heterodimer that is correctly localized at the muscle termini.

Figure 3

The ability of different combinations of the UAS-α_PS2 cytoplasmic mutants (driven by twist-GAL4 and 24B) to rescue the α_PS2 embryonic lethality is indicated by the black bars. The percent rescue of α_PS2 embryonic lethality was calculated as following: (1/4 total number of embryos − unhatched embryos)/ 1/4 total number of embryos × 100). The total number of embryos counted is shown on the left side of each bar. The embryonic lethality of α_PS2 mutants can be rescued with either the UAS-α_PS2 (completely) or the UAS-α_PS2ΔGFFNR (nearly completely) constructs. In contrast, the UAS-α_PS2Δcyt construct is unable to rescue the lethality. A subset of embryos containing 50% mutant and 50% wild-type embryos was selected using a marked balancer chromosome (the number is indicated to the left side of each grey bar). This subset was examined by staining the muscles for muscle myosin (see Fig. 4), and was scored for a loss of function (muscle detachment, light grey bars) or gain of function (formation of ectopic attachment sites, dark grey bars) phenotypes. The percent of embryos with muscle phenotype was calculated as follows: (number of embryos with muscle phenotype)/ 1/2 number of selected embryos × 100). Deletion of the cytoplasmic tail, and to a lesser extent deletion of the GFFNR motif, causes gain-of-function phenotypes.

Figure 4

The cytoplasmic domain of the α_PS2 subunit is required for normal pattern of muscle attachments. Embryos in a–c are stained with an antibody for muscle myosin to reveal the pattern of the muscles. In embryos that lack the α_PS2 subunit, the muscles detach (arrow; b) compared with wild-type (a). Expression of α_PS2Δcyt in the muscles of mutant embryos rescues the muscle detachment phenotype (c). However, the transverse muscles look broader at their tips (see inset in c and compare with inset in a), and in 37% of the mutant embryos (Fig. 3) the ventral lateral muscle (VL) form ectopic muscle attachments with the ventral acute muscles (VA; shown in c, and in higher magnification in d and e). These new attachment sites contain the mutant PS2 integrin (d; revealed with an antibody against β_PS) and the extracellular PS2 ligand, Tiggrin (e). d and e show one segment with the segment borders marked by arrows.

Figure 5

The cytoplasmic domain of the α_PS2 subunit is required to prevent ectopic processes from the lateral surfaces of the muscles. An antibody against the transmembrane protein gp150 has been used to show the muscle surface. a and b show the outline of the muscles of a wild-type embryo at two different magnifications. The higher magnification panel shows that wild-type muscles at stage 16 have smooth lateral surfaces. In contrast, muscles that express α_PS2Δcyt (c and d) have processes emerging from their lateral surface. d shows how ventral (VL) and longitudinal (LT) muscles in mutant embryos with α_PS2Δcyt send out growth cone– like processes to each other (black arrow).

Figure 7

The α_PS2 cytoplasmic domain is required for PS2 integrin function in the proventriculus, but is not required for morphogenesis of the gastric caeca. At this stage, a wild-type proventriculus is a three-layered valvelike structure at the junction with the midgut. Four long, thin gastric caeca (three of them are marked with dots, and the remaining one is out of the plane of focus) are formed (a). Embryos lacking the α_PS2 subunit show an abnormal proventriculus in which the inner layer has been pulled out, and only two blunt gastric caeca are seen (dots; b). Both phenotypes are fully rescued by a GAL4-driven wild-type α_PS2 construct (not shown), but only the gastric caeca, and not the proventriculus, are rescued by α_PS2ΔGFFNR or α_PS2Δcyt.

Figure 6

The α_PS2 cytoplasmic domain is not essential for PS2 integrin function in the visceral mesoderm. Dissected guts stained for actin with phalloidin conjugated to rhodamine show the visceral mesoderm surrounding the gut. The gut musculature is severely disrupted in animals that lack the α_PS2 subunit (b) in contrast to the visceral mesoderm surrounding the gut in wild-type individuals (a). In α_PS2 mutants that carry either the α_PS2ΔGFFNR construct (c) or the α_PS2Δcyt construct (d), the visceral mesoderm phenotype is almost completely rescued, with only a mild detachment around the proventriculus remaining.

Figure 8

The GFFNR motif is required for wild-type levels of cell surface expression of the PS2 integrin. In this case, expression of PS2 integrins has been detected using an antibody against α_PS2. Using the GAL4 system to express the α_PS2ΔGFFNR construct leads to an excess of α_PS2. One fraction of it is localized to the attachment sites (arrow), and the rest remains inside the cell. b shows how α_PS2 made by a minigene (minigene_wt) is localized at the surface of the muscles at the attachment sites (arrow) at levels indistinguishable from the wild-type α_PS2 (a). In contrast, a deletion of the GFFNR motif within the minigene results in greatly reduced surface expression (c).