Cell – ECM Interactions Play Multiple Essential Roles in Aortic Arch Development

Abstract

Rationale:

Defects in the morphogenesis of the 4^th pharyngeal arch arteries (PAAs) give rise to lethal birth defects. Understanding genes and mechanisms regulating PAA formation will provide important insights into the etiology and treatments for congenital heart disease.

Objective:

Cell-ECM interactions play essential roles in the morphogenesis of PAAs and their derivatives, the aortic arch artery (AAA) and its major branches; however, their specific functions are not well-understood. Previously, we demonstrated that integrin α5β1 and fibronectin (Fn1) expressed in the Isl1 lineages regulate PAA formation. The objective of the current studies was to investigate cellular mechanisms by which integrin α5β1 and Fn1 regulate AAA morphogenesis.

Methods and Results:

Using temporal lineage tracing, whole-mount confocal imaging, and quantitative analysis of the second heart field (SHF) and endothelial cell (EC) dynamics, we show that the majority of PAA EC progenitors arise by E7.5 in the SHF and contribute to pharyngeal arch endothelium between E7.5 and E9.5. Consequently, SHF-derived ECs in the pharyngeal arches form a uniform plexus of small blood vessels, which remodels into the PAAs by 35 somites. The remodeling of the vascular plexus is orchestrated by signals dependent on the pharyngeal ECM microenvironment, extrinsic to the endothelium. Conditional ablation of integrin α5β1 or Fn1 in the Isl1 lineages showed that signaling by the ECM regulates AAA morphogenesis at multiple steps: 1) accumulation of SHF-derived ECs in the pharyngeal arches, 2) remodeling of the uniform EC plexus in the 4^th arches into the PAAs; and 3) differentiation of neural crest-derived cells adjacent to the PAA endothelium into vascular smooth muscle cells.

Conclusions:

PAA formation is a multi-step process entailing dynamic contribution of SHF-derived ECs to pharyngeal arches, the remodeling of endothelial plexus into the PAAs, and the remodeling of the PAAs into the AAA and its major branches. Cell-ECM interactions regulated by integrin α5β1 and Fn1 play essential roles at each of these developmental stages.

Graphical Abstract

AAA morphogenesis is critical for neonatal survival; however, cellular mechanisms underlying AAA development are not well-understood. Using lineage tracing, we demonstrate temporal and quantitative differences in the contribution of the SHF to the PAA endothelium that can explain, at least in part, the differential sensitivity of the 4^th PAAs to perturbations. We show that cell-ECM interactions mediated by Fn1 and integrin α5β1 play pleiotropic and cell-type-specific functions at multiple steps of PAA development. Initially, Fn1 and integrin α5β1 regulate the accrual of SHF-derived endothelial progenitors in the pharyngeal arches. Following the formation of small blood vessels in the 4^th arch, Fn1 and integrin α5β1 regulate blood vessel remodeling into the 4^th PAA in an endothelial non-cell-autonomous manner. In addition, and independent of their roles in PAA formation, Fn1 and integrin α5β1 regulate 4^th PAA stability by mediating the differentiation of neural crest-derived cells into vascular smooth muscle cells. Combinatorial expression of integrin α5β1 and Fn1 in the pharyngeal mesoderm and the neural crest is critical for this latter process. The significance of our work lies in identifying cellular dynamics underlying PAA formation, and intricate temporal and cell-type-specific roles of cell-ECM interactions in AAA morphogenesis at multiple steps of its formation and remodeling.

INTRODUCTION

The aortic arch artery (AAA) and its major branches comprise an asymmetrical vascular tree that routes oxygenated blood from the heart into the systemic circulation¹. Defects in the development of the AAA cause devastating forms of congenital heart disease (CHD) due to interruption(s) in the aortic arch, among which the interrupted aortic arch type B (IAA-B) is more prevalent^{2, 3}. Non-lethal defects in aortic arch morphogenesis such as vascular rings can impact the quality of life by causing constriction of the trachea and esophagus, resulting in difficulties with eating and breathing, or leading to dizziness, vertigo, or tinnitus⁴.

The AAA and its major branches develop from the remodeling of three bilaterally symmetrical pairs of pharyngeal arch arteries (PAAs), numbered 3, 4, and 6⁵. It is important to note that defects in PAA formation or remodeling result in phenotypically identical AAA defects⁶. PAAs arise by vasculogenesis from endothelial precursors originating in the second heart field (SHF)^7–13. Experiments in zebrafish and mice have demonstrated that PAA formation is a multi-stage process that entails endothelial specification in the SHF, accrual of SHF-derived endothelial progenitors in the pharyngeal region, differentiation into ECs, and the assembly of SHF-derived ECs into a plexus of small blood vessels^{9, 13–16}. Thereafter, the pharyngeal endothelial plexus becomes connected with the ventral and dorsal aortae. The endothelium of the ventral aortae also forms by vasculogenesis from SHF-derived vascular progenitors and is contiguous with the cardiac outflow tract and the PAAs^{9, 11}. Following pharyngeal arch segmentation, the plexus endothelium within each arch is remodeled into a PAA⁹. The 3^rd PAA is evident by E9.5, before the 4^th and 6^th PAAs are formed. By the evening of E10.5, all three symmetrical pairs of PAAs are formed. Defects in the left 4^th PAA formation lead to IAA-B, which is lethal unless corrected by surgery soon after birth². Following PAA formation, neural crest-derived cells closest to the PAA endothelium differentiate into vascular smooth muscle cells (VSMCs) and surround the PAA endothelium with a VSMC coat by E12.5^17–21. While not essential for PAA formation, the differentiation of neural crest (NC)-derived cells into VSMCs is essential for the stability of the PAAs, and their eventual remodeling into the asymmetrical AAA and its branches. Defects in NC differentiation in the left 4^th pharyngeal arch lead to arch artery regression, and IAA-B^{19, 20, 22}. In summary, IAA-B can arise due to defects in the formation of the left 4^th PAA or due to its regression.

Morphogenesis of distinct organs and structures proceeds within niches comprised of distinct complements of extracellular matrix (ECM) proteins, and alterations in the ECM microenvironment can severely affect embryogenesis^23–27. We discovered that the pharyngeal arch microenvironment is enriched in the ECM glycoprotein fibronectin (Fn1) both at the mRNA and protein levels²⁸. Fn1 is highly expressed in the pharyngeal endoderm, ectoderm, endothelium, and the second heart field (SHF) mesoderm between E8.5 and E10.5, the period coinciding with PAA formation^{28, 29}. Between E10.5 and E11.5, Fn1 becomes highly upregulated in the NC-derived cells adjacent to the 4^th PAA endothelium, corresponding with the onset of VSMC differentiation. Our previous studies demonstrated that local depletion of Fn1 in the pharyngeal microenvironment using the Isl1^Cre/+ knock-in mice or in NC-derived cells, using various NC-expressing Cre lines, resulted in the IAA-B and RERSA^{29, 30}. However mechanistically, IAA-B in these mutants had distinct etiology. Ablation of Fn1 in the Isl1^Cre/+ knockin strain led to defective formation of the 4^th PAAs,²⁹ whereas the ablation of Fn1 in the NC resulted in the regression of originally well-formed 4^th PAAs³¹.

Integrins constitute a major class of transmembrane receptors transducing ECM signals. Integrins are heterodimers of α and β chains. There are 18 α and 8 β subunits encoded by mammalian genomes, giving rise to 24 different αβ combinations³². Integrin α5 complexes with integrin β1, forming the integrin α5β1 heterodimer³³. Integrin α5β1 binds the ECM glycoprotein fibronectin (Fn1) and regulates Fn1 assembly in vivo³⁴. Phenotypes resulting from either global or cell-type-specific ablations of integrin α5 (MGI gene symbol: Itga5) or Fn1 in mice are similar^{26, 28, 29, 31, 34–41}, supporting the notion that integrin α5β1 is a central Fn1 signal transducer in vivo. Previously, we demonstrated that the expression of integrin α5β1 and Fn1 in the Isl1 lineages was required for the formation of the 4^th PAA and that the deletion of either integrin α5 or Fn1 using the Isl1^Cre/+ knock-in strain resulted in IAA-B²⁹. To understand the mechanisms by which integrin α5β1 and Fn1 regulate AAA development, we analyzed SHF and endothelial cell dynamics in integrin α5^flox/-; Isl1^Cre/+ and Fn1^flox/-; Isl1^Cre/+ mutants during the 4^th PAA formation and remodeling, spanning embryonic days (E) E9.5 - E11.5 of development. Our studies point to the essential and distinct roles of integrin α5β1 and Fn1 at multiple stages of PAA formation and remodeling.

METHODS

A detailed description of materials and methods is provided in the Expanded Materials section in the Supplemental Data and in the Methods and Major Resources Table.

Major Resources Table.

In order to allow validation and replication of experiments, all essential research materials listed in the Methods should be included in the Major Resources Table below. Authors are encouraged to use public repositories for protocols, data, code, and other materials and provide persistent identifiers and/or links to repositories when available. Authors may add or delete rows as needed.

Animals (in vivo studies)
Species	Vendor or Source	Background Strain	Sex	Persistent ID / URL
Mus Musculus	Jackson Labs, cat # 10664	C57BL/6J	M, F
Genetically Modified Animals
Strains	Vendor or Source	Background Strain	Other Information	Persistent ID / URL
B6N.129S6-Gt(ROSA)26Sortm1(CAG-tdTomato*,-EGFP*)Ees/J	Jackson Labs, cat # 023537	C57BL/6J		https://www.jax.org/strain/023537
B6.129(Cg)-Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J	Jackson Labs, cat # 007676	C57BL/6J		https://www.jax.org/strain/007676
C57BL/6J	Jackson Labs, cat # 0664	C57BL/6J		https://www.jax.org/strain/000664
Fn1f/f; RosamTmG/mTmG	Jackson Labs Fn1f/f mice: B6;129-Fn1tm1Ref/J Cat # 029624	Mixed 129/C57BL/6J	Fn1f/f mice were crossed to RosamTmG/mTmG mice to homozygocity, then double homozygous mice were maintained by intercrossing	Sakai T; Johnson KJ; Murozono M; Sakai K; Magnuson MA; Wieloch T; Cronberg T; Isshiki A; Erickson HP; Fassler R. 2001. Plasma fibronectin supports neuronal survival and reduces brain injury following transient focal cerebral ischemia but is not essential for skin-wound healing and hemostasis. Nat Med 7(3):324–30 PubMed: 11231631 MGI: J:67961
Itga5f/f; RosamTmG/mTmG	Jackson Labs Itga5f/f mice: B6.129-Itga5tm2Hyn/J cat # 032299 RosamTmG mice: Jackson Labs, cat # 007676	Mixed 129S4/C57BL/6J	Itga5f/f mice were crossed to RosamTmG/mTmG mice to homozygocity, then double homozygous mice were maintained by intercrossing	van der Flier A; Badu-Nkansah K; Whittaker CA; Crowley D; Bronson RT; Lacy-Hulbert A; Hynes RO. 2010. Endothelial alpha5 and alphav integrins cooperate in remodeling of the vasculature during development. Development 137(14):2439–49 PubMed: 20570943 MGI: J:161850
Itga5f/f; RosanTnG/nTnG	Jackson Labs: Itga5f/f mice: B6.129-Itga5tm2Hyn/J cat # 032299 RosamTmG mice: Jackson Labs, cat # 023537	Mixed 129S4/C57BL/6J	Itga5f/f mice were crossed to RosanTnG/nTnG mice to homozygocity, then double homozygous mice were maintained by intercrossing
Itga5+/−; ISl1Cre/+	Jackson Labs: Itga5+/− are B6.129S-Itga5tm1Hyn/J cat # 002274 Isl1Cre/+ are a gift from Sylvia Evans	C57BL/6J	Maintained by crossing to C57BL/6J for 10 years	Yang JT; Rayburn H; Hynes RO. 1993. Embryonic mesodermal defects in alpha 5 integrin-deficient mice. Development 119(4): 1093–105 PubMed: 7508365 MGI: J:16248 Cai, C.L., Liang, X., Shi, Y., Chu, P.H., Pfaff, S.L., Chen, J., Evans, S., 2003. Isl1 identifies a cardiac progenitor population that proliferates prior to differentiation and contributes a majority of cells to the heart. Dev Cell 5, 877–889.
Fn1+/−; Isl1Cre/+	Jackson Labs: Fn1+/− are B6.129S-Fn1tm1Hyn/2J, Cat# 008445 Isl1Cre/+ were from Dr. Sylvia Evans	C57BL/6J	Maintained by crossing to C57BL/6J for 10 years	George EL; Georges-Labouesse EN; Patel-King RS; Rayburn H; Hynes RO. 1993. Defects in mesoderm, neural tube and vascular development in mouse embryos lacking fibronectin. Development 119(4): 1079–91PubMed: 8306876MGI: J:16247
Itga5+/−; Mef2C-AHF-Cre	The mouse strain used for this research project, STOCK Tg(Mef2c-cre)2Blk/Mmnc, RRID:MMRRC_030262-UNC, was obtained from the Mutant Mouse Resource and Research Center (MMRRC) at University of North Carolina at Chapel Hill, an NIH-funded strain repository, and was donated to the MMRRC by Brian Black, Ph.D., University of California, San Francisco	C57BL/6J	Maintained by crossing to C57BL/6J for 10 years	Verzi MP, McCulley DJ, De Val S, Dodou E and Black BL. The right ventricle, outflow tract, and ventricular septum comprise a restricted expression domain within the secondary/anterior heart field. Dev Biol. 2005;287:134–45.
Fn1+/−; Mef2C-AHF-Cre		C57BL/6J	Maintained by crossing to C57BL/6J for 5 years
Itga5+/−; Sox172A-iCre	Dr. Heicko Lickert	C57BL/6J	Maintained by crossing to C57BL/6J for 5 years	Engert, S., Liao, W.P., Burtscher, I., Lickert, H., 2009. Sox17-2A-iCre: a knock-in mouse line expressing Cre recombinase in endoderm and vascular endothelial cells. Genesis 47, 603–610.
Isl1mER-Cre-mER/+	Jackson Labs Isl1tm1 (cre/Esr1*)Krc/SevJ Cat # 029566	C57BL/6J	Maintained by crossing to C57BL/6J for 3 years	Sun, Y., Liang, X., Najafi, N., Cass, M., Lin, L., Cai, C.L., Chen, J., Evans, S.M., 2007. Islet 1 is expressed in distinct cardiovascular lineages, including pacemaker and coronary vascular cells. Dev Biol 304, 286–296.
Mef2C-AHF-DreERT2	Dr. Benoit Bruneau	129/ C57BL/6J	Maintained by intercrossing homozygous mice	Devine, W.P., Wythe, J.D., George, M., Koshiba-Takeuchi, K., Bruneau, B.G., 2014. Early patterning and specification of cardiac progenitors in gastrulating mesoderm. eLife 3.
Ai9	Jackson labs B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J Cat # 007909	C57BL/6J	Maintained by intercrossing homozygous mice	Madisen L; Zwingman TA; Sunkin SM; Oh SW; Zariwala HA; Gu H; Ng LL; Palmiter RD; Hawrylycz MJ; Jones AR; Lein ES; Zeng H. 2010. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat Neurosci 13(1):133–40 PubMed: 20023653 MGI: J:155793
Ai66 with recombined loxP sites	Jackson labs B6;129S-Gt(ROSA)26Sortm66.1(CAG-tdTomato)Hze/J Cat#021876	129/C57BL/6J	LoxP sites in this strain were recombined converting this strain to a Dre reporter strain. This strain was maintained by intercrossing homozygous mice	Madisen L; Garner AR; Shimaoka D; Chuong AS; Klapoetke NC; Li L; van der Bourg A; Niino Y; Egolf L; Monetti C; Gu H; Mills M; Cheng A; Tasic B; Nguyen TN; Sunkin SM; Benucci A; Nagy A; Miyawaki A; Helmchen F; Empson RM; Knopfel T; Boyden ES; Reid RC; Carandini M; Zeng H. 2015. Transgenic mice for intersectional targeting of neural sensors and effectors with high specificity and performance. Neuron 85(5):942–58 PubMed: 25741722 MGI: J:219930

Antibodies
Target antigen	Vendor or Source	Catalog #	Dilution factor	Lot # (preferred but not required)	Persistent ID / URL
VEGFR2	R&D	AF644	1:200
ERG	Abcam	ab214341	1:1000
Pecam 1	BD Pharmingen	550274	1:200
GFP	Aves	GFP1020	1:300
mCherry	Abcam	ab167453	1:1000
BrdU	Abcam	ab6326	1:100
Cleaved caspase 3	Cell Signaling Technologies	9661	1:100
Goat IgG	ThermoFisher	A11058	1:300
Rabbit IgG	ThermoFisher	A31573	1:300
Rat IgG	ThermoFisher	A18745	1:300
Mouse IgG	ThermoFisher	A18745	1:300
Chicken IgY	Jackson Immunoresearch	703-545-155	1:300
Chemicals and kits	Vendor	Cat #	Stock solution	Amount Used	Dilution
BrdU	Fisher Scientific	BP2508250	5 mg/ml in PBS	30mg/kg body weight
In Situ Cell Death Detection Kit, Fluorescein TUNEL	Roche	11684795910	Used as directed in the manual
DRAQ5	Cell Signaling Technology	4084			1:1000
DAPI	Sigma	32670-5MG-F	5 mg/ml in water	5 μg/ml

DNA/cDNA Clones
Clone Name	Sequence	Source / Repository	Persistent ID / URL
Cultured Cells
Name	Vendor or Source	Sex (F, M, or unknown)	Persistent ID / URL

Data & Code Availability
Description	Source / Repository	Persistent ID / URL
Other
Description	Source / Repository	Persistent ID / URL

All animal experiment protocols and procedures were reviewed and approved by the Rutgers Institutional Animal Care and Use Committees.

Data Availability.

All supporting data is presented in the figures and the supplement. Additional data and unique reagents are available upon request from the corresponding author.

RESULTS

SHF harbors PAA progenitors between E7.5 and E9.5.

Previous work from our lab demonstrated that in the mouse, the majority of PAA endothelium is derived from the mesoderm encompassing Mef2C-AHF-Cre- or Isl1- expressing lineages⁹. To define the temporal window during which the SHF mesoderm harbors endothelial progenitors of the PAAs, we used Isl1^{MerCreMer /+} knock-in mice⁴² and Mef2C-AHF-DreERT2 transgenic mice⁴³ combined with pulses of tamoxifen to lineage-label the SHF mesoderm at different developmental times (Fig. 1). Tamoxifen was injected once at a discrete time point between E6.75 – E9.75. Embryos were dissected at E10.5 and stained to detect lineage labeling in the pharyngeal arches. Entire pharyngeal regions were imaged using confocal microscopy to quantify the contribution of lineage-labeled cells to the entire PAA endothelium (Fig. 1, panels A-A2). The expression of VEGFR2 and ERG was used to mark EC cell membranes and nuclei⁴⁴. The labeling of the cardiac outflow tract and the right ventricle using Isl1^{MerCreMer /+} mice was evident at all stages tested, indicating that our labeling technique was consistent with previous studies (data not shown)⁴². Myocardial cells derived from the SHF are labeled when tamoxifen is injected as early as E6.5 in Isl1^MerCreMer/+ knock-in mice⁴²; however, no PAA ECs were labeled when tamoxifen was injected at E6.75 in this strain (Fig. 1B, n=4) or E7.25 in Mef2C-AHF-DreERT2 transgenic strain (Fig. 1C, n=4), suggesting that PAA EC progenitors are specified the SHF later than cardiomyocyte progenitors. The peak labeling of the PAA endothelium occurred when tamoxifen was injected at E7.25 in Isl1^MerCreMer strain (Fig. 1B, n=5) and E8.0 in Mef2C-AHF-DreERT2 strain (Fig. 1C, n=7). While tamoxifen injection into Isl1^MerCreMer/+ resulted in sparse labeling of PAA ECs, the injection of tamoxifen into Mef2C-AHF-DreERT2 transgenic mice led to the labeling of a much larger proportion of ECs in the PAAs (compare Fig. 1B with Fig. 1C). These differences likely reflect that the Mer-Cre-Mer transgene is present as a single copy as it is knocked into the Isl1 locus⁴², while Mef2C-AHF-DreERT2 is a transgenic strain presumably containing multiple copies of the Mef2C-AHF-DreERT2 transgene⁴³. Moreover, the expression of Is1 is downregulated commensurate with endothelial differentiation⁴⁵. Thus, potentially low levels of MerCreMer expression in EC precursors could have resulted in low labeling of endothelial progenitors in Isl1^MerCreMer/+ mice relative to Mef2C-AHF-DreERT2 strain. The difference in the timing of peak EC labeling in the PAAs between Isl1^MerCreMer/+ and Mef2C-AHF-DreERT2 strains is likely to be due to the earlier onset of Isl1 expression compared with the expression of the Mef2C-AHF-DreERT2 transgene. Isl1 regulates the expression of Mef2C and the activation of the Mef2C-AHF enhancer^{46, 47}. Correspondingly, our experiments demonstrate that the peak endothelial labeling of PAAs in Isl1^MerCreMer/+ strain precedes that of the Mef2C-Dre-ERT2 strain by 18 hours (compare Fig. 1B with Fig. 1C). Interestingly, endothelial progenitors destined to contribute to the 4^th PAA continue to be present after E8.5 as more SHF-derived cells are labeled in the 4^th PAAs than in the 3^rd and 6^th when tamoxifen was injected at E8.5 and E9.5 (Fig. 1B1, 1C1, n=7). Thus, our labeling experiments show that the SHF harbors PAA endothelial progenitors between approximately E7.5 and E9.5 of embryonic development.

Figure 1. Endothelial PAA progenitors are present in the SHF as early as E7.25.

Isl1^MerCreMer/+ and Mef2C-AHF-DreERT2 males were mated with reporter females, treated with tamoxifen, dissected at E10.5, and stained to detect VEGFR2 (blue), ERG (red), or tdTomato (orange) (see Methods). A. Sagittal view and 3D reconstruction through the left pharyngeal region. Inset- 3D reconstruction of PAAs. A1 – A2. Sagittal optical sections through the embryo shown in A. B. Highest labeling of PAA endothelium occurred when tamoxifen was injected at E7.25 in Isl1^MerCreMer/+ knock-in mice; B1. SHF-derived cells continue to be added to the 4^th PAA after E9.5. n=8 arches of each type were analyzed at each time point. The difference between the curves was assessed by one-way ANOVA, p=10E-6. C. Peak labeling of PAA ECs occurred when tamoxifen was injected at E8.0 in MEF2C-AHF-DreERT2 strain; n=7 arches of each type were analyzed for each time point, except 4 arches were analyzed for E6.75. C1. Injection of tamoxifen at E8.5 led to a more efficient labeling of the 4^th PAAs than the 3rd and 6^th. n=7 arches, p was calculated using one-way ANOVA with Tukey’s correction for multiple testing.

To analyze the contribution of the SHF to the PAA endothelium quantitatively and to compare two mouse strains commonly used to label the SHF, we imaged the entire pharyngeal arch region containing the arches 3, 4, and 6. We then quantified the proportion of SHF-lineage labeled ECs in the PAAs of E10.5 embryos from Isl1^Cre/+ knock-in and Mef2C-AHF-Cre transgenic lines (Fig. 2). The majority of SHF-derived cells in the pharyngeal arches 3 – 6 are found in the endothelium at 37 somites, as seen in optical sections through the pharyngeal arch region (Fig. 2A – B, n=14). Each PAA is comprised of a similar number of ECs (Fig. 2C, p>0.1, one-way ANOVA with Tukey’s correction for multiple testing, n=14). However, there were differences in PAA labeling among embryos isolated from Isl1^Cre/+ and Mef2C-AHF-Cre mice (Fig. 2D, E). In the constitutive Isl1^Cre/+ knock-in strain, the SHF contribution to the 3^rd and 4^th PAA endothelium was 79±6% and 77±10%, respectively, and 57±12% to the 6^th PAA (Fig. 2D, n=8). However, in the Mef2C-AHF-Cre transgenic line, the SHF contribution to the 3^rd PAA was only 45±8% (Fig. 2E), which is significantly different from the SHF contribution to the 3^rd PAA in the Isl1^Cre/+ knock-in strain (p<1E-15, mixed model logistic regression with Bonferroni correction for multiple testing). The difference in the contribution of the SHF to the 3^rd PAA between the two strains likely reflects the earlier onset of Cre expression in the Isl1^Cre/+ knock-in strain relative to the onset of Cre expression in the Mef2C-AHF-Cre transgenic line⁴⁶. These data suggest that in ~ 50% of progenitors giving rise to the 3^rd PAA, EC fate is specified before the onset of Cre expression from the Mef2C-AHF-Cre transgene, and that following EC specification, the Mef2C-AHF-Cre transgene is not expressed in these cells. The SHF contribution to the PAA endothelium of the 4^th and 6^th PAAs was not statistically distinguishable between the two strains (p=0.29, mixed model logistic regression with Bonferroni correction for multiple testing). Furthermore, in both strains, SHF contribution to the 4^th PAAs was higher than the 6^th PAA (Fig. 2D–E). Our data also show that the deletion of one Isl1 allele, as in the Isl1^Cre/+ knock-in strain, does not impair the contribution of the SHF to the PAA endothelium. In summary, the 4^th PAAs differ from the 3^rd and 6^th PAAs in the timing of the activation of the Isl1 and Mef2C-AHF enhancers. Furthermore, the 4^th PAAs differ from the 6^th in the proportion of SHF-derived cells.

Figure 2. Differences in the contribution of the SHF to the PAA endothelium depend on the mouse strain used.

Mef2C-AHF-Cre; ROSA^tdTomato embryos (35 – 37 somites) were stained with antibodies to VEGFR2 (cyan) to detect endothelial cells, tdTomato (orange) to detect SHF-derived cells, and DAPI (blue) to detect nuclei. A – A2. 3D reconstructions of PAAs and their connections with the dorsal aorta (DA) and the aortic sac. B – B2. Sagittal optical sections show the distribution of all SHF-derived cells in the pharyngeal arches. PAAs 3 – 6 are labeled. All scale bars are 100 μm. C. The number of VEGFR2⁺EGR⁺ cells in the pharyngeal arches was quantified, E10.5 embryos at 35 – 37 somites, n=14 arches of each type were analyzed. Each dot is one arch. Red line marks the median. Black lines mark quartiles. Differences among the three PAA pairs are not significant, p=0.12 by one-way ANOVA with Tukey’s correction for multiple testing. D – E. The percentage of SHF-lineage+ cells among the total number of VEGFR2⁺EGR⁺ cells is plotted, statistical significance was evaluated using mixed model logistic expression method with correction for multiple testing, as described in Methods. D. The use of constitutive Isl1^Cre/+ strain resulted in the labeling of more than 80% of ECs in the 3^rd and 4^th PAAs, n=10 arches of each type were analyzed. E. The labeling of the 4^th PAAs is significantly higher than that of the 3^rd and the 6^th in Mef2C-AHF-Cre strain. The difference in labeling efficiency of the 3^rd PAAs between Isl1^Cre/+ and Mef2C-AHF-Cre is significant, p=2E-16. The difference in the labeling efficiency of 6^th PAAs between the two strains was not significant, p= 0.29; n=8 arches of each type were analyzed.

Cell-ECM interactions mediated by integrin α5β1 specifically regulate the accrual of SHF-derived cells in the pharyngeal region.

Studies described above, together with our previous work⁹ have established a framework for the analysis of EC dynamics and their genetic regulation during the morphogenesis of the AAA and its major branches. Our previous studies demonstrated that the deletion of either integrin α5 or Fn1 in the Isl1 lineages resulted in the defective formation of the 4^th PAAs at E10.5, and consequently, gave rise to IAA-B and retro-esophageal right subclavian artery (RERSA) in these mutants²⁹. IAA-B and RERSA are anomalies resulting from defective formation or remodeling of the left and right 4^th PAAs, respectively^{1, 3}. To determine the mechanisms by which integrin α5β1 and Fn1 regulate the formation of the 4^th PAAs, we analyzed PAA development at distinct stages of embryonic development using whole-mount immunofluorescence, followed by quantitative analyses of SHF-derived populations and their dynamics. PAAs form through the coalescence of pharyngeal arch EC plexus, a network of small blood vessels^{9, 10}: 4^th pharyngeal arch ECs are located within the plexus at E9.5. At E10.5 (33 – 34 somite stage), 50% of the 4^th arch endothelium is found within the PAA (Fig. 3A – C, the vessel surfaced in green in (Fig. 3B – C), and 50% is in the plexus (pink in Fig. 3B – C)⁹. About 50% of integrin α5^flox/-; Isl1^Cre/+ mutants have defective 4^th PAAs, and consequently, 50% of these mutants develop IAA-B and RERSA²⁹. We found that the 4^th PAA was absent in 50% of mutants at 32 – 34 somites (Fig. 3D – F). Instead, the endothelium in the 4^th arches was in the form of a plexus of small blood vessels (marked in pink in Fig. 3E, F). Small 4^th PAAs eventually formed in these mutants by 36 – 39 somites (Fig. 3J, marked in green in Fig. 3K, L; compare with the 4^th PAA surfaced in green in control Fig. 3G–I). Similarly, the 4th PAA formation was delayed in Fn1^flox/-; Isl1^Cre/+ mutants (Online Fig. IA, D). This defect was specific to the 4^th PAA, as the 3^rd and 6^th PAAs formed normally in the mutants (vessels surfaced in white and red, respectively, in Fig. 3). We hypothesized that the defective formation of the 4^th PAAs in our mutants could be due to insufficient EC numbers, defective EC proliferation, or survival. To test these hypotheses, we evaluated the total number of ECs in the 4^th pharyngeal arches of controls and mutants. To quantify EC number, we stained E10.5 embryos with the antibodies to ERG, a transcription factor enriched in the endothelium, and either VEGFR2 or Pecam1, expressed on EC surface⁴⁸. These experiments showed integrin α5^flox/-; Isl1^Cre/+ and Fn1^flox/-; Isl1^Cre/+ mutants had decreased total number of ECs in the 4^th arches at 32 – 33 somites relative to controls (Fig. 3M and Online Fig. IA–C). Despite this decrease in EC numbers, the size of the 4^th arches, the tissues within which PAAs form, was not significantly affected (Fig. 3N). EC proliferation in the 4^th arch was also not affected in the mutants (Fig. 3O), and neither was cell survival, measured by the expression of cleaved caspase 3 or TUNEL (Online Fig. II). Thus, EC deficiency in the 4^th pharyngeal arches of the mutants was not due to defective EC proliferation or survival. The majority of VEGFR2⁺ cells in the pharyngeal region of E9.5 – E10.5 Fn1^flox/-; ROSA^mTmG/+; Isl1^Cre/+ mutant embryos were lineage-labeled with GFP (Online Fig. IIIA – D), indicating that SHF cells in the pharyngeal region of the mutants were not impaired in the acquisition of EC fate. The 4^th pharyngeal arch endothelium expresses VEGFR2 at E9.5, which is a day earlier than Pecam1¹⁰. To determine whether the maturation of pharyngeal arch ECs was affected in the mutants at E10.5, we co-stained embryos with antibodies to VEGFR2 and Pecam1. Despite defective 4^th PAA formation, all VEGFR2⁺ cells in the mutants’ pharyngeal arches also expressed Pecam1 at E10.5 (Online Fig. IIIE – F1), ruling out EC maturation as a cause for decreased EC numbers in the 4^th pharyngeal arch.

Figure 3. Formation of the 4^th PAA is delayed in integrin a5^flox/-; Isl1^Cre/+ mutants.

Integrin a5^flox/+; Isl1^Cre/+ control and a5^flox/-; Isl1^Cre/+ mutant embryos were dissected at different somite stages at E10.5 and stained to detect Pecam1. PAAs are numbered and somite stages are indicated in the first row. A, D, G, J. 3D reconstructions of whole-mount Pecam 1 staining (blue). B, E, H, K. PAA endothelium in the 3^rd, 4^th and 6^th arches shown in the row above was surface-rendered white, green and red, respectively. The plexus endothelium in the 4^th arch was surface-rendered in pink. C, F, I, L. Left side and ventral views of surface-rendered PAAs and the plexus. Development of the 4^th PAAs was specifically affected in the mutants (E, F, K, L). Magnification is the same in all panels. Scale bar is 100 μm. M. Total number of ECs was quantified, as described in Methods. Mutants (n=7) have EC deficiency in the 4^th arch compared with controls (n=12) at 32 – 33 somites. N. The sizes of the 4^th arches are comparable between controls (n=12) and mutants (n=7). O. EC proliferation in the PAA and plexus in the 4^th arches was similar in controls (C) and mutants (M). In all plots, red lines mark medians, dashed lines mark quartiles. Each dot marks one arch. Statistics for M and N were evaluated using 2-tailed, unpaired Student’s t test with Welch’s correction for unequal variance between groups. Statistical comparisons in O were evaluated using Mann-Whitney’s non-parametric tests, p=046 for the comparison of EC proliferation in the PAA, and p>0.999 for the comparison of EC proliferation in the plexus in the 4^th arch.

Since maturation, proliferation, and survival of ECs were not affected in our mutants, we tested whether defective accrual of progenitor cells in the pharyngeal arches prior to the 32 – 33 somite stage was the cause for decreased EC numbers. As we established before, the majority of PAA ECs arise from the SHF (Fig. 2D, E)⁹, and the accrual of EC progenitors derived from the Isl1 and Mef2C-AHF lineages into the pharyngeal arches is nearly complete by E9.5 (Fig. 1B–C). To quantify the number of SHF-derived cells in the pharyngeal mesenchyme, we used ROSA^nT-nG reporter mice, in which nuclear localization sequences are fused with tdTomato and EGFP, leading to the expression of nuclear-localized EGFP upon Cre-induced recombination. We found that the deletion of integrin α5 in the Isl1 lineages resulted in decreased numbers of SHF-derived cells in the pharyngeal mesenchyme of all arches, while the number of SHF-derived cells contributing to the heart was not affected (Fig. 4A–E, Online Fig. IV for the method of surfacing).

Figure 4. The expression of integrin α5β1 in the Isl1 lineages is required for the accrual of SHF-derived cells in the pharyngeal mesenchyme.

Control and mutant embryos carrying one ROSA^nTnG reporter allele were dissected at E9.5 (18 – 20 somites) and stained with DAPI and anti-GFP antibodies. Whole embryos were imaged and the number of GFP⁺ cells in the pharyngeal mesoderm and in the heart was quantified as described in Online Fig. IV. Controls A – A3, Mutants B – B3. Representative images of control and mutant embryos showing SHF components: pharyngeal arch mesoderm (orange) and dorsal pericardial wall (DPW, magenta), ECs - VEGFR2 (blue). The surfaced region extends from the 3^rd pharyngeal arch (white arrows) at the anterior extent to the common cardinal vein (CCV) at the posterior extent. Isl1-lineage DA – dorsal aorta. Anterior is up, ventral is to the right. C. The number of GFP⁺ cells in the mesenchyme of the 1^st and 2^nd arches was decreased in the mutants (n=10) relative to controls (n=6). D. The total number of GFP⁺ cells in the pharyngeal mesenchyme corresponding with the future arches 3 – 6 was decreased in the mutants (n=10) relative to controls (n=10). E. The number of SHF-derived cells in the hearts of controls (n=7) and mutants (n=11) were comparable; F. The number of SHF-derived cells in the DPW was comparable in controls (n=5) and mutants (n=5). G. The proportions of GFP⁺ cells in the heart relative to GFP⁺ cells in splanchnic mesoderm were comparable in controls (n=6) and mutants (n=6). H. The proportion of GFP⁺ cells in the posterior pharyngeal mesenchyme relative to the number of GFP⁺ cells in splanchnic mesoderm was significantly decreased in the mutants (n=7) relative to controls (n=6). Red lines mark medians, dotted lines mark quartiles; For C – D, p values were determined using unpaired, 2-tailed Student’s t tests. For all other panels, non-parametric Mann-Whitney tests were used.

The SHF is composed of the splanchnic mesoderm contained within the distal portions of the 1^st and 2^nd pharyngeal arches and in the dorsal pericardial wall (DPW)^{49, 50}. To test whether the deficiency in the pharyngeal SHF-derived mesoderm was accompanied by the decrease in SHF cell numbers in the DPW, we used IMARIS to surface the DPW and quantified the number of GFP⁺ cells (see Online Fig. IVA – B3 for the details on surfacing). These experiments showed that the numbers of GFP⁺ cells in the splanchnic mesoderm within the DPW were similar between controls and mutants (Fig. 4F). Next, we computed the proportion of GFP⁺ cells in the pharyngeal mesenchyme or the heart relative to the number of GFP⁺ cells in the DPW for each embryo. While the latter ratio was not affected in the mutants (Fig. 4G), the former was significantly decreased (Fig. 4H), suggesting that there may be a defect in EC progenitors’ specification within the SHF mesoderm or their migration to the posterior arches. These experiments indicate that the ECM microenvironment sensed by integrin α5β1 is essential for the accrual of the SHF-derived EC progenitors in the pharyngeal arch mesenchyme.

Integrin α5β1 and fibronectin regulate the remodeling of pharyngeal plexus into the 4^th PAAs independently of EC numbers.

By the 34–35 somite stage, the numbers of SHF-derived cells and ERG⁺ ECs in the mutants recovered and were similar to those of controls (Fig. 5A). The total number of GFP⁺ cells in the posterior pharyngeal arches also recovered (Fig. 5B). The percentage of GFP⁺ ECs in the pharyngeal arches of controls and mutants were comparable (Fig. 5C), indicating that the recovery of EC numbers was not due to the recruitment of ECs from an alternative mesodermal source. The recovery of pharyngeal EC numbers was likely mediated through the proliferation of SHF-derived ECs. The basis for this conclusion is the following. The proliferation index of pharyngeal arch ECs measured by BrdU incorporation was unaltered in the mutants (Fig. 3O), and the proliferation index of ECs in the pharyngeal plexus is 2-fold higher than that of PAA ECs, in controls and mutants (Fig. 3O, plexus). Since the proportion of ECs in the pharyngeal plexus is higher in the mutants than in controls (Online Fig. ID–E), the higher proliferation index of plexus ECs in the mutants is likely responsible for the recovery of EC numbers.

Figure 5. Recovery of EC numbers in integrin α5^flox/-; Isl1^Cre/+ mutants.

A. Total EC number has recovered in integrin a5^flox/-; Isl1^Cre/+ mutants by the 34^th somite stage. B. Total number of SHF-derived mesodermal cells has recovered in the pharyngeal arches in integrin a5^flox/-; Isl1^Cre/+ mutants by the 34^th somite stage. C. The fraction of SHF-derived ECs in pharyngeal arches is comparable among control and mutant embryos. This fraction was calculated by quantifying the number of GFP⁺ERG⁺ cells and dividing by the total number of ERG⁺ cells in the entire pharyngeal arches (e.g. ECs in both PAA and plexus were quantified). In A, statistical significance was evaluated using one-way ANOVA with Sidak’s correction for multiple testing. In A, n=6 controls and n=12 mutants at 34 – 35s; n=7 controls and n=9 mutants at 36 – 37s; In B – C, n=6 controls and n=8 mutants. Open circles mark 4^th arches without a patent PAA (these arches contain EC plexus only). Thick lines mark medians, dotted lines mark quartiles. Statistical significance in B – C was evaluated using Kruskal-Wallis tests with Dunn’s correction for multiple testing.

Our quantitative analyses indicate that defects in PAA formation in integrin α5^flox/-; Isl1^Cre/+ and Fn1^flox/-; Isl1^Cre/+ mutants are indistinguishable from each other (Online Fig. I), suggesting that integrin α5β1 is a major receptor transducing Fn1 signals within the pharyngeal microenvironment. Despite the recovery of EC numbers (Fig. 5A), PAAs remained thin in integrin α5^flox/-; Isl1^Cre/+ and Fn1^flox/-; Isl1^Cre/+ mutants (Fig. 3J – L), and there was a 2 – 3-fold decrease in the proportion of all 4^th pharyngeal arch ECs in the 4^th PAAs at all stages analyzed at E10.5 (Fig. 6A and Online Fig. IE). The 4^th PAA size increases between 32 – 39 somites as more ECs are added to the PAA from the plexus (Online Fig. V). This is reflected in the percent of pharyngeal arch ECs in the PAA⁹. In controls, plexus ECs in the 4^th arch begin coalescing into the PAA when embryos reach between 31 and 32 somites⁹ (Online Fig. V). These rearrangements result in an initially thin 4^th PAAs, in which approximately 50% of the pharyngeal arch ECs are in the plexus, and 50% are in the PAA at 32 – 34 somites⁹. As the development proceeds, by 36–39 somite stage, > 60% of the 4^th pharyngeal arch endothelium becomes incorporated into the 4^th PAA⁹. Thus, the proportion of the pharyngeal arch endothelium in the 4^th PAA can be taken as a measure of PAA formation. The higher the proportion, the larger the PAA⁹.

Figure 6. Integrin α5β1 and Fn1 regulate the remodeling of EC plexus during the formation of the 4^th pharyngeal arch arteries.

A. The proportion of ECs constituting the 4^th PAAs in the mutant is significantly lower than in controls, at all stages analyzed at E10.5, including the stages when the EC population in the 4^th pharyngeal arch has recovered in the mutants; p values were calculated by using Kruskal-Wallis test with Dunn’s correction for multiple testing. Thick black lines mark medians, dashed lines mark quartiles, each dot marks one arch. B – C. Least squares regression analyses indicate the absence of linear correlation between the size of the 4^th PAA and EC number (B) or density (C). PAA size is expressed as the percentage of pharyngeal arch endothelial cells in the 4^th PAA on the y-axis. R² is Pearson’s correlation coefficient. R²=1 indicates perfect correlation. D. Total EC number (x-axis) in mutants with defective (open symbols) or unaffected 4^th PAA (closed symbols) were plotted against the size of the 4^th PAAs on the y -axis. Least squares regression analysis indicated low correlation between these properties, R²<<1. Circles: 32 – 33 somite embryos, rhombi: 34 – 35 somite embryos, triangles: 36 – 39 somite embryos. E. The rearrangement of the endothelial plexus into the 4^th PAAs is defective in mutants relative to controls with the same number of endothelial cells in the 4^th arch (boxes). EC – endothelial cell(s). Controls: α5^flox/+; Isl1^Cre/+ and Fn1^flox/+; Isl1^Cre/+ embryos; Mutants: α5^flox/-; Isl1^Cre/+ and Fn1^flox/-; Isl1^Cre/+ embryos. In A, 32 – 33 s, n=12 controls, n=7mutants; 34 – 35s, n=8 for each group, 36 – 39s, n=4 controls, and n=5 mutants; in B n=26; in C n=12; in D, n=33; in E, n=10 controls, n=9 mutants.

To understand the mechanisms by which integrin α5β1 and Fn1 regulate the remodeling of the 4^th arch endothelial plexus into the PAA, we examined EC dynamics in control and mutant embryos at three-time points, 32 – 33 somites, 34 – 35 somites, and 36 – 39 somites. These stages span about 6 hours on the 10^th day of mouse embryonic development. The formation of the 4^th PAAs lagged in mutants relative to controls at all time points tested during E10.5 (Fig. 6A and Online Fig. ID–E), and 7 of the 16 embryos analyzed contained only a plexus of ECs and lacked a patent 4^th PAAs at the 32 – 34 somite stage (Online Fig. IE), a stage at which over 50% of the 4^th arch endothelium in controls is located within the 4^th PAAs⁹ (Fig. 6A and Online Fig. IE).

Since mutant embryos had fewer ECs in the 4^th pharyngeal arches than controls before the 36^th somite stage, we performed linear least squares regression analyses to test whether the formation of the 4^th PAAs had a linear dependency on the total EC number or EC density in the 4^th. As described above, the percentage of pharyngeal arch ECs in the PAA relative to the plexus can be taken as a measure of PAA formation (Online Fig. V)⁹. Thus, for these analyses, we quantified EC numbers in control embryos isolated between 32 to 39 somite stages and plotted them against the percent of ECs in the 4^th PAAs (Fig. 6B). Despite the sharp, over a 3-fold increase in the number of ECs in the 4^th arches between these stages⁹, the formation of the 4^th PAAs in controls was independent of the total EC number in the 4^th pharyngeal arch tissue (Fig. 6B, n=26, Pearson’s correlation coefficient r²=2E-6) or EC density (Fig. 6C, n=12, r²=0.03). Similarly, least squares regression analyses of PAA formation in the mutants showed that similar to controls, the rearrangement of plexus ECs into the PAA did not depend on the total number of ECs in the 4^th pharyngeal arches (Fig. 6D, n=33, r²=0.10).

Next, we compared PAA formation in controls and mutants with a similar number of ECs in the 4^th pharyngeal arches (Fig. 6E). These analyses showed that the percent of ECs in the PAAs was lower in the mutants (boxes in Fig. 6E). These data indicate that the reorganization of the plexus ECs into the PAA in the 4^th pharyngeal arch does not depend on the EC number at E10.5 and is regulated by factors extrinsic to the pharyngeal arch endothelium. In summary, our studies indicate cell – ECM interactions mediated by integrin α5β1 and Fn1 are essential for the remodeling of the vascular plexus into the PAA in the 4^th pharyngeal arches.

The expression of integrin α5 in the Isl1 lineage is required for the differentiation of neural crest cells into vascular smooth muscle cells.

In the Tbx1^+/− mouse model of 22q11 deletion syndrome, PAA formation recovers in 50 – 68% of the mutant mice^{51, 52}. To determine whether the rearrangement of the endothelial plexus in the 4^th arch was blocked or delayed in our mutants, we examined E11.5 embryos. The incidence of IAA-B and RERSA in integrin α5^flox/-; Isl1^Cre/+ mutants is 50%, which is the same as the incidence of defective 4^th PAA formation. Therefore, we expected to find absent or thin 4^th PAAs in the mutants at E11.5. Contrary to our expectations, the formation of the 4^th PAAs recovered by E11.5, and PAA perimeters in the mutants were comparable with controls (Fig. 7A, n=8). Consistent with the recovery of SHF-derived ECs numbers by 33 – 35s, PAA ECs were GFP⁺ in the mutants as in controls (compare Fig. 7C1 with 7D1, arrowheads). Regression of left 4^th PAAs results in IAA-B, and regression of the right 4^th PAA results in RERSA^{53, 54}. Since 50% of integrin α5^flox/-; Isl1^Cre/+ mutants develop IAA-B and RERSA²⁹, these data indicated that the 4^th PAAs eventually regress in the mutants. Arch artery regression is commonly caused by the defective differentiation of neural crest (NC) cells surrounding the PAA endothelium into vascular smooth muscle cells (VSMCs)^54–59. In the pharyngeal arches, VSMCs exclusively arise from NC-derived cells^{51, 60, 61}. To determine whether the differentiation of NC-derived cells into VSMCs was affected in our mutants, we analyzed VSMC differentiation in the pharyngeal arches. For these experiments, we calculated the fraction of vessel perimeter covered by alpha-smooth muscle actin (αSMA)-expressing cells, using previously-developed methodology³¹. We found that the differentiation of NC-derived cells into VSMCs was severely diminished around the left 4^th PAAs in the mutants (quantified in Fig. 7B; compare sections in Fig. 7C, D, magnified in Fig. 7C2, D2; zoom-out panels are in Online Fig. VI). The decrease in αSMA expression was not due to NC cell death (Online Fig. II).

Figure 7. The expression of integrin α5β1 in the Isl1 lineages regulates the differentiation of neural crest-derived cells into VSCMs at E11.5.

A. PAA perimeter has recovered in size in the mutants by E11.5. No statistically significant differences were found using one-way ANOVA with Sidak’s correction for multiple testing; n=6 controls and n=8 mutants. B. Smooth muscle coverage of the left 4^th PAA was deficient in the mutants, p values were calculated using one-way ANOVA with Sidak’s correction for multiple testing; with the exception of the left 4^th PAAs no other comparisons yielded statistically significant differences. n=6 controls and n=8 mutants (left sides) and n=6 mutants (right sides) were analyzed. C – D. The activation of Notch is not sufficient to induce the differentiation of NC cells into VSMCs. PAAs are numbered. PAA ECs at E11.5 are derived from the Isl1 lineage, green (arrowheads in C1, C2 and D1, D2). C2, D2. VSMC differentiation assayed by the expression of alpha smooth muscle actin (aSMA, blue) is specifically affected around the 4^th PAAs in the mutants (compare regions marked by the arrows in C2 and D2). The activation of Notch assayed by the expression of NICD is not altered in the mutants with defective VSCM differentiation (arrows in C3,C4 and D3,D4). E. Fate map using TFAP2a^IRESCre shows the location of NC-derived cells in the pharyngeal arches. Note extensive colonization of the mesenchyme between the endodermal pouches (endo) by the TFAP2a^IRESCre lineage. aSMA⁺ cells are GFP-negative in the Isl1^Cre/+ strain (arrows in C1-C2). E2. aSMA⁺ cells are GFP+ in TFAP2a^IRESCre strain. All scale bars are 50 μm. Additional zoom-out views are in Online Fig. VI.

Even though the Isl1 lineage marks a subset of NC-derived cells⁶², Isl1 protein is not expressed in NC-lineage cells in the pharyngeal arches, and Isl1 lineage does not label cells adjacent to the PAA endothelium (Fig. 7C1, D1)²⁹. Moreover, a comparison of NC lineage (Fig. 7E – E4) and Isl1 lineage maps at E11.5 demonstrates that αSMA expression coincides with the NC lineage (Fig. 7E2), but not with Isl1 lineage-labeled cells (small arrows in Fig. 7C1 and C2 point to αSMA-expressing cells; arrowheads point to GFP⁺ PAA endothelium). These studies indicate that the expression of integrin α5 in the Isl1 lineage(s) regulates the differentiation of NC cells into VSMCs in a non-cell-autonomous manner. These results are consistent with our previous experiments demonstrating that the expression of integrin α5 in the Mesp1 lineage, marking the anterior mesoderm, regulates the differentiation of NC cells into VSMCs around the 4^th PAA⁶¹. Since the deletion of integrin α5 in the Mesp1 lineage does not result in defective or delayed PAA formation⁶¹, these data indicate that integrin α5β1 regulates differentiation of NC cells into VSMCs around the 4^th PAA independently of its role in PAA formation (summarized in Fig. 8A–B).

FIGURE 8. CELL – ECM INTERACTIONS MEDIATED BY FN1 AND INTEGRIN A5B1 PLAY ESSENTIAL ROLES AT MULTIPLE STAGES OF AORTIC ARCH DEVELOPMENT.

DEPICTIONS OF CORONAL SECTIONS THROUGH THE 4^TH PHARYNGEAL ARCH AT E10.5. LINEAGES MARKED IN SPECIFIC STRAINS ARE IN GREEN, NC-DERIVED CELLS ARE GRAY CIRCLES, NC-DERIVED CELLS NEXT TO THE PAA ENDOTHELIUM ARE MARKED IN ORANGE. A. BLUE RECTANGLE HIGHLIGHTS THE NEW DATA RESULTING FROM THIS MANUSCRIPT: THE EXPRESSION OF INTEGRIN α5β1 AND FN1 IN THE ISL1 LINEAGES REGULATES THE MORPHOGENESIS OF THE AORTIC ARCH BY MEDIATING THE ACCRUAL OF SHF-DERIVED PROGENITORS IN THE PHARYNGEAL ARCHES, THE REMODELING OF THE VASCULAR PLEXUS INTO THE 4^TH PAAS, AND THE DIFFERENTIATION OF NC-DERIVED CELLS INTO VSMCS. THE EXPRESSION OF INTEGRIN α5β1 IN THE MESODERM (B) AND THE EXPRESSION OF FN1 OR INTEGRIN α5β1 IN THE NC (C) IS ESSENTIAL FOR THE DIFFERENTIATION OF NC-DERIVED CELLS INTO VSMCS. DEFECTS IN ANY ONE OF THE STEPS MARKED BY THE ARROWS LEAD TO ABERRANT MORPHOGENESIS OF THE AAA AND ITS MAJOR BRANCHES, RESULTING IN IAA-B AND RERSA.

The differentiation of NC cells into VSMCs is orchestrated in part by a relay of Notch signaling transduced from the PAA endothelium to the adjacent layers of NC-derived cells⁵⁸. The activation of Notch signaling in the NC is required for the differentiation of NC cells into VSMCs^{54, 58}. We demonstrated that this pathway was regulated by the expression of integrin α5 and fibronectin specifically in NC-derived cells at E11.5³¹. To test the possibility that the expression of integrin α5 in the Isl1 lineages regulates the lateral propagation of Notch from the PAA endothelium to the adjacent NC-derived cells, we stained control and mutant sections with an antibody to Notch Intracellular Domain (NICD), an activated form of Notch. However, despite the severe deficiency in the NC cell differentiation into VSMCs in the mutants Notch signaling was activated comparably in controls and mutants, judged by the nuclear localization of NICD (compare Fig. 7C2 – C4 with Fig 7D2 – D4, arrows). These experiments indicate that the expression of integrin α5 in the pharyngeal arch mesoderm regulates the differentiation of NC cells into VSMCs independently of Notch. Furthermore, these experiments indicate that while the activation of Notch is necessary for the differentiation of NC-derived cells into VSMCs, it is not sufficient. Taken together with our previous studies^{31, 61}, our work demonstrates that cell-ECM interactions regulated by integrin α5β1 and Fn1 pleiotropic and stage-specific functions during the morphogenesis of the 4^th PAAs (Fig. 8).

Combinatorial expression of integrin α5 and fibronectin from multiple lineages in the pharynx regulates aortic arch morphogenesis.

The Isl1 lineages encompass splanchnic mesoderm including the SHF, pharyngeal endoderm, surface ectoderm, and some NC-derived cell populations, but not the NC-derived cells in the pharyngeal arches^{29, 42, 50, 61, 62}. Our previous studies indicated that the combined expression of integrin α5β1 or Fn1 in the surface ectoderm and the NC was not required for the formation of the 4^th PAAs^{29, 31}. However, even though PAA formation occurred normally in these mice, the 4^th PAAs regressed later due to defects in the differentiation of NC-derived cells into VSMCs, resulting in RERSA and IAA-B³¹ (Fig. 8). The expression of either integrin α5β1 or Fn1 in the SHF lineage marked by the expression of the Mef2C-AHF-Cre transgene was also not required for PAA formation (Online Tables I and II), indicating that the expression of integrin α5β1 or Fn1 in the SHF alone is not required for cardiovascular development. Consistent with these findings, the expression of integrin α5β1 in the Mesp1 lineage or the endothelium was not required for PAA formation^{61, 63} (Fig. 8). Instead, the expression of integrin α5β1 in the Mesp1 lineage was required for NC-to-VSMC differentiation, and the deletion of integrin α5 in Mesp1 lineage, which includes the PAA endothelium, resulted in IAA-B and RERSA (Fig. 8)^{61, 63}. Taken together, these studies also indicate that the accrual of SHF-derived ECs in the posterior arches and the remodeling of the 4^th pharyngeal arch EC plexus into the PAA are regulated by Fn1 and integrin α5β1 expressed in the pharyngeal tissues other than the mesoderm and endothelium.

The difference in the phenotypes resulting from the deletion of integrin α5 using Mef2C-AHF-Cre and Mesp1^Cre are likely the result of differences in the timing of Cre expression (e.g., the later onset of Mef2C-AHF-Cre may have allowed the perdurance of integrin α5β1 protein through the stages when it is required for mesoderm-NC interactions). Alternatively, the expression of integrin α5β1 in the Mesp1 lineage-derived mesodermal cells before E8.5 is essential for regulating NC cell fate in the pharyngeal arches⁶¹.

Lastly, we tested whether the expression of integrin α5 in the endoderm regulated PAA formation. For these experiments, we used the constitutive Sox17^2A-iCre knock-in mouse line, in which Cre is expressed in the endoderm and pharyngeal arch ECs (Online Fig. VIIA – A4)⁶⁴. However, PAAs formed correctly in α5^flox/-; Sox17^2A-iCre mutants, n=6 (Online Fig. VIIB, B1, C, C1). Together, these data suggest that expression of integrin α5β1 or Fn1 from at least one of the pharyngeal arch cell types is sufficient to mediate the formation of the 4^th PAAs. In contrast, combinatorial signaling by integrin α5β1 in the mesoderm and the NC is essential for the differentiation of NC-derived cells into VSMCs and the 4^th PAA stability (Fig. 8).

DISCUSSION

Proper development of the 4^th PAAs is central to a newborn’s ability to survive and thrive (Karunamuni et al., 2014; Moon, 2008). Unraveling the dynamics of EC progenitors and their descendants during PAAs formation is vital to understanding the genetic and cellular mechanisms regulating PAA formation and how they are altered in congenital heart disease. The reasons for the heightened susceptibility of the 4^th PAAs to genetic and environmental insults have not been well-understood. Our data highlight several key differences in the dynamics of EC progenitors and their contributions to the 4^th PAAs relative to the 3^rd and 6^th. We have demonstrated that the SHF is the primary source of the PAA endothelium and that the majority of endothelial progenitors giving rise to the PAAs are already present in by E7.5. Progenitors marked by the activation of the Isl1 gene and Mef2C-AHF enhancer contribute to the pharyngeal arch endothelium over a span of about two days, from E7.5 to E9.5. However, the contribution to the 4^th PAA continues for a longer period of time than to the 3^rd and the 6^th PAAs.

Although the labeling efficiency of the 6^th PAA was significantly less than the 4^th PAA, the labeling efficiencies of these PAAs were consistent between the Isl1^Cre/+ and Mef2C-AHF-Cre strains. In contrast, ECs in the 3^rd PAA were labeled 50% more efficiently in the Isl1^Cre/+ knock-in line than in the Mef2C-AHF-Cre transgenic strain. This difference likely reflects the earlier onset of Cre expression in the Isl1^Cre/+ strain relative to the Mef2C-AHF-Cre strain^{42, 47, 65}. This difference in the labeling efficiency suggests that about half of endothelial progenitors of the 3^rd PAAs are specified before the onset of Mef2C-AHF-driven Cre expression, and do not activate the Mef2C-AHF enhancer at a later point in development. Thus, if one were to use the Mef2C-AHF-Cre line to generate mutations, the 4^th PAAs could be more affected than the 3^rd and the 6^th because the contribution of the Mef2C-AHF lineage cells to the 4^th PAA endothelium is the highest among the PAAs in this strain. Together, our labeling experiments demonstrate that the 4^th PAAs differ from the 3^rd and 6^th in several aspects: 1) a large fraction of EC progenitors of the 3^rd PAAs arises before that of the 4^th and 6^th; 2) the contribution of the SHF to the 4^th PAAs continues for a longer period of time than to the 3^rd and the 6^th PAAs; and 3) the contribution of SHF-derived ECs to the 4^th PAA is greater than that to the 6^th PAA. These findings suggest that the heightened susceptibility of the 4^th PAAs to genetic and environmental insults may stem from the differences in the dynamics of gene expression in the SHF and the extent of SHF contribution to the 4^th PAAs relative to the 3^rd and the 6^th PAAs.

Our studies show that integrin α5β1 and Fn1 are important for the initial accumulation of SHF-derived EC progenitors in the 4^th pharyngeal arches and that the absence of integrin α5 or Fn1 in the Isl1 lineages results in EC deficiency until 32 – 34 somite stage. Despite the initial EC deficiency in the 4^th arch, EC numbers recover in integrin α5^flox/-; Isl1^Cre/+ and Fn1^flox/-; Isl1^Cre/+ mutants by the 34 – 35 somite stage. We demonstrate that the recovery of EC cell numbers in the pharyngeal arches is not due to compensation from an alternative endothelial source. Instead, we show that the proliferation index of plexus endothelium is 2-fold higher than that of ECs in the 4^th PAA (Fig. 3O). This difference in the proliferation index is maintained in the mutants (Fig. 3O). We hypothesize that since most ECs are in the plexus at 32 – 33 somites in the mutants, their proliferative advantage over ECs in the PAA allows the EC number in the mutant arches to recover by the end of E10.5.

The Isl1 lineage encompasses multiple cell types within the pharynx, including pharyngeal epithelia, mesoderm, and a population of NC cells in the cardiac outflow tract^{62, 65}. Pharyngeal endoderm and the ectoderm are important signaling centers regulating intercellular communications among the germ layers composing the arches^{27, 66, 67}. Despite the recovery of EC populations, 50% of the 4^th PAAs were either thin or absent at 36 – 39 somite stages²⁹. Linear regression analyses showed that the rearrangement of the 4^th pharyngeal arch ECs into the PAA was not dependent on the number or density of ECs in the 4^th arch. Our genetic data show the importance of Fn1 and integrin α5β1 expressed in the pharyngeal tissues extrinsic to the pharyngeal arch mesoderm and endothelium in mediating the remodeling of the pharyngeal arch vascular plexus into the 4^th PAAs (Online Tables I–II)⁶¹.

We investigated the tissues wherein signaling by Fn1 is necessary for PAA formation. Integrin α5β1 is a primary Fn1 receptor during embryogenesis^{28–30, 34, 38, 39, 41}, and the deletion of integrin α5 or Fn1 in the Isl1 lineages results in comparable phenotypes²⁹. To determine the cell type(s) in which signaling by Fn1 regulates PAA formation, we ablated integrin α5 in each of the tissues comprising the Isl1 lineage individually or in combination. The deletion of integrin α5 in the SHF (Mef2C-AHF-Cre strain), the entire anterior mesoderm (Mesp1^Cre/+), the NC (Wnt1-Cre2, P3Pro-Cre), the NC and surface ectoderm (TFAP2α^IresCre/+), or the endoderm and endothelia (Sox17^2A-iCre/+) did not alter the development of the 4^th PAAs (this study and^{29, 31, 61}). Therefore, we conclude that signaling by integrin α5β1 either from the surface ectoderm or the pharyngeal endoderm is sufficient to mediate the formation of the 4^th PAAs, while the combinatorial signaling by integrin α5β1 in the mesoderm and the NC is essential for the 4^th PAA stability and VSMC differentiation. Our experiments also suggest that the pharyngeal endoderm may be an essential source of Fn1 in regulating 4^th PAA formation (Fig. 8).

The PAAs form within the NC-derived pharyngeal mesenchyme and signaling from the PAA endothelium induces the differentiation of the adjacent, NC-derived cells into VSMCs⁵⁸. Despite the initial delay in the 4^th PAA formation, the size of the 4th PAAs in integrin α5^flox/-; Isl1^Cre/+ mutants recovered by E11.5. At this time, we observed a profound deficiency in the expression of αSMA by NC-derived cells around the 4^th PAAs in the mutants. Deficiency in VSMC differentiation causes vessel regression^{31, 54, 59, 68, 69}, consistent with our finding that integrin α5^flox/-; Isl1^Cre/+ and Fn1^flox/-; Isl1^Cre/+ mutants develop IAA-B and/or RERSA, defects that are caused by the aberrant morphogenesis of the left and right 4^th PAAs, respectively²⁹. Our previous studies using integrin α5^flox/-; Mesp1^Cre/+ mice demonstrated that the expression of integrin α5 in the mesoderm regulates NC differentiation into VSMCs without affecting the 4^th PAA formation, and integrin α5^flox/-; Mesp1^Cre/+ mice develop IAA-B and/or RERSA⁶¹ (Fig. 8). Thus, the roles of integrin α5β1 and Fn1 in the formation of the 4^th PAAs are separate from their roles in the differentiation of NC-derived cells into VSCMs³¹.

Mechanisms that lead to IAA-B are complex but generally arise due to either of the following two broad categories of defects: a) defects in the formation of the left 4^th PAA or b) defects in the stability of an otherwise well-formed 4^th PAA. NC ablation studies in chick and genetic manipulation of the NC demonstrate that NC is not required for PAA formation^{53, 54}. However, aberrant differentiation of NC-derived cells into VSMCs leads to PAA regression resulting in various malformations in the aortic arch and its branches, including IAA-B and RERSA^{31, 54, 68}. Our studies show that the expression of integrin α5β1 in the pharyngeal mesoderm and the NC are required for NC-to-VSMC differentiation, and the expression of integrin α5β1 in either of these lineages alone is not sufficient for this process^{31, 61}.

Defects in the formation of the 4^th PAAs often occur in conjunction with 22q11 deletion syndromes⁷⁰. Cumulatively, four prospective studies found that between 40 and 90% of interrupted aortic arch type B (IAA-B) cases diagnosed in fetuses, neonates, and children can be attributed to deletions in the 22q11 region⁷¹. Studies using Tbx1^+/− mice that model 22q11 deletion syndrome indicated that defective formation of the left 4^th PAA underlies IAA-B in these patients^72–74. Intriguingly, several independent publications demonstrated that Tbx1 regulates the expression of integrins and extracellular matrix (ECM) components and showed that defects in cell-ECM interactions downstream of Tbx1 precede pathological sequelae and cardiovascular defects in Tbx1 mutants^75–77. Interestingly, about 50% of Tbx1^+/− mice recover from the initial defect in PAA formation and are viable and fertile⁵²; However, decreased expression of Fn1 was associated with the impeded 4^th PAA recovery⁵¹. Thus, alterations in cell-ECM interactions and pharyngeal ECM microenvironment may underlie lethal AAA defects in 22q11 deletion syndrome downstream of Tbx1. Our work’s significance lies in the identification of cellular dynamics regulating PAA formation and the intricate temporal and cell-type-specific roles of cell-ECM interactions in the regulation of aortic arch morphogenesis at multiple steps of its formation and remodeling (Fig. 8).

NOVELTY AND SIGNIFICANCE.

What Is Known?

• Formation of the aortic arch artery (AAA) is essential for neonatal survival, and aberrations in this process give rise to severe congenital heart disease;

• The left 4^th pharyngeal arch artery (PAA) gives rise to the aortic arch;

• Fibronectin (Fn1) and its major receptor integrin α5β1 regulate the formation and remodeling of the 4^th PAAs.

What New Information Does This Article Contribute?

• The 4^th PAAs differ from the 3^rd and the 6^th in the timing and extent to which second heart field (SHF) progenitors contribute to the PAA endothelium;

• Signaling by Fn1 and integrin α5β1 is critical at multiple stages of PAA formation and remodeling;

• Combinatorial signaling by Fn1 and integrin α5β1 from multiple pharyngeal lineages regulates dynamic cell behaviors during PAA formation and remodeling.

Nonstandard Abbreviations and Acronyms in the Alphabetical Order:

AAA

aortic arch artery

CHD

congenital heart disease

ECs

endothelial cells

Fn1

fibronectin

IAA-B

interrupted aortic arch type B

Itga5

integrin α5

PAA

pharyngeal arch arteries

RERSA

retro-esophageal right subclavian artery

SHF

second heart field

VEGFR2

vascular endothelial growth factor receptor 2