Role of Epha4 Receptor Signaling in Thyroid Development: Regulation of Folliculogenesis and Propagation of the C-Cell Lineage

THYROID-TRH-TSH

Role of EphA4 Receptor Signaling in Thyroid Development: Regulation of Folliculogenesis and Propagation of the C-Cell Lineage

Louise Andersson, Jessica Westerlund, Shawn Liang, Therese Carlsson, Elena Amendola, Henrik Fagman, and Mikael Nilsson Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021 Department of Medical Biochemistry and Cell Biology (L.A., J.W., S.L., T.C., M.N.), Institute of Biomedicine, Sahlgrenska Academy at University of Gothenburg, SE-40530 Go¨ teborg, Sweden; and Biogem IRGS (H.F., E.A.) Istituto di Ricerche Genetiche Gaetano Salvatore, Ariano Irpino 83031, Italy

Transcriptome analysis revealed that the tyrosine kinase receptor EphA4 is enriched in the thyroid bud in mouse embryos. We used heterozygous EphA4-EGFP knock-in mice in which enhanced green fluorescent protein (EGFP) replaced the intracellular receptor domain (EphA4ϩ/EGFP) to localize EphA4 protein in thyroid primordial tissues. This showed that thyroid progenitors originating in the pharyngeal floor express EphA4 at all embryonic stages and when follicles are formed in late development. Also, the ultimobranchial bodies developed from the pharyngeal pouch endoderm express EphA4, but the ultimobranchial epithelium loses the EGFP signal before it merges with the median thyroid primordium. Embryonic C cells invading the thyroid are exclusively EphA4-negative. EphA4 expression continues in the adult thyroid. EphA4 knock-out mice and EphA4-EGFP homozygous mutants are euthyroid and have a normal thyroid anatomy but display subtle histological alterations regarding number, size, and shape of follicles. Of particular interest, the pattern of follicular abnormality differs between EphA4Ϫ/Ϫ and EphA4EGFP/EGFP thyroids. In addition, the number of C cells is reduced by Ͼ50% exclusively in animals lacking EphA4 forward signaling (EphA4EGFP/EGFP). Heterozygous EphA4 mutants have no apparent thyroid phenotype. We conclude that EphA4 is a novel regulator of thyroid morphogenesis that impacts on postnatal development of the two endocrine cell lineages of the differentiating gland. In this process both EphA4 forward signaling (in the follicular epithelium) and reverse signaling mediated by its cognate ligand(s) (A- and/or B-ephrins expressed in follicular cells and C cells, respectively) are probably functionally important. (Endocrinology 152: 1154–1164, 2011)

ph receptors (originally named for the expression in an of signal effectors in the ligand-expressing cell (3). This EErythropoetin-Producing Hepatocellular carcinoma provides novel means by which cells communicate in con- cell line) and their cognate ligands, the ephrins, are in- cert with classical adhesion molecules (e.g., cadherins and creasingly recognized as regulators of embryonic organo- integrins) and external signals (viz., hormones, growth genesis and of homeostatic control in adult organs (1, 2). factors, and transmitter substances) to jointly regulate cell In mammalians this ligand-receptor system consists of behavior. Ephs and ephrins are both classified as A- and eight ephrins and 14 Eph members; Ephs therefore con- B-types based on their binding preferences. Recently, stitute the largest receptor tyrosine kinase family. The EphA–ephrinA bidirectional signaling was found to mod- common mode of action is typically bidirectional in that ify insulin secretion in adult pancreatic ␤-cells (4), pro- the activated Eph receptor mediates forward signaling viding the first compelling evidence for a local regulatory through its tyrosine kinase domain and the receptor- circuit of endocrine function involving this new class of bound ephrin transduces reverse signaling by recruitment molecules.

ISSN Print 0013-7227 ISSN Online 1945-7170 Abbreviations: EGFP, Enhanced green fluorescent protein; Eph, erythropoetin-producing Printed in U.S.A. hepatocellular; UB, ultimobranchial bodies. Copyright © 2011 by The Endocrine Society doi: 10.1210/en.2010-0232 Received February 25, 2010. Accepted December 14, 2010. First Published Online January 25, 2011

1154 endo.endojournals.org Endocrinology, March 2011, 152(3):1154–1164 Endocrinology, March 2011, 152(3):1154–1164 endo.endojournals.org 1155

Putative functions of the Eph–ephrin system in endo- Materials and Methods crine development are largely unknown; so far only Animals and genotyping growth and differentiation of the mammary gland during Mouse embryos (embryonic age: E9.5–E17.5) and adult off- puberty (5–7) and lactation (8) have been thoroughly in- ϩ/Ϫ Ϫ/Ϫ spring of EphA4 knockouts (EphA4 ; EphA4 ) and EphA4 vestigated. Most knowledge on morphogenetic traits reg- mutants with the coding sequence for the intracellular receptor ulated by Ephs and ephrins in the embryo is obtained from domain replaced by EGFP (EphA4ϩ/EGFP; EphA4EGFP/EGFP) investigations of the developing nervous system. This has were collected along with wild-type siblings (animals for breed- revealed significant contributions of either forward or re- ing kindly provided by Klas Kullander, Uppsala University). All animals were anesthetized by isoflurane and thereafter killed by verse signaling or both conjointly in many different bio- carbon dioxide or cutting the aorta (for blood sampling) before logical processes as neural progenitor cell proliferation, surgical removal of whole embryos or adult thyroid (en bloc with Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021 axonal path-finding, topographic brain region organiza- the associated tissues; e.g., trachea). All tissues were instantly tion and synapse formation, and synaptic plasticity (9). At fixed in formaldehyde before further processing (see below); embryos older than E11 were decapitated before fixation. Geno- the cellular level Eph–ephrin interactions regulate tissue typing of wild-type and mutant mice was performed on DNA patterning essentially by promoting either adhesion or re- obtained from tail biopsies, as previously described (15). Animal pulsion. Importantly, the effect of a particular Eph or eph- handling and experiments were approved by the local ethic com- rin is often limited by a restricted spatiotemporal expres- mittee at the University of Gothenburg. sion pattern. On the other hand, multiple Ephs and ephrins are often coexpressed exerting cooperative and RT-PCR analysis partly redundant functions. Tissues samples of thyroid, brain, thymus, and kidney from wild-type and EphA4-null adult (Ͼ8 weeks old) mice were col- The thyroid develops from different primordia in the lected and immediately submersed in RNA later solution (Am- foregut endoderm. Budding, migration, and fusion of the bion, Applied Biosystems, NJ). Before homogenization, tissues median and lateral anlagen eventually forms the compos- were cut into smaller sizes and transferred into RLT buffer (Qia- ite gland (10). This process brings together progenitors of gen, CA). Upon tissue homogenization and sonication, total RNA were isolated according to the manufacture’s protocol the two endocrine cell types, the follicular cells and the C (RNeasy Mini Kit, Qiagen, CA). DNA digestion using DNase cells, to the developing thyroid. Conceptually, thyroid (Roche, Switzerland) was carried out before RT reaction to elimi- morphogenesis resembles much of a developmental pro- nate genomic DNA contamination. RT of RNA into cDNA was cess suggestive of Eph–ephrin involvement, i.e. the pri- performed using GeneAmp RNA PCR Kit (Applied Biosystems) in mordia are most likely subordinated both attraction and accordance to the manufacturer’s instructions. After completion of the reaction, 1 ␮l cDNA was used for subsequent PCR amplifica- repulsion forces to allow a controlled mergence without tion. EphA4 PCR was carried out for 30 cycles, with denaturation premature or otherwise inappropriate mixing of the dif- for 30 sec at 95 C, annealing for 30 sec at 55 C, and extension for ferent cell populations. To investigate this possibility we 30 sec at 72 C. Resulting PCR products were analyzed by electro- screened a transcriptomic database generated from the phoresis on 1% agarose gels containing ethidium bromide. Nega- tive controls were prepared using RNA plus all reagents except mouse thyroid bud and identified EphA4 to be enriched in MuLV reverse Transcriptase, resulting in no detectable DNA prod- thyroid progenitor cells. EphA4 is a multifunctional re- uct, thus indicating no DNA contamination in RNA samples. ceptor first discovered to be critical for motoneuron path- Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA finding in the spinal cord (hence, EphA4-deficient mice are was used as an internal control. Primer sequences were as follows: Ј Ј Ј characterized by a kangaroo-like gait related to a cortico- EphA4, 5 -CCGAAGCAGCCTACACTACC-3 and 5 -GCCAG- CAGTCCAGCATTAAC-3Ј. GAPDH, 5Ј-AGGCCGGTGCT- spinal tract projection defect) (11). At difference with GAGTATGTC-3Ј and 5Ј-TGCCTGCTTCACCACCTTCT-3Ј. other Eph receptors, EphA4 binds to and is activated by both A- and B-type ephrins (12), indicating a high func- Immunofluorescence microscopy tional diversity. Recent findings have established EphA4 Embryonic tissue samples were immersion-fixed with 4% as a crucial player in epithelialization of embryonic tissues formaldehyde overnight at 4 C and thereafter infiltrated with (13) and developmental growth determining final organ 30% sucrose in PBS overnight at 4 C, embedded in Tissue-Tek Ϫ ° size (14). To elucidate a putative role of EphA4 in thyroid (Sakura, Zoeterwoude, The Netherlands), and frozen at 80 C. Cryostat sections (10 ␮m on a Microm HM 500M) were col- development, we first investigated its expression pattern in lected on polylysine glass slides (Menzel-Gla¨ser, Braunschweig, heterozygous knock-in mouse embryos in which the in- Germany) for immunostaining. Paraffin-embedded (adult) spec- tracellular domain of the coexpressed mutant allele is re- imens sections (8 ␮m) were deparaffinized before HE-staining placed by enhanced green fluorescent protein (EGFP) (15). (hematoxylin; Mayers HTX; 0.2% eosin; Histolab, Go¨ teborg, Sweden) or immunolabeling. Before antibody incubations sec- In addition, thyroid phenotypes were characterized in tions were permeabilized in PBS with 0.1% (0.5% for paraffin) EGFP/EGFP EphA4-null mice and in EphA4 mutants lacking Triton X-100 (PBS/Tx) for 20 min and blocked in PBS with 2% forward signaling ability. normal donkey serum (Jackson ImmunoResearch, West Grove, 1156 Andersson et al. Role of EphA4 in Thyroid Morphogenesis Endocrinology, March 2011, 152(3):1154–1164

PA) for 1 h. Sections were then incubated with primary anti- been detected in the mouse pharyngeal endoderm adjacent bodies overnight at 4 C and, after washing, secondary anti- to the cardiogenic mesoderm (18). In an ongoing tran- bodies for1hatroom temperature (all in blocking buffer). scriptomic study on the developing thyroid, several Antibodies/conjugates used were as follows: rat mAb anti–E- cadherin (ECCD-2) (Calbiochem, Darmstadt, Germany), rab- ephrins (Efna2, Efna5, Efnb1, and Efnb2) and Eph recep- bit pAb anti-Titf1/Nkx2.1 (Biopat, Milan, Italy), rabbit pAb tors (Epha4, Ephb2, and Ephb3) were found to be ex- anti-calcitonin (Dako, Glostrup, Denmark), biotin-conju- pressed in the thyroid bud at E10.5 (Fagman and Amen- gated donkey antirat IgG (Jackson ImmunoResearch), Rho- dola, unpublished observations). In particular, Efna5 and damine Red–X-conjugated donkey antirabbit F(abЈ) frag- 2 Epha4 transcripts were enriched in the thyroid bud com- ments (Jackson ImmunoResearch), and streptavidin-FITC (Dako). Immunofluorescence was evaluated in a Bio-Rad Ra- pared with other embryonic tissues. The availability of diance 2000 Laser Scanning Microscope. Image J software EphA4 null and EphA4EGFP/EGFP (kinase dead) mutant (accessible from http://rsbweb.nih.gov/ij/) and Adobe Photo- mice enabled us to investigate putative roles of EphA4 in Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021 shop CS were used for imaging. Fluorescence of endogenous thyroid development. EphA4-EGFP was analyzed in coimmunostained specimens.

Light microscopic morphometry EphA4 expression in embryonic thyroid Serial transverse sections of thyroid lobes from adult (age: development 2–18 months) wild-type and mutant mice were either HE-stained In conformity with previous studies (15), EphA4-EGFP for quantification of follicles (number and size) or immunola- was found to be readily expressed and detectable with beled for counting of C cells expressing calcitonin. Lobe size was ϩ/EGFP estimated from the total number of sections and the maximal confocal laser-scanning microscopy in EphA4 em- lobe diameter. Follicle geometries were measured using the bryos. As the wild-type and mutant alleles are known to be BioPix iQ software (BioPix AB, Gothenburg, Sweden) as previ- coexpressed (15), it is reasonable to assume that the EGFP ously described (16). Analysis comprised standardized lobe lev- signal safely reproduces the natural expression pattern of els (the middle-most section and two additional sections at half the distance between the lobe equator and the superior and in- EphA4 protein. In heterozygous E9.5 embryos, EphA4- ferior lobe poles, respectively). Images imported to BioPix iQ EGFP was ubiquitously expressed in the anterior foregut were calibrated for recognition of follicles followed by automatic endoderm outlining the entire cell–cell contacts of the measurements of individual follicle size (estimated by the lumen endoderm epithelium (Fig. 1A). A similar distribution area), total follicle area, and total follicle number. Obtained data of EphA4-EGFP was observed in Titf1/Nkx2.1 positive were grouped in size categories as detailed in Fig. 4. The distribution of C cells in the adult thyroid was estimated by counting thyroid progenitor cells gathered in the pharyngeal the number of sections exhibiting calcitonin immunoreactivity floor to form the thyroid placode (pre-bud stage) (Fig. relative to the total number of sections per lobe. In addition, the 1B). The thyroid progenitors continued to express number of calcitonin-positive cells was counted in every ten sec- EphA4-EGFP during subsequent developmental stages ϭ Ϫ/Ϫ ϭ EGFP/EGFP tions (wild-type: n 14, EphA4 :n 12, EphA4 : (i.e., after the thyroid bud had pinched off from the n ϭ 12) after subtraction of auto-fluorescent blood cells. Total C cell number was estimated by assuming that uncounted sec- endoderm) (at E11.5; Fig. 1C) and also when downward tions display even changes of C cell number according to the migration was completed and the primordium extended ϩ ϩ ϩ ϩ ϩ ϩ ϫ formula: [(X1 X2)/2 (X2 X3)/2 (X3 X4)/2 etc.] bilaterally as a result of intense cell proliferation (at 10, in which X indicates the number of C cells encountered in E12.5; Fig. 1D). In all instances adjacent mesoderm was consecutive counted sections. Images for this analysis were cap- EphA4-EGFP–negative. tured in a Nikon eclipse 1000 microscope and further handled by Adobe Photoshop CS. The ultimobranchial bodies (UB) constitute the lateral thyroid anlagen that by fusion with the midline primor- Hormone analysis dium bring C cell precursors to the prospective thyroid Whole blood collected from adult wild-type and EphA4 mu- gland (10). These structures are also endodermal and de- tant mice was allowed to clot for 60 min at room temperature and velop from the fourth pharyngeal pouches close to the ϫ centrifuged for 10 min at 2000 g, after which sera were saved origin of the thymus and parathyroid rudiments. We and stored at Ϫ20 C. Total serum levels of T4 and T3 were found that EphA4-EGFP was uniformly expressed in the measured by ELISA (Alpha Diagnostic, TX). budding epithelium in early UB development (Fig. 1E). However, as the UB delaminated from the pouch Results endoderm and grew larger the EGFP signal became weaker and even undetectable in parts of the UB epithe- There is sparse information on Eph receptors and ephrins lium (Fig. 1F). Moreover, at E13.5 when the UB started to in the endoderm and its budding organs during embryo- merge with the median thyroid primordium EphA4-EGFP genesis. So far Ephb3 has been found to be transcribed in was no longer detected in any UB cells (Fig. 1G). At the the floor of the chick foregut (17), and Efna1 transcript has same time EphA4-EGFP was strongly expressed in pro- Endocrinology, March 2011, 152(3):1154–1164 endo.endojournals.org 1157

2C). Immunolabeling for calcitonin showed that differentiated C cells were all EphA4- EGFP negative (Fig. 2D). Notably, the C cells did not spread randomly in the stromal compartment but seemingly adhered to the parenchymal cords of EphA4-positive epithelium even though distinct follicle profiles were not yet discerned (Fig. 2D).

Function of EphA4 in postnatal thyroid

development Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021 The consistent expression of EphA4 in thyroid progenitor cells derived from the median anlage suggests that EphA4 may be a novel regulator of embryonic thyroid development. However, several lines of evidence indicated that EphA4 is dispensable for the major thyroid developmental traits (i.e., there were no signs of thyroid dysgenesis (data not shown) and folliculogenesis and expression of Tift1/ Nkx2.1 and thyroglobulin were not affected in EphA4 deficient late embryos (Fig. 2, E and F). Thus, either EphA4 is not required for the growth and differentiation of the embryonic thyroid or is a putative involvement supplied FIG. 1. Expression of EphA4-EGFP (truncated EphA4 linked to enhanced green fluorescent protein) in thyroid primordial tissues of heterozygous (EphA4ϩ/EGFP) by redundant actions of coexpressed Ephs. mouse embryos. Thyroid progenitor cells were identified by coimmunostaining for Because of a confounding high background the thyroid transcription factor Titf1/Nkx2.1 (red fluorescence). A, EphA4 expression of fluorescence in EphA4-EGFP adult tissues it in the pharyngeal endoderm (e) at E9.5. Sagittal section of embryo. The anterior part of the thyroid placode (tp) is included. B, EphA4 in the thyroid placode (high was not possible discern with certainty EphA4 magnification of the same tissue section imaged in A). The epithelium is protein expression in the adult thyroid. We pseudostratified indicative of initial bud formation. Note that EGFP decorates the therefore analyzed tissue homogenates for entire cell–cell contacts. C, EphA4 in midline thyroid primordium (th) at E11.5 EphA4 mRNA expression by RT-PCR (Fig. 3). representing the migration stage after budding [i.e., the bud is detached from the endoderm (not shown)]. Sagittal section. Surrounding mesenchyme (unstained) does EphA4 transcript was found to be present in not express EphA4 (*). D, EphA4 in thyroid progenitors at the proliferation stage the mouse thyroid at levels comparable to that (E12.5) when the midline primordium grows bilaterally. Frontal section. E, EphA4 of the thymus and slightly less than found in the expression in the ultimobranchial body (ub) epithelium (viz., lateral thyroid anlagen) that buds from the fourth pharyngeal pouch (not shown) at E11.5. Transverse brain and kidney in which EphA4 is known to section. Note that EGFP delineates the basolateral cell surfaces but is absent apically be abundant. This suggests that thyrocytes toward the lumen (arrow). F, EphA4 in the ultimobranchial body (ub) after continue to express EphA4 after birth and on- delamination from the pouch endoderm (E12.5). Transverse section. The cells that are increased in number compared with E11.5 (see F) exhibit either strong or weak wards to adulthood. (*) EGFP signals. G, EphA4 expression in thyroid progenitor cells after fusion of Thymus is previously known to be hypocel- ultimobranchial body (ub) with the midline thyroid anlage (th) at E13.5. Transverse lular in EphA4-null mice (19). It was therefore section. Note that the EGFP signal is completely absent in the ultimobranchial body of interest to investigate the adult thyroid of remnant. Scale bars,50␮m. EphA4 mutants for possible changes at the cellular level. As shown in Fig. 3, A–C, the gross genitor cells derived from the median anlage as these in- anatomy of the gland and the histoarchitecture at large vested the UB (Fig. 1G). were not different between wild-type, EphA4-deficient, In late development (E17.5–18.5) EphA4-EGFP ex- EGFP/EGFP pression was evident in most of the thyroid lobe paren- and EphA4 animals. However, whereas wild- chyma (Fig. 2A). The EGFP signal was confined to assem- type follicular cells were mostly cuboidal (Fig. 4D), the blies of follicular progenitors and polarized thyrocytes follicular epithelium of EphA mutant mice often consisted forming immature follicles (Fig. 2B). In contrast, central of flat cells characterized by an elongated nucleus and a parts of the lobes, which at this stage still contained UB very thin apical cytoplasm (Fig. 4, E and F). Interestingly, remnants (10), were largely devoid of EphA4-EGFP (Fig. abnormal follicular cell shape was most conspicuous in 1158 Andersson et al. Role of EphA4 in Thyroid Morphogenesis Endocrinology, March 2011, 152(3):1154–1164

Similarly, the total lumen area reflecting the lumen volume of all encountered follicles was not significantly altered in the EphA4 mutants (Fig. 5B). Together, these findings suggest that EphA4 influences the generation and maturation of follicles as the gland grows, however without changing the final size of the gland. Circulating thyroid hormone levels were not significantly altered in EphA4 mutant mice [wild-type: T3 104 Ϯ 40 ng/dl (n ϭ 8) and T4 Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021 6.14 Ϯ 2.27 ␮g/dl (n ϭ 10); EphA4Ϫ/Ϫ:T3 101 Ϯ 27.7 ng/dl (n ϭ 4) and T4 6.9 Ϯ 2.2 ␮g/dl (n ϭ 4); EphA4EGFP/EGFP:T3112Ϯ 6 ng/dl (n ϭ 4) and T4 5.7 Ϯ 0.5 ␮g/dl (n ϭ 4)], indicating that the observed follicular phenotypes were not accompanied by overt hypothyroidism. As C cells appearing in the embryonic thyroid were lacking EphA4, it was of interest to investigate whether their number and distribution were at all affected by deletion or functional impairment (EphA4-EGFP homozygos- ity) of EphA4 expressed in the follicular epithelium. Calcitonin-positive cells were counted in serial sections of the thyroid in late development (E17.5) and adult mice. The total C cell number was Ͻ400 in the embryonic and Ͼ25000 in the adult gland, indicating that most C cells in all probability were generated after FIG. 2. EphA4-EGFP expression in late thyroid development (E17.5–18.5). As for images in Fig. 1, EphA4 was detected by EGFP in EphA4ϩ/EGFP mice. All images are birth. No differences were observed between from transverse section of embryo. A, Overview of prospective thyroid lobe wild-type and EphA4-null adult mice (Fig. 6, up- indicating that the prefollicular epithelium uniformly expresses EphA4. B, Detailed per and middle panels); the estimated number of view of thyroid parenchyma indicating basolateral expression of EphA4 in newly differentiated C cells was for EphA4ϩ/ϩ 26026 Ϯ formed follicles (encircled). C, Central lobe portion crowded with EphA4 negative Ϫ/Ϫ cells (*) of presumed ultimobranchial origin. Note that all parenchymal cells express 1670 (n ϭ 6) and for EphA4 27590 Ϯ 6000 Titf1/Nkx2.1 (red). D, Differentiating C cells with calcitonin accumulated in the (n ϭ 4). However, in EphA4EGFP/EGFP mutants cytoplasm (red) are present close to EphA4-positive parenchyma (green). Unlike the C cell count was only 42% [11020 Ϯ 827 follicular progenitors, the C cells are devoid of AphA4-EGFP in the plasma ϭ membrane (arrows). E, Titf1/Nkx2.1 expression in thyroid follicular cells in EphA4- (n 4); Fig. 6, bottom panel] of the normal, deficient E18.5 embryos. F, Accumulation of thyroglobulin (TG) in follicle lumina and C cells reaching the peripheral parts of Ϫ/Ϫ (arrows) of the EphA4 embryonic thyroid. Scale bars,50␮m. the thyroid lobes were few in these mutants. For comparison, calcitonin-positive cells EGFP/EGFP EphA4 glands (Fig. 4E). Computer-based mor- present in the embryonic thyroid did not differ in num- phometric calculation of individual follicle lumen areas in ber between wild-type (n ϭ 368, mean of two) and serially sectioned thyroids further showed that small-sized EphA4EGFP/EGFP (n ϭ 392, mean of two). This suggests 2 follicles (Ͻ1000 ␮m ) were significantly more frequent in that the generation of C cells during postnatal thyroid Ϫ Ϫ EphA / than in wild-type or EphA4EGFP/EGFP animals growth is modulated by EphA4 forward signaling in the (Fig. 5, A and C). The supernumerary small follicles seem- follicular epithelial cells. ingly appeared on the expense of larger ones. In contrast, very large follicles (Ͼ3000 ␮m2) typically located in the dorsolateral portion of the lobes were present in wild-type Discussion and even more frequently in EphA4EGFP/EGFP glands (Fig. 5, A and C). Intermediate-sized follicles (1000–3000 ␮m2) Transcriptome analysis revealed that EphA4 is enriched in did not differ in number between genotypes (Fig. 5C). the mouse thyroid bud compared with other embryonic Endocrinology, March 2011, 152(3):1154–1164 endo.endojournals.org 1159

EGFP heterozygotes in all probability can be safely used for localization of endogenous EphA4 is further supported by our present observations of a normal morphology in the developing thyroid and other embryonic tissues expressing the construct. We found that EphA4 is ubiquitously expressed in the pharyngeal endoderm, including the thyroid primordia. This suggests that EphA4 expression is a general feature of foregut endoderm cells and therefore likely not implicated in thyroid specification, also supported by the finding of a

normal-sized thyroid bud in EphA4-deficient embryos. Of Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021 FIG. 3. EphA4 transcription in adult mouse thyroid. RT-PCR analysis particular interest, thyroid progenitors destined to a fol- was used to determine mRNA expression of EphA4 in tissue licular fate continued to express EphA4 throughout de- homogenates from thyroid, brain, thymus, and kidney in wild-type and velopment, whereas EphA4 was gradually down-regu- EphA4-null mice. GAPDH mRNA was used as an internal control. lated in the UB epithelium before the fusion of the two anlagen. The formation of a composite gland from merg- tissues. In lack of a reliable EphA4 antibody we used ing primordia was evidently not affected in EphA4 mutant EphA4-EGFP for in vivo mapping of EphA4 protein ex- embryos. However, loss of function studies may not be pression in primordial thyroid tissues. The EphA4-EGFP expected to reveal phenotypes when down-regulation of knock-in mouse was originally created for functional in- putative regulatory proteins is part of the normal process. vestigation of EphA4 in the brain (15). This showed that A possible mechanism requiring loss of EphA4 may nev- the expression of Epha4 is normally bi-allelic and that ertheless be suggested based on current understanding of EphA4-EGFP is coexpressed with EphA4 in heterozygous the biological functions of Ephs in general. Eph–ephrin (EphA4ϩ/EGFP) animals. Moreover, the mutant EphA4 re- interactions conceptually promote either adhesion or receptor, which thus lacks the entire intracellular domain pulsion depending on the cell types and spatiotemporal including the tyrosine kinase, rescues some of the func- features involved (3). It is conceivable that similar dy- tional defects observed in EphAϪ/Ϫ brains (15). Together namic changes may take place in the embryonic thyroid this indicates that EphA4-EGFP faithfully recapitulates when the midline anlage and the paired UB approach each the natural expression of EphA4 and is able to bind to and other and the different cell populations gradually mix. activate cognate ephrin ligands to EphA4. That EphA4- Interestingly, Eph receptor internalization promotes ter- mination of adhesion in favor of repulsion (20). Moreover, it was recently found that the turnover of EphA4 is regulated by endocytosis (21). Clearance of EphA4 from the cell surface along with loss of expression in the entire UB might thus be one of probably several molecular events that govern proper entry of C cell precursors into the prospective thyroid. Adult EphA4-null mice and EphA4 mutants lacking the intracellular receptor domain ex- hibited subtle but distinct thyroid phenotypes: small sized follicles were relatively predomi- nant in EphA4Ϫ/Ϫ, whereas very large follicles were most frequent in EphA4-EGFP homozy- gotes. In addition, the follicular epithelium was more flattened than in the wild type, in particular in mice expressing the truncated receptor. FIG. 4. Cross-sectioned thyroid gland in adult normal and EphA4 mutant mice. A– C, Overview of thyroid lobes (*) present on both sides of the trachea (t) in wild-type Normal circulating thyroid hormone levels as (WT), EphA4 deficient (EphA4Ϫ/Ϫ), and EphA4-EGFP homozygous (EphA4EGFP/EGFP) observed do not exclude the possibility of dys- animals. es, esophagus. Scale bar, 1000 ␮m. D–F, Follicular architecture of the regulated thyroid function in EphA4 mutants. thyroid in corresponding lobes. Note that the follicular epithelium is frequently In fact, a flat epithelium typically designates abnormally flat consisting of cells with elongated nuclei (arrows) in mice lacking functional EphA4, at difference with the wild-type cuboidal epithelium. Scale bar, the inactive gland for example in animals in 100 ␮m. which endogenous TSH is suppressed by high 1160 Andersson et al. Role of EphA4 in Thyroid Morphogenesis Endocrinology, March 2011, 152(3):1154–1164 Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021

FIG. 5. Morphometric analysis of thyroid follicles in wild-type and EphA4-mutant adult mice. Quantitative measurements were conduced on HE- stained serial sections of thyroid lobes using the BioPix software (for detailed see Materials and Methods). A, Size distribution of follicles. For each genotype, the vertical columns indicate number of follicles per size category ranging from Ͻ100 to 12600 ␮m2 (lumen area); more than one follicle per size category is indicated by increased shading. Numbers of animals included the following: wild-type (WT; n ϭ 6), EphA knockout Ϫ Ϫ (A4 / ;nϭ 4), and EphA4-EGFP knock-in (A4EGFP/EGFP;nϭ 6). B, Total area of follicles per lobe in A. Mean Ϯ SD (n ϭ number of animals for each genotype same as indicated in A). C, Total number of small, medium-sized, and large follicles in A. Follicles were divided into three size categories (lumen area Ͻ1000 ␮m2, 1000–3000 ␮m2, and Ͼ3000 ␮m2, respectively) and data were statistically evaluated. Mean Ϯ SD (n ϭ number for each genotype indicated in A); asterisks indicate statistically significant differences (*, P Ͻ 0.05; **, P Ͻ 0.01) between bracketed groups. doses of thyroxin. Ongoing studies in our laboratory aim predominate in the central/medial portion of the lobes, to elucidate whether EphA4 might be a novel modifier of whereas large-sized follicles are mostly located close to the thyroid function analogous to that recently shown for gland capsule. Although not experimentally validated, this ␤ EphA5 in pancreatic -cells regulating glucose-dependent topographic heterogeneity probably reflects a gradual gen- production of insulin (4). Further studies are also needed to eration and maturation of new follicles as the gland grows to better understand the mechanism by which follicle size is its adult size. The characteristic follicular phenotypes of differently affected in EphA4 null and EphA4 kinase dead EphA4 mutants strongly suggest that this process is modu- mutants. Nevertheless, the present observations likely reflect distinct roles of, on one hand, EphA4-mediated forward sig- lated by EphA4, presumably through reciprocal interactions naling and, on the other hand, reverse signaling elicited by with coexpressed ephrins (see Fig. 7, A and B, for a working cognate ephrins to which EphA4 binds. model). This hypothesis gains support from previous reports It is well known that follicle size varies considerably be- on the involvement of EphA-type receptor signaling in epi- tween different regions of the normal thyroid: small follicles thelial morphogenesis (6, 22) Of particular interest, EphA4 Endocrinology, March 2011, 152(3):1154–1164 endo.endojournals.org 1161 Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021

FIG. 6. Quantification of C cell numbers in normal and EphA4-mutant adult mice. The thyroids were serially sectioned and immunostained for calcitonin. Calcitonin-positive cells were counted in every ten sections comprising in total 15–17 levels in one lobe per animal (number of animals included were as follows: WT; n ϭ 6; EphA4Ϫ/Ϫ;nϭ 4; EphA4EGFP/EGFP;nϭ 4). Number of C cells is indicated for each section level and genotype (left). Representative micrographs on the tissue distribution of calcitonin-positive cells are shown for each genotype (right). Scale bar, 100 ␮m. interacting with ephrinB2 was recently shown to promote EphA4-null mice had a normal C cell count. A more severe epithelialization and establishment of cell polarity in the pre- phenotype in mice expressing a truncated EphA4 receptor somitic mesoderm, thereby creating boundaries of impor- has previously been recognized for the involvement of tance for somite segmentation (13). In principal, this seg- EphA4 in neuronal development (23). A plausible expla- mentation process resembles distinct steps of thyroid nation to this is that signal-deficient EphA4 may exert a morphogenesis in which the follicular progenitors first form dominant negative effect on redundant mechanisms, pos- papilla-like epithelial cords that subsequently reorganize into sibly by obstructing functionally important cis interac- multiple premature follicles (10). A novel function of Eph– tions among coexpressed Eph receptors (23). This con- ephrin intercellular signals might thus be to coordinate fol- forms with the suggested paradox of Eph signaling (3), licular histogenesis in the developing thyroid by regulating implicating that global deficiency of a particular Eph will epithelial organization and cell shape. be fully compensated for by the action of closely related Homozygous mice carrying the EphA4-EGFP con- Ephs but that partial deletion or mutation of the same struct were diminished of thyroidal C cells, whereas molecule will block redundancy even if expressed only in 1162 Andersson et al. Role of EphA4 in Thyroid Morphogenesis Endocrinology, March 2011, 152(3):1154–1164

There is ample evidence in support of a growth-stim- ulating effect of Eph receptors (3). This may be achieved either directly by forward signal activation of, for example, the Ras/MAPK pathway or indirectly by forming mul- timeric complexes with peptide growth factor receptors leading to their activation. As a result, Eph receptor mu- tations often lead to reduced cellularity in otherwise normal-shaped organs. This is evident also for EphA4-deficient animals in which hypoplasia has been reported for the thymus (19) and brain (24). Notably, EphA4 has been shown to regulate motor axon guidance during limb de- Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021 velopment in cooperation with Ret (25), a tyrosine kinase receptor proto-oncogene that is also expressed in embryonic and juvenile C cells (26). Interestingly, the thyroid C cell number is reduced by more than 50% in Ret-deficient mice (27). A plausible hypothesis is therefore that EphA4 might act in concert with Ret to promote proliferation of the C cell lineage. A common problem in morphogenesis is to discrimi- nate between cell proliferation and migration as both pro- FIG. 7. Modeling of proposed interactions and suggested functions of cesses, although conveyed by distinct intracellular signal- EphA4 in the thyroid based on findings of the present study. A, ing pathways, often take place concomitantly. For Reciprocal binding of EphA4 and yet uncharacterized ephrins in the thyroid follicular epithelium likely mediates both forward and reverse example, perturbed positioning of stem cells in the intes- signaling in a cell-autonomous manner (i.e., the interaction involves tinal crypts of EphB2/B3 mutant mice obscures a genuine the same cell-type). Altered follicle structure and number in EphA4- growth-promoting effect of EphB2 on mucosal progeni- mutant mice indicate a modulating role in thyroid morphogenesis. This may be accomplished by EphA4 signaling per se or indirectly through tors (28). The restricted C cell distribution in EphA4- EphA4-mediated activation of putative growth factor receptors. B, In EGFP homozygous thyroids may thus be secondary to a prospective thyroid lobes the proliferating thyroid progenitors first decreased proliferation rate. However, this does not rule form papillary cords from which follicles subsequently develop (see also Ref. 10). It is proposed that folliculogenesis taking place postnatally out that C cell migration is also influenced. Studies by may be imprinted by parenchymal rearrangement in late embryonic others have shown that the tyrosine kinase activity is development and that this is modified by EphA4 in progenitor cells. C, mostly required for Eph receptors to promote cell migra- EphA4 is expressed at the basal surface of the thyroid progenitors to which initerstital C cells seemingly adhere. Conceivably, C cell ephrins tion (3). Interestingly, the cellular response to one partic- bind to and activate EphA4 in the follicular progenitors. The phenotype ular Eph may shift from repulsion to adhesion depending of EphA44EGFP/EGFP mice suggests that EphA4 forward signaling on whether the kinase domain is activated or not (29). This promotes C cell proliferation and possibly also the ability of C cells to opens up the possibility of alternating adhesion and re- invade the thyroid. A putative role of reverse signaling by activated ephrins in this process is yet unknown. D, Differentiated C cells pulsion conducted by molecular switches of Eph–ephrin migrate radially along the cords of prospective follicular cells in the bidirectional signaling in adjacent cells. We observed that embryonic thyroid. It is suggested that heterotypic cell–cell interactions C cells invading the embryonic thyroid did not scatter are required for the C cells to invade successfully and reach also the peripheral parts of the gland. EphA4 likely participates in this process, randomly in the stromal compartment but were intimately implicating a noncell autonomous mode of action. associated with the follicular progenitors as these orga- nized into a trabecular network before the formation of a subpopulation of cells. From a mechanistic point of follicles. Juxtaposition of the two cell types was evident as view, it is important to note that EphA4 is not expressed both cell populations increased in number and the lobes in the C cells themselves. A novel noncell autonomous were enlarged. Such a close apposition of cells conceivably mode of action is thus suggested for EphA4 expressed in reflects binding of cell surface molecules implicated in in- the thyroid follicular epithelium promoting propagation tercellular adhesion. If Eph–ephrin interactions are part of of the C cell lineage. Hypothetically, this may be achieved this, activation of EphA4 in the presumptive follicular ep- by reverse signaling of activated cognate ephrin(s) on C ithelium may feed back on the behavior of the C cells cells that bind to EphA4 in the thyrocyte plasma mem- leading to radial migration (Fig. 7D). brane, EphA4 forward signal-mediated release of soluble An interesting issue is whether the poor generation of C factors that exert a paracrine effect on the C cells, or a cells in EphA4EGFP/EGFP mice is entirely a postnatal effect combination of the two (Fig. 7C). or whether a putative primary defect during embryonic life Endocrinology, March 2011, 152(3):1154–1164 endo.endojournals.org 1163 contributes to the phenotype. The fact that the number of 2. Miao H, Wang B 2009 Eph/ephrin signaling in epithelial develop- calcitonin-positive cells encountered in late development ment and homeostasis. Int J Biochem Cell Biol 41:762–770 EGFP/EGFP 3. Pasquale EB 2005 Eph receptor signalling casts a wide net on cell was not different in wild-type and EphA4 mu- behaviour. Nat Rev Mol Cell Biol 6:462–475 tant embryos does not exclude the possibility of a reduced 4. Konstantinova I, Nikolova G, Ohara-Imaizumi M, Meda P, Kucera number of C cell precursors invading the embryonic thy- T, Zarbalis K, Wurst W, Nagamatsu S, Lammert E 2007 EphA- Ephrin-A-mediated beta cell communication regulates insulin secre- roid. First of all, the total number of differentiated C cells tion from pancreatic islets. Cell 129:359–370 at E17.5 was exceedingly low compared with the adult 5. Munarini N, Jager R, Abderhalden S, Zuercher G, Rohrbach V, gland, which makes this unsafe to consider proportional Loercher S, Pfanner-Meyer B, Andres AC, Ziemiecki A 2002 Altered to the precursor number. In fact, because of the lack of a mammary epithelial development, pattern formation and involution in transgenic mice expressing the EphB4 receptor tyrosine kinase. suitable biomarker before calcitonin is expressed, there is J Cell Sci 115:25–37 no information on how many C cell precursors are orig- 6. Vaught D, Chen J, Brantley-Sieders DM 2009 Regulation of mam- Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021 inally specified in early development. Neither is it known mary gland branching morphogenesis by EphA2 receptor tyrosine kinase. Mol Biol Cell 20:2572–2581 whether an undifferentiated C cell precursor pool exists in 7. Haldimann M, Custer D, Munarini N, Stirnimann C, Zurcher G, the thyroid after birth. In this discussion, it is relevant to Rohrbach V, Djonov V, Ziemiecki A, Andres AC 2009 Deregulated also consider the embryonic origin of C cells. Classical ephrin-B2 expression in the mammary gland interferes with the development of both the glandular epithelium and vasculature and quail-chick transplantation studies by LeDourain and col- promotes metastasis formation. Int J Oncol 35:525–536 leges (30, 31) showed that avian C cell precursors destined 8. Weiler S, Rohrbach V, Pulvirenti T, Adams R, Ziemiecki A, Andres for the ultimobranchial glands (C cells do not enter the AC 2009 Mammary epithelial-specific knockout of the ephrin-B2 thyroid in birds) originate from the neural crest. Although gene leads to precocious epithelial cell death at lactation. Dev Growth Differ 51:809–819 not experimentally validated, C cells in higher vertebrates 9. Klein R 2009 Bidirectional modulation of synaptic functions by are also believed to be neuroectodermal. Previous findings Eph/ephrin signaling. Nat Neurosci 12:15–20 of a disturbed migration of neural crest cells into the pha- 10. Fagman H, Andersson L, Nilsson M 2006 The developing mouse thyroid: embryonic vessel contacts and parenchymal growth pattern ryngeal arches of EphA4 kinase-dead mutants (32, 33) during specification, budding, migration, and lobulation. Dev Dyn therefore suggest the possibility of a neural crest defect 235:444–455 leading to a diminutive C cell precursor pool and eventu- 11. Dottori M, Hartley L, Galea M, Paxinos G, Polizzotto M, Kilpatrick ally reduced number of C cells. Whether undifferentiated T, Bartlett PF, Murphy M, Kontgen F, Boyd AW 1998 EphA4 (Sek1) receptor tyrosine kinase is required for the development of the cor- precursors or mature C cells are the source of C cell lineage ticospinal tract. Proc Natl Acad Sci USA 95:13248–13253 propagation during postnatal thyroid growth thus re- 12. Durbin L, Brennan C, Shiomi K, Cooke J, Barrios A, Shanmugal- mains an open question. ingam S, Guthrie B, Lindberg R, Holder N 1998 Eph signaling is required for segmentation and differentiation of the somites. Genes Dev 12:3096–3109 13. Watanabe T, Sato Y, Saito D, Tadokoro R, Takahashi Y 2009 Acknowledgments EphrinB2 coordinates the formation of a morphological boundary and cell epithelialization during somite segmentation. Proc Natl We thank Klas Kullander, Department of Neuroscience, Uppsala Acad Sci USA 106:7467–7472 14. Munoz JJ, Garcia-Ceca J, Alfaro D, Stimamiglio MA, Cejalvo T, University, for providing EphA4 mutant mice and valuable ad- Jimenez E, Zapata AG 2009 Organizing the thymus gland. Ann NY vice at the start of this project, and the CCI facility, Sahlg- Acad Sci 1153:14–19 renska Academy at the University of Gothenburg, for techni- 15. Grunwald IC, Korte M, Adelmann G, Plueck A, Kullander K, Adams cal assistance using the Bio-Rad Radiance 2000 Laser RH, Frotscher M, Bonhoeffer T, Klein R 2004 Hippocampal plas- Scanning Microscope. ticity requires postsynaptic ephrinBs. Nat Neurosci 7:33–40 16. Andersson L, Bostrom P, Ericson J, Rutberg M, Magnusson B, Marchesan D, Ruiz M, Asp L, Huang P, Frohman MA, Boren J, Address all correspondence and requests for reprints to: Olofsson SO 2006 PLD1 and ERK2 regulate cytosolic lipid droplet Mikael Nilsson, Department of Medical Biochemistry and Cell formation. J Cell Sci 119:2246–2257 Biology, Institute of Biomedicine, Sahlgrensk Academy at Uni- 17. Baker RK, Vanderboom AK, Bell GW, Antin PB 2001 Expression versity of Gothenburg, Box 440, SE-40530 Go¨ teborg, Sweden. of the receptor tyrosine kinase gene EphB3 during early stages of E-mail: [email protected]. chick embryo development. Mech Dev 104:129–132 This work was supported by Swedish Research Council 18. McBride JL, Ruiz JC 1998 Ephrin-A1 is expressed at sites of vascular development in the mouse. Mech Dev 77:201–204 (grant no 537-42-3), Swedish Cancer Foundation (Grant No. 19. Munoz JJ, Alfaro D, Garcia-Ceca J, Alonso CL, Jimenez E, Zapata 09-0464) and Sahlgrenska University Hospital (ALF). A 2006 Thymic alterations in EphA4-deficient mice. J Immunol Disclosure Summary: The authors have nothing to declare. 177:804–813 20. Zimmer M, Palmer A, Kohler J, Klein R 2003 EphB-ephrinB bidirectional endocytosis terminates adhesion allowing contact mediated repulsion. Nat Cell Biol 5:869–878 References 21. Deininger K, Eder M, Kramer ER, Zieglgansberger W, Dodt HU, Dornmair K, Colicelli J, Klein R 2008 The Rab5 guanylate exchange 1. Pasquale EB 2008 Eph-ephrin bidirectional signaling in physiology factor Rin1 regulates endocytosis of the EphA4 receptor in mature and disease. Cell 133:38–52 excitatory neurons. Proc Natl Acad Sci USA 105:12539–12544 1164 Andersson et al. Role of EphA4 in Thyroid Morphogenesis Endocrinology, March 2011, 152(3):1154–1164

22. Miao H, Nickel CH, Cantley LG, Bruggeman LA, Bennardo LN, 28. Holmberg J, Genander M, Halford MM, Anneren C, Sondell M, Wang B 2003 EphA kinase activation regulates HGF-induced epi- Chumley MJ, Silvany RE, Henkemeyer M, Frisen J 2006 EphB re- thelial branching morphogenesis. J Cell Biol 162:1281–1292 ceptors coordinate migration and proliferation in the intestinal stem 23. Dufour A, Egea J, Kullander K, Klein R, Vanderhaeghen P 2006 cell niche. Cell 125:1151–1163 Genetic analysis of EphA-dependent signaling mechanisms control- 29. Holmberg J, Clarke DL, Frisen J 2000 Regulation of repulsion ver- ling topographic mapping in vivo. Development 133:4415–4420 sus adhesion by different splice forms of an Eph receptor. Nature 24. North HA, Zhao X, Kolk SM, Clifford MA, Ziskind DM, Dono- 408:203–206 ghue MJ 2009 Promotion of proliferation in the developing cerebral 30. Le Douarin N, Fontaine J, Le Lievre C 1974 New studies on the cortex by EphA4 forward signaling. Development 136:2467–2476 neural crest origin of the avian ultimobranchial glandular cells– 25. Kramer ER, Knott L, Su F, Dessaud E, Krull CE, Helmbacher F, interspecific combinations and cytochemical characterization of C Klein R 2006 Cooperation between GDNF/Ret and ephrinA/EphA4 cells based on the uptake of biogenic amine precursors. Histochem- signals for motor-axon pathway selection in the limb. Neuron 50: istry 38:297–305 35–47 31. Polak JM, Pearse AG, Le Lievre C, Fontaine J, Le Douarin NM 1974

26. Lindfors PH, Lindahl M, Rossi J, Saarma M, Airaksinen MS 2006 Immunocytochemical confirmation of the neural crest origin of Downloaded from https://academic.oup.com/endo/article/152/3/1154/2457430 by guest on 23 September 2021 Ablation of persephin receptor glial cell line-derived neurotrophic avian calcitonin-producing cells. Histochemistry 40:209–214 factor family receptor alpha4 impairs thyroid calcitonin production 32. Smith A, Robinson V, Patel K, Wilkinson DG 1997 The EphA4 and in young mice. Endocrinology 147:2237–2244 EphB1 receptor tyrosine kinases and ephrin-B2 ligand regulate tar- 27. Lindahl M, Timmusk T, Rossi J, Saarma M, Airaksinen MS 2000 geted migration of branchial neural crest cells. Curr Biol 7:561–570 Expression and alternative splicing of mouse Gfra4 suggest roles in 33. Kuriyama S, Mayor R 2008 Molecular analysis of neural crest mi- endocrine cell development. Mol Cell Neurosci 15:522–533 gration. Philos Trans R Soc Lond B Biol Sci 363:1349–1362