• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Apr 3, 2007; 104(14): 5919–5924.
Published online Mar 26, 2007. doi:  10.1073/pnas.0607779104
PMCID: PMC1851592
Developmental Biology

Crucial roles of Foxa2 in mouse anterior–posterior axis polarization via regulation of anterior visceral endoderm-specific genes

Abstract

Anterior visceral endoderm (AVE) plays essential roles with respect to anterior–posterior axis development in the early mouse embryo. To assess the genetic cascade involved in AVE formation, the cis-regulatory elements directing expression of vertebrate Otx2 genes in the AVE were analyzed via generation of transgenic mice. Otx2 expression in AVE is regulated directly by the forkhead transcription factor, Foxa2. Moreover, Foxa2 is essential for expression of the Wnt antagonists, Dkk1 and Cerl, in visceral endoderm during the pre- to early streak stages; however, Foxa2 appears to be dispensable for subsequent Dkk1 expression associated with forebrain induction. Thus, we propose that Foxa2 is crucial in early anterior–posterior axis polarization in terms of regulation of expression of AVE-specific genes. These findings provide profound insights into conserved roles of Foxa2 transcription factors in anterior specification throughout the evolution of the chordate body plan.

Keywords: cis element, gene regulation, Wnt signaling, forebrain induction, Dkk1

The initial anterior–posterior (A–P) axis in mouse embryos is established before gastrulation (1). By embryonic day (E)5.5, the mouse embryo proper, consisting of epiblast and visceral endoderm (VE), develops a clear proximal–distal axis, as marked by expression of Hex in the distal VE (DVE). By E6.0, DVE cells have migrated directionally to the prospective anterior side, and the proximal markers (Cripto, Wnt3, and Nodal) have shifted to the posterior side where primitive streak formation will occur. Several lines of evidence have indicated that the anterior visceral endoderm (AVE) is required for normal anterior patterning (2, 3). Notably, this function depends on the production of Nodal and Wnt antagonists, such as Cerl, Lefty1, and Dkk1, which locally repress the posteriorizing effects of Nodal and Wnt signals (46). Moreover, these antagonists also regulate migration of DVE cells toward the prospective anterior side (79). However, little is known regarding the genetic cascade governing AVE formation, i.e., how transcription factors control antagonists and one another in AVE cells in a spatially and temporally specific manner.

The Otx2 gene, a paired-like class homeobox gene, plays critical roles in the generation and function of AVE (10). Otx2 is expressed in the DVE at E5.5 and subsequently in the AVE. In the absence of Otx2, DVE cells fail to migrate toward the anterior. Consistent with this observation, Dkk1 is not activated in Otx2 mutant VE; consequently, canonical Wnt signaling is up-regulated (8). The current investigation determined that a cis-regulatory element highly conserved among vertebrates is essential for Otx2 expression in AVE. Biochemical and genetic analysis revealed that Foxa2 (previously named Hnf3β), a forkhead family transcription factor, directly transactivates Otx2 expression. More importantly, Foxa2 is necessary for expression of AVE-specific genes including Wnt antagonists in A–P axis polarization.

Results and Discussion

Identification of the Crucial Core Cis Element for Vertebrate Otx2 Expression in AVE.

A 5′ 1.8-kb mouse Otx2 promoter region directs expression in the DVE at E5.5 and in the AVE from the pre- to the midstreak stages (11). To identify the specific cis elements necessary for AVE expression in this region, a series of deletion constructs was generated and tested for activity in transgenic mice (Fig. 1A, #1–6). The activity was unaffected by the deletion of −683 to −594 bp (Fig. 1 A and B, #1–3); however, removal of an additional 51 bp (to −543 bp) abolished reporter activity in AVE (Fig. 1C, #4). Consistently, internal excision of the 51-bp fragment from the 1.8-kb promoter led to complete loss of expression of the transgene (Fig. 1E, #8). However, deletion of −648 to −594 bp did not affect activity (Fig. 1D, #7). These deletion analyses determined that the crucial element for induction of mouse Otx2 expression in AVE consists of 51 nucleotides (from −594 to −543 bp).

Fig. 1.
Identification of crucial cis elements in mouse and Fugu Otx2 genes. (A) The Otx2 promoter region appears at the top; below the promoter region, the mutant constructs linked to a lacZ reporter gene at the translational start site for expression analysis ...

Alignment of the sequence with the corresponding regions from human and chick Otx2 promoters identified a 51-bp stretch characterized by 100% and 90.2% identity, respectively [supporting information (SI) Fig. 5A]. Examination of the reporter activity of the human and chick promoters revealed that these conserved 51-bp segments are also crucial for AVE expression (SI Fig. 5). These findings in concert strongly support the critical nature of the 51-bp sequence with respect to cis activity in AVE.

Comparison of the sequences of different species appears to afford a powerful approach with respect to identification of functional segments for gene regulation (12). Thus, we exploited the compact genome of the pufferfish, Takifugu rubripes (Fugu), which is approximately eight times smaller than the mouse genome (13). More importantly, the nucleotide sequence between mouse and pufferfish genomes in the noncoding region displays a higher degree of divergence relative to that of human or chick genomes (14). Cis activity of the Fugu Otx2 genomic fragments (from −30.5 to +38.5 kb), which directs expression in the AVE, was surveyed via production of transgenic mouse lines (Fig. 1F) (14). Consequently, cis activity in the AVE was detected in the 5-kb fragment (from +10.5 to +15.5 kb) (Fig. 1F, F9). Further subdivision of the 5-kb fragment mapped this activity to a 1.1-kb fragment (from +14.4 to +15.5 kb) (Fig. 1 G and H, F17). Moreover, the 1.1-kb fragment, fused with a heterologous promoter, namely, the mouse Otx2 minimal promoter from −543 bp (#4), could drive AVE-specific expression (Fig. 1I). This observation indicated that the Fugu 1.1-kb fragment possesses cis activity similar to that of the mouse 51-bp sequence.

A search was conducted for conserved motifs between the mouse 51-bp and the Fugu 1.1-kb sequences to identify a crucial core cis element in terms of possible transcriptional factor binding sites; as a result, one AT-rich element, TNTTTNTTT (referred to as PE1), emerged (Fig. 2 A and D). The mouse Otx2 promoter possesses two PE1 sites (PE1-a and PE1-b); PE1-a is located outside the 51-bp fragment, whereas PE1-b occurs within the 51-bp stretch (Fig. 2A). To establish whether the PE1 elements are necessary for cis-regulatory activity, additional expression analysis was performed involving precise mutant constructs (Fig. 2A). A 5′ deletion construct from −586 bp to the translational start site, which contained one PE1 element, continued to direct lacZ expression in the AVE (Fig. 2B, #9). However, a second deletion construct, which lacked the PE1-b element, displayed no cis activity (Fig. 2C, #10). The pufferfish 1.1-kb fragment, which drives AVE expression, contains two PE1 sequences (Fig. 2D, F17); mutant constructs, which exhibited mutations in one or both PE1 elements (F19–21), were examined for cis activity (Fig. 2 D–F). Mutation of both PE1 sites resulted in complete loss of expression; moreover, AVE expression was more greatly reduced by the PE1-c mutation than by the PE1-d mutation (Fig. 2 D–F, F19–21). Consequently, the PE1-c site is more critical with respect to AVE expression in comparison to the PE1-d site in the Fugu transgene. These data clearly demonstrated that both mouse and Fugu PE1 elements are essential for cis activity in AVE.

Fig. 2.
Identification of Foxa2 binding sites that were conserved between mouse and Fugu Otx2 genes. (A) Nucleotide sequences of mouse Otx2 5′-upstream region and schematic diagram of the transgene constructs. (D) Nucleotide sequence within the Fugu genomic ...

Foxa2 Directly Controls Otx2 Expression in AVE.

A search for transcription factor binding motifs related to PE1 revealed that the consensus recognition sequence of Foxa2, KGNATRTTTRYTTW, matches the PE1 sites (http://motif.genome.ad.jp/). Foxa2 expression in the VE has been shown to play crucial roles in A–P axis development, although the mechanism has not been clarified (15, 16). Therefore, the ability of Foxa2 to regulate directly Otx2 expression in AVE through the PE1 element was examined (Figs. 2G, ,3,3, and and4).4). EMSA employing FOXA2 protein synthesized in vitro was performed to determine whether FOXA2 binds directly to the PE1 sequences (Fig. 2G). Precise EMSA experiments involving PE1 oligonucleotides demonstrated that FOXA2 protein can bind directly to PE1-b and PE1-c in vitro (Fig. 2G). The lower affinity of the PE1-a and PE1-d sites for FOXA2 is consistent with our expression analysis of mutant constructs (Fig. 1D and Fig. 2 A–F).

Fig. 3.
Foxa2 regulates Otx2 expression in AVE via the PE1 element. (A–D) Foxa2 mRNA expression was assessed after implantation to the prestreak stage by using whole-mount in situ hybridization. (E–H) X-Gal staining at E5.5 (E and F) and E6.5 ...
Fig. 4.
A–P axis patterning is affected in Foxa2−/− embryos. Whole-mount in situ hybridization of wild-type (A–N), Foxa2−/− (A′N′), and Otx2−/− mutant (O and P) embryos is ...

Endogenous Foxa2 mRNA expression was analyzed by using whole-mount in situ hybridization to assess whether Foxa2 expression coincides with Otx2 expression within the VE (Fig. 3 A–D). Foxa2 expression was apparent in the embryonic portion of the VE before E5.5 as reported (Fig. 3A) (17). Foxa2 expression was restricted to the DVE around E5.5 (Fig. 3B); furthermore, Foxa2 was expressed asymmetrically in the prospective anterior aspect of the VE at E5.75 and E6.0 (Fig. 3 C and D). These findings demonstrated that Foxa2 expression is identical to Otx2 expression in VE from E5.5 to E6.0 (11).

Otx2-lacZ transgene activity in the Foxa2 mutant background was analyzed to evaluate Foxa2 participation in Otx2 expression (Fig. 3 E–H). First, hemizygous transgenic mice carrying construct #3 (Fig. 1B) were crossed with Foxa2+/− mice (16), followed by examination of the cis activity in Foxa2−/− embryos. LacZ expression of transgene #3 was abolished by the Foxa2 null mutation (Fig. 3 F and H). Additionally, endogenous Otx2 transcripts were not detected within the VE of Foxa2−/− embryos as early as E5.5 (Fig. 3I′). At subsequent E6.5, Otx2 expression, which were evident in the epiblast and proximal VE, were absent in the mutant DVE (Fig. 3J′, K′, and L–N). Next, transgenic mice carrying a constitutive active form of Foxa2 cDNA under the control of the CAG promoter were generated to assess whether Foxa2 can transactivate Otx2 expression (Fig. 3 O′Q′). In the active form of Foxa2, a 156th-threonine, a phosphorylation target of Akt kinase, is mutated to alanine; consequently, dephosphorylated FOXA2 is exclusively localized in the nucleus (18). In the Foxa2-misexpressing embryos, Otx2 expression was expanded to the posterior side beyond the endogenous domain at E7.5 (Fig. 3 P′ and Q′). The aforementioned findings demonstrated that Foxa2 mediates Otx2 expression in AVE primarily through the PE1-b element.

Foxa2 Is Essential for Expression of Wnt Antagonists in AVE but Dispensable for Forebrain Induction.

Expression of several molecular markers associated with A–P axis polarization were examined in Foxa2−/− embryos at E6.5 to E8.5 to determine whether Foxa2 had broader roles in A–P axis development than just controlling Otx2 expression (Fig. 4). Consequently, we found that Foxa2−/− embryos fail to express Wnt antagonists and thereby presumably to form the A–P axis correctly (Fig. 4 A′H′ and SI Fig. 6 A–D′). Expression of Cerl, the Nodal, Bmp, and Wnt antagonist (5), was not observed in the Foxa2−/− embryo as early as E5.5 (Fig. 4 A′ and B′). Additionally, expression of Dkk1, another Wnt antagonist, was absent or sharply reduced in Foxa2−/− embryos (Fig. 4C′). Consistent with these results, Dkk1 expression was notably expanded in Foxa2 (T156A) misexpressing embryos at E6.5 (Fig. 3 R′ and S′). On the other hand, expression of Lhx1 and sFRP1, which also mark the AVE of wild-type embryos, was detected in Foxa2−/− DVE (Fig. 4 D and D′ and SI Fig. 6 A–B′). Expression of the posterior markers, T, Cripto, and Nodal, which was coordinated with that of AVE markers, was apparent in the proximal side of Foxa2−/− embryos at E6.5 (Fig. 4 E′H′). Consistent with failure in DVE migration, β-catenin expression was markedly up-regulated in the DVE of Foxa2−/− mutant embryos at E6.5 (C.K.-Y. and I.M., data not shown). These findings indicated that Foxa2 is essential for DVE migration via regulation of the AVE-specific genes including Wnt antagonists.

Previous study with Otx2 and Foxa2 double mutant embryos suggests that both genes genetically interact in forebrain development at later E9.5 (19); however, precise interaction within the VE is poorly understood. Then, expression of Dkk1 was examined in Otx2+/−;Foxa2+/− double mutant embryos, and, consequently, Dkk1 expression was reduced markedly in the AVE of Otx2+/−;Foxa2+/− embryos; in contrast, Cerl expression was not reduced (SI Fig. 6 E–G). These data suggest that Foxa2 and Otx2 genes act cooperatively to induce Dkk1 expression in AVE; additionally, these results indicate that Foxa2 regulation of Cerl expression is largely independent of Otx2.

To determine more precisely whether Foxa2 mediates A–P axis patterning outside the AVE, gene expression in Foxa2−/− embryos was examined at subsequent E7.5. Gsc and Cerl expressions were absent, which is probably attributable to failure in terms of proper formation of anterior mesendoderm (Fig. 4 I′ and J′). However, Dkk1 expression appeared to be up-regulated ectopically in the ectoderm of Foxa2−/− embryos at E7.5 (Fig. 4 K′ and L′; n = 7/9). Consequently, Foxa2−/− embryos were able to form forebrain transiently marked by Fkh2 (n = 3/4) and Six3 (n = 4/5) expression (Fig. 4 M′ and N′) (20), whereas Otx2−/− embryos fail to express Dkk1, Fkh2, and Six3 (Fig. 4 O and P) (10). Because inhibition of canonical Wnt signaling is essential for forebrain induction (2123), unexpected induction of Dkk1 expression may account for the partial rescue of anterior development in Foxa2−/− embryos. Dkk1 expression was reduced substantially in Foxa2 (T156A) misexpressing embryos (Fig. 3T′), and misexpression of Dkk1 cDNA led to anterior expansion in the mouse embryos (C.K.-Y. and I.M., data not shown), which are consistent with this hypothesis; moreover, expression of the active form of β-catenin was down-regulated in Foxa2−/− embryos in a manner similar to that of the wild-type embryo at E7.5, whereas β-catenin expression was markedly up-regulated in Otx2−/− embryos (Fig. 4 Q–V). These observations, in concert, indicated that Foxa2 is essential for A–P axis polarization with respect to regulation of expression of multiple AVE-specific genes within the VE of pregastrula embryos; on the other hand, Foxa2 appears to be dispensable for transient forebrain induction.

Finally, demonstration that pufferfish Otx2 contains cis elements that can control AVE expression in transgenic mice (Figs. 1 and and2)2) suggests the possible presence of tissues equivalent in function to AVE in teleosts. Orthologues of Foxa2 genes are expressed in the endoderm of all chordates examined before gastrulation (24, 25). Consistently, ascidian Otx expression is directly regulated by FoxA-a, an ascidian homologue of Foxa2 (2628). Moreover, ascidian FoxA-a, which is the earliest determinant of the ascidian anterior ectoderm, directly activates Ci-sFRP1/5, a canonical Wnt antagonist (27, 28). Thus, the genetic regulation of Otx and Wnt antagonists by Foxa2 genes for anterior specification is highly conserved throughout the evolution of the chordate body plan.

Materials and Methods

Experimental Animals.

Mouse Otx2 promoter constructs (#1–6, 9, and 10) were produced via deletion of the 5′ flanking regions fused to the lacZ reporter, VEcis-lacZ (10). Fugu genomic constructs (F0–18) were generated as described (14). Internal deletion constructs (#7 and 8, F19–21) were generated by PCR-based mutagenesis. The CAG-Foxa2 T156A transgene construct was produced based on the method of Kimura-Yoshida et al. (8). Details regarding the construction of the transgene vectors are available upon request. Transgenic mice were produced via microinjection of fertilized eggs from CD-1 as described (29). Transgene integrations were identified by PCR analysis. β-catenin mutant mice (30) were obtained from The Jackson Laboratory (Bar Harbor, ME). Mice were housed under the guidelines of Osaka Medical Center and Research Institute for Maternal and Child Health for animal and recombinant DNA experiments.

β-Gal Staining, in Situ Hybridization, and Immunohistochemistry.

Transgenic founder mice were established; subsequently, β-gal activity was analyzed in F1 hemizygous transgenic embryos according to the approach of Kimura et al. (11). In situ hybridization involving digoxygenin-labeled probes was conducted in a manner identical to that of Wilkinson (31). Whole-mount immunohistochemistry of anti-active-β-catenin (8E7; Upstate, Lake Placid, NY) was performed as described (8).

EMSA.

The FOXA2 protein was produced employing an in vitro translational system (Promega, Madison, WI) as described (32). Binding reaction was conducted as reported (33). The competition assay included two molecular ratios of labeled oligonucleotides to unlabelled oligonucleotides, of 1:50 or 1:100. Oligonucleotides used for the assay were PE1-a (5′-GGTGTTATCAGCATTATTTATTTAGCCAAAGA), PE1-b (5′-GGGATTTCTTGAATTGTTTCTTTGTTTCTCAC), PE1-c (5′-GGAGGAGAGAAAGAGAAAAGAAAAACAGAAGG), and PE1-d (5′-GGGCAAAGGCATTTTGTCTTTTTATTCCCCT).

Supplementary Material

Supporting Figures:

Acknowledgments

We are grateful to Drs. E. De Robertis (University of California, Los Angeles, CA), B. Hogan (Duke Unversity Medical Center, Durham, NC), H. Sasaki (RIKEN Center for Developmental Biology, Kobe, Japan), M. M. Shen (UMDNJ–Robert Wood Johnson Medical School, Piscataway, NJ), C. Niehrs (German Cancer Research Center, Heidelberg, Germany), B. G. Herrmann (Max-Planck-Institute for Molecular Genetics, Berlin, Germany), and H. Hamada (Osaka University, Suita, Japan) for plasmids. We also thank Mr. M. Suzuki of the Center for Animal Resources and Development (Kumamoto University) and Ms. R. Nakayama of LARGE (RIKEN CDB) for generating transgenic mice. This work was supported in part by grants-in-aid for Scientific Research on Priority Areas “Comparative Genomes” and (B) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan and the Sumitomo Foundation. J.R. holds a Canadian Institutes of Health Research Distinguished Investigator award.

Abbreviations

A–P
anterior–posterior
En
embryonic day n
VE
visceral endoderm
DVE
distal VE
AVE
anterior VE.

Footnotes

The authors declare no conflict of interest.

This article is a PNAS direct submission.

This article contains supporting information online at www.pnas.org/cgi/content/full/0607779104/DC1.

References

1. Beddington RS, Robertson EJ. Cell. 1999;96:195–209. [PubMed]
2. Varlet I, Collignon J, Robertson EJ. Development (Cambridge, UK) 1997;124:1033–1044. [PubMed]
3. Thomas P, Beddington R. Curr Biol. 1996;6:1487–1496. [PubMed]
4. Glinka A, Wu W, Delius H, Monaghan AP, Blumenstock C, Niehrs C. Nature. 1998;391:357–362. [PubMed]
5. Piccolo S, Agius E, Leyns L, Bhattacharyya S, Grunz H, Bouwmeester T, De Robertis EM. Nature. 1999;397:707–710. [PMC free article] [PubMed]
6. Perea-Gomez A, Vella FD, Shawlot W, Oulad-Abdelghani M, Chazaud C, Meno C, Pfister V, Chen L, Robertson E, Hamada H, et al. Dev Cell. 2002;3:745–756. [PubMed]
7. Yamamoto M, Saijoh Y, Perea-Gomez A, Shawlot W, Behringer RR, Ang SL, Hamada H, Meno C. Nature. 2004;428:387–392. [PubMed]
8. Kimura-Yoshida C, Nakano H, Okamura D, Nakao K, Yonemura S, Belo JA, Aizawa S, Matsui Y, Matsuo I. Dev Cell. 2005;9:639–650. [PubMed]
9. Srinvas S. Genesis. 2006;44:565–572. [PubMed]
10. Simeone A, Acampora D. Int J Dev Biol. 2001;45:337–345. [PubMed]
11. Kimura C, Yoshinaga K, Tian E, Suzuki M, Aizawa S, Matsuo I. Dev Biol. 2000;225:304–321. [PubMed]
12. Hardison RC. Trends Genet. 2000;16:369–372. [PubMed]
13. Brenner S, Elgar G, Sandford R, Macrae A, Venkatesh B, Aparicio S. Nature. 1993;366:265–268. [PubMed]
14. Kimura-Yoshida C, Kitajima K, Oda-Ishii I, Tian E, Suzuki M, Yamamoto M, Suzuki T, Kobayashi M, Aizawa S, Matsuo I. Development (Cambridge, UK) 2004;131:57–71. [PubMed]
15. Dufort D, Schwartz L, Harpal K, Rossant J. Development (Cambridge, UK) 1998;125:3015–3025. [PubMed]
16. Ang SL, Rossant J. Cell. 1994;78:561–574. [PubMed]
17. Perea-Gomez A, Shawlot W, Sasaki H, Behringer RR, Ang S. Development (Cambridge, UK) 1999;126:4499–4511. [PubMed]
18. Wolfrum C, Besser D, Luca E, Stoffel M. Proc Natl Acad Sci USA. 2003;100:11624–11629. [PMC free article] [PubMed]
19. Jin O, Harpal K, Ang SL, Rossant J. Int J Dev Biol. 2001;45:357–365. [PubMed]
20. Klingensmith J, Ang SL, Bachiller D, Rossant J. Dev Biol. 1999;216:535–549. [PubMed]
21. Glinka A, Wu W, Onichtchouk D, Blumenstock C, Niehrs C. Nature. 1997;389:517–519. [PubMed]
22. Piccolo S, Agius E, Leyns L, Bhattacharyya S, Grunz H, Bouwmeester T, DeRobertis EM. Nature. 1999;397:707–710. [PMC free article] [PubMed]
23. Mukhopadhyay M, Shtrom S, Rodriguez-Esteban C, Chen L, Tsukui T, Gomer L, Dorward DW, Glinka A, Grinberg A, Huang SP, et al. Dev Cell. 2001;1:423–434. [PubMed]
24. Olsen CL, Jeffery WR. Development (Cambridge, UK) 1997;124:3609–3619. [PubMed]
25. Taguchi S, Tagawa K, Humphreys T, Nishino A, Satoh N, Harada Y. Dev Genes Evol. 2000;210:11–17. [PubMed]
26. Oda-Ishii I, Bertrand V, Matsuo I, Lemaire P, Saiga H. Development (Cambridge, UK) 2005;132:1663–1674. [PubMed]
27. Imai KS, Levine M, Satoh N, Satou Y. Science. 2006;312:1183–1187. [PubMed]
28. Lamy C, Rothbacher U, Caillol D, Lemaire P. Development (Cambridge, UK) 2006;133:2835–2844. [PubMed]
29. Nagy A, Gertsenstein M, Vinterstein K, Behringer RR. Manipulating the Mouse Embryo. Woodbury, NY: Cold Spring Harbor Lab Press; 2003.
30. Brault V, Moore R, Kutsch S, Ishibashi M, Rowitch DH, McMahon AP, Sommer L, Boussadia O, Kemler R. Development (Cambridge, UK) 2001;128:1253–1264. [PubMed]
31. Wilkinson DG. In Situ Hybridization. New York: Oxford Univ Press; 1998.
32. Nakano T, Murata T, Matsuo I, Aizawa S. Biochem Biophys Res Commun. 2000;267:64–70. [PubMed]
33. Foucher I, Montesinos ML, Volovitch M, Prochiantz A, Trembleau A. Development (Cambridge, UK) 2003;130:1867–1876. [PubMed]
34. Courtois V, Chatelain G, Han ZY, Le Novere N, Brun G, Lamonerie T. J Neurochem. 2003;84:840–853. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...