|Home | About | Journals | Submit | Contact Us | Français|
Mutation in the “nude” gene, i.e. the FoxN1 gene, induces a hairless phenotype and a rudimentary thymus gland in mice (nude mouse) and humans (T-cell related primary immunodeficiency). Conventional FoxN1 gene knockout and transgenic mouse models have been generated for studies of FoxN1 gene function related to skin and immune diseases, and for cancer models. It appeared that FoxN1's role was fully understood and the nude mouse model was fully utilized. However, in recent years, with the development of inducible gene knockout/knockin mouse models with the loxP-Cre(ERT) and diphtheria toxin receptor-induced cell abolished systems, it appears that the complete repertoire of FoxN1's roles and deep-going usage of nude mouse model in immune function studies have just begun. Here we summarize the research progress made by several recent works studying the role of FoxN1 in the thymus and utilizing nude and “second (conditional) nude” mouse models for studies of T-cell development and function. We also raise questions and propose further consideration of FoxN1 functions and utilizing this mouse model for immune function studies.
The nude mutation 1 in the gene (FoxN1, forkhead box N1) 2-4, which encodes a transcriptional factor for the family of forkhead proteins, is responsible for this defect and has been known for a long time. The FoxN1 (former name: Whn or Hfh11) gene, located on chromosome 11 in mice and chromosome 17 in humans 5-9, is mainly expressed in thymic epithelium, distinct keratinocyte populations in the epidermis, and hair follicles. FoxN1 in rodents and FOXN1 in humans are highly conserved in sequence and function 5, 9. Mutations in FoxN1 cause inborn dysgenesis of the thymus (thymic rudiment and lack of lymphocytes) 10-13 and hairless skin (short and bent hair shafts inside the skin) 9, 13-15, which happen in mice, rats, and humans. The FOXN1 mutation in humans causes human nude (alopecia and nail dystrophy) and results in a primary T-cell deficiency 13, 16-18 related to severe infections, whereas the FoxN1 mutation in mice results in the generation of nude mice, which have been widely used as a model 19, 20 for experimental oncological, immunological, dermatological, and transplantation studies due to their immune deficiency in T-cell development and failure in hair follicle development (nude skin). However, in recent years, comprehensive understanding of the nude gene in the thymus and utility of nude mouse models for immunology and cancer studies are just now emerging. For example, with molecular technology moving forward, such as the development of the loxP-Cre/-CreERT system 21-23, it appears that the precise roles of FoxN1 are just beginning to be unveiled. In this review, we summarize recent findings in ongoing attempts to determine the functions of FoxN1 in the thymus and to utilize nude and “second (conditional) nude” mouse models for studies of T-lymphopoiesis and T-cell function.
Generally, FoxN1, a transcription factor, acts through its target genes in order to regulate the differentiation of epithelial cells. Specifically, FoxN1 regulates keratinocytes to differentiate under proliferating conditions 24, 25. The typical phenotypes resulting from an inborn null mutation of FoxN1 are developmental failures in the skin and thymic epithelium 15. Maturation of the thymic epithelial meshwork during thymic organogenesis occurs in two genetic stages 12, 26 - the first stage involves FoxN1-independent induction and outgrowth of the thymic epithelial anlage from the third pharyngeal pouch, controlled by genes such as the Eya1 and Six 27, Hoxa3 28, and Tbx1 29, 30. The second genetic step involves epithelial patterning and differentiation, which is FoxN1-dependent differentiation of the immature epithelial cells into functional cortical TECs (cTECs) and medullary TECs (mTECs). Recent reports emphasize FoxN1 as a powerful regulator that promotes differentiation in both the cTECs and mTECs during thymus organogenesis 31. FoxN1 expression in the thymus is ambiguously believed to be expressed in all fetal TECs but not in all adult TECs 32, 33, while FoxN1-negative TECs are reported to be derived from FoxN1+ TECs 33. Therefore, which TEC subsets lose FoxN1 with age, and why these subsets lose FoxN1 with age has yet to be clearly identified. An inborn null mutation in FoxN1 3 causes a differentiation failure in TECs thereby halting thymic development at a rudimentary stage—the thymic lobe is still present but thymic lymphopoiesis is completely blocked 12, 34. This causes an alymphoid thymus and severe primary T-cell immunodeficiency in nude mice and humans 8, 35, 36 with congenital alopecia and defective immunity, resulting in death in early childhood from severe infections 37, 38. Therefore, the FOXN1 mutation is a severe human primary immunodeficiency disease 13, 16-18, 39, 40.
In the skin, FoxN1 is required for normal hair follicle development regulating the initiation of keratinocyte terminal differentiation, which has been well reviewed 9, 15. FoxN1 expression, mainly in the hair shaft cortex, was reported to peak during anagen—hair growth period, then fall during catagen (destruction) and telogen (rest) 9, 14, 15. Recently, FoxN1 was also reported to regulate pigmentation in the skin (related to skin darkness), demonstrated by using an engineered—keratin-5-driven FoxN1 (K5-FoxN1) transgenic (Tg) mouse 41. These authors found that while the FoxN1-null nude mouse completely lacks pigmentation in the hair cortex, K5-FoxN1 Tg confers ectopic acquisition of pigmentation in hair cortical cells. This is said 41 to be due to regulation via the FoxN1-Fgf2 regulatory axis on pigment transfer from melanocytes to keratinocytes.
Although skin and thymus phenotypes resulting from the inborn FoxN1-null mutation are well-known, central nervous system deterioration, such as anencephaly, during the organogenesis resulting from this mutation has only recently been reported 36. The Amorosi group found that FoxN1 is expressed in the brain choroid plexus of murine embryos by using a FoxN1 heterozygous mouse, in which one copy of FoxN1 bears an inserted β-galactosidase (LacZ) reporter gene 12. However, this leads us to ask why FoxN1-null gene knockout mice do not show a neural tube defect, and why do not all FoxN1 deficient human fetuses have a neural tube defect 42? Consequently, whether FoxN1 mutation really causes congenital brain developmental abnormalities remains to be confirmed.
FoxN1 gene-manipulated mouse models, such as loss-of-function 12, 43 and gain-of-function 41, 44 models, have been available in the studies of FoxN1 gene function. However, with molecular and cellular technology moving forward, many new methodologies have been developed. For example, the loxP-Cre/CreERT-mediated conditional FoxN1 gene “loss-resumption or revert” 31, 45 and the conditional FoxN1 gene knockout 46 mouse models have been developed in recent years. This facilitates the determination of the precise roles of FoxN1.
These systems have been used to artificially (conditionally) control gene expression (conditional knockout or over-expression) in the mouse for a couple of decades 21-23. The loxP-Cre/CreERT system centered on the Cre gene, short for cyclization recombination 47. The Cre gene encodes a site-specific DNA recombinase, which can recombine DNA at specific sites, which are 34-base pairs long, known as loxP (locus of X-over P1) sequences. These sequences act as magnets for Cre to recombine the DNA fragment in between the two loxP sites, resulting in recombination-excision (deletion) of the loxP-flanked DNA fragment. If this excised DNA fragment is a functional part of a gene, its deletion will cause this gene to become dysfunctional. For example, in the conditional FoxN1 gene knockout mouse (this mouse is now available at the Jackson Laboratory, #012941, http://jaxmice.jax.org/strain/012941.html) 46, the DNA binding domain, i.e. functional domain of transcription factor FoxN1, located on exons 5 and 6 of the FoxN1 locus 48, 49, is flanked by two loxP genes (Fig. (Fig.1A).1A). When this domain is deleted (termed ΔE5&6) by Cre or CreERT, the FoxN1 gene loses its function 46. If this DNA fragment is a loxP-flanked STOP cassette (STOPflox), a roadblock sequence positioned upstream of a functional gene or cDNA, the deletion of this STOPflox will cause the gene to be re-expressed (Fig. (Fig.1B)1B) or to resume (Fig. (Fig.1C).1C). For example, in our unpublished novel STOPflox -FoxN1 transgenic mouse (Fig. (Fig.1B,1B, this mouse is available by request), the flag-FoxN1 cDNA (kindly provided by Dr. Brissette 25) carried by a composite of CMV-immediate early gene enhancer/chicken β-actin promoter (pCAG) (kindly provided by Dr. McMahon 50, 51) was inserted into a backbone of the Rosa26 locus. In the front and the end of this fragment a STOPflox cassette and IRES-GFP reporter gene were inserted, respectively. This makes conditional expression of the FoxN1 transgene controlled by Cre/CreERT. Furthermore, in conditional FoxN1 gene “loss-resumption” or reversible mouse models 31, 45, a STOPflox cassette (including two splice acceptors and a hygromycin or neomycine cassette) flanked by loxP sites is inserted into a normal FoxN1 gene, which destroys and silences normal FoxN1 transcription, resulting in an inborn mutant phenotype during organogenesis. After this STOPflox cassette is depleted by introduction of Cre/CreERT, the endogenous FoxN1 expression resumes and the phenotype is reversed (Fig. (Fig.1C1C is an example).
Since Cre can be driven by different tissue-specific promoters, it can be uniquely expressed in certain tissues but not in others. Therefore tissue-specific promoter-driven Cre can achieve tissue-specific loxP-flanked DNA fragment deletion. This is one mechanism of conditional gene expression. The other mechanism is temporally-controlled gene expression in somatic cells rather than in germline cells. This can be achieved by Cre-ERT gene 52, 53, which is the Cre-recombinase fused to a mutated ligand binding-domain of the human estrogen receptor (ER). The estrogen receptor binding-domain represses Cre in an inactive state until de-repressed by Tamoxifen (TM), because the ER binds TM but not estrogens. Therefore, deletion of the loxP-floxed DNA fragment is induced by administration of TM but not mouse or human estrogens 53, 54. By combining a tissue-specific promoter with CreERT, the loxP-flanked DNA fragment deletion can be controlled in a spatio-temporal fashion, thereby facilitating the introduction of a somatic mutation in a given gene, at a chosen time, in a selected cell type 21-23. Particularly, this system benefits the study of the later roles of genes whose mutations cause early embryonic lethal phenotypes. Although a mutation of FoxN1 will not cause lethality in embryos, its roles in the developed postnatal thymus and in different keratin-type epithelial cells would have largely remained unknown without the FoxN1flox mouse model.
A new system for cell lineage ablation, based on transgenic expression of a diphtheria toxin receptor (DTR) carried by cell lineage specific gene and induced cyto-ablation via injection of diphtheria toxin (DT) has been developed in recent years 55-58. Recently, this approach was used in the study of a FoxN1-positive thymic epithelial cell lineage 33. Dr. Boehm's group clearly showed that after specific FoxN1+ TEC lineage was killed (cytoablation) by induction with DT in early embryogenesis, the orthotopic thymus becomes aplastic, and these TECs cannot fully regenerate.
Although the general role of FoxN1 is to regulate the differentiation of epithelial cells in the thymus and skin, it was largely unknown whether the roles of FoxN1 in the thymus and skin are identical. If not, then how might they differ? The overt differences in FoxN1's roles in the thymus and skin were revealed in a recently published paper 59. One important difference is that FoxN1 is involved in morphogenesis and maintenance of the three-dimensional (3D) thymic micro-structure, which is important for a functional thymus. As we know, two-dimensional (2D)-monolayer (non-Notch ligand transformed 60) stromal cells cannot support T-cell development in culture. However re-aggregated stromal cell-constituted 3D pseudo-thymic lobes can fully support T-cell development in a fetal thymic organ culture (FTOC) setting. This is, at least in part, due to the alteration of certain key molecules. For example, dissociated thymic stromal cells lost the Notch ligand Delta-like expression, while re-aggregated thymic stromal cells (3D) regained its expression 61. However, the normal micro-structure in the skin is two-dimensional or polarized, i.e. the epithelial layer (basal layer) on one side expresses keratin (K)5 and K14, and the epithelial layer (apical layer) on the other side expresses K8 and K18. The other important difference is that FoxN1 determines the pigmentation pattern in the skin 41, but this is inapplicable in the thymus.
Additionally, using K14Cre transgenic mice 62 to delete FoxN1flox in K14 promoter-driven epithelial cells seems to have a larger impact on the skin than on the thymus 59. Deletion of FoxN1flox in K14 epithelial cells is sufficient to cause a hair follicle defect resulting in a nude phenotype, similar to that of the natural FoxN1-null mutant mice, but does not induce an alymphoid thymic rudiment, thus differing from the thymic phenotype of the natural FoxN1-null mutant mice. It is unclear whether this phenotype is a result of low versus high expression of K14Cre in the thymus versus the skin. By using a K14Cre-mediated LacZ expression mouse model, which was generated by crossing K14Cre mice with R26-STOPflox-LacZ reporter mice, in which the STOPflox cassette is deleted upon K14Cre expression, thereby subjecting LacZ expression to be controlled by K14 promoter. Jackson Laboratory confirmed that the K14Cre-mediated LacZ is strongly expressed in the postnatal thymus, particularly in the thymic medulla (several images are posted in Jackson Laboratory web site: http://cre.jax.org/Krt14/Krt14-creNano.html). Therefore, the difference observed by Guo et al. should be due to the different impacts of FoxN1 on K14 epithelial cells in the skin and thymus, rather than a result of lower expression of K14Cre in the thymus.
As mentioned previously, there are FoxN1-independet and -dependent genetic stages, during thymic organogenesis and TEC differentiation 12, 26. Owing to the lack of suitable genetic tools to address it, there was a long-running argument centered on whether FoxN1 continues to maintain a functional thymus following the second genetic stage of thymic organogenesis, especially in the adult thymus. Gordon et al. generated a FoxN1-LacZ mouse model 63, in which a LacZ cDNA cassette was inserted into the 3'UTR of the FoxN1 locus. Chen et al. observed that LacZ has an adverse effect on FoxN1 expression with age, via a supposed methylation mechanism, to induce thymic postnatal involution 64. Therefore, FoxN1 was experimentally demonstrated to be required in the postnatal thymus. Because this mouse model cannot be spatio-temporally controlled, precise information of defects in timing and TEC subsets is not available, whereas, the inducible FoxN1flox gene knockout mouse model can be used for addressing these question 46.
Recently, using FoxN1flox-K14Cre mice, the Guo et al. 59 demonstrated that under certain circumstances the postnatal role of FoxN1 may be even more important than its prenatal role. They found that homozygous FoxN1flox/flox mice without the Cre gene have FoxN1flox deletion (Fig. (Fig.2A,2A, genotype case #3) when their mother has the Cre gene. These mice have completely normal phenotypes in the thymus and skin. This deletion record comes from a historic Cre-mediated FoxN1flox deletion and should happen in their prenatal life inside the mother's uterus. However, homozygous FoxN1flox/flox mice carrying their own Cre gene have a FoxN1flox deletion (Fig. (Fig.2A,2A, genotype case #4) and display mutant phenotypes in the thymus and skin. This FoxN1flox deletion should happen in both prenatal and postnatal life. This finding demonstrated that FoxN1 deletion happened inside the mother's uterus, driven solely by the parent's K14Cre (no Cre gene in offspring), which does not induce mutant phenotypes in the thymus and skin of the offspring. Instead, only when FoxN1flox is deleted in both prenatal (mediated by mother's Cre) and postnatal (mediated by self Cre) are the mutant phenotypes in the thymus and skin induced. This confirmed the importance of postnatal FoxN1. However, this phenotype could not have been revealed without the loxP-Cre system because in the naturally occurring FoxN1-null mutation the FoxN1 gene cannot be deleted separately in prenatal and postnatal life. Furthermore, this phenotype is not only found in FoxN1flox-K14Cre mice, but also in other FoxN1flox-Cre (resulting in germline deletion) mice, such as in ubiquitous FoxN1flox-Cre mice (EIIa-Cre, Jackson Lab #003724) (Fig. (Fig.2).2). Therefore, FoxN1 is required not only for prenatal epithelial patterning (previously known) but also crucial for postnatal epithelial homeostasis. Prenatal deletion (mediated by Cre inside the mother's uterus) of FoxN1 alone cannot induce mutant phenotypes, but both prenatal and postnatal deletion (via offspring's own Cre) are able to induce thymic and skin mutant phenotypes.
The importance of FoxN1's role in the postnatal thymus, beyond the second genetic stage 12 of thymic organogenesis in the fetal thymus is indisputable. Furthermore, this raises two intriguing issues: 1) FoxN1's probable role in the postnatal thymus is to regulate epithelial cell homeostasis. 2) Since postnatal TECs continue to undergo homeostasis, this process should be supported by tissue-specific stem/progenitor cells in situ. Therefore, adult thymic epithelial stem/progenitor cells probably exist in the postnatal thymus. This is a glaring issue and is discussed in the following section.
Based on anatomic, keratin type, and functional criteria, thymic epithelial regions can be divided into the cortex and medulla. The TECs in the cortical region are called cTECs, while those in the medullary region are referred to as mTECs. Both epithelia provide different microenvironments that are responsible for distinct stages in thymocyte development 26, 65. As we know, FoxN1 is required for the development of both cTECs and mTECs in the fetal stage 12. However, since FoxN1 is expressed in almost all TECs of the embryonic thymus, but not in all TECs of the adult thymus 32, 33, FoxN1's role in different TEC subsets in the prenatal and postnatal thymus does not seem to be identical. Therefore the question arises, is FoxN1 equally required for homeostasis of both cTECs and mTECs in the postnatal thymus?
In the fetal stage, FoxN1 mainly regulates TEC patterning. Since both cTECs and mTECs arise from the same bi-potential TEC progenitors 45, 66, and FoxN1 regulates the process of differentiation, it is straightforward to conclude that both mTECs and cTECs are equally FoxN1-dependent during fetal thymic organogenesis 67. However, in the postnatal thymus, mTECs with the keratin type K5+ and K14+, which are similar to epithelial stem cell markers and exhibit progenitor activity in the skin and mammary gland 68-70, were more sensitive to the loss of FoxN1. cTECs possess keratin type K8+ and K18+, with the same marker as mature epithelial cells and terminally differentiated epithelial cells in the apical layer of stratified squamous epithelium of the skin. Additionally, cTECs are not as sensitive as mTECs to the loss of FoxN1 using K5- and K18-CreERT-mediated FoxN1flox-deletion mouse models 46. This finding in the postnatal thymus may account for the fact that K8+/K18+ cTECs and K5+/K14+ mTECs are not equally FoxN1-dependent in the postnatal thymus. Although it may be due to a long half-life in cTECs compared to that in mTECs, FoxN1 may not be required for mature/differentiated epithelia, which have K8+/18+ markers in the thymic cortical region and skin epithelial apical layer. However, FoxN1 should regulate the immature/undifferentiated TECs (epithelial progenitor cells), which may be a small subpopulation present in the K5+/K14+ TEC populations, located in the medullary region and/or the corticomedullary junction (CMJ) in the postnatal thymus. Specifically, FoxN1 in the adult thymus is required for adult TEC progenitors 33, which express K5+/K14+ markers, and these progenitors support TEC homeostasis in the adult thymus. A recent report further confirms this, showing that thymopoiesis depends on a FoxN1-positive TEC lineage, while FoxN1-negative TECs are descendants of FoxN1-positive TECs. FoxN1-negative TECs do not contribute to thymopoietic function in the adult thymus 33. Further support of the existence of postnatal TEC progenitors, which are dependent on FoxN1, was made in Osada et al 2010 71, where premature thymic involution was observed after inhibition of Wnt signaling through conditional expression of Dkk1 resulting in a decline in FoxN1 expression and loss of TEC progenitors.
Further progress in the recognition of FoxN1 function in recent years was made by two reports that determined whether thymus development is sensitive to the genetic dosage of FoxN1 and the association with age-related thymic involution. They also determined if heterozygous FoxN1 (a half genetic dose of FoxN1, i.e. FoxN1nu/+), which is known to be sufficient to induce TEC patterning in the thymic organogenesis, is also sufficient to maintain homeostasis for a steady-state normal thymus in the postnatal life. One report 64 showed that the mutant phenotype is dependent on FoxN1 genetic dosage. The thymus in wild type (WT) mice is completely normal, and it is completely abnormal in FoxN1-null mice (natural FoxN1-null nude mouse). The abnormality lies in between these two extremes for FoxN1-null heterozygote (nu/+) and LacZ/nu chimera mice. The degree of severity is: WT < nu/+ < LacZ/nu < null in a genetic dose-dependent manner. Another study 72 using a FoxN1flox mouse carrying a ubiquitous CreERT transgene (uCreERT), that took advantage of a low-dose spontaneous Cre leakage due to incomplete ER blockage in vivo 73, found that spontaneous leakage of uCreERT caused FoxN1flox deletion accompanying a progressive loss of FoxN1+ TECs with accelerated age-related thymic involution. This also occurred in heterozygous FoxN1flox/+ mice (deletion of floxed-FoxN1 in one copy of FoxN1 gene), representing a haplo-insufficient phenotype but related to age, i.e. age-related haplo-insufficiency. This finding extends previous observations in adult natural FoxN1-nu/+ heterozygous mice 74, 75.
Expression of FoxN1 in the thymi of naturally middle-aged and aged WT mice is significantly reduced 76. By increasing the dosage of FoxN1 in these thymi via intrathymic administration of exogenous FoxN1-cDNA, a rapid gain-of-function approach, thymic involution and declining thymic function can be partially rescued 72. Dr. Le's group confirmed this hypothesis at the genetic level with their FoxN1 transgenic mouse model, which showed that an up-regulation of FoxN1 expression in the aged thymus can rejuvenate function of the atrophied thymus 44. In their experiments, they ingeniously selected the FoxN1 transgenic mice with low copy numbers for their observations. We found that highly over-expressed FoxN1 induces adverse effects on thymus development (data unpublished), and even causes a lethal new-born phenotype (Fig. (Fig.3).3). In our newly generated STOPflox-FoxN1 transgenic mice (Fig. (Fig.1B),1B), the FoxN1 cDNA (kindly provided by Dr. Brissette 25) is driven by the Rosa26 promoter and enhanced expression by a composite of CMV-immediate early gene enhancer/chicken β-actin promoter (pCAG) (kindly provided by Dr. McMahon 50, 51). This results in high over-expression of FoxN1. In Figure Figure3,3, we show that K14Cre-mediated STOPflox-FoxN1 transgenic new-born mice died within 24 hours of birth. These neonatal mice share similar phenotypes with the involucrin promoter-driven FoxN1 transgenic neonatal mice, which have ectopic and enhanced expression of FoxN1 25, displaying abnormal skin and possessing open eyes at birth.
The nude mice or second nude (inducible FoxN1 gene knockout) mice provide animal models to facilitate studies of T-cell development and postnatal T-cell function in immunity and autoimmunity related to human disease. Recently, Dr. Boehm's group designed elegant experiments by using a FoxN1-null (nude mouse) model to reveal that thymic epithelia possess synergistic, context-dependent, and hierarchical functions in lymphopoiesis 77. TECs of FoxN1-null mice were transformed by cDNAs of the chemokines Ccl25 and Cxcl12, the cytokine Scf, and the Notch ligand DLL4 carried by the FoxN1-promoter to generate FoxN1-Ccl25, -Ccl12, -Scf, and -DLL4 transgenic mice under the FoxN1-null background. In these transgenic embryonic thymi, they found precise environmental components that can support mast cells, B progenitor cells, and T progenitor cells, respectively 77. Another recent work using a tissue-specific FoxN1flox gene knockout mouse model studied influenza infection in aging 78. They found that K14Cre-mediated FoxN1flox deletion-induced defects in the thymic medulla reduced antigen-specific CD8+ T-cell and IgG responses to influenza virus, combined with increased lung injury, weight loss and mortality. These findings provided the first evidence that defects in the medulla directly causes changes in T-cell function that mimics aging defects during an immune response to an infectious agent 78. A third recent work using the second nude (conditional FoxN1flox gene knockout) mouse model addressed possible mechanisms of increased autoimmune susceptibility in the elderly 79. Age-related disruption of steady-state thymic medulla caused by two-dimensional thymic epithelial cysts, primarily generated in the medulla, was found to perturb thymocyte negative selection. Negative selection is the main mechanism for the generation of central immune tolerance 80 necessary to prevent autoimmune diseases. This disruption was confirmed to provoke autoimmune phenotypes, such as inflammatory cell infiltration in multiple organs and the generation of anti-nuclear antibodies 79.
Although much progress has been made in recent years in unveiling the roles of FoxN1 using advanced technology, quite a few questions in this field still exist. For example, if FoxN1 is involved in the regulation of adult thymic epithelial (TE) stem/progenitor cells, more experiments are required to understand the localization and characteristics of individual TE stem/progenitor cells from the postnatal thymus. There is an ongoing debate 81 about whether the adult thymic epithelial stem cells even exist and where they may be located, even though epithelial stem/progenitor cells have been implicated in the FoxN1 “loss-resumption or revertible” (FoxN1SA2) adult thymus 45. In the FoxN1SA2 mice, a loxP-flanked hygromycin cassette was inserted into FoxN1 introns 6-7 (Fig. (Fig.1C).1C). This insertion destroys normal FoxN1 splicing and silences the gene. The mice have a nude phenotype and a defective thymus. Upon hK14-CreERT transgene activation in the adult thymus the insertion is deleted, and FoxN1 is re-expressed in putative adult TE stem/progenitor cells, which can differentiate into normal cortical and medullary TECs and support normal thymic regeneration and function 45. However, it can be argued that the TEC progenitors in the FoxN1SA2 defective thymus are dormant cells persisting from the fetal stage due to the FoxN1 mutation, which may not represent TE stem/progenitors in the normal adult thymus. Further work is required to resolve the debate.
Another clue for TE stem/progenitors presenting in the natural adult thymus is provided by the FoxN1flox-K5CreERT mouse models, in which the FoxN1 gene was conditionally knocked out in K5+ 46 epithelial cells after TECs fully developed in the adult thymus, resulting in acute thymic atrophy. This is probably due to the disruption of adult TEC homeostasis supported by TE stem/progenitor cells in the adult thymus. Since the K5 and/or K14+ promoters are active in epithelial stem/progenitor cells of the skin and mammary gland, the TE progenitors in the adult thymus may also be present within K5+ and/or K14+ TECs to support TEC homeostasis. However, the proportion and functional characteristics of adult thymic epithelial stem/progenitor cells that are K5+ and/or K14+ is largely unknown. Obtaining direct evidence by lineage-tracking changes in these adult thymic epithelial progenitors is a critical need.
Regulation of TEC homeostasis is possibly co-regulated by FoxN1 and other stem cell-related genes, such as p63. The transcription factor p63, which encodes for multiple isoforms (containing an N-terminal transactivation domain, termed as TAp63, and lacking this domain, termed as ΔNp63) 82, is pivotal for the development of stratified epithelial tissues, including the epidermis, breast, prostate, and thymus 83. The role of p63 in thymic development may be considered to be essential for the proliferation potential of thymic epithelial stem/progenitor cells 84, 85. Specifically, thymic development is considered to be regulated by the ΔNp63 isoform through the maintenance of epithelial progenitor “stemness”. By introducing ΔNp63 and TAp63 transgenes into a p63 gene knockout background, ΔNp63, but not TAp63, is able to rescue defective thymus development 84. Recently, we obtained a clue that the role of TAp63 in the thymus is probably associated with TEC senescence, since it was increased with thymic aging and associated with an age-related increase of senescent cell clusters (Manuscript under preparation). This phenotype may be accelerated by a blockade of TEC differentiation via conditionally knocking out the FoxN1 gene. We suspect that p63 and FoxN1 may form a p63-FoxN1 regulatory axis in TEC homeostasis during aging. However, the mechanism controlling how the proliferation regulator p63 and differentiation regulator FoxN1 work collaboratively in the regulatory axis is still mysterious, and more studies are appreciated.
FoxN1 is a transcription factor whose functions are executed by targeting other genes through its DNA binding domain. Therefore, to understand its functional mechanisms in determining its target genes is important. However, the precise target genes that are regulated by FoxN1 remain ill defined, mostly due to technical difficulties in precisely isolating enough physiologically intact TECs at certain developmental stages. One group performed laser-capture micro-dissection to capture TECs from the E12.5 FoxN1-null nude mouse thymi, and found five FoxN1 target TEC genes in their microarray analysis, of which programmed cell death-1 ligand is the only gene of known function 86. However, emerging studies via immuno-histological methods suggest that changes in these five genes in the FoxN1-null mutant thymus have been undervalued. For example, FoxN1 target Fgf2 has been identified in the skin 41. FoxN1 may target the Notch ligands, DLL-1 and DLL-4 87, and the chemokines, CCL25 and CXCL12 88, in nude thymic anlages. It may also target Notch-1 receptor in the skin as demonstrated in a transgenic mouse model 89. We also preliminarily analyzed FoxN1 targeting genes by using a microarray assay from FACS sorted CD45- EpCAM+ TECs. One TEC group was derived from ubiquitous-CreERT-mediated FoxN1flox knockout induced by TM in the postnatal thymus 46, the other group was WT mouse TECs. We found that at least 5 groups (Fig. (Fig.4A)4A) in over 300,000 genes screened underwent significant changes, either increasing or decreasing. Changes of log2 ratio > 2 or < -2 were observed in 1197 genes (Fig. (Fig.4B).4B). The most interesting gene family is toll-like receptor (TLR), in which TLR4 shows a significant increase (Fig. (Fig.4C).4C). We are conducting further work on this gene family to determine its physiological significance. Chromatin immunoprecipitation (ChIP) on Chip and ChIP on Sequence approaches may be one of the best ways to determine FoxN1 target genes.
Recent progress using advanced technology to study FoxN1's roles in the thymus shows that FoxN1 regulates not only TEC patterning in the fetal stage but also TEC homeostasis in the postnatal thymus. Comparing the thymus with the skin, FoxN1 has its own distinct roles and impacts on organs in the generation and maintenance of three-dimensional microstructure and pigmentation, respectively. FoxN1's role in the neuron has been brought up, but is still obscure. There is still plenty of room to apply nude and secondary nude (conditional FoxN1 gene knockout) mouse models in studies of immunology, hematology, and tumorgenesis. The functional mechanisms of FoxN1's collaborative roles with other genes during thymic development and aging remain to be further determined.
We would like to thank Dr. Yong Zhao (Institute of Zoology, Chinese Academy of Science) for his insightful discussions. We apologize to those colleagues whose work could not be referenced directly due to space constrains.
Funding support: This work was supported by the National Institute of Allergy and Infectious Diseases at the National Institutes of Health, USA (Grant numbers: R01AI081995) to D-M.S.