Informed Clinical Decisions by Outfoxing Human FOXN1 Variants

Abstract

Thymic T cell development is orchestrated by thymic epithelial cells (TECs). The master transcriptional regulator of TECs is Forkhead Box N1 (FOXN1), which controls their differentiation, expansion and function. Biallelic founder mutations in FOXN1 caused a Nude/Severe Combined Immunodeficiency (SCID) phenotype due to congenital thymic aplasia and alopecia universalis. This established the critical role of FOXN1 in TECs and epithelial cells in the skin and nails. The emergence of newborn screening for severe T cell deficiency via the T cell receptor excision circle (TREC) assay along with exome and genome sequencing has led to dramatic increases in the number of FOXN1 variants identified. The consequent impact of the FOXN1 variants ranges from pathogenic to benign. Yet most FOXN1 mutations are listed as variants of unknown significance. Among monoallelic FOXN1 variants are some that act as dominant negatives, resulting in a transient T cell lymphopenia. In this review, the clinical impacts of diverse FOXN1 variants are categorized based on the type and location of the mutation. Knowing how these FOXN1 mutations affect protein function informs on clinical care; laboratory monitoring, prophylactic measures, and allogeneic thymic implant decisions. This review provides key functional insights into FOXN1, enabling better clinical care.

The Impact of FOXN1 Variants on the Thymus, Hair, Nails and Skin

The thymus is the principal organ responsible for T cell development. In infants/toddlers, the thymic weight and cellularity is impressive, reaching an average of 20 g with up to 5 x 10¹⁰ cells, of which >98% are developing thymocytes ^{1, 2}. Prior to understanding its functional role in the 1960’s, a “large” thymus was erroneously linked to sudden infant death syndrome (thymic asthma and later, status thymico-lymphaticus) ³. This clinical belief led to over 100,000 infants having their thymuses therapeutically irradiated to diminish its size. The disastrous consequences were over 10,000 thyroid cancers and a high odds ratio for breast cancer ^{4, 5}. Today, clinicians are rightfully more focused on those infants born with thymic hypoplasia or aplasia. A key transcription factor regulating thymus growth and functionality is Forkhead Box N1 (FOXN1), responsible for thymic epithelial cell (TECs) development, differentiation and function ^6-9. TECs, in turn, are necessary for “educating” T cells in the thymus. FOXN1 arose as a gene duplication of FOXN4, an ancestral paralog expressed in the pharyngeal endoderm of chordates ¹⁰. In vertebrates, FOXN1 is also expressed in epithelial cells and keratinocytes along the pre-cortical and cortical regions of the hair shaft, and in the matrix and bed of the nails ¹¹. These expression patterns nicely explain the original Nude/Severe Combined Immunodeficiency (SCID) phenotype described for patients harboring now 6 distinct autosomal recessive and compound heterozygous FOXN1 loss of function mutations (in chronological order, p.R255Ter, p.R320W, p.S188fs, p.R114Ter, p.V294I, p.P272T, and p.R114X/E139fs in trans) (Table 1) ^12-16. Patients with these bi-allelic mutations have a congenital thymic aplasia (T⁻B⁺NK⁺ phenotype) along with alopecia universalis and nail dystrophy ^{12, 17}. The T⁻B⁺NK⁺ phenotype results from a functional impairment of TECs. While bi-allelic loss of function FOXN1 variants were initially designated as SCID due the serious clinical impacts, current guidelines from the Primary Immune Deficiency Treatment Consortium (PIDTC) restrict the SCID classification for those linked to hematopoietic stem cell defects ¹⁸. Thymic stromal cell disorders caused by FOXN1 mutations are now classified as non-SCID disorders. Other non-SCID thymic stromal cell defects include DiGeorge syndrome (e.g., 22q11.2 deletion syndrome), CHARGE syndrome (due to CHD7 mutations), and pathogenic variants in FOXI3, PAX1, TBX1, and TBX2 ^{18, 19}. The specific impact of FOXN1 variants on the stromal cells in the thymus was best revealed by a characterization of autosomal recessive (AR) Foxn1 mutations that spontaneously arose in Nude/SCID mice (nu/nu, p.P112fs) and rats (rnu/rnu, p.S473Ter) ^{20, 21}. At embryonic day 11.5 (human weeks 7-8), the murine nu/nu embryos had a normal thymic anlage within the 3^rd pharyngeal pouch ^{20, 22}. From e12 onwards, TECs failed to differentiate and proliferate, developing T cells were absent, and the thymic tissue was cystic.

Table 1:

Classification of FOXN1 variants and clinical indicators^a

Category	Mutation	Mode of Inheritance	TREC values at birtha	Circulating CD3+ T cells at birth (cells/μl)	Severe Infections	Alopecia Nail Dystroph y	Recommended or Actual Treatments
Pathogenic	p.R114Ter, p.S188fs, p.R255Ter, p.P272T, p.V294I, p.R320W, p.T313fs/p.W363Ca p.P430S/Q489fsa,b	Autosomal Recessive (AR) or Compound Heterozygousb,c	Low to Absent	0-489 cells/μl	Yes	90%	Thymus implant
	Clinical Case Example64: p.R255Ter/p.R255Ter	AR recessive	Preceded TREC assays	2219 cells/μl <1% naïve	M. bovis Rotavirus	Yes	Thymic implant given
	Clinical Case Example40: p.P430S/Q489fs	Compound Heterozygous	Absent	137 <2% naive	Not reported	No	Ig replacement HCT givenc
Likely Pathogenic (Dominant negative)	p.P401Afs, p.P402Lfs, p.L404Cfs, p.P432fs, p.L439Cfs, p.Y455Cfs, p.H457Pfs, p.P465Rfs, p.P473Hfs, p.Q489Rfs	Monoallelic	Low to Absent	11-1500 cells/μl	Variable	40-60%	Monitor T cell levels over time Thymus implant for persistent severe T cell deficiency
	Clinical Case Example52: p.Q489Rfs	Monoallelic	Absent	465 9% naive	None		Ig replacement Prophylaxis for Pneumocystis jirovecii pneumonia
Likely Pathogenic	p.V294L, p.H321N, p.H321R, p.L325P, p.C328R, p.P350L, p.L456Ter	Monoallelic	Low	14-1500 cells/μl	Variable	40-60%	Monitor T cell levels over time Thymus implant for persistent severe T cell deficiency
	Clinical Case Example42: p.H321N42	Monoallelic	Not reported	14 cells/μl	Rinovirus Otitis Atopic dermatitis	Yes	HCT (remained T cell lymphopenic)
Uncertain Significance	p.A121T, p.A283V, p.R341C, p.W346R, p.Y617fs, splice variants	Monoallelic	Low	>1200 cells/μl	Variable	None	None
Likely Benign	p.A121V, p.R128W, p.G238D, p.P395L, p.S569Y, p.C586Y, p.T629M	Monoallelic	Variable	>1200 cells/μl	No	None	None
Benign	p.R69C, p.P430S, p.L519Q	Monoallelic	Variable	>1200 cells/μl	No	None	None

^aData obtained from ClinGen database along with publications from Bosticardo, Du, Giardino, Moses, Pasternak, Chong-Deto

^bCompound heterozygous patients

^cPatient received a bone marrow transplant prior to FOXN1 variant identification

In a human thymus, FOXN1 enables TECs to differentiate from two bipotent progenitors into cortical TECs (cTECs) and medullary TEC (mTECs) subsets ^{23, 24}. These two TEC subsets are responsible for the positive and negative selection of T cells ^{25, 26}. This is initiated as developing CD4⁺CD8⁺ thymocytes begin co-expressing the alpha and beta chains of the T cell receptor (TCR) complex ²⁶. FOXN1 regulated transcripts, expressed by both cTECs and mTECs, control the developmental fate of the thymocytes. This involves interactions between the αβ TCR and peptide/HLA complexes expressed on TECs. In this process, T cells are selected to survive by recognizing but not overtly reacting to self-peptide/HLA molecules (positive and negative selection) ²⁵. During thymopoiesis, the CD4⁺CD8⁺ develop into CD4 and CD8 single positive (SP) cells, moving from the thymic cortex to the medulla. CD4⁺ and CD8⁺ SP interactions with medullary TECs enforces the selection process. The mTECs promiscuously express and process diverse tissue specific proteins into peptides loaded onto HLA molecules and presented to the SP cells. Potentially autoreactive T cells are purged due to strong intracellular signals. Concomitantly, the mTECs also support regulatory T cell (Treg) development ^27-30. The mTECs develop tissue specific identities, with the number of mTEC subsets currently ~15 based on single cell RNA sequencing ^{28, 29, 31}. Commonly referred to as mimetics, some of these mTEC subsets lose FOXN1 expression over time ²⁸. Importantly, the thymocyte interactions with the various TEC subsets establishes a T cell repertoire unique to each individual ^{26, 32}.

So how does FOXN1 participate in T cell development? First, FOXN1 positively regulates the expression of ~500 genes (targets the GACGC nucleotide sequence), controlling the development, differentiation and expansion of TECs into the various subsets ^33-35. Second, FOXN1 gene targets in cTECs include the chemokines (CCL19, CCL21, CCL25) that recruit ETPs. Third, FOXN1-regulated transcripts such as peptidases and components of the proteosome (thymoproteome) are expressed ^{33 36}. These enzymes and transporters degrade proteins into peptides bound by HLA class I ³⁷. Finally, FOXN1 positively regulates a cTEC specific serine protease (encoded by Prss) required for CD4 lineage commitment via MHC class II ^{33, 38}. Given such critical roles of FOXN1 in T cell development, loss of function variants can have a devasting impact on thymopoiesis.

With regards to the hairless phenotype, a subset of keratin genes (e.g., Krt32, Krt33b, Krt84) are positively regulated by FOXN1 in the skin ^{35, 39, 40}. FOXN1-deficiency diminishes the levels of these and other keratins along with keratin associated protein (KAP) within the hair follicle. The keratin genes do not appear to be regulated by Foxn1 in TECs ³³. In the hair follicle, some Foxn1-regulated keratins overlap with those controlled by Hoxc13, which, in turn, positively regulates Foxn1 expression ¹¹. The consequent reduction in keratins prevents the hair shaft from forming sufficient tensile strength to extrude through the dermal layers. Lower keratin levels likewise impacts the growth and strength of the nail bed. Complete loss-of-function FOXN1 variants cause severe onychodystrophy, with a thinner matrix that does not have a defined border with the epithelium in the nail bed ⁴¹. Of note, a monoallelic FOXN1 deficiency can result in nail dystrophy, albeit less severe ⁴². An interesting and less-appreciated aspect of a FOXN1-deficiency is scar-less wound healing ³⁹. Thus, absence of Foxn1 in mice (nu/nu) diminishes the fibrotic tissue that forms on the surface of the skin after wounding ³⁹. In the basal layers, there is also limited deposition of extracellular matrix (ECM) proteins such as keratins ⁴³.

Patients with biallelic loss-of-function FOXN1 variants lack T cells due to congenital thymic aplasia and face recurring life threatening viral, bacterial, and fungal infections along with a lack of hair and nail dystrophy ^{12, 16, 17, 42, 44}. Laboratory measures that are specified in the European Society for Immunodeficiency guidelines for defining athymia include low circulating CD3⁺ T cell counts <50 cells/μl, low naïve T cells (<50 cells/μl; <5% of total T cells), and low to undetectable T cell receptor excision circle (TRECs) values (Figure 1, Table 1) ⁴⁵. The TREC assay is a PCR screen to identify recent thymic emigrants ^46-50. An expanding number of patients have been identified with single allelic and/or compound heterozygous FOXN1 variants. In many, peripheral T cell levels do not meet congenital athymia criteria, although values remain below the mean relative to age-matched controls (Table 1). These variants are sometimes found in patients via newborn TREC assays (Table 1) ^{40, 42}. Additionally, nail dystrophy and hair growth abnormalities are less pronounced and/or absent in those with single allelic and/or compound heterozygous FOXN1 variants ^{16, 40, 42, 44, 51, 52}. For a health care provider, patients with a single allelic or compound heterozygous FOXN1 variants are challenging due to incomplete information on the impact of the FOXN1 mutation on thymic development. Adding to this challenge is the variable clinical penetrance among patients with the same variant, sometimes inherited from a healthy parent. This makes prognosis and appropriate management for such patients difficult. With targeted exome and whole genome sequencing now commonly used to identify causal variants producing in-born errors of immunity, >820 human FOXN1 variants are listed in the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar/). Use of this database can sometimes provide clarity for the clinician, particularly when a FOXN1 variant is designated pathogenic. These are usually loss of function variants known to cause congenital athymia. Unfortunately, many of the cataloged FOXN1 variants are listed as “variants of unknown significance” (VUS). These VUS may be further classified as likely benign, uncertain significance, or likely pathogenic. What concerns exist for patients with monoallelic FOXN1 VUSs? How does one manage these patients? To improve on the accuracy of the ClinVar classifications, the ClinGen consortium (https://clinicalgenome.org/) is actively curating gene variants linked to severe combined immunodeficiency diseases including ADA, DCLRE1C, IL2RG, IL7R, JAK3, RAG1, RAG2 and recently, FOXN1 ⁵³. ClinGen is an FDA-recognized human genetic variant database using expert curated assertions and evidence-based summaries. Forty-nine FOXN1 variants have so far been curated in the ClinGen database (Figure 2A). In the following sections, practical considerations are provided for identifying and managing patients with the diverse types of FOXN1 variants, based in part on understanding how protein is function is impacted.

Figure 1.

Hierarchical clinical decision tree for patients with diverse FOXN1 variants. This process diagram illustrates the clinical course after a low TREC is reported following newborn screening. The chart follows the decision-making plan following various clinical orders and outcomes. Since the clinical outcomes can include other causes of thymic hypoplasia, such as 22q11.2 deletion syndrome, these are incorporated into the tree. Note <1% of 22q11.2DS patients have thymic aplasia and are often referred to as complete DiGeorge. Red diamond = patient characteristics; Blue box = clinical assay or intervention; Grey box = post-clinical plan; Green diamond = normal result. Information on clinical decision tree design ⁹². ATO = Artificial thymic organoid, used to determine if hematopoietic cell intrinsic defect or putative stromal cell defect ^93-95; AR= autosomal recessive, CID= Combined immunodeficiency; CMA= chromosomal microarray; Het= heterozygous; HCT= hematopoietic cell transplant; TES= targeted exome sequencing; WES=whole genome sequencing; WGS= whole genome sequencing.

Figure 2.

Human FOXN1 variant classifications. A. The ClinGen database has 49 curated human FOXN1 variants. These are designated as pathogenic, likely pathogenic, uncertain significance, likely benign and benign. A pie chart was used to analyze the number of missense, frameshift, nonsense, and splice cite variants in the different categories, with other included in the likely benign group. B. The exon organization of human FOXN1 is shown, with exons 2-9 of the nine exons forming the cDNA. The cDNA encodes a 648 amino acid protein that has defined domains that control thymopoietic potential (green), DNA binding (Forkhead domain; blue), nuclear localization sequence (dark blue) and transactivation region (yellow), the latter containing an aspartic acid rich region (red). The autosomal recessive complete loss of function variants are listed in red under the protein domains. An example of diverse variants and their locations within the various domains is shown in the dotted boxes. The information is a small sampling of the >760 variants listed in ClinVar. C. Different human FOXN1 variants were tested in functional assays with a luciferase reporter assay using the beta5t promoter region, a specific target of FOXN1. The relative activities of normal FOXN1 and several benign variants were compared to selected missense, frameshift, and nonsense mutations. Since the cDNA was used in this assay, splice variants are not shown. Noteworthy, there is considerable variation in the relative activity levels of the different FOXN1 variants, including two with gain-of-function features.

Identifying Patients with FOXN1 variants

A first indicator of severe thymic dysfunction is low TRECs levels. Two consecutive low TREC values are strongly suggestive of T cell deficiency. These patients need expeditious referral for confirmatory flow cytometry, either through central laboratories associated with a newborn screening program, or via an immunologist (Figure 1). Clinical practice guidelines for infants with defects in thymic function have been published by various groups ^{45, 49, 54}. At our center, recommendations include holding the infant out of daycare, avoiding crowded public places, sequestering the patient from sick family contacts and young siblings, and pausing breastfeeding to avoid potential CMV exposure while awaiting confirmatory testing ⁵⁵. These strategies minimize infectious exposures. A differential diagnosis is formulated based on the flow cytometry results (Figure 1). At our center, infants who are confirmed to have severe T cell deficiency (consistent with SCID or athymia) are admitted to the bone marrow transplant unit at the time of confirmation. This enables onboarding of medications, performing additional tests to identify genetic causes and addressing infectious or inflammatory complications. Various referral centers and providers differ with regard to whether patients are subsequently managed at home or hospitalized while awaiting definitive therapy for SCID or athymia. These decisions often take non-medical factors into account, although criteria for determining which patients are appropriate for home management are not well established ⁵⁵. For those patients in whom flow cytometry and other analyses confirm severe T cell deficiency (whether it is caused by congenital athymia or SCID), proactive clinical interventions are required. These can prevent community acquired and opportunistic infections that can cause organ damage and limit the efficacy of potentially curative therapies. The latter include thymic implants, hematopoietic cell transplants (HCT), or gene therapy (Figure 1). At our facility, established clinical recommendations for antimicrobial prophylaxis are provided along with monitoring while awaiting definitive therapy ⁵⁵. The genetic causes for congenital athymia, CID or SCID cause are assessed using targeted exome panels or whole exome/genome sequencing (Figure 1) ^{18, 45, 56}. Thymic stromal cell defects can arise from mutations in FOXN1 or FOXI3, PAX1, TBX1, TBX2, CHD7 (in the case of CHARGE syndrome) along with 22q11.2 deletion syndrome ¹⁹.

Outfoxing FOXN1 Variants for Patient Centric Clinical Care

Patients who present with either a T⁻B⁺NK⁺ congenital athymia or a T^loB⁺NK⁺ presentation have been identified with different FOXN1 variants^{15, 16, 40, 42, 51, 52, 56-60}. Congenital athymia is usually seen in those with biallelic pathogenic variants and a few with compound heterozygous variants (e.g., p.D313fs/W363Cdel5aa and p.R114X/E139fs, both in trans) ^{14-16, 40, 52, 60}. For patients with congenital athymia caused by FOXN1 biallelic or compound heterozygous mutations, the recommended treatment is a donor thymic implant (referred to as thymus transplant in the United Kingdom) ^{45, 49, 61}. Early thymic progenitors from host hematopoietic cells will repopulate the donor thymus tissue, restoring naïve T cell development, which can take months to years ^61-63. (Table 1, Figure 1) ^{54, 63, 64}. Availability of this therapy is limited in the United States due to the high costs and insurance issues. There are certain cases where FOXN1 deficient (and other athymic) patients have received a T cell replete allogeneic hematopoietic stem cell transplant (HCT) ^{40, 65}, which is commonly done for CID or SCID patients (Table 1)^{18, 45, 49, 54}. While HCT can provide some level of immune reconstitution for patients with congenital athymia, the recipient will remain TREC negative as there are no host thymically derived T cells. Immune reconstitution comes from an expansion of mature donor T cells. This is not ideal as the donor T cells represent a relatively small pool of clones and thus provide limited protection from infections⁶⁶ Mature donor T cells also mediate graft-versus host disease, which is a risk in this scenario ^{67 68}.

The explosion of next generation DNA sequencing for patients has dramatically expanded the number of FOXN1 variants in the databases ^{15, 16, 40, 42, 51, 52, 56-60}. These include nucleotide deletions, insertions, missense mutations and/or splice site variants in the FOXN1 locus. The impact of single allelic and compound heterozygous FOXN1 mutations on TEC functions is difficult to ascertain ⁵¹. Some heterozygous FOXN1 variants are loss of function, yet their impact on the normal FOXN1 protein can vary considerably. Even more challenging is deciding whether a VUS accounts for a patient’s T cell deficiency. Wavering depressions in autologous naïve T cells may occur ^{16, 40, 42, 52}. Atypical or Omenn-like presentations have been reported, wherein autologous oligoclonal T cell populations expressing activation/memory markers (CD45RO⁺, HLA-DR) have expanded in the periphery ^{16, 40, 42, 52}. Diminished T cell proliferation results can be somewhat misleading in these patients. While T cell proliferation assays can be lower, this does not provide a good indicator of T cell function ¹⁸. FOXN1 is expressed in TECs and not T cells, so T cell proliferation in response to mitogenic stimulus should not be affected. This is evident if one normalizes for absolute T cell numbers or uses a flow cytometry-based assay in which proliferation can be assessed specifically in T cells. Adding to the difficulty in deciding on a clinical course, heterozygous loss-of-function FOXN1 alleles can have a damaging impact in early infancy due to very low TRECs and a T cell lymphopenia ^{16, 42, 51, 52}. This is quite variable among the cohorts studied and for many, T cell output from the thymus can improve over time ⁵¹. In this case, the clinician should monitor the changes in T cell numbers in the first few years of life. The same FOXN1 variant can sometimes be identified in healthy parents who have only mild T cell deficiencies, as will be discussed below.

As mentioned, many of the >820 FOXN1 variants listed in ClinVar are VUSs and likely benign. To outfox the FOXN1 variant for better clinical insights into patient management, an understanding of the type and location of the mutation is beneficial. FOXN1 is located on chromosome 17q21.2 and spans 9 exons. The coding region spans exons 2-9, forming a 648 amino acid protein (Figure 2A). There are 4 key domains required for protein function, an NH2-terminal region supporting T cell development (thymopoietic potential), the DNA binding region, a nuclear localization sequence and a COOH-terminally located transactivation domain. Mutations in specific amino acids in any of these 4 domains can cripple protein function (Figure 2B). First, the NH2-terminal domain (exon 3) confers thymopoietic capabilities to FOXN1 (amino acids 42-198) ⁶⁹ Patient variants identified in this region include p.R114Ter and p.S188fs. Second is the Forkhead box domain, which lies between amino acids 271-378. The Forkhead domain contains the DNA binding region, which targets the nucleotide sequence GACGC ³³. The crystal structure of this domain with the GACGC sequence revealed several key polar interactions involving the amino acids K271, Y276, K285, N317, R320, H321, S323, K331, C344, and W346 and the DNA ⁷⁰ N322 had polar interactions with a water molecule needed to interact with GACGC ⁷⁰. Mutations in any of these key amino acids will affect either the DNA binding affinity and/or the nucleotide recognition specificity (Table 1). Variants identified in patients at these key positions include p.R320W, p.N322S, p.R341C, and p.W346R. Third is a nuclear localization signal between amino acids 314-344 within the Forkhead domain that enables the protein to enter the nucleus ⁵¹. The FOXN1 variants p.R320W, p.L325P and p.C328R had diminished nuclear localization ⁵¹. Fourth is the COOH-terminal region that contains a transactivation domain. Encompassing amino acids 485-548, this region has several critically spaced aspartic acids essential for transactivation activity, p.D517, p.D521, p.D524, p.D528, p.D530 ⁷¹. One patient with a FOXN1 variant in this region is p.L519Q. While listed as benign, the impact of this variant on the patient was not reported. An effective procedure to determine the functional impact of a mutation on FOXN1 is a luciferase reporter assay ³³. In this assay, a promoter target of FOXN1 (Psmb11) is linked to the reporter (luciferase). This is transfected into a cell line along with an expression vector containing FOXN1. By comparing the levels of luciferase activity achieved with a normal FOXN1 control to the variant, a relative activity measure can reveal the functional impact of the mutation (Figure 2C) ^{40, 72}. This approach has uncovered complete and partial loss of function and two gain of function FOXN1 mutations (p.H321R and p.N322S)⁵¹. Clinically, any individual presenting with biallelic loss-of-function variants established in the luciferase reporter assay would be pathogenic. However, a single allelic loss-of-function variant can have different outcomes. For example, a cluster of heterozygous FOXN1 variants can antagonize the function of the normal protein due to a dominant negative impact ⁵⁹. Evidence suggests that FOXN1 variants with a single nucleotide deletion after the DNA binding domain, causing a frameshift with a stop codon at amino acid 550, will function as a dominant negative ^{59 51}. While the frameshift can start at varying locations near the end of the DNA binding domain, the key change is the loss of the aspartic acid residues involved in transactivation activity. The group of dominant negative FOXN1 variants include any with nucleotide deletions (or insertions) between the coding nucleotides from c.1127 to c.1538 (p.378 to p.518). These frameshifted, truncated FOXN1 variants form a relatively stable mRNA, bypassing nonsense mediated RNA decay ⁵⁹. The dominant negative protein is proposed to complex with wildtype FOXN1, forming punctate aggregates in the nucleus that appear to antagonize normal protein function ⁵⁹. Patient FOXN1 variants that form such truncated dominant negative proteins include p.P401Afs, p.P402Lfs, p.L404Cfs, p.P432fs, p.L439Cfs, p.Y455Cfs, p.H457Pfs, p.P465Rfs, p.P473Hfs, and p.Q489Rfs (Table 1, Figure 2B). With the luciferase reporter assay, the ability of any FOXN1 variant to act as a dominant negative can be determined by co-transfecting both the variant and the normal control. If dominant negative, an equimolar amount of variant will completely antagonize the wildtype FOXN1 protein (Figure 2C) ⁵¹ .

Patients that express such FOXN1 frameshift dominant negative variants often have low TRECs at birth along with a corresponding T cell lymphopenia. For the majority, the T cell counts will gradually improve over time ^{42, 51, 59}. Yet, the clinical impact of these dominant negative mutations can vary widely, even within several members of the same family. Some patients have minimal infectious phenotypes. Other patients with the same mutation (sometimes within the same family) can have recurring respiratory viral infections, pneumonia, or opportunistic infections such as disseminated CMV, and Epstein Barr Virus (EBV)-driven lymphoproliferative disorders or lymphoma. Patients with heterozygous FOXN1 mutations do not require thymic implant as either their T cell deficiency is not severe, or their family history suggests that the magnitude of their T cell deficiency will improve over time. Such patients may be categorized, using recently established recommendations for congenital athymia, as either “thymic hypoplasia” or “borderline cases”. This pattern warrants careful follow-up, supportive care, infectious prophylaxis or immunomodulatory therapies depending on the degree of immunodeficiency or immune dysregulation, and consideration of thymic implantation for the most severe cases ⁴⁵. Examples of such scenarios include one study where 5 infants out of 47 patients with heterozygous FOXN1 variants had severe recurrent infections ⁴². Four of these five had a frameshift variant after the DNA binding domain, forming a dominant negative protein ⁴². A distinct group of 7 infants with an identical p.P401Afs144 frameshift variant had varied phenotypes. One had absent TRECs, correspondingly low CD3 T cells (<100 cells/μL) and absent CD8 cells. Four of the 7 had TREC levels recorded as “below State threshold,” with CD3 cells ranging from 327 to 928 cells/μL. B cells were between 81 and 135 cells/μL. Two were adults, one with a prior history of infections and the 2^nd with no clinical manifestations ⁴². In another report, a father and his 3 children shared the same p.P465Rfs82 heterozygous FOXN1 frameshift variant ¹⁶. All 3 siblings had very low T cell counts at younger ages, and one passed away due to severe graft versus host disease, after receiving an allogeneic umbilical cord blood stem cell transplant for severe immunodeficiency. For the remaining children and adult, T cell counts trended to below or at the lower end of normal ranges. A third study reported on two patients the single allelic p.Q489Rfs frameshift variant who had low T cells but no infectious history ⁴⁰. A distinct patient had a compound heterozygous mutation, p.Q489Rfs and p.P430S. CD3 T cells were 137 cells/μl in the second week of life and dropped to 79 cells/μl at 2.5 months of age ⁴⁰. Prior to identifying the two FOXN1 variants, this patient received a HCT and has been doing well since. A FOXN1 p.Q489Rfs patient was also reported in another study ⁵². At time of assessment, that patient’s T cell levels were at 46 cells/μl, which increased to 483 cells/μl after 2 years. Another patient was described with a heterozygous frameshift variant within the Forkhead domain of FOXN1, p.N349Ifs201 ⁵². This created a truncated protein with a coding portion of the DNA binding domain lost. The patient had low TRECs (13 copies/μl) with CD3 T cell counts at 2270 cells/μl. The T cells dropped to 1400 cells/μl after 30 months. It remains unclear whether this variant acted as a dominant negative.

There are single allelic FOXN1 variants that are not dominant negative and still result in low TRECs and naïve T cell numbers in newborns ⁴². It remains unclear why these act in this manner. In affected individuals, the T cell numbers have been shown to improve over time, suggesting thymic implant would not be appropriate. This may simply reflect a gene dosage requirement. A second explanation is monoallelic expression of FOXN1 in TECs, which has been reported for other transcription factors ^73-75. Alternatively, the mutation could affect protein-protein interactions between FOXN1 and other transcription factors that coordinately regulate its expression in a feedforward loop. For example, the SIX1 and SOX9 transcription factors coordinately assemble with FOXN1 to control its expression ⁵⁷.

Clinical Approaches in the Future

The TREC newborn screen has been incredibly valuable in identifying infants with congenital athymia during their first weeks of life ⁴⁷. For those with congenital athymia due to biallelic FOXN1 loss of function variants, a thymic implant is clinically approved and appropriate (Figure 1) ^{54, 63}. As this relies on allogenic thymic tissues from living donors (typically infants), availability of such tissue may become more limiting given the increasing understanding of the negative long term consequences when donors are thymectomized in infancy ^{76, 77}. Research into alternative approaches will be crucial to meet the clinical needs of those with congenital athymia or severe thymic hypoplasia. One alternate approach garnering attention is the use of pluripotent thymic epithelial progenitors (iTEPs), developed from patient cells. Generating patient specific iTEPs with cDNAs encoding wildtype FOXN1 could enhance the formation and expansion such cells due to the central role of FOXN1 in TEC growth. In animal models, human iTEPs can restore thymopoiesis ^78-81. Advances in this field include the development of human thymic organoids ^82-85. The current impediment to develop such organoids is the limited ability to re-create an effective, large 3-dimentional meshwork needed for thymus functions. This has been partly overcome by using donor thymuses that are stripped of all cells, with the remaining collagen/extracellular matrix protein scaffold enabling the formation of larger 3-d structures ^{86, 87}. The use of host derived human thymic organoids, as opposed to allogeneic thymic tissue, may also resolve the issue of autoimmunity often occurring in patients with thymic implants ⁶¹.

Patients with single allelic loss-of-function and/or dominant negative activities remain a difficult cohort to manage. Thymic implants are not recommended. Future curative strategies include either gene editing technologies to correct the FOXN1 mutation in iTEPs or introduce high levels of an unmutated FOXN1 gene into the thymus to overcome the non-functional or dominant negative FOXN1 variant. In dose response studies, increasing the levels of normal FOXN1 relative to a dominant negative can restore sufficient transcriptional activity to drive target gene expression ⁵¹. Several strategies exist to introduce FOXN1 into thymic stromal cells in vivo. First is direct intrathymic injection of a recombinant TAT-FOXN1 fusion protein ⁸⁸. The TAT domain enables FOXN1 to directly translocate into the stromal cells, and intrathymic injections of this fusion protein restored TEC functionality in irradiated mice whose thymic stromal cells are damaged ^{88, 89}. Incorporating the CCR9 protein domain at the N-terminal end of FOXN1 to create a fusion protein (rCCR9-TAT-FOXN1) that enables it to be injected IV. Done in mice, the recombinant FOXN1 accumulates in the thymus ⁹⁰. This improved TEC cellularity in aged mice when FOXN1 levels begin to diminish ⁹⁰. A similar recombinant protein strategy could be applied to patients carrying dominant negative or single allelic variants of FOXN1 as a means to improve TEC functionality in the thymus. Second, advances in nanotechnologies could soon enable a directed distribution of hFOXN1 stabilized mRNAs (like mRNA vaccines) to the thymus ⁹¹. New regulators that can be introduced as mRNA nanoparticles include the SIX1 and SOX9 transcription factors, which positively regulate FOXN1 expression in TECs ⁵⁷. Such strategies may be a turning point for clinical management of patients with transient or permanent disruption of TEC functions due to FOXN1 mutations.

Summary

The expanding number of monoallelic and biallelic FOXN1 variants makes clinical decisions on patient care beguiling. Understanding the location and impact of the variant on FOXN1 function alleviates some of these concerns given newfound knowledge of the 4 distinct domains within the protein (Figure 2B). As this information continues to be compiled in databases such as ClinVar and ClinGen, more informed clinical decisions, such as the need for a thymic implant, are becoming more manageable.

Abbreviations:

ATO

Artificial thymic organoid

AR

Autosomal recessive

BMT

Bone marrow transplant

CID

Combined immunodeficiency

CMA

Chromosomal microarray

DN

Dominant negative

FOXNI

Forkhead Box N1

HCT

Hematopoietic Cell Transplantation

Het

Heterozygous

SCID

Severe combined immunodeficiency

SCT

Stem cell transplant

TES

Targeted exome sequencing

TECs

Thymic epithelial cells

TCR

T cell receptor

TRECs

T cell receptor excision circles

WES

Whole exome sequencing

WGS

Whole genome sequencing