Identification of cellular retinoic acid binding protein 2 (CRABP2) as downstream target of nuclear factor I/X (NFIX): implications for skeletal dysplasia syndromes

Abstract Nuclear factor I/X (NFIX) mutations are associated with 2 skeletal dysplasias, Marshall-Smith (MSS) and Malan (MAL) syndromes. NFIX encodes a transcription factor that regulates expression of genes, including Bobby sox (BBX) and glial fibrillary acidic protein (GFAP) in neural progenitor cells and astrocytes, respectively. To elucidate the role of NFIX mutations in MSS, we studied their effects in fibroblast cell lines obtained from 5 MSS unrelated patients and 3 unaffected individuals. The 5 MSS NFIX frameshift mutations in exons 6-8 comprised 3 deletions (c.819-732_1079-948del, c.819-471_1079-687del, c.819-592_1079-808del), an insertion (c.1037_1038insT), and a duplication (c.1090dupG). Quantitative reverse transcription polymerase chain reaction (qRT-PCR) and western blot analyses using MSS and unrelated control fibroblasts and in vitro expression studies in monkey kidney fibroblast (COS-7) cells showed that frameshift mutations in NFIX exons 6-8 generated mutant transcripts that were not cleared by nonsense-mediated-decay mechanisms and encoded truncated NFIX proteins. Moreover, BBX or GFAP expression was unaffected in the majority of MSS fibroblasts. To identify novel NFIX downstream target genes, RNA sequencing and proteomics analyses were performed on mouse embryonic fibroblast (MEF) cells derived from control Nfix+/+, Nfix+/Del2, Nfix+/Del24, NfixDel24/Del24, Nfix+/Del140, and NfixDel140/Del140 mice, compared with NfixDel2/Del2 mice which had developmental, skeletal, and neural abnormalities. This identified 191 transcripts and 815 proteins misregulated in NfixDel2/Del2 MEFs with ≥2-fold-change (P <0 .05). Validation studies using qRT-PCR and western blot analyses confirmed that 2 genes, cellular retinoic acid binding protein 2 (Crabp2) and vascular cell adhesion molecule 1 (Vcam1), were misregulated at the RNA and protein levels in NfixDel2/Del2 MEFs, and that CRABP2 and VCAM1 expressions were altered in 60%–100% of MSS fibroblast cells. Furthermore, in vitro luciferase reporter assays confirmed that NFIX directly regulates CRABP2 promoter activity. Thus, these altered genes and pathways may represent possible targets for drugs as potential treatments and therapies for MSS.


Introduction
Marshall-Smith syndrome (MSS; MIM 602 535) is a rare autosomal dominant disorder, characterized by growth retardation, short stature, distinctive facial features (comprising of a high forehead, proptosis, blue sclerae, anteverted nares, small and retracted mandible, gingival hypertrophy, and hypertrichosis), skeletal abnormalities, delayed motor and neural development, respiratory complications, and postnatal failure to thrive. 1,2][5] NFIX, located on chromosome 19p13.2,contains 11 exons and encodes 14 transcripts through differential splicing and the use of different transcription initiation sites, of which 11 are protein coding.NFIX isoform1, representing the canonical sequence, is a ubiquitously expressed 502 amino acid protein, which contains a conserved 194 amino acid N-terminal DNA binding and dimerization domain, and a C-terminal transactivation/repression domain of variable length, due to alternative splicing of exons 7 and 9.In mammals, the NFI gene family consists of 4 closely related genes (NFIA, NFIB, NFIC, and NFIX), which encode transcription factors that bind as homoor heterodimers to the consensus palindromic sequence 5'-TTGGC(N5)GCCAA-3' present in the promoter regions of viral and cellular genes, to either activate or suppress transcription. 64][5] In contrast, entire gene deletions or heterozygous missense, nonsense, and frameshift NFIX mutations that predominantly affect exon 2, which encodes the N-terminal DNA binding and dimerization domain, typically lead to transcripts that are cleared by NMD.The resulting NFIX haploinsufficiency is associated with an overgrowth disorder called Malan (Sotos-like) syndrome (MAL; MIM 614 753), which is characterized by unusual facial phenotype, skeletal dysplasia, intellectual disability, and behavioral problems. 3,5,7he role of NFIX in MSS is currently poorly understood.NFIX has been reported to bind to the promoter region of the glial fibrillary acidic protein (GFAP) gene to activate gene transcription in astrocytes, 8 while binding of NFIX to the Bobby Sox (BBX) enhancer element suppresses transcription in neural progenitor cells. 9GFAP encodes intermediate filaments and is important for differentiation of cortical precursor cells into mature astrocytes during brain development 8 and NFI family members have previously been reported to be important factors driving astrocyte differentiation during development of the central nervous system. 10BBX encodes a transcription factor that binds to DNA to promote cell cycle progression from the G1 to S phase. 9In order to elucidate the role of NFIX mutations in MSS, we initially studied their effects on GFAP and BBX expression in fibroblast cell lines obtained from 5 MSS patients, and then utilized these fibroblasts and mouse embryonic fibroblasts (MEFs) derived from wild-type (Nfix +/+ ) mice, heterozygous (Nfix +/Del2 , Nfix +/Del24 and Nfix +/Del140 ) and homozygous (Nfix Del2/Del2 , Nfix Del24/Del24 and Nfix Del140/Del140 ) mutant mouse models 11 in hypothesisfree approaches of RNA sequencing and proteomic analyses (Figure S1).Our approach led to the identification of 2 genes as novel downstream targets of NFIX, which may represent cellular pathways that could be used for targeted drug discovery as potential treatments for MSS patients.

Study approval
Written informed consent was obtained from patients or their legal guardians, using protocols approved by the local and national ethics committees.All animal studies were approved by the Medical Research Council Harwell Institute Ethical Review Committee and were licensed under the Animal (Scientific Procedures) Act 1986, issued by the UK Government Home Office Department (PPL30/2433 and PPL30/3271).

Nucleic acid analyses, in vitro expression assays, RNA sequencing, proteomics, and in silico analyses
Extraction of DNA and RNA, DNA and RNA sequence analyses; generation of NFIX expression constructs, immunofluorescence, quantitative reverse transcription polymerase chain reaction (qRT-PCR), and western blot analyses; and dual luciferase reporter activity assays, chromatin immunoprecipitation (ChIP) followed by real time-PCR (ChIP-RT-PCR), RNA sequencing, proteomics, and in silico analyses, were performed as described in Supplemental Materials and Methods.

Statistical analysis
Data were expressed as mean and SD or SEM.All analyses were performed using Prism (GraphPad), and a value of P <0 .05 was considered significant for all analyses, as described in Supplemental Materials and Methods.

Confirmation of NFIX mutations in MSS patient fibroblasts and effects on NFIX, BBX , and GFAP expression
The fibroblast cell lines from the 5 MSS patients were confirmed to have heterozygous NFIX mutations in exons 6-8, and these comprised 3 deletions (c.819-732_1079-948del (Del1), c.819-471_1079-687del (Del2), c.819-592_1079-808del (Del3)), an insertion (c.1037_1038insT), and a duplication (c.1090dupG), resulting in truncated NFIX proteins 3,4 (Figure S2).No coding NFIX mutations were detected in the fibroblasts from the 3 unrelated controls.The NFIX mutations did not affect NFIX mRNA levels in the 5 MSS fibroblast cell lines, which were not significantly different to those in the unrelated controls (Figure S3A), thereby confirming that the MSS mutant NFIX transcripts, as expected, are not cleared by NMD.However, NFIX protein expression could not be detected in any of the MSS or unrelated control fibroblasts by western blotting, indicating that NFIX protein levels were below the level of detection.
NFIX is reported to suppress BBX transcription but activate GFAP expression, in neural progenitor cells and astrocytes, respectively.However, there was no significant difference in BBX expression at either the mRNA (Figure S3B) or protein (Figures S3D-E) levels between fibroblasts from MSS patients or unrelated controls.In addition, there was also no significant difference in GFAP expression in 4 of the 5 MSS fibroblast cell lines, when compared with unrelated controls, while the MSS c.1090dupG was associated with reduced GFAP expression (-12 fold, n = 4, P <0 .05) at the mRNA (Figure S3C) and protein (-24 fold, n = 4, P <0 .01, Figure S3D-F) levels, when compared with the unrelated controls.
These effects of MSS-associated NFIX mutations were further investigated by in vitro expression assays in which Nterminal-Flag tagged NFIX wild-type (NFIX WT) and MSSmutant (NFIX InsT, NFIX DupG, and NFIX Del comprising the loss of exons 6 and 7 that is representative of the deletions found in MSS fibroblast cell lines Del1, Del2 and Del3) cDNA constructs were transiently transfected into COS-7 cells.Immunofluorescence analysis showed that the NFIX mutations had no effect on the cellular localization of the NFIX protein with WT, InsT, DupG and Del NFIX proteins all predominantly located within the nucleus (Figure 1A).qRT-PCR analysis showed that there was no significant difference in NFIX mRNA expression between WT and the MSSmutants (Figure 1B), thereby confirming the results obtained in the MSS fibroblasts.The combined use of NFIX overexpression and antibodies against the N-terminal-Flag tag allowed NFIX WT and mutant protein expression to be detected in the transiently transfected COS-7 cells by western blot analysis (Figure 1C and D).In each case the MSSassociated NFIX mutations resulted in the expression of truncated proteins of reduced molecular weight (<50 kDa) compared with WT NFIX (55 kDa) (Figure 1C).However, the levels of all mutant NFIX proteins were significantly lower than the level of WT NFIX (n = 4, P <0 .001, Figure 1C and  D).These results further confirm that the MSS-associated NFIX mutations are transcribed and translated to produce mutant truncated proteins, and that mutations in the Cterminal part of the NFIX gene do not result in transcripts being cleared by NMD.
The observed reduction in GFAP expression associated with the NFIX c.1090dupG MSS mutation in the fibroblasts (Figure S3C, D, and F) was further assessed using reporter constructs comprising the luciferase reporter gene downstream of the GFAP promoter, which were transiently cotransfected with WT or MSS-mutant NFIX cDNA constructs into COS-7 cells.WT NFIX activated the GFAP promoter and caused a ∼12-fold increase (n = 4, P <0 .0001, Figure 1E) in luciferase activity in cells with the GFAP promoter cloned in the forward orientation compared with cells with the GFAP promoter cloned in the reverse orientation.In contrast, the NFIX DupG mutation caused a 2.5-fold reduction (n = 4, P <0 .0001) in luciferase activity compared with WT NFIX (Figure 1E), confirming a partial loss of function of NFIX transactivation activity at the GFAP locus, and consistent with results obtained in the MSS NFIX c.1090dupG patient fibroblasts.Luciferase reporter activity was unaffected by the MSS-associated NFIX InsT and Del mutations when compared with WT NFIX (Figure 1E).

Identification of novel NFIX downstream target genes
The majority of the examined MSS-associated NFIX mutations did not act via the previously reported target genes, BBX or GFAP, and we therefore hypothesized that other downstream target genes may be involved in skeletal biology.To identify such novel NFIX downstream target genes that may be misregulated in MSS patients, we chose to initially undertake RNA sequencing and proteomics studies in MEFs because there is reduced genotypic variability in mice generated on the same genetic background compared with fibroblasts derived from unrelated MSS patients, thereby maximizing our chances of identifying statistically significantly altered pathways.We used MEFs derived from wild-type (WT; Nfix +/+ ) mice, and previously established Nfix +/Del2 , Nfix +/Del24 , Nfix +/Del140 , Nfix Del2/Del2 , Nfix Del24/Del24 and Nfix Del140/Del140 mutant mice 11 that had deletions in exon 7 comprising: a 140 nucleotide deletion (Del140) that caused skipping of exon 7 due to alternative splicing of exon 6 to exon 8 and production of wild-type short Nfix isoforms that lack exon 7; an in-frame 24 nucleotide deletion (Del24) which caused loss of 8 amino acids; and a 2 nucleotide deletion (Del2) that caused a frame shift and a premature termination. 11Only the Nfix Del2/Del2 mice showed a range of MSS phenotypes, while Nfix +/+ , Nfix +/Del2 , Nfix +/Del24 , Nfix Del24/Del24 , Nfix +/Del140 and Nfix Del140/Del140 mice were phenotypically normal. 11Expression of Bbx was found to differ by <2-fold between WT and mutant MEFs at the RNA level (Table S1), consistent with results obtained in the MSS human fibroblasts, but was absent at the protein level, while Gfap expression was below the level of detection in the WT and mutant MEFs.

Validation of altered MEF gene expression in MSS human fibroblasts
To validate the changes observed in the expression of the Crabp2, Vcam1, Kctd12 and Idi1 genes revealed by the RNA and proteomic analyses in the MEFs, their expression in the MEFs and human fibroblasts was assessed by qRT-PCR ( Figure 2A-H) and western blot (Figure 3A-J) analyses, respectively.Crabp2 expression was confirmed by qRT-PCR analysis to be significantly increased in the Nfix Del2/Del2 MEFs (2.4-fold, P <0 .0001),compared with the mean expression in the Nfix +/+ , Nfix +/Del2 , Nfix +/Del24 , Nfix Del24/Del24 , Nfix +/Del140 and Nfix Del140/Del140 MEFs (Figure 2A).Western blot analysis also showed significantly increased CRABP2 expression in Nfix Del2/Del2 MEFs (6fold, P <0 .0001, Figure 3A and C).Furthermore, CRABP2 expression was significantly increased in the MSS human fibroblasts that had the c.819-592_1079-808del, Del3 (2-fold (P <0 .01)at the RNA level and 6.6-fold (P <0.0001) at the protein level), and c.1037_1038insT (3.3-fold (P <0 .0001)at the RNA level and 5.6-fold (P <0 .0001)at the protein level), when compared with the mean of the unrelated control human fibroblasts (Figures 2B and 3B and D).CRABP2 expression was also increased in the MSS human fibroblasts that had the c.1090dupG (4.5-fold change, P <0 .001),compared with unrelated controls, although this was only significant at the protein level (Figures 2B and 3B and D), possibly due to the high degree of variability in expression between the 3 unrelated control human fibroblasts.Surprisingly, CRABP2 expression in the MSS human fibroblasts with the c.819-471_1079-687del (Del2) was reduced, compared with the unrelated control fibroblasts, although this was only significant at the RNA (0.2-fold change, P <0 .05, Figure 2B), and not the protein level (Figure 3B and D).These findings suggest that defects in NFIX generally lead to increased CRABP2 expression.
Kctd12 expression, assessed by qRT-PCR and western blot analyses, in the MEFs did not reveal a significant difference between Nfix Del2/Del2 MEFs and all the other MEFs (Figures 2E and 3A and G), thereby not supporting the RNA sequencing and proteomic data, and differences in KCTD12 expression were also not detected in the human MSS and unrelated control fibroblasts (Figures 2F and 3B and H).
Overall, expression of the CRABP2 (Crabp2) and VCAM1 (Vcam1) genes was found to be consistently altered in the Nfix Del2/Del2 MEFs compared with all other MEFs, and in the majority of MSS human fibroblast cells, and we therefore focused further investigations on the promoters of these 2 genes.

Effects of NFIX on CRABP2 and VCAM1 expression in vitro
In silico analysis of the 5' untranslated regions of the CRABP2 and VCAM1 genes identified putative nuclear factor I (NFI) binding sites (palindromic sequence 5'-TTGGC(N5)GCCAA-3'), to which NFIX has been reported to bind. 12These comprised one potential NFI binding site in the human and mouse genomes at positions −2176 to −2163 and −1822 to −1809 upstream of the ATG start site of the CRABP2 and Crabp2 genes, respectively, which were conserved (Figure 4A); and another at positions −238 to −224 and −202 to −188 upstream of the ATG start site of the VCAM1 and Vcam1 genes, respectively, which were conserved (Figure 4A').
To investigate the effects of these identified putative NFI binding sites on promoter activity, luciferase reporter constructs under the transcriptional control of either the wildtype or mutant CRABP2 (Figure 4B) or VCAM1 (Figure 4B') promoters, consisting of wild-type (WT) or mutated (MUT) NFI binding sites, respectively, were transfected into COS-7 cells.Mutation of the NFI binding site in the CRABP2 promoter resulted in a 1.3-fold (P <0 .001)increase in luciferase expression compared with the wild-type CRABP2 promoter (Figure 4C), while mutation of the NFI binding site in the VCAM1 promoter resulted in 0.1-fold (P <0 .001)decrease in luciferase expression compared with the wild-type VCAM1 promoter (Figure 4C').Moreover, transient co-transfection of N-terminal-FLAG tagged wild-type (WT) or MSS-mutant NFIX cDNA constructs (InsT, DupG or Del) with these luciferase reporter constructs under the transcriptional control of wild-type CRABP2 or VCAM1 promoter in COS-7 cells revealed all the NFIX mutants to significantly increase luciferase reporter activity driven from the CRABP2 promoter (2.0-fold (P <0 .0001),1.7-fold (P <0 .05),and 1.6-fold (P <0 .05),respectively), but not from the VCAM1 promoter, when compared with wild-type NFIX (Figure 4D and D').Furthermore, chromatin immunoprecipitation (ChIP) followed by real time polymerase chain reaction (ChIP-RT-PCR) using DNA samples from the 3 unrelated control fibroblast cell lines revealed significant enrichment of NFIX binding to the -2176 to −2163 genomic site upstream of the ATG start site of CRABP2 (P <0 .0001, Figure 4E and F), but not of the −238 to −224 genomic site upstream of the ATG start site of VCAM1, when compared with the non-specific IgG antibody (Figure 4E' and F').Thus, our studies confirm that NFIX can directly modulate the CRABP2 promoter by suppressing its activity, and that NFIX may indirectly affect VCAM1 expression, potentially via the retinoic acid (RA) pathway (Figure 5).

Discussion
Our studies provide further insights of the roles of NFIX mutations in causing MSS.Thus, the fibroblast and in vitro expression studies confirmed that frameshift mutations in NFIX exons 6-8 generate transcripts that escape NMD and encode truncated NFIX proteins.Moreover, the majority of the studied MSS-associated NFIX mutations did not act via   the known downstream targets, BBX or GFAP genes ( Figure 1 and Figure S3).However, our RNA sequencing and proteomics studies identified CRABP2 and VCAM1 as new NFIX downstream target genes that were misregulated in 60%-100% of the MSS human fibroblasts and in Nfix Del2/Del2 MEFs (Figures 2 and 3).Finally, in vitro expression and luciferase reporter assays confirmed that NFIX directly regulated the activity of the CRABP2 promoter, and that NFIX may indirectly affect VCAM1 expression (Figure 4), potentially via the RA pathway (Figure 5).These findings may help to provide some explanations for the occurrence of the range of skeletal and neuronal phenotypes in MSS patients as follows.
CRABP2, encoded by CRABP2, is a 138 amino acid (16 kDa) protein with strong binding affinity for RA and is a key component of the RA signaling pathway.RA is a metabolite and biologically active form of vitamin A, with complex and pleiotropic functions throughout life but particularly during embryogenesis where it regulates cell lineage and stem cell differentiation in developing tissue and organ systems, including the facial region, hindbrain, forebrain, eye and inner ear. 13CRABP2 regulates the expression of downstream genes in RA pathway by binding to and transporting RA from the cytoplasm to the nucleus, where RA binds to its nuclear retinoic acid receptors (RARs) and peroxisomal proliferation-activated receptors, which form heterodimers with retinoid X receptors and associate with retinoic acid response elements in the promoter regions of RA-responsive genes to modulate transcription of target genes.RA degradation occurs via the action of cytochrome P450 (CYP) enzymes such as Cyp26A1 and Cyp26C1 that clear RA from the body. 13RA's well-established effects during the development of numerous organs and tissues include skeletal growth, bone remodeling, and brain development. 13A negatively influences osteoblast function, proliferation and differentiation as well as deposition of mineralized matrix. 13RA is required for normal bone development, acting as a morphogen during limb development 14 as homozygous Crabp2 −/− knockout mice were reported to be viable, normal and fertile with the exception for minor limb malformation, with variable penetrance depending on the genetic background. 15Since CRABP2 is important for RA function, its expression has been reported to be upregulated through the RA pathway in a permissive cellular context, 16 and increased CRABP2 expression is generally linked to the activation of the RA pathway. 13High retinol (vitamin A1) intake is associated with decreased bone mineral density and increased fracture risk in humans, [17][18][19] and high dietary RA in mice has been shown to reduce bone mass by increasing osteoclastogenesis and decreasing osteoblastogenesis, without affecting bone mineralization. 20Moreover, shortterm treatment of murine primary calvarial osteoblasts with RA induced an increase in Crabp2 expression and other genes involved in retinol-dependent signaling and skeletal remodeling. 20In addition, homozygous Crabp2 em1(IMPC)Mbp mice (C57BL/6N background) generated via CRISPR-Cas9mediated deletion of exons 2-3 had heart abnormalities and decreased bone mineral content (International Mouse Phenotyping Consortium).Furthermore, the establishment of an RA gradient across the brain is required during development.In the forebrain, rostral-most tissues at the gastrula and neurula stages are believed to be devoid of RA signaling due to the action of Cyp26A1 and Cyp26C1. 13dministration of exogenous RA at such stages leads to drastic phenotypes with complete anencephaly or severe microcephaly due to inappropriate activation of RARs, which normally function as transcriptional repressors in anterior regions. 21Thus, altered CRABP2 expression due to MSSassociated NFIX mutations could potentially disrupt RA signaling and lead to the skeletal abnormalities and increased fracture risk, as well as CNS anomalies that are observed in MSS patients. 22CAM1, encoded by VCAM1, is a 739 amino acid (81 kDa) cell surface glycoprotein adhesion molecule that is a member of the Ig-superfamily of transmembrane proteins, and is expressed by cytokine-activated endothelial cells 23 with its expression being upregulated in response to inflammatory cytokines, chemokines and growth factors.24 Increases in VCAM1 expression in macrophages, fibroblasts and endothelial cells have been reported in rheumatoid arthritis, 25 while Vcam1 expression in murine osteoblasts is important for cellular adhesion and activation of osteoblasts to binding partners in bones, which are required for maintaining the signals regulating the balance between bone formation and bone resorption.26 VCAM1 is also important for the regulation of hematopoietic stem cells self-renewal, proliferation, differentiation, trafficking, mobilization and function in the bone marrow, 27,28 and conditional Vcam1 deletion in mice is reported to impair lymphocyte migration to the bone marrow.29 However, Vcam-1-deficient mouse embryos are not viable and die due to severe defects in placental and heart development.30 VCAM1 is also reported to maintain the integrity of the mesenchymal and neural stem cell niches and to act as an environmental sensor to injury and disease.31,32 For example, mesenchymal stem cells that reside in the perivascular niche of many organs can proliferate after kidney, lung, liver or heart injury to generate myofibroblasts 32 and are a major source of osteoblast-like cells during vascular calcification in a mouse model of chronic kidney disease.33 Finally, elevated VCAM1 levels have been reported in patients with extensive cerebral small vessel disease, 34 dementia, 35 Alzheimer's disease, 36 Parkinson's disease 37 and small vessel stroke.38 This suggests that increased VCAM1 expression in the endothelial cells of the CNS may have pathological actions, which can potentially exacerbate cognitive dysfunction. Incontrast, in experimental neurodegenerative disease and age-related cognitive dysfunction models, treatment with dietary supplements of alpha lipoic resulted in downregulation of inflammatory cell infiltration into the CNS and improved memory.39 Therefore, increased expression VCAM1 due to NFIX mutations potentially contribute to CNS and bone abnormalities as well as other anomalies such as renal cysts, nephrocalcinosis, hydronephrosis, congestion of the liver and spleen, cardiomegaly, right ventricular hypertrophy and heart failure, 1,11,40,41 which are reported in MSS patients.22 VCAM1 appears to be an indirect target of the RA pathway, possibly via the actions of intermediate transcription factors, non-classical associations of receptors with other proteins, or even more distant mechanisms.16 The VCAM1 promoter has restricted activity due to the presence of strong negative regulatory elements (silencers) situated between -1.7 kb and -288 bp upstream of the transcription start site.[42][43][44] In addition, the VCAM1 promoter has 2 important tandem nuclear factor kappa-β (NF-κβ) elements (activators) that are required to overcome transcriptional inhibition, which are responsible for cytokine-dependent activation of the VCAM1 promoter.[42][43][44] Reported inducers of the VCAM1 promoter include thrombin, tumor necrosis factor alpha (TNF-α) and RA.[42][43][44] However, the responsiveness of the VCAM1 promoter to RA appears to be cell-specific due to the variable distribution, number and combination of silencers and activators between cell types. Fo example, RA has been reported to activate p50/p65 induced Vcam1 expression in mouse neuroepithelial cells, which are precursors for neurons and glia, 43 while in human dermal microvascular epithelial cells RA has been shown to inhibit TNF-α induced VCAM1 expression.45 Moreover, RA has been reported to significantly increase VCAM1 antigen expression in human neuroblastoma cells.46 Our results showed variability in the expression of the NFIX downstream target genes CRABP2 and VCAM1 (Figures 2 and 3), between mouse and human fibroblasts, which might not be indicative of changes in gene expression in other skeletal or non-skeletal cell types and in vivo, and this could be due to cell autonomous, monoallelic, and stochastic variation in NFIA, NFIB, NFIC and NFIX expression.Since the NFI family members are ubiquitously expressed in partially overlapping patterns and have the same conserved N-terminal DNA binding and dimerization domain, all 4 related genes may recognize the same consensus sequence present in the promoter region of downstream target genes, and may compensate for one another in case of mutation by changing their expression pattern and acquiring new regulatory capabilities in order to provide functional redundancy for the mutation.10,11,22 Moreover, MSS patients are heterozygous for NFIX mutations, and this contrasts with Nfix +/Del2 mice which are normal, while developmental, skeletal and neural abnormalities are observed in Nfix Del2/Del2 mice.][48][49] An example of allelic variations is provided by the autosomal dominant disorder spondyloepimetaphyseal dysplasia, Missouri type (SEMD MO ) which in humans is due to a heterozygous matrix metalloproteinase 13 (MMP13) missense F56S mutation, but in mice occurs only in homozygous Mmp13 -/-mice deleted for exons 3, 4 and 5 that have defects in growth plate cartilage and delayed endochondral ossification, while heterozygous Mmp13 +/- mice have normal growth plates and no skeletal abnormalities.The phenotypic differences between humans and mice in this case were demonstrated to be due to auto-catalytic MMP13 enzyme activity induced by the mutant which degraded the WT MMP13, thereby exerting a dominant effect and leading to a deficiency in MMP13 that was similar in Mmp13 -/-mice, which developed skeletal abnormalities resembling those observed in patients SEMD MO.

50
In summary, we report the identification of CRABP2 and VCAM1 as NFIX downstream target genes that are misregulated in 60%-100% of the MSS human and mouse fibroblasts.Thus, NFIX may directly regulate the activity of the CRABP2 promoter and NFIX mutations may alter CRABP2 expression in MSS patients, and NFIX may indirectly affect VCAM1 expression via the RA signaling pathway (Figure 5).Interestingly, CRABP2 expression has been reported to be upregulated through the RA pathway, while VCAM1 appears to be an indirect target of the RA pathway in a permissive cellular context. 16This suggests that dysfunctional NFIX proteins in MSS may directly increase CRABP2 expression through a reduced ability to suppress CRABP2 promoter activity which in turn results in the activation of the RA pathway, which may then indirectly misregulate VCAM1 expression in a tissuespecific manner, and contribute to the pathology observed in MSS (Figure 5).This may indicate a possible misregulation of the RA pathway in MSS patients, and the possibility that drugs targeting the RA pathway might be of therapeutic benefit to MSS patients.

Figure 1 .
Figure 1.Cellular localisation and functional studies of MSS mutants.COS7 cells were transiently transfected with N-terminal Flag tagged WT (NFIX WT) or MSS-mutant (NFIX InsT, NFIX DupG, NFIX Del) NFIX cDNA constructs.pCMVTag2C was used as a negative control.(A) Cellular localisation of NFIX following immunofluorescence studies using an anti-Flag primary antibody and an Alexa Fluor 488 secondary antibody.NFIX proteins are predominantly located within the nucleus as determined by the overlap of the signal (teal colour) from the green fluorescent Flag tag on the NFIX protein and the blue DAPI stained nuclei.Scale bar represents 50 μm.(B) qRT-PCR analysis of NFIX expression normalised to GAPDH expression.(C) Western blots analysis using anti-Flag primary antibodies and HRP conjugated secondary antibody, and (D) quantified by densitometry analysis.Antibodies against GAPDH (37 kDa) were used as loading control.Untreated cells (UT) were used as an additional control.(E) In vitro dual luciferase reporter assays, in which the luciferase reporter gene is under the control of the GFAP promoter containing 3 NFIX binding sites, co-transfected with WT or mutant NFIX cDNA constructs.Data are represented as mean ± SEM, n = 4-12, * * * * P <0 .0001,* * * P <0 .001,* * P <0 .01.

Figure 4 .
Figure 4. Effects of NFIX on activities of CRABP2 and VCAM1 promoters.Locations of a putative NFI binding sites (underlined) in the human and mouse genomes at positions: (A) −2176 to −2163 and −1822 to −1809 upstream of the ATG start site of the CRABP2 and Crabp2 genes, respectively; and (A') −238 to −224 and −202 to −188 upstream of the ATG start site of the VCAM1 and Vcam1 genes, respectively.Luciferase reporter constructs were generated by cloning: (B) 2.8 kb of the CRABP2 promoter, containing an NFI binding site located at position −2176 to −2163 upstream of the ATG start site; and (B') 1.7 kb of the VCAM1 promoter, containing an NFI binding site located at position −238 to −224 upstream of the ATG start site, upstream of the firefly luciferase reporter gene.In vitro dual luciferase reporter assays in COS-7 cells, in which the luciferase reporter gene is under the transcriptional control of the (C) CRABP2 and (C') VCAM1 promoter, containing the wild-type (WT) or mutated (MUT; generated via the transition of 7 nucleotides) NFI binding site.A fourth construct lacking the entire promoter region (pGL4.10)was included as a control.In vitro dual luciferase reporter assays in COS-7 cells, in which the luciferase reporter gene is under the transcriptional control of the WT (D) CRABP2 and (D') VCAM1 promoter, co-transfected with WT or MSS-associated mutant (NFIX InsT, NFIX DupG, NFIX Del) NFIX cDNA constructs.Chromatin immunoprecipitation using an anti-NFIX antibody or a non-specific IgG antibody compared to total input, coupled with real-time PCR (RT-PCR) was used to determine binding of NFIX to the NFI motifs identified in the CRABP2 and VCAM1 promoters.Two primers sets were designed to encompass either the identified: (E) −2176 to −2163 NFI motif or an unrelated −442 to −430 genomic site upstream of the ATG start site of CRABP2 as an additional control; or (E') −238 to −224 NFI motif or an unrelated −2154 to −2142 genomic site upstream of the ATG start site of VCAM1 as an additional control.Enrichment of NFIX occupancy at the (F) CRABP2 and (F') VCAM1 promoter relative to the IgG control was quantified as a percentage of total input.Data are represented as mean ± SEM, n = 3-8, * P <0 .05,* * * P <0 .001,* * * * P <0 .0001.

Figure 5 .
Figure 5. Hypothesized role of NFIX on CRABP2 expression in MSS patients.NFIX mutations in MSS patients result in the production of dysfunctional NFIX proteins that directly bind to NFI binding sites (5'-TTGGC(N5)GCCAA-3') in the CRABP2 promoter.Since mutant NFIX proteins have a reduced ability to suppress CRABP2 promoter activity, this results in increased CRABP2 expression, which in turn activates the RA pathway.Misregulation of CRABP2 expression contributes to the skeletal and neurological phenotypes observed in MSS patients.