Abstract
Background
Cell-free (cf) RNA in amniotic fluid supernatant (AFS) reflects developmental changes in gene expression in the living fetus, including genes specific to the central nervous system (CNS). Although it has been previously shown that CNS-specific transcripts are present in AFS, it is not known whether changes in the AFS transcriptome reflect the specific pathophysiology of fetal CNS disorders. In myelomeningocele, there is open communication between the CNS and amniotic fluid.
Objectives
To identify molecular pathophysiologic changes and novel disease mechanisms specific to myelomeningocele by analyzing AFS cfRNA in fetuses with open myelomeningocele.
Study Design
AFS was collected from 10 pregnant women at the time of the open myelomeningocele repair in the second trimester (24.5+/−1.0 wks) and 10 archived AFS from sex and gestational age-matched euploid fetuses without myelomeningocele were used as controls (20.9+/−0.9 wks). Differentially regulated gene expression patterns were analyzed using Human Genome U133 Plus 2.0 arrays.
Results
Fetuses with myelomeningocele had 284 differentially-regulated genes (176 up- & 108 down-regulated) in AFS. Known genes associated with myelomeningocele (PRICKLE2, GLI3, RAB23, HES1, FOLR1) and novel dysregulated genes were identified in association with neurodevelopment and neuronal regeneration (up-regulated, GAP43 and ZEB1) or axonal growth and guidance (down-regulated, ACAP1). Pathway analysis demonstrated a significant contribution of inflammation to pathology and a broad influence of Wnt signaling pathways (Wnt1, Wnt5A, ITPR1).
Conclusion(s)
Transcriptomic analyses of living fetuses with myelomeningocele using AFS cfRNA demonstrated differential regulation of specific genes and molecular pathways relevant to this CNS disorder, resulting in a new understanding of pathophysiological changes. The data also suggested the importance of pathways involving secondary pathology, such as inflammation, in myelomeningocele. These newly identified pathways may lead to hypotheses that can test novel therapeutic targets as adjuncts to fetal surgical repair.
Introduction
In the US, 1,500 infants are born annually with myelomeningocele.1 A relatively recent therapeutic breakthrough for the condition was the development of fetal surgery that can improve motor function and reduce the risk of developing hydrocephalus in affected infants.2 Although the long-term outcomes for children who underwent fetal surgery have yet to be determined, some children are still expected to have significant motor and cognitive impairments based on a cohort study implemented before a randomized trial.3 Additionally, not all fetuses are candidates for fetal surgery, and there are considerable maternal and fetal risks.3,4 Alternative and/or complementary medical approaches to treatment are warranted.
In order to develop medical treatments, the fetal molecular pathophysiology of myelomeningocele needs to be rigorously investigated. This has not yet been done, partly due to its complexity and heterogeneity.5 The understanding of pathological mechanisms in the fetus has been limited by a lack of tools to evaluate its dynamic molecular interactions in living human fetuses. Transcriptomic analysis of cell-free (cf) RNA in amniotic fluid (AF) reflects fetal development in real time. Gene expression can be assessed from various organs including brain and spinal cord.6–9 Previous reports of the AF cfRNA transcriptome analysis using microarrays identified specific and molecular mechanisms in fetal diseases such as aneuploidies10,11, twin-twin transfusion syndrome12, maternal obesity9 and fragile X syndrome13. We hypothesized that analysis of the AF transcriptome using cfRNA will provide novel insights into myelomeningocele pathophysiology that will lead to the identification of novel therapeutic targets. We also hypothesized that myelomeningocele AF transcriptome analysis will provide more transcripts originating from the central nervous system than healthy fetuses or other conditions, as fetal central nervous system tissue is exposed directly to the AF. Myelomeningocele has heterogeneous genetic and environmental pathophysiology.5,14,15 A systems biology approach may be ideal to analyze the disease in order to identify common molecular pathways relevant to disease mechanisms. In this study, we aimed to determine specific molecular pathways and genes that are differentially expressed in fetuses with myelomeningocele. We also tested if genes from the central nervous system are over-represented in AF from fetuses with myelomeningocele, possibly originating during the time when the neural tube was open.
Methods
Study design
This study compared AF cfRNA transcription between second trimester fetuses with myelomeningocele and sex- and relatively gestational age-matched control fetuses. While case samples (myelomeningocele) were collected prospectively, the control samples were from previous studies performed in our laboratory.8,9 The study was approved by the Institutional Review Boards of Tufts Medical Center (#8908) and Children’s Hospital for Philadelphia (CHOP) (#14-010958).
Subjects and samples
Ten pregnant women undergoing fetal surgery between 22 and 27 weeks of gestation for myelomeningocele at CHOP were prospectively enrolled into the study between December 2014 and September 2015. All subjects met the criteria of the prior randomized controlled study for myelomeningocele repair.2
The control AF samples consisted of residual discarded second trimester clinical samples from the Cytogenetics Laboratory at Tufts Medical Center, originally obtained and analyzed for prior studies from June 2011 to April 2012.8,9,16 RNA was extracted within six months after collection. These samples were collected from pregnant women carrying structurally normal fetuses who underwent diagnostic amniocenteses for standard indications (advanced maternal age, abnormal serum screening test). The control fetuses were confirmed to have normal karyotypes with cytogenetic tests. Archived samples were used due to the nationwide decrease in the frequency of diagnostic amniocentesis after the introduction of non-invasive prenatal testing. The gestational age of the control fetuses could not be precisely matched since the majority of the diagnostic amniocenteses occur before 22 weeks, and fetal surgery occurs at 23~26 weeks.
Under general anesthesia, pregnant women underwent open fetal surgery, as previously described.2 After the uterus was exposed and incised, 5~10 ml of amniotic fluid (AF), which would otherwise have been discarded, were aspirated through a blunt plastic cannula attached to a 20-mL syringe through intact amniotic membranes. Upon completion of the surgery, the lost AF volume was replaced with warmed normal saline using standardized surgical techniques. AF samples were immediately frozen at −20°C to be sent to Tufts Medical Center, where the rest of the analyses were performed. Upon arrival at Tufts Medical Center, the samples were stored in a −80°C freezer until RNA extraction.
All control samples were processed for RNA extraction within six months after sample collection between December 2011 and June 2012 using the same method as in our prior study.16 Among ten controls, five were processed on microarrays within six months after collection. The extracted RNAs from the other five samples were stored in a −80°C freezer until hybridized on microarrays in June 2015. Microarray quality was not different between the five freshly processed samples and the five samples processed later (t-tests of the scale factors and % present calls between the two groups have p-values of 0.25 and 0.28, respectively).
RNA extraction
Within six months after collection, the AF samples from the fetuses with myelomeningocele were thawed and centrifuged at 300g at 4C for 10 minutes, and separated into AF supernatant (AFS) and pellet. Control samples were centrifuged with the same method but right after the collection at the Cytogenetics Laboratory. RNA was extracted from AFS using Qiagen Isolation of Circulating Cell-Free Nucleic Acids and the Qiagen RNeasy MinElute (Qiagen, Venlo, Netherlands) cleanup step as previously described.16
Transcriptome microarray and quality control
Extracted RNA was amplified using the Ovation Pico WTA System V2 (NuGEN, San Carlos, CA). The quantity and purity of amplified cDNA was measured using Nanodrop® (Thermo Scientific, Wilmington, ED) at the ssDNA setting to assure a sufficient amount of cDNA and quality. The amplified cDNA was fragmented and biotin-labeled using the FL-Ovation cDNA Biotin Module V2 according to the manufacturer’s instructions (NuGEN, San Carlos, CA). Biotin-labeled and fragmented cDNAs were hybridized to the AffymetrixGeneChip® Human Genome U133 Plus 2.0 Array (Affymetrix, Santa Clara, CA) as previously described.6
We chose gene expression microarrays over RNA sequencing for two reasons. Prior studies in our laboratory showed that micorarrays detected a greater number of genes than RNA sequencing from the same sample.17 Also, microarray data can be directly used to test hypotheses regarding novel therapies using the Connectivity Map.18
After quality control assessment using Affymetrix® 5.0 software (Affymetrix, Santa Clara, CA), the raw microarray data were normalized using the affyPLM package in Bioconductor (http://www.bioconductor.org) in R (ver. 3.2.5, R Foundation for Statistical Computing, Vienna, Austria), for further analysis.19 The normalization was performed with the threestep() function using ideal mismatch as the background correction technique, quantile normalization, and summarization using the Tukey-Biweight algorithm. Microarray data was deposited in GEO (accession number, GSE101141). Differentially regulated genes between fetuses with myelomeningocele and controls were identified using paired t-tests with the Benjamini-Hochberg adjustment (adjusted p value < 0.05) to control the false discovery rate despite multiple tests.20,21. We have previously demonstrated that using paired t-tests, provided the pairings are fixed before the analysis and based solely on clinical characteristics, can allow us to see phenotypic differences despite the variability introduced by gestational age.10,22
Identification of differentially regulated genes, their functional annotations and functional pathway analysis
Genes that were significantly up- or down-regulated as defined above, and that showed consistent up- or down-regulation in the indicated direction in at least 8 of the 10 pairs, were identified as “common” differentially regulated genes in myelomeningocele. We chose to use this additional directional filter to identify truly common pathways relevant to myelomeningocele and avoid false conclusions based on outliers. These genes were uploaded to the Ingenuity Pathway Analysis (IPA)™ tool (Version 28820210, Ingenuity Systems, Redwood City, CA) to determine functional annotations of differentially expressed genes and specific genetic pathways that were significantly different between the case and control AFS samples. Based on a manually curated gene data base (mostly adult and some fetal) of functional annotations and biological interactions, IPA identifies molecular pathways and predicts downstream effects on biological and disease processes.23 Analyses identified differentially regulated pathways that were reported with false-discovery rates and bias-corrected Z-scores, where calculated. Z-scores provided both magnitudes and directions of differential regulation, i.e. how much the pathways were activated or inhibited.23 The functional annotation of individual genes was also manually investigated using available online databases such as PubMed.
Tissue specificity analysis
BioGPS is a gene expression annotation portal and data resource, including gene “atlases” measuring genome-wide expression in a range of tissues and platforms.24 To identify tissue-specific transcripts, we downloaded the human gene expression atlas from BioGPS in which samples were profiled using Affymetrix U133A arrays.24 We computed the median expression level of each gene across all non-malignant samples. We then defined tissue-specific genes as those in which maximum expression in a particular tissue type was at least 30 multiples of the median (MoMs), and whose next highest tissue type had expression levels no more than one third as high as the maximum for that gene.
To determine if AFS of myelomeningocele contains more CNS transcripts than that of controls, we identified CNS-specific probes, meaning probes specific to CNS-derived tissues in BioGPS as described above, in each “core transcriptome”, the set of probes called “present” in at least eight of 10 samples in either the cases or the controls. Using Fisher’s exact test, we determined whether the probability of CNS-specific probes' presence in the core transcriptome is different for the myelomeningocele samples and the controls.6
Results
Clinical characteristics of subjects
The clinical characteristics of study subjects are listed in Table 1. The gestational age of fetuses could not be matched in absolute since the majority of diagnostic amniocenteses occur before 22 weeks in control fetuses (mean ± SD (range); 20.9 ± 0.8 (20.0 – 22.0)), and fetal surgery occurs at 24~26 weeks (24.5 ± 1.0 (22.9 – 25.7) (p-value <0.01, paired t-test). Fetal sex (female myelomeningocele 60.0%, control 60.0%, P = 1.00, Fisher’s exact test) and maternal age (28.3 ± 4.3, 30.4 ± 4.2, P = 0.28, t-test) were not significantly different between fetuses with myelomeningocele and controls.
Table 1.
Subject demographics
| Myelomeningocele | Control | ||||
|---|---|---|---|---|---|
| Case ID | Gestational age [weeks] |
Fetal sex | Case ID | Gestational age [weeks] |
Fetal sex |
| AF011 | 22 6/7 | M | N13 | 19 1/7 | M |
| AF005 | 23 2/7 | F | N6 | 19 6/7 | F |
| AF010 | 23 5/7 | F | MATCON17 | 20 4/7 | F |
| AF008 | 24 2/7 | F | MATCON2 | 20 5/7 | F |
| AF004 | 24 6/7 | F | MATOB7 | 21 0/7 | F |
| AF009 | 24 6/7 | M | MATCON23 | 20 3/7 | M |
| AF002 | 25 0/7 | M | MATCON9 | 21 3/7 | M |
| AF001 | 25 1/7 | F | N12 | 21 5/7 | F |
| AF006 | 25 3/7 | M | N8 | 21 5/7 | M |
| AF007 | 25 5/7 | F | N3 | 22 0/7 | F |
Differential gene expression profile of amniotic fluid
The efficiency of microarray hybridization (%, mean ± SD) did not significantly differ between myelomeningocele (39.4 ± 5.3) and control groups (33.1 ± 10.2) (p-value = 0.11, paired t-test) (data not shown). The second trimester fetuses with myelomeningocele had significantly distinct gene expression profiles compared to control fetuses. Using a Benjamini-Hochberg adjusted p-value of 0.05 as a cut-off, and the consistency criteria defined above, we found 368 probe sets showing consistent differential expression between myelomeningocele and controls. When analyzed in IPA, these corresponded to 284 genes (176 up-regulated and 108 down-regulated). We have performed batch analysis in Combat to test if batch effects reflecting these differences influenced our results.25 There did appear to be some batch differences between most of the immediately processed controls and most of the controls that were hybridized later, but there were outliers in each group, raising questions about the possible cause of this effect given the small numbers of samples in each group. Combat normalization mitigated but did not eliminate this effect as we observed all 284 differentially regulated genes were still among the top differentially expressed genes after Combat normalization. We have also run a differential expression analysis comparing the control samples that were processed within six months to those processed after three years. None of the 284 genes that we have identified as differentially expressed in myelomeningocele were seen to be differentially regulated between control batches. Therefore, we do not believe that this unavoidable technical factor meaningfully influenced our results.
Functional annotations of genes differentially expressed in fetuses with myelomeningocele
Differential gene expression analysis between cases and controls was carried out in R. Gene annotation of probesets is based on the IPA database. Table 2 lists the top ten most up- and down-regulated genes in myelomeningocele cases versus controls. Eight of these 20 genes had known roles in spinal cord and neuronal development but not in association with myelomeningocele (GFAP, GAP43, ZEB1, C1orf6, ACAP1, SKI, SEMA5A, TSHZ3). The most dramatically up-regulated gene (172 fold) was SERPINB6.
Table 2.
Top 10 up- and down-regulated genes
| a. Top 10 up-regulated genes | ||
|---|---|---|
| Gene symbol | Molecular function | Expression fold change |
| SERPINB6 | Serine protease inhibitor: Inhibits neuropsin (synaptic modulator), inner hair cells function | 172.9 ↑ |
| GFAP | Intermediate filament protein of the astroglial skeleton, astrogliosis | 79.63 ↑ |
| GAP43 | Neuronal cone regeneration | 44.0 ↑ |
| TNNC1 | Regulates striatal muscle contraction | 33.1 ↑ |
| CLDN18 | Integral membrane proteins and components of tight junction strands | 28.2 ↑ |
| RPTN | Calcium binding protein | 25.7 ↑ |
| ZEB1 | Dorsal root ganglion development | 22.5 ↑ |
| C1orf61 | Brain specific transcriptional activator of c-fos | 22.4 ↑ |
| KIAA1211 | (Unknown function) | 20.5 ↑ |
| LYZ | Lysozyme which has antibacterial activity against bacteria | 20.2 ↑ |
| b. Top 10 down-regulated genes | ||
|---|---|---|
| Gene symbol | Molecular function | Expression fold change |
| SC5D | C-5 sterol desaturase | 18.8 ↓ |
| RIPK2 | Receptor-interacting protein (RIP) family of serine/threonine protein kinases | 10.9 ↓ |
| ACAP1 | Axonal growth, anterograde axonal transport | 9.0 ↓ |
| CDK14 | Cell cycle progression | 8.5 ↓ |
| SKI | TGIF, midline development, anterior/posterior axis specification, BMP signaling pathway | 8.2 ↓ |
| SEMA5A | Attractive/permissive effects on dorsal root ganglion axons | 8.1 ↓ |
| HIC2 | Cardiac development | 7.3 ↓ |
| SYDE1 | GTPase activator | 6.9 ↓ |
| S1PR3 | Angiogenesis, vascular endothelial cell function | 6.7 ↓ |
| TSHZ3 | Brainstem respiratory center | 6.5 ↓ |
Among the entire list of differentially-expressed genes identified by IPA, 20 with significant roles in brain and spinal cord development were identified. The genes involved planar cell polarity, spinal cord ventral-dorsal axis development, neuronal differentiation, axonal development, Schwann cell development, motor neuron survival, synapse formation and neuronal injury and repair (Table 3). In the same list, 33 genes associated with inflammation and neurodegeneration were identified (Table 4). These genes involve migration of inflammatory cells, accumulation of neutrophils, neuroinflammation and neurological injury.
Table 3.
Example of differentially expressed genes in fetuses with myelomeningocele with significant roles in brain and spinal cord development and function identified by IPA (* novel genes associated with myelomeningocele)
| Molecular function | Genes |
|---|---|
| Planer cell polarity | WNT5A (↓ 3.7 folds), PRICKLE2 (↓ 3.9) |
| Spinal ventral-dorsal regionalization | WNT1 (↓ 1.7), WNT5A (↓ 3.7), SKI* (↓ 8.2), GPR37* (↓ 3.9), HNF1B* (↑ 4.4) |
| Neuronal differentiation | GFAP (↑ 79.6), C1orf61* (↑ 22.4) |
| Dorsal horn development | NEUROD*6 (↓ 1.9), ZEB1* (↑ 22.5) |
| Axonal growth, guidance | GAP43* (↑ 44.0), ACAP1* (↓ 9.0), SEMA5A* (↓ 8.1), DRAXIN* (↓ 1.7) |
| Schwann cell development | POU3F1* (↓ 6.5) |
| Motor neuron survival | PQBP1* (↓ 4.8) |
| Synaptic transmission, plasticity | SERPINB6 (↑ 172.9), TSHZ3 (↓ 6.5), ERBB3* (↓ 1.8) |
| Neuronal injury, repair | GFAP (↑ 79.6), GAP43* (↑ 44.0), HNF1B* (↑ 4.4), HMGB1* (↑ 2.3) |
Table 4.
Example of differentially expressed genes in fetuses with myelomeningocele with significant roles in inflammation and neurodegeneration identified by IPA
| Molecular function | Genes |
|---|---|
| Migration of inflammatory cells | CCL3 (↑ 6.1), CD44 (↑ 19.0), DEFB1 (↑ 14.3), HMGB1 (↑ 2.3), LGALS1 (↑ 2.4), PRKCQ (↑ 10.3), S100A8 (↑ 3.8), S100A9 (↑ 3.6), S1PR3 (↓ 6.7), SLPI (↑ 2.9), TIMP2 (↑ 2.9), WNT5A (↓ 3.7) |
| Accumulation of neutrophils | CCL3 (↑ 6.1), HMGB1 (↑ 2.3), S100A8 (↑ 3.8), S100A9 (↑ 3.6), SLPI (↑ 2.9), TIMP2 (↑ 2.9) |
| Neuroinflammation | GFAP (↑ 79.6), SREBF2 (↑ 3.4), AQP4 (↑ 15.8) |
| Other immunological functions | SERPINB6 (↑ 172.9) |
| Neurological injury | ADRB3 (↓ 3.3), CA1 (↑ 6.7), CA12 (↑ 2.5) GADD45G (↓ 3.3), GAP43 (↑ 44.0), GFAP (↑ 79.6), GLI3 (↓ 5.1), HES1 (↑ 3.5), HNF1B (↑ 4.4), ITPR1 (↑ 4.4), MALAT1 (↑ 5.1), MCL1 (↓ 2.3), NAMPT (↑ 4.4), PITPNA (↑ 2.8), PRICKLE2 (↓ 3.9), SNTA1 (↑ 4.3), SYNJ1 (↑ 5.6), TNNC1 (↑ 33.2), TRIB1 (↑ 2.6), ZFP36 (↑ 2.4) |
Functional pathway analysis
Ingenuity Pathway Analysis identified 24 pathways that were significantly associated with differentially regulated genes (as calculated by right-tailed Fisher's exact test with p-value < 0.05). Among these pathways, three pathways were measured with Z-scores which confirm the predicted directionality of pathway regulation (that is, they were up- or down-regulated as predicted) in second trimester fetuses with MMC. Specifically, inflammatory response was up-regulated (leukocyte extravasation; Z-score 2.45), while dorsal horn function (neuropathic pain signaling in dorsal horn; Z-score −0.45) and Wnt signaling pathways (glioblastoma multiforme signaling; Z-score −1.34) were down-regulated (Table 5, supplementary figure).
Table 5.
Dysregulated IPA pathways in fetuses with myelomeningocele
| Pathways | Molecules | No. of Genes | Z-score |
|---|---|---|---|
| Leukocyte extravasation | CD44, CLDN18, DLC1, MAPK8, MSN, MYL6, PRKCQ, TIMP2 | 8 | +2.4 |
| Neuropathic pain signaling in dorsal horn | GPR37, ITPR1, PLCB3, PLCE1, PRKCQ | 5 | −0.4 |
| Glioblastoma multiforme signaling | ITPR1, NF1, PLCB3, PLCE1, WNT1, WNT5A | 6 | −1.3 |
Tissue specific expression of genes
Among 284 differentially-regulated genes, tissue specificity analysis identified 28 genes (10%) that were highly specific in tissue origin. Among them, 12 (42 %) were specifically expressed in the central nervous system (CNS), brain and spinal cord (Table 6). GFAP, C1orf61, FEZ1, and SCRN1 are known to be expressed in the spinal cord. GAP43, TCEAL2, TUBB2B, FABP7, C1orf61, and DOK5 are known to be expressed in the fetal brain (FABP7 and DOK5 are solely expressed in fetal brain). These genes encode diverse categories of proteins, which carry diverse functions in CNS such as neuronal migration (TUBB2B, FABP7), neuronal differentiation (GFAP, C1orf61, MTUS1)26, neuronal resilience to various stresses (GFAP, MT3)27, axon development (GAP43, FEZ1), synaptic transmission and intercellular signaling in neurons (SYNJ1, MTUS1, SCRN1)28 (Table 6).
Table 6.
Central nervous system (CNS) specific genes differentially expressed in fetuses with myelomeningocele
|
Gene symbol |
Molecular function | Tissues with specific expression | Expression fold change |
|---|---|---|---|
| GFAP | Intermediate filament proteins of mature astrocytes, resilience to cellular stress to glial cells26 | Spinal cord, prefrontal cortex, hypothalamus, thalamus, olfactory bulb, retina | 79.6 ↑ |
| GAP43 | Neuronal cone regeneration, presynaptic protein | Fetal brain, amygdala, whole brain, hypothalamus, thalamus, prefrontal cortex, cingulate cortex, occipital lobe | 44.0 ↑ |
| TCEAL2 | Member of the transcription elongation factor A (SII)-like (TCEAL) gene family | Prefrontal cortex, pineal body (day and night), hypothalamus, amygdala, whole brain, pituitary gland, fetal brain, occipital lobe, temporal lobe, subthalamic nucleus | 9.5 ↑ |
| TUBB2B | Neuronal migration, microtuble | Fetal brain, amygdala | 8.8 ↑ |
| FABP7 | Establishment of the radial glial fiber in the developing brain, fatty acid binding protein | Fetal brain | 8.0 ↑ |
| SYNJ1 | Phosphoinositide phosphatase that regulates synaptic transmission and membrane trafficking | Pineal body (day and night), amygdala, prefrontal cortex, whole brain | 5.6 ↑ |
| C1orf61 | Brain specific transcriptional activator of c-fos involved in neuroprogenitor proliferation, cellular remodeling46 | Caudate nucleus, hypothalamus, spinal cord, temporal lobe, thalamus, fetal brain, whole brain, cerebellum peduncles, occipital lobe, prefrontal cortex, cerebellum, amygdala, cingulate cortex, medulla oblongata, parietal lobe, pons | 5.1 ↑ |
| FEZ1 | Intercellular protein, regulates axonal bundling and elongation within axon bundles and axon overgrowth | Spinal cord, hypothalamus, whole brain, parietal lobe, thalamus, retina, occipital lobe, medulla oblongata, amygdala, caudate nucleus, prefrontal cortex, cerebellum peduncles | 4.9 ↑ |
| MT3 | Neuron specific metal binding protein, protection against reactive oxygen species, or adaptation to stress27,47 | Whole brain, prefrontal cortex, temporal lobe, thalamus, amygdala | 4.0 ↑ |
| MTUS1 | Neuronal differentiation via interaction with angiotensin II receptor48 | Pineal body (day and night), cerebellum peduncles, occipital lobe | 3.6 ↑ |
| SCRN1 | Control of fear conditioning, reward learning, cognitive function and feeding behavior28 | Olfactory bulb, pineal body (day and night), prefrontal cortex, cerebellum peduncles, hypothalamus, occipital lobe, parietal lobe, cingulate cortex, spinal cord | 2.1 ↓ |
| DOK5 | Membrane protein mediating neurite outgrowth by activation of the MAP kinase pathway | Fetal brain | 3.2 ↓ |
Of 595 CNS-specific probes on the Affymetrix U133A arrays, the myelomeningocele “core transcriptome” expressed more CNS-specific probes than the “core transcriptome” of the controls (135 vs. 84 probes, p-value = 0.00017, Fisher’s exact test).
Discussion
In this study, human gene expression analysis of living fetuses using AFS RNA identified novel and known genes involved in the molecular pathophysiology of myelomeningocele, such as pathways associated with spinal cord and neuronal development and robust inflammation. In particular, the extent of inflammation in the affected fetuses is a novel observation. The tissue specificity analysis identified CNS-specific genes. Our overall findings suggest that a systems biology approach through analysis of the amniotic fluid cfRNA transcriptome could provide insights into the fetal molecular pathophysiology of myelomeningocele, along with the detection of CNS-specific genes. Given that myelomeningocele is a heterogeneous disorder, understanding gene regulation may help us understand disease mechanisms and potential drug targets.
Dysregulation of spinal and neuronal developmental genes and pathways in the second-trimester fetuses with myelomeningocele
It is notable that the original annotations in IPA were made based on adult tissue. In contrast, the functional annotations in this study were interpreted here in the context of embryonic and fetal physiology. This showed that various spinal and neuronal development genes and pathways were differentially-regulated in fetuses with myelomeningocele. Among the three significantly altered pathways, those denoted as “Glioblastoma multiforme signaling” and “Neuropathic pain signaling in dorsal horn” in adult tissue also represent embryonic and fetal Wnt signaling pathways (Table 3). One of the Wnt signaling sub-pathways, planar cell polarity pathway, induces the earliest steps in neurulation “conversion extension” (lengthening and narrowing of the initially disc-shaped neural plate), the failure of which is speculated to cause neural tube defects.5 In this study, the fetuses with myelomeningocele had significantly down-regulated pathways and associated genes (Wnt5A, PRICKLE2).
Surprisingly, this study also identified novel genes and pathways differentially-regulated in fetuses with myelomeningocele that involve multiple distinct steps of spinal and neural development beyond neurulation processes, such as spinal cord ventral-dorsal axis development (Wnt1, Wnt5A, SKI, GPR3729,30), neuronal differentiation, axonal development (Wnt5A, ITPR1, PLCB3, PLCE1, ACAP1, SEMA5A, and DRAXIN29,30), dorsal root ganglion development (SEMA5A, NEUROD, and ZEB131), Schwann cell development, motor neuron survival, synapse formation and neuronal injury and repair (Table 3). Disturbances in formation of the cortico-spinal tract may explain how progressive motor impairment occurs in myelomeningocele.32
Certain genes that have no known association with spinal or neural development were also significantly up-regulated, such as SERPINB6. In humans, SERPINB6 mutations are associated with sensorineural hearing loss.33 In human fetuses at 21 weeks’, SERPINB6 is diffusely expressed in the developing brain (www.brainspan.org database; no information available for spinal cord) but its function in fetal development is unknown. SERPINB6 is also known as proteinase inhibitor 6 or placental thrombin inhibitor, and may contribute to the secondary pathology of myelomeningocele for its role in anti-viral, anti-apoptotic effects, and fibrinolysis.34–36 SERPINB6 also endogeneously inhibits neuropsin, which modifies synaptic plasticity and long-term potentiation.37,38 Many other genes with previously unknown associations with myelomeningocele or spinal development were up- or down-regulated, but none to the extent of SERPINB6 (Table 2).
While some of these pathways identified function in an early embryonic phase, such as neurulation, some function in later gestation, when axonal development and synapse formation are active. Considering such broad temporal distribution, cell free (cf) RNA in AF may have the nature of a real-time identification of currently functioning genes and pathways, as well as being a potential “time capsule” of accumulated information of formerly expressed genes in living fetuses.7 Further investigation is needed to test this hypothesis.
Inflammation and neurodegeneration as the critical fetal neuropathology of myelomeningocele
Among all analyses in this study, either pathways (Table 5) or individual molecular analyses (Table 4), inflammation and neurological injury stand out consistently as the most prominent transcriptomic alterations in fetuses with myelomeningocele. There were differential-regulations in neuroinflammation (GFAP, SREBF2 and AQP4), microglia activation (ERBB3 and HMGB1 via receptor of advanced glycation end-products (RAGE) cascade), axonal regeneration (GAP43)39,40 and synaptogenesis (ITPR141). GFAP, the second most up-regulated gene (80-fold) in this study, is an intermediate filament protein and a marker of mature/reactive astroglia, one of the primary responders to neural injuries. GFAP is known to increase in the AF of a retinoic acid induced rat myelomeningocele model, potentially reflecting severity of the spinal defect and neurological impairments.42 Not only the injury process, but also the regenerative responses may be activated in fetuses with myelomeningocele, such as axonal regeneration (GAP43)39,40 and synaptogenesis (ITPR141). Former studies also reported abnormal neuronal development, inflammation and gliosis in placode tissue of fetuses with myelomeningocele by immunohistology in fetopsy specimens43 and surgical excised tissues44. Neuro-inflammation, injuries and regeneration may constitute secondary pathologies in fetuses with myelomeningocele that may deserve further investigation as potential modifiable therapeutic targets.
Transcripts specific to CNS
Here we sought less invasive ways to identify tissue origins of differentially-regulated genes in AF without collecting fetal tissue. Our analysis showed that a high portion (40%) of the tissue-specific transcripts were specific to the CNS. The majority of them were not expressed in the spinal cord, although this might be a biased view, as less in situ expression data are available for the fetal spinal cord than brain in the BIOGPS database. Genes that are highly specifically expressed in the CNS have relevant developmental functions in the second trimester such as neuronal differentiation, migration, axonal development and maintenance of neurons (Table 6). These genes originated broadly from various CNS tissues, including cerebrum and cerebellum beyond the local spinal cord malformation (Table 6). Such broad distribution of differentially-regulated genes may explain cognitive impairments observed in adults with myelomeningocele that cannot be explained by the primary spinal lesion or complications of hydrocephalus.45
Analysis of the core transcriptomes showed that AFS of fetuses with myelomeningocele contains more transcripts specific to CNS compared to that of controls. It is not known if CNS transcripts in AFS may have originated at a time when neural tube was open or if they are actively expressed in the second trimester fetus. In the case of myelomeningocele, there is another possibility that CNS transcripts may continue to “leak out” with cerebrospinal fluid. Further study is needed to answer these questions.
Study limitations
One limitation of transcriptomic analysis utilizing cfRNA in AFS is the fact that the tissue origins of the cell-free RNAs cannot be traced, resulting in a lack of in situ information.7 Although our tissue specificity analysis may have demonstrated neuronal tissue origin of these altered genes, further in situ analysis is warranted. It is encouraging that former studies have shown similar pathological processes (expression of GFAP and pro-inflammatory molecules) in excised fetal placode tissues.43,44 In future efforts, excised fetal placode tissues can be an ideal model to test our hypotheses derived from this study. Transcriptomic analysis of AFS samples has an advantage over tissue analysis since it can be applied to fetuses with myelomeningocele that are not candidates for fetal surgery.
Our study is also limited by the fact that we could not completely match gestational ages between groups because of inevitable differences in the timing of amniocentesis between control (diagnostic) and case (surgical) groups. This may contribute to some of the changes observed in this study, which may be caused by different developmental stages. However, for practical and ethical reasons, relative matching was considered to be the most reasonable alternative.
In summary, our study has shown dysregulation of multiple distinct steps in spinal and neural development, and systemic activation of neuroinflammation, neurodegeneration and neuronal repair in pre-surgical fetuses with myelomeningocele, which may contribute to progressive impairment of motor and cognitive functions during the pregnancy. The study has also shown that fetuses with myelomeningocele have greater number of CNS-specific transcripts in the AFS compared to controls. This study provides proof of principle that a systems biology approach using the AFS cfRNA transcriptome identifies common molecular pathophysiology in living fetuses with myelomeningocele. The list of differentially regulated genes and pathways from this study will also be used to identify potential drurg targets. For example, by using the connectivity map (portals.broadinstitute.org/cmap/)18, we plan to identify compounds that could reverse the gene signature associated with myelomeningocele. Insights from this study may therefore lead to identification of potential targets for fetal treatment in combination with fetal surgery.
Supplementary Material
Acknowledgments
Role of the funding source
NICHD award K23HD079605 supported the primary role in study design; the collection, experiment, analysis and interpretation of data; and the writing of the report.
The Susan Saltonstall Award supported study design, the collection, and the experiments to generate preliminary data. NICHD award R01HD076140 supported efforts in computational data analysis and interpretation.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosure
The authors report no conflict of interest.
- Tarui T, Flake A, Kim A, McClain L, Fried I, Bianchi D. Transcriptomic analysis of living fetuses with myelomeningocele. (selected as a Top Abstract). Oral Presentation at the Annual Meeting of International Society for Prenatal Diagnosis (ISPD), Berlin, Germany, July 2016.
- Tarui T, Flake A, Kim A, McClain L, Fried I, Bianchi D. Transcriptomic analysis of living fetuses with myelomeningocele. Poster Presentation at the Annual Meeting of American Academy of Neurology, Boston, MA, April 2017.
Disclaimer
No disclaimer.
Contributor Information
Tomo Tarui, Mother Infant Research Institute, Pediatrics, Floating Hospital for Children, Tufts Medical Center, Boston, MA, USA.
Aimee Kim, Center for Fetal Research, Pediatric Surgery, Children’s Hospital of Philadelphia, Philadelphia, PA, USA.
Alan Flake, Center for Fetal Research, Pediatric Surgery, Children’s Hospital of Philadelphia, Philadelphia, PA, USA.
Lauren Mcclain, Center for Fetal Research, Pediatric Surgery, Children’s Hospital of Philadelphia, Philadelphia, PA, USA.
John Stratigis, Center for Fetal Research, Pediatric Surgery, Children’s Hospital of Philadelphia, Philadelphia, PA, USA.
Inbar Fried, Department of Biomedical Informatics, Harvard Medical School, Boston, MA, USA.
Rebecca Newman, Department of Computer Science, Tufts University, Boston, MA, USA.
Donna K. Slonim, Department of Computer Science, Tufts University, Boston, MA, USA.
Diana W. Bianchi, Medical Genetics Branch, National Human Genome Research Institute, Bethesda, MD, USA.
References
- 1.Canfield MA, Honein MA, Yuskiv N, et al. National estimates and race/ethnic-specific variation of selected birth defects in the United States, 1999–2001. Birth Defects Res A Clin Mol Teratol. 2006;76(11):747–756. doi: 10.1002/bdra.20294. [DOI] [PubMed] [Google Scholar]
- 2.Adzick NS, Thom EA, Spong CY, et al. A randomized trial of prenatal versus postnatal repair of myelomeningocele. N Engl J Med. 2011;364(11):993–1004. doi: 10.1056/NEJMoa1014379. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Danzer E, Thomas NH, Thomas A, et al. Long-term neurofunctional outcome, executive functioning, and behavioral adaptive skills following fetal myelomeningocele surgery. Am J Obstet Gynecol. 2016;214(2):269, e261–268. doi: 10.1016/j.ajog.2015.09.094. [DOI] [PubMed] [Google Scholar]
- 4.Moise KJ, Jr, Moldenhauer JS, Bennett KA, et al. Current Selection Criteria and Perioperative Therapy Used for Fetal Myelomeningocele Surgery. Obstet Gynecol. 2016;127(3):593–597. doi: 10.1097/AOG.0000000000001296. [DOI] [PubMed] [Google Scholar]
- 5.Copp AJ, Greene ND. Genetics and development of neural tube defects. J Pathol. 2010;220(2):217–230. doi: 10.1002/path.2643. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Hui L, Slonim DK, Wick HC, Johnson KL, Bianchi DW. The amniotic fluid transcriptome: a source of novel information about human fetal development. Obstet Gynecol. 2012;119(1):111–118. doi: 10.1097/AOG.0b013e31823d4150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Zwemer LM, Bianchi DW. The amniotic fluid transcriptome as a guide to understanding fetal disease. Cold Spring Harb Perspect Med. 2015;5(4) doi: 10.1101/cshperspect.a023101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Edlow AG, Slonim DK, Wick HC, Hui L, Bianchi DW. The pathway not taken: understanding 'omics data in the perinatal context. Am J Obstet Gynecol. 2015;213(1):59, e51–172. doi: 10.1016/j.ajog.2015.03.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Edlow AG, Vora NL, Hui L, Wick HC, Cowan JM, Bianchi DW. Maternal obesity affects fetal neurodevelopmental and metabolic gene expression: a pilot study. PLoS One. 2014;9(2):e88661. doi: 10.1371/journal.pone.0088661. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Slonim DK, Koide K, Johnson KL, et al. Functional genomic analysis of amniotic fluid cell-free mRNA suggests that oxidative stress is significant in Down syndrome fetuses. Proc Natl Acad Sci U S A. 2009;106(23):9425–9429. doi: 10.1073/pnas.0903909106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Koide K, Slonim DK, Johnson KL, Tantravahi U, Cowan JM, Bianchi DW. Transcriptomic analysis of cell-free fetal RNA suggests a specific molecular phenotype in trisomy 18. Hum Genet. 2011;129(3):295–305. doi: 10.1007/s00439-010-0923-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hui L, Wick HC, Moise KJ, Jr, et al. Global gene expression analysis of amniotic fluid cell-free RNA from recipient twins with twin-twin transfusion syndrome. Prenat Diagn. 2013;33(9):873–883. doi: 10.1002/pd.4150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Zwemer LM, Nolin SL, Okamoto PM, Eisenberg M, Wick HC, Bianchi DW. Global transcriptome dysregulation in second trimester fetuses with FMR1 expansions. Prenat Diagn. 2017;37(1):43–52. doi: 10.1002/pd.4928. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Boyles AL, Hammock P, Speer MC. Candidate gene analysis in human neural tube defects. Am J Med Genet C Semin Med Genet. 2005;135C(1):9–23. doi: 10.1002/ajmg.c.30048. [DOI] [PubMed] [Google Scholar]
- 15.Shimoji K, Kimura T, Kondo A, Tange Y, Miyajima M, Arai H. Genetic studies of myelomeningocele. Childs Nerv Syst. 2013;29(9):1417–1425. doi: 10.1007/s00381-013-2197-2. [DOI] [PubMed] [Google Scholar]
- 16.Dietz JA, Johnson KL, Massingham LJ, et al. Comparison of extraction techniques for amniotic fluid supernatant demonstrates improved yield of cell-free fetal RNA. Prenat Diagn. 2011;31(6):598–599. doi: 10.1002/pd.2732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zwemer LM, Hui L, Wick HC, Bianchi DW. RNA-Seq and expression microarray highlight different aspects of the fetal amniotic fluid transcriptome. Prenat Diagn. 2014;34(10):1006–1014. doi: 10.1002/pd.4417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Lamb J, Crawford ED, Peck D, et al. The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease. Science. 2006;313(5795):1929–1935. doi: 10.1126/science.1132939. [DOI] [PubMed] [Google Scholar]
- 19.Gentleman RC, Carey VJ, Bates DM, et al. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol. 2004;5(10):R80. doi: 10.1186/gb-2004-5-10-r80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Statist Soc Ser B (Methodological) 1995;57:289–300. [Google Scholar]
- 21.IPA Rei. [Website] 2016 http://ingenuity.force.com/ipa/articles/Feature_Description/Regulator-Effects.
- 22.Peat JK, Barton B. Medical statistics : a guide to data analysis and critical appraisal. 1. Malden, Mass.: Blackwell Pub.; 2005. [Google Scholar]
- 23.Kramer A, Green J, Pollard J, Jr, Tugendreich S. Causal analysis approaches in Ingenuity Pathway Analysis. Bioinformatics. 2014;30(4):523–530. doi: 10.1093/bioinformatics/btt703. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Su AI, Wiltshire T, Batalov S, et al. A gene atlas of the mouse and human protein-encoding transcriptomes. Proc Natl Acad Sci U S A. 2004;101(16):6062–6067. doi: 10.1073/pnas.0400782101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Johnson WE, Li C, Rabinovic A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics. 2007;8(1):118–127. doi: 10.1093/biostatistics/kxj037. [DOI] [PubMed] [Google Scholar]
- 26.Hol EM, Pekny M. Glial fibrillary acidic protein (GFAP) and the astrocyte intermediate filament system in diseases of the central nervous system. Curr Opin Cell Biol. 2015;32:121–130. doi: 10.1016/j.ceb.2015.02.004. [DOI] [PubMed] [Google Scholar]
- 27.Hidalgo J, Aschner M, Zatta P, Vasak M. Roles of the metallothionein family of proteins in the central nervous system. Brain Res Bull. 2001;55(2):133–145. doi: 10.1016/s0361-9230(01)00452-x. [DOI] [PubMed] [Google Scholar]
- 28.Pang YY, Chen XY, Xue Y, Han XH, Chen L. Effects of secretin on neuronal activity and feeding behavior in central amygdala of rats. Peptides. 2015;66:1–8. doi: 10.1016/j.peptides.2015.01.012. [DOI] [PubMed] [Google Scholar]
- 29.Domanitskaya E, Wacker A, Mauti O, et al. Sonic hedgehog guides post-crossing commissural axons both directly and indirectly by regulating Wnt activity. J Neurosci. 2010;30(33):11167–11176. doi: 10.1523/JNEUROSCI.1488-10.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Aviles EC, Wilson NH, Stoeckli ET. Sonic hedgehog and Wnt: antagonists in morphogenesis but collaborators in axon guidance. Front Cell Neurosci. 2013;7:86. doi: 10.3389/fncel.2013.00086. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Sala P, Prefumo F, Pastorino D, et al. Fetal surgery: an overview. Obstet Gynecol Surv. 2014;69(4):218–228. doi: 10.1097/OGX.0000000000000061. [DOI] [PubMed] [Google Scholar]
- 32.Liu Y, Shi J, Lu CC, et al. Ryk-mediated Wnt repulsion regulates posterior-directed growth of corticospinal tract. Nat Neurosci. 2005;8(9):1151–1159. doi: 10.1038/nn1520. [DOI] [PubMed] [Google Scholar]
- 33.Sirmaci A, Erbek S, Price J, et al. A truncating mutation in SERPINB6 is associated with autosomal-recessive nonsyndromic sensorineural hearing loss. Am J Hum Genet. 2010;86(5):797–804. doi: 10.1016/j.ajhg.2010.04.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Strik MC, Wolbink A, Wouters D, et al. Intracellular serpin SERPINB6 (PI6) is abundantly expressed by human mast cells and forms complexes with beta-tryptase monomers. Blood. 2004;103(7):2710–2717. doi: 10.1182/blood-2003-08-2981. [DOI] [PubMed] [Google Scholar]
- 35.Bernier A, Cleret-Buhot A, Zhang Y, et al. Transcriptional profiling reveals molecular signatures associated with HIV permissiveness in Th1Th17 cells and identifies peroxisome proliferator-activated receptor gamma as an intrinsic negative regulator of viral replication. Retrovirology. 2013;10:160. doi: 10.1186/1742-4690-10-160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Izuhara K, Ohta S, Kanaji S, Shiraishi H, Arima K. Recent progress in understanding the diversity of the human ov-serpin/clade B serpin family. Cell Mol Life Sci. 2008;65(16):2541–2553. doi: 10.1007/s00018-008-8049-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Kato K, Kishi T, Kamachi T, et al. Serine proteinase inhibitor 3 and murinoglobulin I are potent inhibitors of neuropsin in adult mouse brain. J Biol Chem. 2001;276(18):14562–14571. doi: 10.1074/jbc.M010725200. [DOI] [PubMed] [Google Scholar]
- 38.Shiosaka S, Ishikawa Y. Neuropsin--a possible modulator of synaptic plasticity. J Chem Neuroanat. 2011;42(1):24–29. doi: 10.1016/j.jchemneu.2011.05.014. [DOI] [PubMed] [Google Scholar]
- 39.Nguyen L, He Q, Meiri KF. Regulation of GAP-43 at serine 41 acts as a switch to modulate both intrinsic and extrinsic behaviors of growing neurons, via altered membrane distribution. Mol Cell Neurosci. 2009;41(1):62–73. doi: 10.1016/j.mcn.2009.01.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Hung CC, Lin CH, Chang H, et al. Astrocytic GAP43 Induced by the TLR4/NF-kappaB/STAT3 Axis Attenuates Astrogliosis-Mediated Microglial Activation and Neurotoxicity. J Neurosci. 2016;36(6):2027–2043. doi: 10.1523/JNEUROSCI.3457-15.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Sugawara T, Hisatsune C, Le TD, et al. Type 1 inositol trisphosphate receptor regulates cerebellar circuits by maintaining the spine morphology of purkinje cells in adult mice. J Neurosci. 2013;33(30):12186–12196. doi: 10.1523/JNEUROSCI.0545-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Danzer E, Zhang L, Radu A, et al. Amniotic fluid levels of glial fibrillary acidic protein in fetal rats with retinoic acid induced myelomeningocele: a potential marker for spinal cord injury. Am J Obstet Gynecol. 2011;204(2):178, e171–111. doi: 10.1016/j.ajog.2010.09.032. [DOI] [PubMed] [Google Scholar]
- 43.George TM, Cummings TJ. The immunohistochemical profile of the myelomeningocele placode: is the placode normal? Pediatr Neurosurg. 2003;39(5):234–239. doi: 10.1159/000072867. [DOI] [PubMed] [Google Scholar]
- 44.Kowitzke B, Cohrs G, Leuschner I, et al. Cellular Profiles and Molecular Mediators of Lesion Cascades in the Placode in Human Open Spinal Neural Tube Defects. J Neuropathol Exp Neurol. 2016;75(9):827–842. doi: 10.1093/jnen/nlw057. [DOI] [PubMed] [Google Scholar]
- 45.Roach JW, Short BF, Saltzman HM. Adult consequences of spina bifida: a cohort study. Clin Orthop Relat Res. 2011;469(5):1246–1252. doi: 10.1007/s11999-010-1594-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Jeffrey PL, Capes-Davis A, Dunn JM, et al. CROC-4: a novel brain specific transcriptional activator of c-fos expressed from proliferation through to maturation of multiple neuronal cell types. Mol Cell Neurosci. 2000;16(3):185–196. doi: 10.1006/mcne.2000.0866. [DOI] [PubMed] [Google Scholar]
- 47.Atrian S, Capdevila M. Metallothionein-protein interactions. Biomol Concepts. 2013;4(2):143–160. doi: 10.1515/bmc-2012-0049. [DOI] [PubMed] [Google Scholar]
- 48.Kantarci S, Al-Gazali L, Hill RS, et al. Mutations in LRP2, which encodes the multiligand receptor megalin, cause Donnai-Barrow and facio-oculo-acoustico-renal syndromes. Nat Genet. 2007;39(8):957–959. doi: 10.1038/ng2063. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.