Whole genome expression profiling of normal human fetal and adult ocular tissues

Abstract

To study growth and development of ocular tissues, gene expression patterns in normal human fetal versus adult eyes were compared. Human retina/retinal pigment epithelium, choroid, sclera, optic nerve* and cornea* tissues were dissected from fetal (24 week gestational age) (N = 9; *N = 6), and adult (N = 6) normal donor eyes. The Illumina^® whole genome expression microarray platform was used to assess differential expression. Statistical significance for all comparisons was determined using the Benjamin and Hochberg False Discovery Rate (FDR, 5%). Significant gene expression fold changes > 1.5 were found in adult versus fetal retina/RPE (N = 1185), choroid (N = 6446), sclera (N = 1349), and cornea (N = 3872), but not optic nerve. Genes showing differential expression were assessed using Ingenuity Pathway Analysis (IPA) for enriched functions and canonical pathways. In all tissues, development, cell death/growth, cancer functions, and signaling canonical pathways were enriched. There was also a general trend of down-regulation of collagen genes in adult tissues.

1. Introduction

The human visual system is complex and requires numerous tissue and cell types to communicate with each other and the brain throughout the process of development (Kolb et al., 2011). In humans, rapid axial growth of the eye globe is seen in fetal development with slowing toward the end of gestation (Fledelius and Christensen, 1996). However, despite this rapid fetal growth, the human eye is not full sized at birth nor is the visual system fully developed; visual signals and nutritional factors contribute to postnatal ocular development in the early years of life (Bremond-Gignac et al., 2011). During childhood development, an active regulatory process of ocular growth, emmetropization, aims to match the optical power of the cornea and lens to the axial length of the eye (Gordon and Donzis, 1985). Failure of emmetropization in postnatal ocular development commonly results in either myopic or hyperopic refractive error, or blurred vision (Gordon and Donzis, 1985).

Given the impracticality of human sample collection, researchers have utilized animal models to study developmental and late-onset ocular diseases. These animal models have demonstrated that visual cues interpreted by the retina and signaled through the choroid and sclera locally control the shape and size of the eye through unknown mechanisms (Faulkner et al., 2007; Tkatchenko et al., 2009; Wallman and Winawer, 2004; Wildsoet and Wallman, 1995). Human eyes with high degrees of refractive error have distinct clinical phenotypes including changes to their size, shape and tissue structure (Xu et al., 2007). Highly myopic eyes tend to be larger than their emmetropic counterparts, with the elongation primarily localized axially (Atchison et al., 2004), suggesting that changes in any or all of those tissues (central retinal, retinal pigment epithelium [RPE] and sclera) may be responsible for failed emmetropization and myopic development.

A better understanding of the tissue-specific expression differences during normal growth and development is key to identify ocular growth and development mechanisms in human tissues. As collection of postnatal eyes undergoing emmetropization is impractical, fetal ocular tissue types of central retina/RPE, choroid, sclera, optic nerve and cornea were compared to their adult counterparts from normal donor eyes to study patterns of ocular growth during development. Although it is unclear to what extent the mechanisms controlling postnatal emmetropization are active in prenatal development, one gene identified in a study of extreme hyperopic refractive error, MFRP (MIM 606227), has been demonstrated to be necessary for both prenatal ocular growth and postnatal emmetropization (Sundin et al., 2008). To our knowledge this is the first whole genome expression analysis comparing human adult versus fetal ocular tissues. This data can provide an understanding of normal changes these tissues undergo during prenatal growth and development in humans and it also may contain clues to understand how diseases such as myopia may result from disruptions to normal growth processes.

2. Materials and methods

2.1. Ocular sample selection

The tissues selected for this study were central retina/RPE, choroid, sclera, optic nerve and cornea. Selection of tissues was based on relevance to disorders studied in our lab such as micro-phthalmia and myopic exaggerated eye growth, and feasibility of collection. The posterior wall tissues were selected and prioritized for larger sample sizes based on their relation to a focal disease studied in our lab, high myopia. Additionally, the cornea and optic nerve tissues were selected due to their relations with other diseases under study in our lab including glaucoma and corneal abnormalities.

To compare growth and development gene expression in ocular tissue types, normal samples from two age groups were used: fetal eyes and adult eyes. The fetal donor eyes were obtained from Advanced Biosciences Resources (Alameda, CA, USA), while the adult eyes were obtained from the North Carolina Eye Bank (Winston–Salem, North Carolina, USA). Fetal gestational age was determined by most recent menstruation in addition to fetal foot measurements. The group of fetal eyes consisted of late prenatal fetal eyes of approximately 24-weeks gestational age from elective abortions with no known defects or abnormalities. 24-weeks gestational eyes are the oldest prenatal eyes readily available and are undergoing rapid growth and axial elongation (Fledelius and Christensen, 1996). Nine fetal donor eyes (four male and five female samples), and six fully grown adult donor eyes (three of each gender) were used for microarray analyses. Space and cost limitations required that optic nerve and cornea sample sizes were reduced to six adult and six fetal samples each. All adult donors were Caucasian, and donors with known ocular disorders were excluded (Table 1). Ethnic and health information was not available for fetal donors. The study was approved by Duke University's Institutional Review Board and adhered to the tenets of the Declaration of Helsinki guidelines.

Table 1. Donor Information for Adult Ocular Whole Globes Used in Tissue Expression Analyses.

Preservation time interval (PTI) is listed in hours. Age is listed in years.

Individual	Race	Sex	Age	PTI	Cause of death
2809	Caucasian	Female	76	4:00	Leukemia/acute myeloid leukemia
2835	Caucasian	Female	55	4:55	Dementia
2828	Caucasian	Male	80	5:20	Pneumonia
2834	Caucasian	Male	56	5:22	Abdominal aortic aneurysm rupture/heart disease
2217	Caucasian	Female	77	6:24	Chronic obstructive pulmonary disease/emphysema
2836	Caucasian	Male	67	5:10	Heart disease

2.2. Ocular dissection

All adult whole globes were immersed in RNAlater^® (Qiagen, Hilden, Germany) within 6.5 h (Table 1) of collection, and shipped overnight on ice. Several studies have shown that the postmortem delay in preservation of samples within this time frame have limited effects on RNA integrity from brain tissue (Durrenberger et al., 2010; Ervin et al., 2007). Fetal whole globes were collected and preserved in RNAlater^® within minutes of collection and shipped overnight on ice. Prior to immersion in RNAlater^®, a 2 mm incision was cut equatorially into all whole globes to allow permeation of the solution to the inner tissues while minimizing unwanted physical changes to the tissue, such as retinal tearing. All whole globes were dissected on the same day as arrival. The retina, RPE, choroid and scleral tissues were isolated at the posterior pole using a circular, double embedded technique using round 7 mm and 5 mm biopsy punches. To reduce contamination of retina to the other tissues samples, the second biopsy punch of 5 mm was used in the center of the 7 mm punch after retinal removal. The adult RPE was collected in RNAlater^® by gentle brushing from the choroid, pipetting the solution, and centrifuging at 4°C to remove the RNAlater^®. Fetal eye samples proved difficult to separate the RPE from the retina. Consequently, the retina and RPE were collected in total. Additionally, optic nerve and corneal samples were isolated from each eye in each age group. Central corneal samples were isolated using a clean 5 mm biopsy punch. The whole optic nerve was collected using clean dissection scissors. The fibrous sclera, optic nerve, and cornea samples were cut into smaller portions (about 1 mm²) using a scalpel to aid in subsequent homogenization. After dissection all tissues were immediately frozen in liquid nitrogen for storage at −80°C until RNA extraction.

2.3. RNA extraction and whole genome expression processing

RNA was extracted from each tissue sample using the mirVanaTM total RNA extraction kit (Ambion, Austin, Texas, USA) following the manufacturer's protocol. The tissue samples were homogenized at 4°C in Ambion lysis buffer using a Bead Ruptor Tissue Homogenizer (Omni, Kennesaw, Georgia, USA) with 2.38 mm metal bead tubes using the following machine settings: 4 cycles × 30 s at speed 4.0 with 30 s break between cycles. Quality control for each sample included measuring RNA concentration and 260/280 nm ratios using a Nanodrop^® (Invitrogen, Carlsbad, California, USA). The ocular tissue RNA samples were labeled and amplified using the Illumina^® Total Prep kit (Ambion, Austin, Texas, USA). RNA samples were hybridized to Illumina^® HumanHT-12 v4 Expression BeadChips (San Diego, California, USA), which target over 25,000 annotated genes (over 48,000 probes). All protocols were performed following the manufacturer's recommendations. Twelve samples per chip were processed and chips were run in two batches, each containing samples of at least two tissue types. The first batch (N = 51) included all adult and fetal retina/RPE, choroid and sclera samples. The second batch included all adult and fetal cornea and optic nerve samples (N = 24).

2.4. Whole genome expression analyses

The microarray data preprocessing procedures were conducted separately for each of the two batches (See Section 2.3). After the data was generated for each batch, background noise was subtracted from the intensity values using the Illumina^® GenomeStudio program. The data was exported from GenomeStudio and log2 transformed. Sample outliers were determined by principle component analyses using the Hoteling's T2 test (Hotelling, 1931) (at 95% confidence interval) and removed from further analyses. The data intensity was normalized by Quantile normalization followed by Multichip Averaging (Irizarry et al., 2003) to reduce chip effects. The exact Wilcoxon rank sum test (Wilcoxen, 1945) was used to identify differentially expressed genes because of the relatively small sample size. Fetal ocular tissues were compared to their adult counterparts, and assessed for relative fold changes per probe. Adult retina and RPE samples were averaged for comparison to fetal retina/RPE. The Benjamin and Hochberg False Discovery Rate (Benjamini and Hochberg, 1995) (FDR) procedure was applied, and FDR was controlled at 0.05 to determine statistical significance for all comparisons.

2.5. Quantitative real-time PCR (qPCR)

A total of 1.5 μg of RNA for each sample was converted into double –stranded cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystem, Austin, Texas) according to the manufacturer's instructions. qPCR reactions were carried out with TaqMan primers and probe sets from Applied Biosystem, (Austin, Texas). GAPDH was used as the endogenous control house-keeping gene. PCR reactions were carried out with three technical duplicates for each sample. The comparative threshold cycle (C_T) method (McCurdy et al., 2008), was used to calculate the relative gene expression difference between adult and 24-week groups, and the results were expressed as fold changes in gene expression.

2.6. Pathway analyses

In total, five adult versus fetal expression comparisons were made, representing each layer of tissue. Functional and pathway enrichment assessment was conducted using Ingenuity^® Pathway Analysis ([IPA], Ingenuity^® Systems, www.ingenuity.com) for each tissue comparison using lists of genes with differentially expressed probes meeting the following criteria. In the retina/RPE, choroid, sclera and cornea comparisons, only probes meeting FDR significance (See Section 2.4) were included. The adult versus fetal optic nerve comparisons did not produce any probes meeting FDR significance; consequently, probes with raw p-values <0.05 were included. To focus on changes that we were confident were biologically important, only those probes with fold changes ≥1.5 in each tissue were included. For each tissue comparison, Table 2 contains the number of probes, representative unique genes, as well as the range and average fold changes.

Table 2. Filtered Number of Probes and Unique Genes.

Retina/RPE, Choroid, Sclera and Cornea were filtered by False Discovery Rate (FDR) <0.05. *Optic Nerve were filtered by raw p-value <0.05. All tissues were filtered by a fold change (FC) >1.5.

Tissue	FC range	Average FC	# probes	# unique genes
Retina/RPE	−8.50 ↔ 21.09	1.86	1236	1185
Choroid	−617.57 ↔ 550.49	5.56	7124	6446
Sclera	−125.11 ↔ 6.40	2.28	1423	1349
Optic nerve*	−205.22 ↔ 1695.34	5.90	2399	2179
Cornea	−1252.45 ↔ 205.77	7.34	4354	3872

For each tissue comparison, the lists of differentially expressed genes, their significance and fold change were imported into IPA. Each gene was mapped to its corresponding object In the Ingenuity^® Knowledge Database (www.ingenuity.com). Functional analysis of these genes identified biological functions and/or diseases that were most significant to the molecules (genes) in the group. A right-tailed Fischer's exact test was used to calculate a p-value determining the probability that each biological function and/or disease assigned was due to chance alone. Additionally, IPA calculated a z-score, or standard score, for each functional classification. A z-score of <−2.00 indicates that function has a significantly decreased predicted activation in the data set. Conversely, a z-score of >2.0 indicates a predicted increased activation of that functional group in the data set.

Canonical pathway analysis identified pathways from the IPA library of canonical pathways that were most significant to the group. The significance of the association between the data set and the canonical pathway was measured in two ways: 1) A ratio of the number of genes from the data set that map to the pathway divided by the total number of genes that map to the pathway; 2) Fischer's exact test was used to calculate a p-value determining the probability that the association between the genes in the data set and the canonical pathway was due to chance (Ingenuity^® Systems, www.ingenuity.com).

3. Results

3.1. Microarray analyses

The number and fold change of genes differentially expressed varied by tissue type. The retina/RPE had the lowest average fold change and the fewest number of differentially expressed genes, while the cornea had the largest fold changes and the highest number of differentially expressed genes (Table 2). The optic nerve comparisons failed to yield any FDR significantly differentially expressed genes.

The microarray chips were redundant, thus, many genes had multiple tagging probes within them. Given the stringent quality control and fold change criteria used in our testing, all probes for a given gene did not always meet that threshold. Consequently, some genes have differing numbers of probes in different tissue types. Additionally, probes with near/below background intensity readings in one or both age groups tended to have very high fold changes (Table 3). In particular, the most differentially expressed genes in the retina/RPE and sclera included a number of genes whose expression detection levels were low (bottom 5% intensity of differentially expressed genes) for one or both age groups (Table 3). These low readings may have resulted either from failed probes or absent expression. The degree of expression fold changes in these genes (noted in Table 3) with low readings in both groups may appear higher due to the relative calculation and the normalization process and should be considered when interpreting the data.

Table 3. Highest Fold Changes.

Genes (represented by probes) from each tissue with the highest fold changes (FC). Negative FC indicates increased expression in fetal relative to adult tissues.

Retina/RPE			Choroid			Sclera			Optic nerve			Cornea
Gene	FC	P-value	Gene	FC	P-value	Gene	FC	P-value	Gene	FC	P-value	Gene	FC	P-value
CRHR1b	−8.5	0.0003	HBG1	−617.57	0.0007	ACOX1b	−125.11	0.0065	COL1A1	−205.22	0.0043	LOC100134134b	−1252.45	0.0043
FLJ40244a	−8.11	0.0012	COL9A1b	−165.29	0.0007	EYA2a	−96.93	0.0016	COL3A1	−171.4	0.0043	OGNb	−612.53	0.0043
LOC100133528a	−5.25	0.0025	HBG2	−159.85	0.0007	KRTAP5-6a	−91.07	0.0003	COL1A2	−100.87	0.0043	MEG3b	−451.89	0.0043
CYP24A1a	−4.65	0.0026	HCG2P7	−117.26	0.0013	GRIA1	−32.41	0.0012	C5ORF13	−97.75	0.0043	COL5A1	−321.1	0.0043
ETFBb	−4.31	0.000027	HBA1	−98.16	0.0007	CDONa	−9.68	0.0011	CPNE4b	−73.78	0.0075	FAM180Ab	−277.35	0.0043
C13ORF23b	−4.1	0.0059	GRIK1	−92.66	0.0007	CDC45Lb	−9.61	0.000037	MMP15b	−63.99	0.0067	OGN	−274.14	0.0043
NUP155a	−3.49	0.0036	APLNR	−75.93	0.0007	MRI1	−9.43	0.0009	PRNDb	−63.09	0.0043	THBS2	−265.3	0.0043
ZNF92b	−3.4	0.0002	HBM	−74.74	0.0007	SNIPa	−8.94	0.0001	SEMA5Bb	−60.69	0.008	DPTb	−256.95	0.0043
MAGED4b	−3.38	0.0001	DUXAP3	−66.52	0.0027	MAL2	−8.59	0.000037	COL1A2	−60.34	0.0043	SOCS1	−250.99	0.0043
C20ORF118b	−3.36	0.0055	DMC1	−65	0.0027	FOXL2a	−8.04	0.0007	CDH7b	−60.03	0.0067	THRA	−228.18	0.0043
KRT6Ab	−3.28	0.0011	NLRP8	−62.8	0.0027	KCNIP2	−7.92	0.0016	C20ORF103b	−59.18	0.008	PXDN	−217.78	0.0043
SNIPa	−3.2	0.0068	GRIK1b	−62.17	0.0007	COL11A1a	−7.81	0.0044	PCDH12b	−46.03	0.0075	IGDCC4b	−211.63	0.0043
ASB6b	−3.17	0.0064	LOC100128084	−55.69	0.0013	UBASH3Aa	−7.43	0.0001	COL5A1	−45.92	0.0043	COL1A1	−209.79	0.0043
KIF4Ab	−3.16	0.0008	TOP2A	−55.64	0.0007	S100A12a	−7.15	0.000037	TBC1D10Cb	−44.02	0.0067	COL3A1	−161.78	0.0043
FAM19A1b	−3.09	0.0042	ARMCX6	−54.79	0.0024	NEU3	−7.05	0.000037	KAL1	−43.22	0.0043	RASL11Bb	−157.22	0.0043
MMD2b	−3.08	0.0001	LOC645452	−50.32	0.0027	KIAA1549	−6.56	0.000037	APOA1b	−36.38	0.0075	MMP23B	−154.42	0.0043
LRRIQ3b	−3.05	0.0008	IGF1b	−45.62	0.0007	DEFA4a	−6.3	0.0027	HS.204481b	−35.77	0.0055	KCNIP4b	−148.3	0.0043
SGOL2a	−3.01	0.0018	FCAR	−45.56	0.0027	TCF2a	−6.26	0.0046	TNC	−34.88	0.0043	PCDH17b	−147.06	0.0043
KLK12b	−2.97	0.0068	KCNK4b	−45.45	0.0007	PTGER3a	−6.14	0.0011	FNDC1	−33.21	0.0043	MFAP2	−143.13	0.0043
TDGF1b	−2.96	0.0001	CRTAP	−43.96	0.0007	DDX25	−6.01	0.0021	RSPO4b	−31.26	0.0055	NTMb	−131.38	0.0043
NNAT	−2.94	0.000027	CXCL12	−43.05	0.0007	RASSF5a	−5.98	0.0046	FOXL2b	−29.21	0.0054	COCHb	−131.04	0.0043
LOC651746b	−2.92	0.0002	UBE2C	−42.86	0.0007	BUB1b	−5.94	0.0005	COL5A2	−26.92	0.0043	UNQ1940b	−126.31	0.0043
OTOSb	−2.92	0.0008	CDKN2AIPNL	−41.93	0.0047	RPEa	−5.92	0.0046	TNFRSF25b	−24.92	0.0043	SRPX	−126.27	0.0043
COL6A6b	−2.89	0.0046	COL4A5b	−39.77	0.0007	MATN4a	−5.91	0.0027	CDC20Bb	−24.3	0.0043	CAPN6	−103.79	0.0043
ZWILCHb	−2.88	0.0008	KIAA1751	−36.97	0.0027	KLK3b	−5.75	0.0004	SLIT1b	−23.74	0.008	HBG1b	−99.54	0.0043
SSX2b	−2.86	0.0068	SFRP4	−36.3	0.0007	CACNA1G	−5.73	0.0021	THSD1b	−23.04	0.008	FBN2	−93.03	0.0043
NSBP1b	−2.84	0.000027	TRIM24b	−36.18	0.0007	METTL1a	−5.71	0.0004	SLC16A14b	−22.9	0.008	HBA2b	−89.18	0.0043
RIPPLY2b	−2.81	0.000027	CCT7	−35.79	0.0024	HBMa	−5.71	0.0027	OCA2b	−22.81	0.0075	THY1b	−86.78	0.0043
COL2A1b	−2.74	0.0002	NCAPG	−35.44	0.0007	ARPP-21	−5.67	0.0054	ARSEa	−21.69	0.0066	SOX11	−81.66	0.0043
TTTY17Ba	−2.74	0.005	MCART1	−34.88	0.0027	AGR2b	−5.61	0.0012	SPON1	−20.66	0.0043	SMOb	−80.53	0.0043
TTTY9Aa	−2.73	0.0036	COL1A2	−34.43	0.0024	ALDH4A1	−5.44	0.0035	MYH7b	−20.65	0.008	C1QTNF1	−79.43	0.0043
BPY2Ca	−2.68	0.0072	LRRC37B2	−34.37	0.0027	TOP2Ab	−5.43	0.000037	UNC13C	−20.63	0.0043	MMP23A	−78.59	0.0043
KIAA0514b	−2.65	0.000027	LOC100128505	−34.36	0.0027	HRH3	−5.37	0.0072	FBN2	−20.57	0.0043	COL11A1b	−75.57	0.0043
LOC653889b	−2.64	0.0002	ARHGAP28	− 33.11	0.0007	TBC1D29a	−5.35	0.0016	NKD2	−20.31	0.0043	PMEPA1b	−74.45	0.0043
MATKb	−2.62	0.0006	AP1S1b	−32	0.0024	OR1E2a	−5.31	0.0033	MMP25b	−20	0.0043	CCND2	−73.66	0.0043
TREML2Pa	−2.6	0.0025	GPC3	−31.98	0.0007	MIR1258b	−5.3	0.000074	MXRA5	−19.37	0.0043	HBG2b	−70.49	0.0043
BTBD17	−2.59	0.000027	LOC399900	−31.51	0.0027	GPR44a	−5.22	0.0072	C20ORF46b	−18.85	0.0075	SCXAb	−69.87	0.0043
SLC18A2b	−2.54	0.0008	HBZ	−30.9	0.0007	COL6A6b	−5.18	0.000037	VCAN	−18.8	0.0043	SNCAIPb	−63.58	0.0043
PAMR1a	−2.54	0.0036	LAMA1	−30.8	0.0007	TRIM29a	−5.17	0.0062	DACT1	−18.02	0.0043	PYCR1	−62.75	0.0043
ABHD11b	−2.53	0.0019	IP6K2b	−30.51	0.0007	RHBDF2a	−5.12	0.0007	NMUa	−18.02	0.0067	C1QTNF5	−61.52	0.0043
INMTb	21.09	0.000027	SERPINA3	550.49	0.0007	MYOCb	6.4	0.000037	BCAS1b	1695.34	0.0067	MAMDC2	205.77	0.0043
C5ORF48a	9.79	0.0004	SAA1	344.94	0.0007	ABCF2a	5.47	0.0004	BCAS1b	1105.48	0.0075	AKR1C2	192.43	0.0043
GCGRb	5.33	0.000053	MT1Mb	221.47	0.0007	FMO3a	4.1	0.0021	MAGb	111.21	0.0043	KRT24	181.18	0.0043
ASCL4a	5.17	0.0046	PLA2G2A	220.99	0.0007	SLPIa	3.92	0.000037	S1PR5b	79.37	0.0055	ADH7	174.83	0.0043
HAND1b	4.37	0.0018	SNTNb	179.32	0.0027	PLA2G2A	3.64	0.000037	ZNF488b	50.7	0.0075	ZBTB16b	156.59	0.0039
LOC399706b	4.08	0.0003	C19ORF12	174.68	0.0007	GLDN	3.26	0.000037	GPX3	47.64	0.0043	ANGPTL7	142.31	0.0043
HNMTa	3.9	0.0044	LOC100128693	139.44	0.0013	AKR1B10b	3.14	0.000037	RAB11FIP4b	43.7	0.008	CA3b	114	0.0039
KIR2DL3b	3.8	0.000053	LOC100132620	137.85	0.0007	SLC26A4a	3.14	0.000037	FA2Hb	38.01	0.0075	IL20RAb	111.17	0.0043
NRP2b	3.72	0.0001	TTC21B	123.34	0.0013	SCNN1Aa	3.08	0.000037	MAL	25.43	0.0043	AKR1C4	109.18	0.0043
DLX6b	3.52	0.0005	ZNF273	112.7	0.0027	CCL15b	2.94	0.0012	CENTA1	24.84	0.0043	SLURP1b	99.65	0.0039
FOXR1b	3.42	0.0015	LOC100132518	108.2	0.0007	KCNE1b	2.9	0.0001	SEPT4	22.55	0.0043	SLC47A1b	89.77	0.0039
OR52A1a	3.41	0.0001	IL1RL1b	96.02	0.0007	COMP	2.78	0.000037	COBLb	19.78	0.0043	IFI27	82.06	0.0043
LOC100127983b	3.36	0.0068	GTF2IRD2B	86.41	0.0007	CD69	2.72	0.0021	MAP6D1	18.84	0.0043	INMTb	81.8	0.0039
HERV-FRDb	3.35	0.0005	LOC100132540	85.5	0.0024	C10ORF81b	2.59	7.37E-05	SEPT4	18.7	0.0043	GJB4	81.42	0.0043
40610b	3.33	0.0025	NNMTb	84.35	0.0007	TCEAL6	2.58	0.000037	COL4A3	17.72	0.0067	MME	74.43	0.0043
IL1RNb	3.3	0.0036	PPIL3	80.97	0.0007	SCGB3A2	2.57	0.000037	APOD	16.26	0.0043	MYOC	74.2	0.0043
DEFB119b	3.14	0.000027	RUNDC2B	80.95	0.0007	CHI3L2b	2.56	0.000037	QDPR	14.35	0.0043	EFCBP1b	73.56	0.0039
EDNRBb	3.12	0.0002	LOC729652	80.46	0.0007	COL4A3	2.56	0.0011	PRODH	12.89	0.0043	TMEM45Bb	68.26	0.0039
MAGEA11b	3.1	0.0005	STAG3L1	74.06	0.0007	LIN28b	2.54	0.0007	MIR1974	12.05	0.0173	STEAP4b	66.43	0.0039
MT1M	3.02	0.000027	PRIM2	69.86	0.0013	SAA1	2.54	0.0007	APLP1	12.02	0.0043	PLK5Pb	66.16	0.0039
LOC389834b	3	0.000027	SNHG10	68.88	0.0007	MMP3	2.53	0.000037	CAPN3	11.78	0.0087	HSD11B1b	63.84	0.0039
FAM55A	2.99	0.0003	BCAS4	67.49	0.0007	ABCC2	2.51	0.000037	PPP1R14A	11.55	0.0043	ECM1	63.1	0.0043
DCST2b	2.97	0.000053	LOC100134172	66.89	0.0007	KRTAP9-3b	2.49	0.0003	PCTK3	11.43	0.0043	NQO1	58.54	0.0043
LOC646085b	2.94	0.0003	LOC100129445	65.3	0.0007	MMP12a	2.43	0.0011	MIR1978	11.06	0.0043	CRTAC1	54.78	0.0043
RHO	2.93	0.000027	PTERb	65.12	0.0027	MT1Mb	2.42	0.000037	SLC7A2	9.89	0.0043	KRT27	47.54	0.0043
CAMPb	2.93	0.0037	GNPTAB	64.42	0.0047	CYP4F12b	2.39	0.000037	TP53INP2	9.58	0.0043	CALML3	46.96	0.0043
BEST3b	2.91	0.0003	TRIM16L	60.82	0.0047	OR7D2	2.39	0.0003	C7ORF41	9.54	0.0043	PTPN22b	46.1	0.0039
MICBb	2.86	0.000027	CHI3L2	59.38	0.0007	SERPINA3	2.39	0.0005	MT1X	8.93	0.0043	COL8A1b	43.63	0.0039
LGSNb	2.86	0.005	LOC202134	59.22	0.0007	ASB11b	2.35	0.0027	BCYRN1	8.92	0.0043	PSCA	40.4	0.0043
HUS1b	2.82	0.005	LOC100128562	58.77	0.0007	GPR88	2.34	0.000037	COL4A4a	8.89	0.0067	ZBTB16b	40.19	0.0043
LRRC25b	2.8	0.0068	TAF8	58.48	0.0007	PDPN	2.34	0.0024	DBNDD2	8.63	0.0043	ECM1	38.83	0.0043
FAM55Db	2.77	0.0025	PPID	58.47	0.0007	LOC100130988b	2.29	0.0046	FAM107A	8.21	0.0043	SOD3	37.61	0.0043
EPYCb	2.76	0.0012	LOC651192	53.93	0.0007	ANGPTL7	2.27	0.000037	CDKN1A	8.09	0.0173	MICALCLb	37.6	0.0043
KCNJ14b	2.74	0.0008	LOC728054	52.7	0.0013	NNMTa	2.27	0.000037	SYNM	7.74	0.0087	AGXT2L1b	36.92	0.0039
VCAM1	2.73	0.005	LOC647592	52.36	0.0007	GBP2	2.25	0.000037	ENPP2	7.67	0.0173	CYP26A1	35.79	0.0043
GJA8b	2.69	0.0008	STAP2	49.61	0.0027	INMTb	2.25	0.000037	C10ORF116	7.64	0.0043	FAM164Cb	35.05	0.0039
KRTAP19-4b	2.65	0.0018	ZNF100	49.02	0.0007	MYOC	2.22	0.000037	CHN2	7.47	0.0043	ALDH3A1	34.93	0.0043
OR13F1b	2.64	0.0068	TEX11	48.84	0.0013	KRT28b	2.22	0.0062	S100A8b	7.38	0.0353	SYT8	34.61	0.0043
LOC653075b	2.63	0.0005	SLA2	47.56	0.0007	ZBTB16	2.21	0.000037	ENPP2	7.22	0.0173	RGS7BPb	34.48	0.0039
CYP2A7b	2.63	0.0068	ST20	46.86	0.0007	KGFLP1a	2.21	0.0007	RGS1b	7.05	0.0043	CLCA4	34	0.0043
Gene			FC			P-value		Gene			FC			P-value
Adult versus fetal retina/RPE
CRHR1b			−8.5			0.0003		INMTb			21.09			2.65E-05
FLJ40244a			−8.11			0.0012		C5ORF48a			9.79			0.0004
LOC100133528a			−5.25			0.0025		GCGRb			5.33			5.29E-05
CYP24A1a			−4.65			0.0026		ASCL4a			5.17			0.0046
ETFBb			−4.31			2.65E-05		HAND1b			4.37			0.0018
C13ORF23b			−4.1			0.0059		LOC399706b			4.08			0.0003
NUP155a			−3.49			0.0036		HNMTa			3.9			0.0044
ZNF92b			−3.4			0.0002		KIR2DL3b			3.8			5.29E-05
MAGED4b			−3.38			0.0001		NRP2b			3.72			0.0001
C20ORF118b			−3.36			0.0055		DLX6b			3.52			0.0005
KRT6Ab			−3.28			0.0011		FOXR1b			3.42			0.0015
SNIPa			−3.2			0.0068		OR52A1a			3.41			0.0001
ASB6b			−3.17			0.0064		LOC100127983b			3.36			0.0068
KIF4Ab			−3.16			0.0008		HERV-FRDb			3.35			0.0005
FAM19A1b			−3.09			0.0042		40610b			3.33			0.0025
MMD2b			−3.08			0.0001		IL1RNb			3.3			0.0036
LRRIQ3b			−3.05			0.0008		DEFB119b			3.14			2.65E-05
SGOL2a			−3.01			0.0018		EDNRBb			3.12			0.0002
KLK12b			−2.97			0.0068		MAGEA11b			3.1			0.0005
TDGF1b			−2.96			0.0001		MT1M			3.02			2.65E-05
NNAT			−2.94			2.65E-05		LOC389834b			3			2.65E-05
LOC651746b			−2.92			0.0002		FAM55A			2.99			0.0003
OTOSb			−2.92			0.0008		DCST2b			2.97			5.29E-05
COL6A6b			−2.89			0.0046		LOC646085b			2.94			0.0003
ZWILCHb			−2.88			0.0008		RHO			2.93			2.65E-05
SSX2b			−2.86			0.0068		CAMPb			2.93			0.0037
NSBP1b			−2.84			2.65E-05		BEST3b			2.91			0.0003
RIPPLY2b			−2.81			2.65E-05		MICBb			2.86			2.65E-05
COL2A1b			−2.74			0.0002		LGSNb			2.86			0.005
TTTY17Ba			−2.74			0.005		HUS1b			2.82			0.005
TTTY9Aa			−2.73			0.0036		LRRC25b			2.8			0.0068
BPY2Ca			−2.68			0.0072		FAM55Db			2.77			0.0025
KIAA0514b			−2.65			2.65E-05		EPYCb			2.76			0.0012
LOC653889b			−2.64			0.0002		KCNJ14b			2.74			0.0008
MATKb			−2.62			0.0006		VCAM1			2.73			0.005
TREML2Pa			−2.6			0.0025		GJA8b			2.69			0.0008
BTBD17			−2.59			2.65E-05		KRTAP19-4b			2.65			0.0018
SLC18A2b			−2.54			0.0008		OR13F1b			2.64			0.0068
PAMR1a			−2.54			0.0036		LOC653075b			2.63			0.0005
ABHD11b			−2.53			0.0019		CYP2A7b			2.63			0.0068
Adult versus fetal choroid
HBG1			−617.57			0.0007		SERPINA3			550.49			0.0007
COL9A1b			−165.29			0.0007		SAA1			344.94			0.0007
HBG2			−159.85			0.0007		MT1Mb			221.47			0.0007
HCG2P7			−117.26			0.0013		PLA2G2A			220.99			0.0007
HBA1			−98.16			0.0007		SNTNb			179.32			0.0027
GRIK1			−92.66			0.0007		C19ORF12			174.68			0.0007
APLNR			−75.93			0.0007		LOC100128693			139.44			0.0013
HBM			−74.74			0.0007		LOC100132620			137.85			0.0007
DUXAP3			−66.52			0.0027		TTC21B			123.34			0.0013
DMC1			−65			0.0027		ZNF273			112.7			0.0027
NLRP8			−62.8			0.0027		LOC100132518			108.2			0.0007
GRIK1b			−62.17			0.0007		IL1RL1b			96.02			0.0007
LOC100128084			−55.69			0.0013		GTF2IRD2B			86.41			0.0007
TOP2A			−55.64			0.0007		LOC100132540			85.5			0.0024
ARMCX6			−54.79			0.0024		NNMTb			84.35			0.0007
LOC645452			−50.32			0.0027		PPIL3			80.97			0.0007
IGF1b			−45.62			0.0007		RUNDC2B			80.95			0.0007
FCAR			−45.56			0.0027		LOC729652			80.46			0.0007
KCNK4b			−45.45			0.0007		STAG3L1			74.06			0.0007
CRTAP			−43.96			0.0007		PRIM2			69.86			0.0013
CXCL12			−43.05			0.0007		SNHG10			68.88			0.0007
UBE2C			−42.86			0.0007		BCAS4			67.49			0.0007
CDKN2AIPNL			−41.93			0.0047		LOC100134172			66.89			0.0007
COL4A5b			−39.77			0.0007		LOC100129445			65.3			0.0007
KIAA1751			−36.97			0.0027		PTERb			65.12			0.0027
SFRP4			−36.3			0.0007		GNPTAB			64.42			0.0047
TRIM24b			−36.18			0.0007		TRIM16L			60.82			0.0047
CCT7			−35.79			0.0024		CHI3L2			59.38			0.0007
NCAPG			−35.44			0.0007		LOC202134			59.22			0.0007
MCART1			−34.88			0.0027		LOC100128562			58.77			0.0007
COL1A2			−34.43			0.0024		TAF8			58.48			0.0007
LRRC37B2			−34.37			0.0027		PPID			58.47			0.0007
LOC100128505			−34.36			0.0027		LOC651192			53.93			0.0007
ARHGAP28			−33.11			0.0007		LOC728054			52.7			0.0013
AP1S1b			−32			0.0024		LOC647592			52.36			0.0007
GPC3			−31.98			0.0007		STAP2			49.61			0.0027
LOC399900			−31.51			0.0027		ZNF100			49.02			0.0007
HBZ			−30.9			0.0007		TEX11			48.84			0.0013
LAMA1			−30.8			0.0007		SLA2			47.56			0.0007
IP6K2b			−30.51			0.0007		ST20			46.86			0.0007
Adult versus fetal sclera
ACOX1b			−125.11			0.0065		MYOCb			6.4			3.69E-05
EYA2a			−96.93			0.0016		ABCF2a			5.47			0.0004
KRTAP5-6a			−91.07			0.0003		FMO3a			4.1			0.0021
GRIA1			−32.41			0.0012		SLPIa			3.92			3.69E-05
CDONa			−9.68			0.0011		PLA2G2A			3.64			3.69E-05
CDC45Lb			−9.61			3.69E-05		GLDN			3.26			3.69E-05
MRI1			−9.43			0.0009		AKR1B10b			3.14			3.69E-05
SNIPa			−8.94			0.0001		SLC26A4a			3.14			3.69E-05
MAL2			−8.59			3.69E-05		SCNN1Aa			3.08			3.69E-05
FOXL2a			−8.04			0.0007		CCL15b			2.94			0.0012
KCNIP2			−7.92			0.0016		KCNE1b			2.9			0.0001
COL11A1a			−7.81			0.0044		COMP			2.78			3.69E-05
UBASH3Aa			−7.43			0.0001		CD69			2.72			0.0021
S100A12a			−7.15			3.69E-05		C10ORF81b			2.59			7.37E-05
NEU3			−7.05			3.69E-05		TCEAL6			2.58			3.69E-05
KIAA1549			−6.56			3.69E-05		SCGB3A2			2.57			3.69E-05
DEFA4a			−6.3			0.0027		CHI3L2b			2.56			3.69E-05
TCF2a			−6.26			0.0046		COL4A3			2.56			0.0011
PTGER3a			−6.14			0.0011		LIN28b			2.54			0.0007
DDX25			−6.01			0.0021		SAA1			2.54			0.0007
RASSF5a			−5.98			0.0046		MMP3			2.53			3.69E-05
BUB1b			−5.94			0.0005		ABCC2			2.51			3.69E-05
RPEa			−5.92			0.0046		KRTAP9-3b			2.49			0.0003
MATN4a			−5.91			0.0027		MMP12a			2.43			0.0011
KLK3b			−5.75			0.0004		MT1Mb			2.42			3.69E-05
CACNA1G			−5.73			0.0021		CYP4F12b			2.39			3.69E-05
METTL1a			−5.71			0.0004		OR7D2			2.39			0.0003
HBMa			−5.71			0.0027		SERPINA3			2.39			0.0005
ARPP-21			−5.67			0.0054		ASB11b			2.35			0.0027
AGR2b			−5.61			0.0012		GPR88			2.34			3.69E-05
ALDH4A1			−5.44			0.0035		PDPN			2.34			0.0024
TOP2Ab			−5.43			3.69E-05		LOC100130988b			2.29			0.0046
HRH3			−5.37			0.0072		ANGPTL7			2.27			3.69E-05
TBC1D29a			−5.35			0.0016		NNMTa			2.27			3.69E-05
OR1E2a			−5.31			0.0033		GBP2			2.25			3.69E-05
MIR1258b			−5.3			7.37E-05		INMTb			2.25			3.69E-05
GPR44a			−5.22			0.0072		MYOC			2.22			3.69E-05
COL6A6b			−5.18			3.69E-05		KRT28b			2.22			0.0062
TRIM29a			−5.17			0.0062		ZBTB16			2.21			3.69E-05
RHBDF2a			−5.12			0.0007		KGFLP1a			2.21			0.0007
Adult versus fetal optic nerve
COL1A1			−205.22			0.0043		BCAS1b			1695.34			0.0067
COL3A1			−171.4			0.0043		BCAS1b			1105.48			0.0075
COL1A2			−100.87			0.0043		MAGb			111.21			0.0043
C5ORF13			−97.75			0.0043		S1PR5b			79.37			0.0055
CPNE4b			−73.78			0.0075		ZNF488b			50.7			0.0075
MMP15b			−63.99			0.0067		GPX3			47.64			0.0043
PRNDb			−63.09			0.0043		RAB11FIP4b			43.7			0.008
SEMA5Bb			−60.69			0.008		FA2Hb			38.01			0.0075
COL1A2			−60.34			0.0043		MAL			25.43			0.0043
CDH7b			−60.03			0.0067		CENTA1			24.84			0.0043
C20ORF103b			−59.18			0.008		SEPT4			22.55			0.0043
PCDH12b			−46.03			0.0075		COBLb			19.78			0.0043
COL5A1			−45.92			0.0043		MAP6D1			18.84			0.0043
TBC1D10Cb			−44.02			0.0067		SEPT4			18.7			0.0043
KAL1			−43.22			0.0043		COL4A3			17.72			0.0067
APOA1b			−36.38			0.0075		APOD			16.26			0.0043
HS.204481b			−35.77			0.0055		QDPR			14.35			0.0043
TNC			−34.88			0.0043		PRODH			12.89			0.0043
FNDC1			−33.21			0.0043		MIR1974			12.05			0.0173
RSPO4b			−31.26			0.0055		APLP1			12.02			0.0043
FOXL2b			−29.21			0.0054		CAPN3			11.78			0.0087
COL5A2			−26.92			0.0043		PPP1R14A			11.55			0.0043
TNFRSF25b			−24.92			0.0043		PCTK3			11.43			0.0043
CDC20Bb			−24.3			0.0043		MIR1978			11.06			0.0043
SLIT1b			−23.74			0.008		SLC7A2			9.89			0.0043
THSD1b			−23.04			0.008		TP53INP2			9.58			0.0043
SLC16A14b			−22.9			0.008		C7ORF41			9.54			0.0043
OCA2b			−22.81			0.0075		MT1X			8.93			0.0043
ARSEa			−21.69			0.0066		BCYRN1			8.92			0.0043
SPON1			−20.66			0.0043		COL4A4a			8.89			0.0067
MYH7b			−20.65			0.008		DBNDD2			8.63			0.0043
UNC13C			−20.63			0.0043		FAM107A			8.21			0.0043
FBN2			−20.57			0.0043		CDKN1A			8.09			0.0173
NKD2			−20.31			0.0043		SYNM			7.74			0.0087
MMP25b			−20			0.0043		ENPP2			7.67			0.0173
MXRA5			−19.37			0.0043		C10ORF116			7.64			0.0043
C20ORF46b			−18.85			0.0075		CHN2			7.47			0.0043
VCAN			−18.8			0.0043		S100A8b			7.38			0.0353
DACT1			−18.02			0.0043		ENPP2			7.22			0.0173
NMUa			−18.02			0.0067		RGS1b			7.05			0.0043
Adult versus fetal cornea
LOC100134134b			−1252.45			0.0043		MAMDC2			205.77			0.0043
OGNb			−612.53			0.0043		AKR1C2			192.43			0.0043
MEG3b			−451.89			0.0043		KRT24			181.18			0.0043
COL5A1			−321.1			0.0043		ADH7			174.83			0.0043
FAM180Ab			−277.35			0.0043		ZBTB16b			156.59			0.0039
OGN			−274.14			0.0043		ANGPTL7			142.31			0.0043
THBS2			−265.3			0.0043		CA3b			114			0.0039
DPTb			−256.95			0.0043		IL20RAb			111.17			0.0043
SOCS1			−250.99			0.0043		AKR1C4			109.18			0.0043
THRA			−228.18			0.0043		SLURP1b			99.65			0.0039
PXDN			−217.78			0.0043		SLC47A1b			89.77			0.0039
IGDCC4b			−211.63			0.0043		IFI27			82.06			0.0043
COL1A1			−209.79			0.0043		INMTb			81.8			0.0039
COL3A1			−161.78			0.0043		GJB4			81.42			0.0043
RASL11Bb			−157.22			0.0043		MME			74.43			0.0043
MMP23B			−154.42			0.0043		MYOC			74.2			0.0043
KCNIP4b			−148.3			0.0043		EFCBP1b			73.56			0.0039
PCDH17b			−147.06			0.0043		TMEM45Bb			68.26			0.0039
MFAP2			−143.13			0.0043		STEAP4b			66.43			0.0039
NTMb			−131.38			0.0043		PLK5Pb			66.16			0.0039
COCHb			−131.04			0.0043		HSD11B1b			63.84			0.0039
UNQ1940b			−126.31			0.0043		ECM1			63.1			0.0043
SRPX			−126.27			0.0043		NQO1			58.54			0.0043
CAPN6			−103.79			0.0043		CRTAC1			54.78			0.0043
HBG1b			−99.54			0.0043		KRT27			47.54			0.0043
FBN2			−93.03			0.0043		CALML3			46.96			0.0043
HBA2b			−89.18			0.0043		PTPN22b			46.1			0.0039
THY1b			−86.78			0.0043		COL8A1b			43.63			0.0039
SOX11			−81.66			0.0043		PSCA			40.4			0.0043
SMOb			−80.53			0.0043		ZBTB16b			40.19			0.0043
C1QTNF1			−79.43			0.0043		ECM1			38.83			0.0043
MMP23A			−78.59			0.0043		SOD3			37.61			0.0043
COL11A1b			−75.57			0.0043		MICALCLb			37.6			0.0043
PMEPA1b			−74.45			0.0043		AGXT2L1b			36.92			0.0039
CCND2			−73.66			0.0043		CYP26A1			35.79			0.0043
HBG2b			−70.49			0.0043		FAM164Cb			35.05			0.0039
SCXAb			−69.87			0.0043		ALDH3A1			34.93			0.0043
SNCAIPb			−63.58			0.0043		SYT8			34.61			0.0043
PYCR1			−62.75			0.0043		RGS7BPb			34.48			0.0039
C1QTNF5			−61.52			0.0043		CLCA4			34			0.0043

^aIndicates fold probes/genes whose intensity values for both adult and fetal samples were in the lowest 5% among the differentially expressed genes for that tissue.

^bIndicates the adult (negative FC) or fetal (positive FC) intensity value was in the lowest 5% among the differentially expressed genes for that tissue.

In each tissue, some of the most differentially expressed genes have near background readings in one of the two age groups, which may indicate expression being turned on or off (Table 3). Also the most differentially expressed genes per tissue type included many genes that have either previously been implicated or suggested for involvement in ocular diseases particular to these tissues. As a group, extracellular matrix collagens were among the most differentially expressed genes during growth and development in all tissue types tested (Table 3). Fig. 1 is a heat map of 31 extracellular matrix, ocular developmental, and glaucoma-associated genes comparatively expressed in fetal relative to adult ocular tissues.

Fig. 1.

Heat map of 31 extracellular matrix, ocular development, and glaucoma-associated genes comparatively expressed in fetal relative to adult ocular tissues.

3.2. Real-time quantitative PCR

Real time quantitative PCR was performed on 6 genes for cornea, 7 genes for optic nerve, and three genes for retina/retinal pigment epithelium, choroid, and sclera. All demonstrated verification of the microarray results with respect to up- or down-regulation (Fig. 2).

Fig. 2.

Quantitative real-time polymerase chain reaction validation results of select genes.

3.3. Pathway analyses

IPA functional and canonical pathway analysis was performed separately for each tissue comparison. In all tissue comparisons there was overlap in the most significant functional categories identified (Table 4). Conversely, the most significant canonical pathways identified were more variable between tissue types (Table 5).

Table 4. Ten Most Significant Functional Assignments.

Ten most significant functional groups for adult versus fetal differentially expressed genes.

Tissue	Functions annotation	P-value	# genes
Retina/RPE	Differentiation	7.15E-16	156
	Tumorigenesis	2.48E-12	260
	Proliferation of cells	8.05E-12	182
	Cell movement	8.36E-11	133
	Tissue development	9.28E-11	182
	Function of blood cells	4.33E-10	57
	Immune response	1.14E-09	116
	Cell death of blood cells	3.18E-09	60
	Development of lymphocytes	4.35E-09	60
	Quantity of cells	1.06E-08	118
Choroid	Tumorigenesis	3.19E-28	1317
	Cell death	1.37E-23	1058
	Cell cycle progression	3.81E-20	376
	Proliferation of cells	1.07E-16	835
	Growth of cells	1.22E-14	588
	Expression of RNA	2.95E-14	658
	Transcription	1.63E-13	610
	Differentiation	2.10E-13	617
	Transcription of RNA	2.37E-13	599
	Organization of cytoplasm	6.20E-13	329
Sclera	Tumorigenesis	6.17E-18	365
	Cell division of chromosomes	3.73E-12	43
	Cell cycle progression	1.14E-10	110
	Tissue development	1.35E-10	237
	Proliferation of cells	2.13E-10	231
	Differentiation	4.42E-10	179
	Cell death	2.39E-09	272
	Cell movement	6.59E-09	164
	Interphase	1.30E-08	73
	Morphology of organ	8.54E-08	132
Optic Nerve	Tumorigenesis	5.22E-10	548
	Tissue development	1.31E-08	371
	Expression of RNA	8.54E-08	288
	Proliferation of cells	9.32E-08	357
	Microtubule dynamics	8.38E-07	100
	Cell cycle progression	9.49E-07	152
	Transcription	9.68E-07	263
	Morphology of basement membrane	1.24E-06	9
	Immortalization	4.28E-06	17
	Morphology of blood vessel	1.18E-05	44
Cornea	Tumorigenesis	9.15E-28	1060
	Tissue development	6.71E-19	702
	Cell death	7.63E-19	831
	Proliferation of cells	1.51E-18	684
	Growth of cells	1.05E-17	492
	Encephalopathy	3.18E-17	363
	Cell movement	1.96E-16	483
	Invasion of cells	2.00E-16	199
	Movement disorder	2.52E-14	289
	Organization of cytoplasm	3.77E-14	275

Table 5. Top Five Canonical Pathways.

Significance (p-value) of pathways is determined by Fischer's exact test of probability of genes in the dataset fit into the pathway by chance alone and by proportion (ratio) of genes in pathway that are included in the dataset (# genes).

Tissue	Ingenuity canonical pathways	–log(p)	Ratio	# genes
Retina/RPE	Atherosclerosis signaling	3.99	1.09E-01	14
	LXR/RXR activation	3.25	9.56E-02	13
	p38 MAPK signaling	3.25	1.13E-01	12
	MIF regulation of innate immunity	3.15	1.40E-01	7
	Hepatic fibrosis/hepatic stellate cell activation	3.14	9.52E-02	14
Choroid	Axonal guidance signaling	7.03	2.98E-01	128
	Molecular mechanisms of cancer	5.55	2.88E-01	109
	CXCR4 signaling	5.33	3.39E-01	57
	Ephrin receptor signaling	5.09	3.12E-01	62
	EIF2 signaling	4.85	3.22E-01	66
Sclera	ATM signaling	5.31	2.20E-01	13
	Role of CHK proteins in cell cycle Checkpoint control	3.68	2.29E-01	8
	Atherosclerosis signaling	3.53	1.24E-01	16
	Mitotic roles of polo-like kinase	2.37	1.38E-01	9
	Cell cycle: G2/M DNA damage Checkpoint regulation	2.17	1.43E-01	7
Optic Nerve	Wnt/β-catenin signaling	5.61	2.03E-01	35
	Hereditary breast cancer signaling	4.99	2.05E-01	26
	Human embryonic stem cell pluripotency	4.58	1.80E-01	27
	Cell cycle: G1/S checkpoint regulation	3.64	2.31E-01	15
	RAR activation	3.47	1.60E-01	30
Cornea	Breast cancer regulation by Stathmin1	6.67	2.93E-01	61
	Mitochondrial dysfunction	6.35	2.64E-01	46
	Wnt/β-catenin signaling	6.16	3.14E-01	54
	Axonal guidance signaling	5.89	2.37E-01	102
	Molecular mechanisms of cancer	5.79	2.41E-01	91

3.3.1. Functional annotation enrichment

To identify overrepresentation of genes within known functional pathways involved in growth and development, we performed IPA functional annotation analyses. Probes including their associated gene, p-value, and fold change were analyzed separately for each comparison. The significance of each functional group is given as a p-value. The ten most significant functional assignments for each tissue type by p-value are shown in Table 4. For presentation purposes, functional assignments containing redundant genes and functions were consolidated. Full data is available in Supplementary Table S1.

In all comparisons of ocular tissues, the most significant functional assignments included development, cell growth/death and cancer categories. Tumorigenesis was one of the two most significantly associated functional assignments in all tissues. Other common categories included cell cycle, movement, function, morphology and expression. These significant functional assignments showed that genes involved in basic cellular regulations were differentially activated in rapidly growing and developing tissues relative to their adult counterparts.

In the retina/RPE, the most significant functional assignments also included differentiation, cell proliferation, cell movement and tissue development. In the choroid, cell death, proliferation and growth were among the most significantly associated functional annotations. In the sclera, the most significant functional groups included cell division and progression, as well as tissue development. In the optic nerve, tissue development, RNA expression, cell proliferation and microtubule dynamics were some of the most significantly enriched functions. The lower significance for the optic nerve in the array data was reflected in its lower significance for functional classifications. Despite large variances in the number and significance of genes between tissue type comparisons, the most significant functional assignments across all adult versus fetal comparisons, including the optic nerve, were comparable (Table 4). Lastly in the cornea, the most significantly associated functional annotations included tissue development, cell death, proliferation and growth.

Supplementary Table S2 contains the significant functional assignments with predicted activation changes for each adult versus fetal tissue comparison (absolute value [z-score] > 2.0). Adult versus fetal compared tissues still shared major functional categories in common, but with some differences from those most significantly present. When taking direction of fold changes and their predicted effects on a functional assignment, all tissues had predicted activation changes for cancer and cell movement categories. Many tissue types also had development and cell growth/death categories in common. Even within the same tissue type these shared categories were not consistently predicted to have the same directional activation; however, those shared categories had differing functional assignments and contained distinct genes. Functionally assigned groups containing redundant genes which were removed from the data for publication did not contradict one another in the direction of predicted activation. In the adult retina/RPE, choroid and sclera, functions predicted to have increased activation included cell proliferation, while those predicted to have decreased activation included lymphocyte movement and/or proliferation. In the adult optic nerve, predicted increased functions included cell growth, formation, and organization. Lastly in the adult cornea, predicted functional increases included development, differentiation, and cell cycle progression, while decreased included cell invasion and survival.

3.3.2. Canonical pathway enrichment

To identify specific pathways involved in the rapid growth and development seen in prenatal development we performed canonical pathway analyses using IPA. Probes including their associated gene, p-value, and fold change were analyzed simultaneously for each comparison to identify overrepresentation of genes within known canonical pathways. The significance of each pathway was assigned a p-value. The five most significant canonical pathways by p-value are shown in Table 5. Additionally, IPA calculates a ratio of genes differentially expressed within each canonical pathway. Full data is available in Supplementary Table S3.

There was some overlap in pathways between tissue types and all tissues included signaling pathways among the most significantly associated (Table 5). The retina/RPE and sclera both included atherosclerosis signaling among pathways with the highest significance. The most significant pathways in the choroid and cornea both included axonal guidance signaling. In the sclera and optic nerve tissues both included significant pathways involving cell cycle regulations. Other significant pathways in retina/RPE involved expression, regulation and immune responses of the cell. In the choroid, significant pathways also included growth, expression and immune responses. In the sclera, many of the most significantly associated pathways involved cell cycle regulation. In the optic nerve, the most significant pathways were cell fate, growth, regulation and expression. For the cornea, the pathways with the highest significance were cell fate and growth.

3.4. Disease gene overlap

The most differentially expressed genes in each of the tissues included genes previously implicated in ocular diseases such as myopia, glaucoma and syndromes with an ocular component (Table 3). Examples from the 40 most differentially expressed probes in each direction for each tissue include: MYOC (MIM 601652), which has been implicated in the development of myopia (Tang et al., 2007; Vatavuk et al., 2009), had probes with 6.40 and 2.22 fold higher expression in the adult relative to fetal scleral tissue and 74.2 fold higher expression in the adult corneal tissue. LAMA1 (MIM 150320), which has also previously been implicated in myopic development (Zhao et al., 2011), had 30.80 lower fold expression in the adult (relative to fetal) choroid. NTM (MIM 607938), which has been implicated in central corneal thickness in primary open angle glaucoma (Ulmer et al., 2012), had −131.38 lower fold expression in adult corneal tissue relative to the fetal tissue. PTPN22 (MIM 600716), implicated in Behçet disease (Baranathan et al., 2007; Sahin et al., 2007), has 46.10 higher fold expression in the adult cornea. Also, all tissues tested included collagens in the most differentially expressed genes (Table 3), many of which have been implicated in ocular diseases.

Additionally, a number of genes previously identified as involved with ocular disease belonged to the most significantly enriched functional annotations for all tissues tested. In particular, there were a large number of genes implicated in either non-syndromic or syndromic high myopia in various degrees, differentially expressed in these tissues and present in the most significantly enriched functional annotations. Fourteen differentially expressed genes associated with non-syndromic myopia were present in the ten most significant functional classifications for the retina/RPE, choroid and/or sclera (Table S4). Many of these myopia-associated genes were found in the same functional annotations with concentrations in those groups related to cell growth and proliferation (including tumorigenesis).

4. Discussion

To our knowledge, this is the first whole genome expression study comparing human adult and fetal retina/RPE, choroid, sclera, optic nerve and cornea. This study investigated gene expression patterns during ocular growth and development in humans. We assessed differentially expressed genes between rapidly growing fetal tissues relative to their mature adult counterparts. Then, we examined the functions and canonical pathways of those differentially expressed genes to extrapolate information regarding the mechanisms of prenatal ocular growth and development. This study identified genes, biological functions and canonical pathways that were most active or inactive during normal growth and development in human ocular tissues. In addition to not having been investigated in this way before, some of the tissues investigated in this study also are involved in clinical features of highly myopic individuals. Investigation of normal growth and development mechanisms in these human tissues may also help to interpret biological mechanisms of ocular disease by providing a reference framework.

4.1. Microarray analyses

Whole genome expression analysis comparing human fetal and adult ocular tissues revealed over 1000 genes that were FDR significantly differentially expressed in each of the tissues tested, except the optic nerve. Tissues-specific gene expression patterns that are likely attributable to their distinct functions were identified for growth and development, such as rhodopsin up-regulation in the adult retina/RPE or hemoglobin down-regulation in the adult vascular choroid. Despite having different biological functions, the ocular tissues examined in this study had overlap in their most differentially expressed genes (Table 3). Some of the overlap shared by all or most tissue types, such as up-regulation of methyltransferases in adult tissues, may be attributed to the fundamental similarities in development between tissues. Other overlaps, such as extracellular matrix genes in the fibrous tissues, may reflect structural similarities between specific tissues. In all tissues examined – especially the fibrous sclera, optic nerve and cornea – there was a general trend of down-regulation of collagen genes in adult tissues, which is consistent with previous findings in mouse postnatal versus adult sclera (Zhou et al., 2006).

4.2. Pathway analyses

All tissues examined in this study had overlap in biological functions the differentially expressed genes were most significantly associated with. While tumorigenesis was one of the most significantly enriched biological functions in all tissues, this finding does not indicate abnormal growth in the tissues samples under investigation. Rather, we believe it suggests that there was an enrichment of genes that regulate normal cell proliferation and/or death mechanisms (also shared by all tissues) that are frequently targeted in the development of cancer. There was considerable variation between tissue types in the most significantly enriched canonical pathways in the sets of differentially expressed genes (Table 5). Yet despite these differences, there was broad consistency in the presence of signaling pathways among the tissues. As with the enrichment for tumorigenesis functional genes, these significantly enriched canonical pathways included disease pathways. The genes involved in these diseases (as reported in the literature) were likely enriched in the data sets not because these tissues are diseased, but because these genes may have normal functional roles in these tissues.

In the retina/RPE, the enrichment of biological functions relating to blood cells and the immune system may be due to differences in the proportion of foveal vs. non-foveal tissues and consequently differences in vasculature in the collected tissue, or indicative of vascular formation in the fetal tissue. A deeper analysis of the direction of expression of these genes and/or differential expression analyses of foveal versus non-foveal retina/RPE may clarify this finding. Similarly, the most significant canonical pathways in the retina/RPE broadly related to vasculature. Again, further study would be necessary to determine the cause of these findings. Interestingly, axonal guidance signaling was the most significantly enriched pathway in the choroid. It is possible that known signaling pathways such as this may also function as the inter-tissue signaling pathways that locally regulate ocular growth in postnatal emmetropization. In the sclera and optic nerve, the enrichment of canonical pathways involving the regulation of cell growth may be because these fibrous tissues undergo substantial growth and remodeling as they develop. Finally, the cornea had the most variation within a tissue in the most significant canonical pathways; though similarly to the other fibrous tissues examined, three of the top five involved the regulation of cell growth.

It is not surprising that all tissues shared functional categories involving cell growth, death and development given that they were in a state of rapid growth and development. These significantly enriched functional categories were identified from IPA's network derived from scientific findings, suggesting that many of the prenatal growth and development mechanisms in the eye are similar to other mechanisms found in other tissues and systems. Likewise, these same genes and pathways involved in normal growth may contribute to abnormal growth when dysfunctional. This abnormal growth could arise in the form of a tumor when restricted to a specific tissue or cell type, or could accompany corresponding changes in nearby tissues. Our findings suggest that the prenatal growth and development seen in these ocular tissues is largely influenced by standard mechanisms and this data provides a useful framework to study candidates. However, we are still unsure how these tissues are communicating with one another and how prenatal and postnatal ocular growth signals may differ. While, these inter-tissue signaling molecules may not be among the most significant functional categories if they are indeed specific to the eye and not well established in the literature; other molecules in the mechanistic pathway could still be identified as a starting point.

4.3. Disease gene overlap

In all tissues, a number of genes previously implicated in ocular disease were identified among the most differentially expressed genes and the most significantly associated biological functions. It is not unreasonable to think that the disruption of these genes in their normal functions within these tissues may contribute to or cause disease. For example, as a gene that regulates cellular growth may be mutated to allow unrestricted cancerous growth, a mutation in a gene regulating local control of eye growth may result in failure of the emmetropization process leading to refractive error. While these differentially expressed genes presented were not identified in diseased versus normal tissue, our findings suggest that many disease genes also have roles in normal prenatal growth and development of the eye. This information on the normal roles of these genes in the tissues with phenotypic changes in diseases eyes may help to elucidate their mechanistic roles in disease development.

4.4. Consideration of limitations

Since the fetal and adult eyes were obtained from different sources, expression differences could in part be attributed to different processing methods and time frames for procurement.

None of the genes in the optic nerve met FDR statistical significance-However, other qualitative metrics such as probe concordance and pathway assessment were comparable to the other tissues examined in this study, lending some credibility to their validity. It is unlikely that there are no truly differentially expressed genes during growth and development in the optic nerve given the physical and developmental changes the tissue undergoes, the fold changes seen in expression, and the overlap with other tissues in genes and enriched functional annotations of the set of genes. The optic nerve may have failed to reach statistical significance for differential expression of any of its genes for several reasons, such as its small sample size (due to sample drop out) or biological variation (due to variation in the proportion of optic nerve stem to head in biological replicates). It is more likely that the high variability between samples was the major contributor, as we did note large expression changes (Table 2).

There were also potential confounding factors in the assessment of differentially expressed genes during growth and development in the retina/RPE in this study. Combining the retina and RPE may have reduced the power to detect expression differences, which may account for the relatively low number of differentially expressed genes and their corresponding fold changes. Another potential caveat is the possible differences in contribution of foveal tissues to the posterior retina/RPE in the fetal versus adult tissues. Given that the same size biopsy punch was used in both age groups, it may be assumed that the fetal tissue included a higher proportion of non-foveal retina/RPE than the adult tissue.

4.5. Conclusions

We present the first whole genome expression analysis comparing human adult and fetal retina/RPE, choroid, sclera, optic nerve and cornea. This data provides a wealth of information regarding the genes, biological functions and canonical pathways that are up- or down-regulated during growth and development. A deeper understanding of the expression patterns in the developing human eye, as gained by this information, was a missed early step in understanding how genetic defects or damage may manifest as diseases. This information is not only beneficial for congenital diseases arising from developmental failures, but may also be used to study other diseases resulting from failures to the normal visual system such as myopia. This data set provides a network with which to study the normal expression, function and pathway interactions of genes in ocular tissues that are relevant to disease development. This information could help generate testable hypotheses for the mechanistic roles of these genes in disease development. Although not emphasized in this manuscript, there were limited microRNAs (approximately 200) that were incidentally expressed in various ocular tissues. A future study of a purposeful microRNA microarray or RNASeq study of ocular tissues would be useful for subsequent microRNA target identification and functional validation of predicted targets or to study the regulatory effects of miRNAs on target genes. We propose that the information in this data set may be used to prioritize candidate genes by providing insights into how they might function in disease development as well as confirming their presence in the relevant tissues. However, further genetic and/or functional validation of candidate genes would be necessary to make specific claims regarding their involvement in the development of myopia.

This expression information may be particularly useful when other genetic means, such as association mapping or sequencing, fail to reduce the number of candidate genes to a cost and resource effective number. The expression data presented in this study for the retina/RPE, choroid and sclera was also used to prioritize candidate genes from a genetic association mapping study for the high myopia locus MYP3 (MIM 603221) (Hawthorne et al., 2013). While expression data alone is not sufficient to determine candidates for ocular disease, it may be used interpret and prioritize candidates when validation of a large number of genes in not feasible or practical, as was the case with the MYP3 locus. The candidates identified in our MYP3 association mapping study were found to be nominally significantly associated in an independent cohort and one of the candidates, PTPRR (MIM 602853), was also recently identified in a large scale meta-analysis of myopia GWA study (Verhoeven et al., 2013). This replication of a candidate gene for myopic development supports its validity and also provides preliminary support of the utilization of this expression data for disease candidate prioritization.

4.6. Future directions

While these hypotheses require further testing, we have provided a novel framework (normal ocular growth and development) in which to interpret and contextualize candidate genes for ocular disease. Early analyses of the functional classifications of these genes may help elucidate their potential roles in disease progression, particularly their tissue-specific roles. Next, a deeper understanding of the mechanisms connecting these tissues may be better studied using the enriched signaling pathways identified. Genes whose mutations may confer an increased risk for ocular disease via known or undetermined pathways may be components of these established pathways regardless of their differential expression in the tissues. This data could be used to prioritize candidate genes within previously identified disease loci containing large numbers of genes. Supplemental Table S5 contains a list of genes within MIM recognized myopia loci (a focus disease in our lab) which are differentially expressed in the retina/RPE, choroid, and/or sclera. We have only scratched the surface of the possibilities with this vast resource by interpreting and prioritizing candidate genes for one disease study of myopia. The use of this information for other loci and common ocular diseases associated with these tissues should be explored further.