Transcriptome Analysis of Newt Lens Regeneration Reveals Distinct Gradients in Gene Expression Patterns

Regeneration of the lens in newts is quite a unique process. The lens is removed in its entirety and regeneration ensues from the pigment epithelial cells of the dorsal iris via transdifferentiation. The same type of cells from the ventral iris are not capable of regenerating a lens. It is, thus, expected that differences between dorsal and ventral iris during the process of regeneration might provide important clues pertaining to the mechanism of regeneration. In this paper, we employed next generation RNA-seq to determine gene expression patterns during lens regeneration in Notophthalmus viridescens. The expression of more than 38,000 transcripts was compared between dorsal and ventral iris. Although very few genes were found to be dorsal- or ventral-specific, certain groups of genes were up-regulated specifically in the dorsal iris. These genes are involved in cell cycle, gene regulation, cytoskeleton and immune response. In addition, the expression of six highly regulated genes, TBX5, FGF10, UNC5B, VAX2, NR2F5, and NTN1, was verified using qRT-PCR. These graded gene expression patterns provide insight into the mechanism of lens regeneration, the markers that are specific to dorsal or ventral iris, and layout a map for future studies in the field.


Introduction
Amphibians, especially newts, possess regenerative capabilities that are missing in higher vertebrates. Newts can regenerate their limbs, brain, heart, tail with spinal cord and other tissues. Interestingly, newts can also regenerate the lens after its complete removal (lentectomy). This system provides many advantages for regenerative studies because the whole organ (lens) is being removed. Lens regeneration occurs from the iris by a process that involves the transdifferentiation of pigmented epithelial cells (PECs) to lens cells. Another interesting aspect of this process is that lens regeneration occurs only from the dorsal and never from the ventral iris. This allows the use of the ventral iris as a natural non-regenerative control in lens regeneration experiments [1,2,3].
Two major hallmarks of lens regeneration are the re-entry of the cell cycle 4 days post-lentectomy (dpl) and the formation of a dedifferentiated vesicle 8 dpl [4]. Interestingly, both dorsal and ventral iris cells re-enter the cell cycle [5]. In the past, limited expression studies, either using individual gene probes or smallscale microarray analysis have indicated that dorsal and ventral irises show no major differences in gene expression. In other words, most of the examined genes were expressed in both irises. Thus, to date no clear expression pattern has emerged to account for the ability of the dorsal iris to be the source of the regenerating lens. More recently a microarray analysis during early stages of lens regeneration was performed. In that study expression in 1, 3 and 5 dpl from the dorsal or the ventral iris was compared with the corresponding intact iris (0 day). While that study indicated regulation in genes related to DNA repair, extracellular matrix and redox homeostasis, direct comparisons between gene expression in dorsal and ventral iris could not be assessed. Thus, a direct comparison of transcriptomes was needed to delineate global gene expression differences between dorsal and ventral iris during lens regeneration in newts [6,7].
Next-generation high-throughput techniques allow transcriptome analysis based on de novo assemblies, making them extremely useful for non-model systems like the newt. Here, we investigate transcriptional changes during newt lens regeneration in an attempt to identify patterns that provide clues for the ability of dorsal iris but not the ventral to transdifferentiate. We focused on 4 dpl and 8 dpl for both dorsal and ventral iris as these time points are crucial stages for lens regeneration because these time points encompass the events of cell cycle re-entry and dedifferentiation. For RNA-seq we used a de novo assembled trancriptome making use of short Illumina and longer 454 and Sanger reads [8].
Here we report and for the first time the expression of more than 38,000 annotated transcripts in dorsal and ventral iris during lens regeneration. The analysis has been focused on the quantitative and qualitative differences between dorsal and ventral iris of those transcripts. We found very few genes to be dorsal or ventral iris-specific. However, certain cohorts of genes grouped according to their function were found to be preferentially upregulated in the dorsal iris. Genes involved in the cell cycle, transcriptional apparatus, cytoskeleton and immune response are among those with much higher expression in the dorsal than the ventral iris. This graded expression might provide robust regulation that allows the dorsal iris to ''win'' over the ventral iris.

Animals -Lentectomy
Handle and operations on Notophthalmus viridescens have been described previously [6]. Briefly, newts were purchased from Charles Sullivan Inc. Newt Farm. Newts were anesthetized in 0.1%(w/v) ethyl-3-aminobenzoate methanesulfonic acid (MS222; Sigma) in phosphate buffered saline. Lentectomy was performed using a scalpel to incise the cornea and tweezers to pull out the lens through the incision. For the present study newts were kept for 4 and 8 dpl before tissue harvest.

Ethics Statement
All procedures involving animals were approved by the University of Dayton Institutional Animal Care and Use Committee (IACUC; Protocol ID: 011-02). All surgical procedures were performed in anesthetized with MS222 newts. All appropriate procedures were used in order to alleviate pain and distress while working with newts.
Tissue Harvest and RNA Extraction for qRT-PCR 4 or 8 dpl newts were anesthetized in MS222. Whole eye balls were removed and placed in dishes filled with RNAlaterH Solution (Applied Biosciences). Using fine scissors and tweezers, eye balls were dissected first by separating the anterior from the posterior part ( Figure 1A) and then by removing remaining neural retina and the ciliary body from the anterior part ( Figure 1C). Dorsal or ventral iris sectors were collected in approximately 135u of the whole iris leaving out a board area between dorsal and ventral side which has a black-colored pigmentation ( Figure 1B). Dorsal and ventral sectors were then collected in microcentrifuge tubes filled with RNAlaterH Solution. The tubes were briefly centrifuged and RNAlaterH Solution was completely removed. RNA extraction was performed following TRIzolH Reagent protocol (Applied Biosciences) for 500 ml of reagent or the aqueous phase was transferred to RNA Clean & Concentrator TM (Zymo Research) columns and the recommended protocol was followed. Quality of isolated RNA was determined by Nanodrop 2000 spectrophotometer (Thermo Scientific). Good quality samples had A260/ A280 ratio greater than 2 and a peak at 260 nm.

Reverse Transcription Reaction (RT)
200 ng total RNA was used for RT reactions. First-strand cDNA synthesis kit (GE healthcare) was used following the recommended protocol for oligo(dt) primers. Half the volumes were used for negative RT reaction without using oligo(dt) primers and the samples were incubated for 5 min at 98uC for enzyme inactivation. All the samples had a clear band after Polymerase Chain Reaction (PCR) with RPL27 gene (housekeeping gene). Dorsal samples needed to be positive for TBX5 and negative for VAX2. Ventral samples needed to be positive for VAX2 and negative for TBX5. qRT-PCR was performed using iQ TM SYBRH Green Supermix (Bio-Rad) and Bio-Rad iCycler (Bio-Rad) following company's protocol for 25 ml. Primer specificity was determined using melt curve analysis. An extra cycle of 6 sec was added to genes that were showing detectable signal from primer dimers and the temperature was determined by the melt curve (usually 2-4uC lower than the melting temperature; see methods below). Amplification cycles (Ct) of samples were compared to Ct of a standard curve created by the cDNA of the gene used. Gene expressions were then normalized to the expression of the reference gene (RPL27).

Primers, PCR and qRT-PCR Settings
For the present study the following primers were used (written from 59

Statistical Analysis for qRT-PCR Results
Statistical analysis was performed using two-way analysis of variance (ANOVA) and Student's t-test for independed samples. Samples were run in triplicates (n = 3). Statistical significance was determined with 95% confidence (p,0.05). Equal variances for student's t-test were assumed when Levene's test p value was greater than 0.05. The newt transcriptome [8] was annotated with the BLAS-T2GO tool [9] using the nr database. We used a cutoff of (e210) for sequence assignments, collected annotations and corresponding GO terms [10]. Transcripts were selected depending on their expression and location in the iris as described in Figure 2 for comparing dorsal iris versus ventral iris groups and as described in Figure 4 for comparing day 4 and day 8 groups. Fisher's exact test corrected for multiple selections (feature available in BLAST2GO tool) was used for the different groups and statistically significant enriched GO terms were identified (FDR ,0.05). Transcripts assigned to enriched terms were selected. Human homologues of those transcripts were found using the BLAST tool [11].

RNA Expression Calculation
Illumina reads were mapped to the newt transcriptome using BWA [12]. Reads per kilobase per million mapped reads (RPKM) values were calculated for each transcript having at least one unique mapping read. Since intron/exon data are missing for the newt, we used a recently published microarray experiment for RPKM cutoff estimation [6]. We selected transcripts presented on the microarray having a valid spot structure (value for circularity .80%, intensity variation within the spot very small, flagged as valid) but lacking a significant signal during microarray analysis (signal to noise ratio (snr) ,1, significant spots are used that have a snr ratio .3). We detected 101 spots to be valid for these parameters. Mapping of array coordinates to the transcriptome resulted in 81 non redundant individual transcripts. We assumed these candidates to be a good estimate for RPKM cutoff selection and calculated the average RPKM value for this transcript set for each investigated timepoint. We received a RPKM value of 0.64 for 4dv, 1.14 for 4dd, 1.13 for 8dv and 1.1 for 8dd. Illumina sequencing raw reads can be found in the NCBI sequence read archive under the accession: ERP001353 [8]. Assembled transcripts and annotation are located in the newtomics database [8,13].

Transcriptome and Analysis Overview
The recently assembled newt transcriptome contains nearly 38,000 annotated genes [8]. After Illumina sequencing, the reads from the dorsal and ventral iris 4 and 8 dpl were mapped to the newt transcriptome and RPKM values were calculated. Since the newt genome is not yet available and reads that map to noncoding areas can not been found, we calculated the RPKM cutoff using micorarrays that were previously used during lens regeneration (see methods for more details). The cutoff in the RPKM values were 0.64 for ventral iris 4 dpl, 1.14 for dorsal iris 4 dpl, 1.13 for ventral iris 8 dpl and 1.1 for dorsal iris 8 dpl. Transcripts Table 1. List of up-regulated (.2 times) transcripts related to cell cycle*.

Function
Dorsal Ventral  Table 2. List of up-regulated (.2 times) transcripts related to gene regulation*.

Function Dorsal Ventral
HTT + found to have RPKM more than the cutoff were considered to be expressed and were taken into account for the comparisons below.

Dorsal Iris Shows Enrichment of Up-regulated Genes Involved in Cell Cycle, Cytoskeleton, Gene Expression and Immune Response
Since the major purpose of this study is to identify patterns of gene expression in the dorsal and ventral iris that might correlate with the ability of the dorsal iris for transdifferentiation, we first looked into transcripts that were regulated either in the dorsal or in the ventral iris at both collection points (4 dpl and 8 dpl). Analysis of the results was performed as outlined in Figure 2.
Interestingly, we only saw a handful of genes that are either exclusively expressed (no reads were mapped to them) in the dorsal or the ventral iris. This finding was not surprising because in the past our laboratory had seen, using limited expression data, that  Table 3. List of up-regulated (.2 times) transcripts related to cytoskeleton*. the ventral and dorsal irises show similar patterns of gene expression (see below). Next, we investigated potential patterns of up-regulated genes. Specifically we asked which genes are consistently up-regulated in the dorsal or the ventral iris (both at 4 and 8 dpl). Fisher's exact test with multiple testing correction for Gene Ontology (GO) from transcripts that are up-regulated at least 2 fold in dorsal iris 4 and 8 dpl compared to the ventral iris, and transcripts that are upregulated at least 2 fold in ventral iris 4 and 8 dpl compared to the dorsal iris are shown in Table S1. Interestingly, we found that more GO terms are enriched in the dorsal samples than in the ventral. In particular, GO terms related to cell cycle, regulation of gene expression, cytoskeleton and immune response were overrepresented. In contrast, ventral samples, which generally generated fewer enriched GO terms, primarily showed GO terms related to transposons like RNA-directed DNA polymerase activity, RNA-directed DNA replication and DNA integration.
Although we did not find major qualitative differences in the expression patterns between dorsal and ventral irises, a clear-cut difference in the expression level of several genes were evident: GO terms related to cell cycle (indicating proliferation), cytoskeleton (indicating cell migration and morphological changes), transport of molecules and cytokinesis (indicating altered gene regulation), and immune response (indicating responsiveness to injury) were enriched at the site where lens regeneration commences. Strikingly, in these gene categories 609 genes were up-regulated in the dorsal iris and only 66 in the ventral iris, which will provide much needed insights into the mechanisms that allow the dorsal to ''win'' over the ventral iris.

Cell Cycle-related Transcripts
Most of the factors related to cell cycle were up-regulated in the dorsal iris when compared to ventral iris (124/7) ( Table 1). In detail: Mitosis: Table 1 shows factors related to all steps of mitosis including spindle formation, microtubule-associated, chromosomal movement and mitosis progression, which are up-regulated in dorsal samples. Interestingly, proteins that act as complexes are concomitantly up-regulated including NDC80 complex, cohesin complex and chromosomal passenger complex (CPC). NDC80 complex is essential for chromosome segregation and spindle formation [14], and NDC80, SPC24, SPC25 and NUF2 are upregulated. Cohesin complex is required for the packing of the chromosomes [15], and STAG1, RAD21, SMC3, SMC4, PLK1 and SGOL1 are up-regulated. CPC is related to centromere functions during mitosis [16], and AURKB and CDCA8 are upregulated.
Tumor suppression, proliferation-related, p53/TP53-associated and RB1-associated proteins are up-regulated in dorsal iris. These proteins promote or repress proliferation and cell cycle. It has been previously shown that newt muscle cells re-enter the cell cycle after inactivation of RB. After entering the S phase these cells were resting in G2 phase. So, it was hypothesized that factors should be expressed, which promote the G2/M checkpoint after phosphorylation of RB [31]. Our data suggests that transcripts that regulate the G2/M transition and thereby proliferation include CDC25A, CDK1, CCNB1, the APC/C complex and PLK1.
Factors playing a role in DNA repair need to be seen in the context of cell proliferation and cell cycle, since the failure to repair DNA damage will prevent proliferation. In previous studies using qRT-PCR, it was shown that rad1 is up-regulated at 3 and 5 dpl in the dorsal compared to ventral iris showing activation of robust DNA repair mechanism to prevent accumulation of mutations in dividing cells [6]. A similar pattern emerged in the current RNA-seq analysis: many factors that play a role in DNA repair are at least 2 fold up-regulated in the dorsal compared to the ventral iris. Among them RAD1 along with CHEK1, a kinase that has a key role in cell cycle arrest and apoptosis decisions after DNA damage [32].
Likewise, most of the transcripts that were identified to act on translation (and miRNA processing and function) are up-regulated in the dorsal iris. Only 1 transcript, eIF2B, was found to be upregulated in the ventral iris. Table 3 shows transcripts that are up-regulated in dorsal iris at least 2 fold than the ventral iris and the opposite. Most of the transcripts are shown to be up-regulated in the dorsal iris (134/18).
Actin-related: Transcripts up-regulated in the dorsal group and included in this category play roles in actin polymerization and organization in order to support cell shape, movement and cell adhesion with the extracellular matrix, and the linkage of actin with other proteins in order to facilitate transport or contraction. WASH1 [116], ARPC1B, ARPC5L [117], SSH2 [118] SCIN [119], IQGAP1 [120], PPP1R9A [121] and DIAPH1 [122] are playing roles in actin polymerization and organization. CNN2, MYO10, MYO1E, MYO9A, MYO1G, MYO6, MYO9B, MYO7A, MYO1D and ROCK1 are involved in contraction [123,124]. CTNNA1 and MTSS1 are related to cell adhesion with the extracellular matrix and cell-cell contact [125,126]. KLHL3 and MYO5A are playing roles in molecular transport [127,128]. PLEK2 and VASP are related to cell movement [129,130] and SYNE1 and WDR1 are linking cytoskeleton with other proteins [131,132].
In addition to factors that are actin or tubulin-related, Table 3 shows other proteins that are related to cell adhesion, movement and linkage of plasma membrane with cytoskeleton that are upregulated in the dorsal iris. For lens regeneration, cell adhesion and locomotion is very important since PECs need to change their environmental behavior to transdifferentiate and change their cell fate. Previous studies have found that extracellular matrix is being remodeled and matrix metalloproteinases are up-regulated already 1 day post-lentectomy to prepare the environment for the onset of lens regeneration [6]. Our data clearly show changes in the molecules that determine the interaction of PECs with the environment and remodeling of cytoskeleton components and networks of PECs. Another interesting aspect is that many factors involved in tumor metastasis are up-regulated in the dorsal iris which indicates a role of these molecules at the onset of cell locomotion.
The role of immune response and its involvement in the initiation of regeneration has been extensively investigated in the past [141,142]. It has been hypothesized that molecules involved in the regulation of the immune response have a novel role in regeneration or that the immune response itself is crucial for regeneration. The issue has not been settled yet. Nevertheless, complement components seem to be important for liver regeneration [143] and have also found to be expressed in limb and lens regeneration [144,145]. The present results provide strong evidence of a crucial role of injury response in regeneration, which needs to be investigated further.

Transposon-related Transcripts
Interestingly, transposons are the only transcripts that are enriched in ventral compared to dorsal samples (Table S2). Transposons have many types and they do not have an assigned biological function. They can be transcribed, reverse-transcribed and integrated back to the genome (retrotransposons) or not. So far, we have no specific role of transposons in regeneration (or rather, inhibition of it?). Since transposon-related transcripts are enriched in ventral samples it would be interesting to learn more about a specific role in repressing specific programs.

Highly Regulated Transcripts
Summing up, we have identified patterns of gene expression that are predominant in the dorsal iris. Genes that are involved in cell cycle, gene regulation, cytoskeleton and immune response show a graded expression along the dorsal/ventral iris. Thus, our study is the first to show how the dorsal iris differs from the ventral iris, and how specific patterns of gene expression correlates with the dorsal iris' regenerative ability. Comparisons comprised two critical time points, 4 and 8 dpl. Interestingly, comparison of gene expression patterns at each time point separately, recapitulates our primary finding that most transcripts of these gene categories are up-regulated at 4 dpl in the dorsal iris in comparison to 4 dpl in the ventral iris, or in 8 dpl in the dorsal iris in relation to 8 dpl in the ventral iris. In Tables 5 and 6 we show a selected group of genes to exemplify this point. The tables also allow a view on the top regulated transcripts. It becomes clear that a few of them are either dorsal-specific or ventral-specific. Only 3 transcripts were found to be exclusively present in the dorsal iris. These transcripts correspond to protein-1 like (ras associated and pleckstrin domains-containing), transmembrane protein 185A-like (TMEM family) and to chromatin assembly factor 1 (CAF1). Other transcripts that show very high expression in dorsal iris were TBX5, TMEM185A, E3 ubiquitin-protein ligase HERC2-like (HERC2) (.32 times), TMEM116, ephrin-B2, and netrin receptor (UNC5B) (.16 times). In the ventral iris, except transposons, we find that netrin-1 (NTN1), nuclear receptor 2F5-like (NR2F5), and VAX2 are expressed 32 times higher than in the dorsal iris. The function of TBX5 and VAX2 were discussed above, but it is interesting to note here that they might provide a dorsal or ventral identity to the adult iris. Currently, it is not known to what extent these genes control regeneration but functional assays will settle this issue. Nevertheless, TBX5 and VAX2 can be used as markers for dorsal and ventral iris, respectively. Interestingly, we find NTN1 in the ventral iris but its receptor (UNC5B) in the dorsal iris. Likewise, we find ephrin-B2 in the dorsal and its receptor in the ventral iris, which might reveal a so-far unsuspected communication between dorsal and ventral iris. Despite its role in axon guidance, NTN1 has also been shown to inhibit leukocyte migration. Thus, NTN1 up-regulation might protect injured tissues [146]. UNC5B is responsible for apoptosis and because NTN1 is up-regulated by p53 it is considered as an oncogene [147]. Ephrin receptors activated by ephrins have been shown to inhibit signaling by oncogenes. The case of TMEM proteins is interesting as well. Even though not much is known for TMEM185A and TMEM116, TMEM16F is known to form a Ca 2+ -activated channel, which plays a role in blood coagulation that is also mediated by thrombin activation [148]. In turn, blood coagulation has been implicated in the induction of lens regeneration from the dorsal iris [142].
We have also verified some of these patterns via qRT-PCR, which confirm this remarkable difference in expression along the dorsal/ventral iris. The up-regulation of TBX5, FGF10 and UNC5B in the dorsal iris and the up-regulation of VAX2, NR2F5 and NTN1 in the ventral iris are shown in Figure 3. Interestingly, the expression of these genes is also dependent on time, suggesting a potential role of those genes during regeneration (ANOVA; a,0.05). In addition, we further verified the up-regulation of NTN1 in the ventral iris and its receptor (UNC5B) in the dorsal iris.
What Genes are Regulated Specifically at 4 or 8 dpl?
We showed only minor qualitative differences in gene expression between the dorsal and ventral iris. Quantitative changes are dominant and seem to correlate with regenerative abilities. To answer the question whether there are any differences at day 4 (both dorsal and ventral) versus day 8 (dorsal and ventral), which might uncover the importance of timing rather than spatial regulation, we compared the 4 day group with the 8 day group as shown in Figure 4.
Fisher's exact test with multiple testing corrections for GO from transcripts that are up-regulated in both dorsal and ventral iris 4 dpl compared to 8 dpl versus the whole reference transcripts reveals that DNA polymerase activity and nucleotidyltransferase activity GO terms to be significantly enriched (Table S3). This indicates the initiation of the cell cycle re-entry which has been found to be the major event 4 dpl [149].
Fisher's exact test with multiple testing corrections for GO from transcripts that are up-regulated in both dorsal and ventral iris 8 dpl compared to 4 dpl versus the whole reference transcripts revealed many interesting patterns (Table S4): as expected GO terms in the cellular component category and related to extracellular matrix are over-represented in the group. Furthermore over-represented terms include extracellular matrix structural constituents and metalloendopeptidase activity in the molecular function category (Table S4). Collagen catabolic process, collagen metabolic process, cell adhesion, extracellular matrix and structure organization, and peptide secretion are overrepresented in the biological process category (Table S4). In addition, many of the GO terms in the biological process category related to differentiation, movement, development and patterning are over-represented in the group. Finally, many GO terms that are over-represented in the group are related to macromolecule transport and synthesis. These results indicate active remodeling, transcription and metabolism at day 8, which was expected because the process of dedifferentiation and specification of the lens vesicle peaks at this time point.

Conclusion
The transcriptome analysis during lens regeneration revealed much needed and useful information. First, we were able to identify quantitative patterns of gene expression that create gradients along the dorsal/ventral iris. This finding is of particular importance since it establishes a molecular framework that drives the ability of the dorsal iris for lens regeneration. Second, our analysis identified genes that might be critical for the induction of lens regeneration. For the first time, we now know factors that can be studied in functional assays, such as in trangenesis or knockdown [150,151], to establish their potency in inducing/ inhibiting regeneration. In addition, this knowledge allows us to perform comprehensive comparisons to other animal models that lack the ability for lens regeneration, which might unveil fundamental differences and similarities between regenerating and non-regenerating species.

Supporting Information
Table S1 GO terms that are over-represented in dorsal iris 4 and 8 dpl versus ventral iris 4 and 8 dpl. (XLSX)