|Home | About | Journals | Submit | Contact Us | Français|
The serine protease thrombin is generated from its precursor, prothrombin, in the coagulation cascade and plays a central role in fibrin deposition and platelet activation mediated through the protease activated receptors. Knockdown of prothrombin in the zebrafish was previously shown to recapitulate the phenotype observed in prothrombin knockout mice, such as an absence of blood pericardial edema, and hemorrhage. However, the role of thrombin during embryogenesis is not fully understood. To find genes affected by potential thrombin signaling in embryogenesis before blood circulation microarray analysis was performed using total RNA prepared from antisense-injected, knockdown embryos versus mismatch-injected at 20 hours post fertilization. A total of 63 upregulated and downregulated genes were identified with duplicate microarrays using dye reversal and a two-fold difference limitation. Real time RT-PCR for 10 selected genes identified by the microarray confirmed the expression changes in these genes. One particular gene, phlda3, was at least eleven fold upregulated, and in situ hybridization revealed expansion of phlda3 expression in the central nervous system, branchial arches, and head endoderm in knockdown embryos. The identification of these genes regulated by thrombin according to microarray analysis should provide a greater understanding of the effects of thrombin activity in the early vertebrate embryo.
Thrombin is the central serine protease of the blood coagulation cascade that converts fibrinogen to fibrin and is generated from its zymogen precursor prothrombin by the action of Xa. Thrombin also mediates a cellular response primarily through cleavage and activation of the G-protein-coupled, protease activated receptor-1 (PAR-1) that has been extensively characterized in platelet activation . In addition to being a potent platelet agonist, thrombin also exerts activity in fibroblasts, smooth muscle, neurons, endothelium and other cell types [1–5].
Targeted gene inactivation of prothrombin in mice leads to embryonic lethality at approximately midgestation [6–8]. A prothrombin-deficient phenotype exhibited many developmental defects related to yolk sac vascular integrity such as enlarged and dilated capillary structure, vessels devoid of blood, and flattening of visceral yolk sac endoderm. In the zebrafish, knockdown of prothrombin produced an early phenotype in which greater than one third of all embryos have abnormalities in their overall growth with defects in the anterior and posterior regions . When grown to 48 hours post fertilization (hpf), these embryos recapitulated the prothrombin deficient phenotype observed in the mouse embryo. They exhibited an absence or reduced number of blood cells, reduced blood flow, pericardial edema, and blood clots in the trunk region. These results showed a conserved role for thrombin in vertebrate embryonic development.
To elucidate changes in gene expression affected by prothrombin knockdown, hybridizations on microarrays consisting of oligonucleotides representing 14,000 genes were performed using RNAs made from 20 hpf antisense morpholino-injected (ASMO) embryos displaying the previously characterized abnormalities versus mismatch control-injected (MMMO) embryos. Hybridization results revealed a total of 63 upregulated and downregulated genes using a twofold expression change limit. Real time quantitative RT-PCR (QRTPCR) was used to confirm microarray results for 10 genes identified. A gene that was 11-fold upregulated, called phlda3, encodes a small protein containing a pleckstrin homology domain that could be involved in regulating IP3 release, recruitment of proteins to the intracellular membrane surface, and membrane shape changes. Phlda3 was shown to be upregulated in the CNS with possible expansion of expression in the branchial arches and anterior endoderm. Additionally, the sry-related HMG box transcription factor, sox21, was shown to be downregulated greater than twofold. Sox21 is duplicated in the zebrafish genome and is termed as sox21a or sox21b . Sox21a is expressed throughout the forebrain, midbrain and hindbrain, but shows greatest expression in the midbrain-hindbrain boundary (MHB) . Reduction of sox21a MHB expression in knockdown embryos further verified microarray results. All of these results taken together indicate a potential role for thrombin signaling in the embryonic brain. The identification of sox21a, phlda3, and the other genes regulated by thrombin according to microarray analysis will provide a greater understanding of the effects of thrombin activity in the early vertebrate embryo.
A prothrombin antisense morpholino oligonucleotide, 5′ GTTTGGCTCCCATCCTTGAGAGTGA-3′ (ASMO) against the 5′-UTR to target the translational start site of the zebrafish prothrombin mRNA and a control oligonucleotide 5′GTTTCGCTCGCATGCTTCAGACTGA-3′ (MMMO) with 5 base mismatches (mismatches indicated by underlines) were purchased from Gene-Tools LLC, Philomath, OR. Embryos were microinjected with 4.5 nL ASMO (1 mg/mL) or MMMO diluted in Danieau buffer into the yolk of one to four cell stage embryos . Injected embryos were maintained in embryo medium at 28°C until 20 hpf. Embryos that exhibited early phenotype (depicted in Fig. 1) as previously reported were collected .
Total RNA was isolated from 20 hpf embryos collected from 4 independent microinjection experiments. The first batch of RNA was made from 16 embryos (each embryo was injected with 4.5ng ASMO) displaying the early phenotype, and 16 embryos injected with 4.5 ng MMMO that were normal in appearance. The same selection procedure was used for 3 more experiments using 12 embryos each for ASMO and MMMO groups. Total RNA was isolated from embryos, and independent RNA preparations were pooled, run through columns provided in the RNeasy® Mini Kit (Qiagen Inc., Valencia, CA), and final RNA integrity, purity, and concentration were again estimated by UV spectrophotometry and gel electrophoresis.
The University of Texas Southwestern Medical Center at Dallas Microarray Core Facility performed the following steps using the high quality total RNA prepared. Five micrograms of pooled ASMO and MMMO RNAs were used with the Superscript II Reverse Transcriptase kit (Invitrogen, Carlsbad, CA) for first strand cDNA synthesis. Second strand was synthesized using the SuperScript™ Double-Stranded cDNA Synthesis Kit (Invitrogen). Reactions were incubated at 16°C for 2h, then 1 μl T4 polymerase was added, incubated for 10 min at 16°C, and finally reactions were stopped by adding 10 μl of 0.5M EDTA to inhibit all enzymes.
Complementary DNA was purified using the GeneChip Sample Cleanup Module (Affymetrix, Santa Clara, CA) according to the protocol supplied by the manufacturer. Antisense RNA (aRNA) was amplified by T7 in vitro transcription using the MessageAmp aRNA Kit (Ambion, Austin, TX) with unlabeled NTPs, 16 μl cDNA, and incubated at 37°C overnight according to protocol. The GeneChip Sample Cleanup Module (Affymetrix) was then used again for aRNA purification, and aRNA was quantified by spectrophotomety. Probes were labeled using the ASAP RNA Labeling Kit (Perkin Elmer, Boston, MA) in two separate reactions using 2μg of each aRNA with ASAP Cyanine-3 Reagent or ASAP Cyanine-5 Reagent according to protocol. Each labeled probe was then combined, mixed into the same tube, and purified using a Microcon YM-30 filter column. RNA probe synthesis was also repeated by reverse labeling ASMO and MMMO RNA samples.
Five microliters of probe were then mixed with preheated ASAP Hybridization buffer (Perkin Elmer) and placed on to two Zebrafish 14K arrays (MWG Biotech Inc., High Point, NC) with each array being divided into two slides (array A and array B) each representing 14,067 total genes. Probes containing the MMMO-Cy3 and ASMO-Cy5 mixtures were used for two microarray A slides and the reverse label mixture (MMMO-Cy5 and ASMO-Cy3) was used for two microarray B slides for performing the hybridizations in duplicate. Probes were hybridized for 14–16h at 62°C. Slides were then washed, dried, and scanned for data acquisition using a GenePix® 4000B scanner (Axon Instruments, Inc., Union City, CA).
Scanned GenePix data was imported into the online GeneTraffic™ DUO two-color microarray data analysis software version 2.8–9 (Iobion Informatics LLC, La Jolla, CA) for data analysis. LOWESS sub-grid normalization was used for each slide, and background was automatically subtracted from a generated correction value. Two fold change criteria were used to select differentially expressed genes from the microarray to complete an annotated list of results with fold level changes greater than or equal to 2-fold or less than or equal to 0.5-fold. Data was sorted by selection of successful hybridizations. All genes were manually grouped according to function or subcellular localization.
Total RNA was made from early phenotype and MMMO-injected reference embryos at 20hpf similar to RNAs prepared for the microarray as described below. Three independent RNA preparations were made from a pool of embryos for each ASMO and MMMO-injected group by homogenization in RNAzol™ B Isolation of RNA solution (Leedo Medical Laboratories, Houston, TX) using a Brinkman polytron homogenizer. RNA pellets were resuspended in nuclease-free water at 60°C for 10 min. RNAs were treated with DNAse using the DNA-free™ DNase Treatment kit (Ambion, Austin, TX) at 37°C for 30 min to eliminate any potentially contaminating genomic DNA, and 500 ng of treated total RNA were used for cDNA synthesis using Taqman® Reverse Transcription Reagents (Applied Biosystems, Foster City, CA) according to protocol.
Primers were designed online with Primer3 (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3.cgi/). All primer sequences used for validation experiments are reported in Table 2, and were used with β-actin primers: forward 5′ TTCTGGTCGGTACTACTGGTATTGTG 3′, and reverse 5′ ATCTTCATCAGGTAGTCTGTCAGGT 3′. Primers used for the detailed confirmation to specifically detect sox21a and b duplicated genes in their expression in knockdown embryos were: 5′ CCGCATTATTCCGTGCTC 3′ and 5′ ATGCCAGGTAAGGTTCATGC 3′ for sox21a; 5′ AAGGACAAATTCGCGTTCC 3′ and 5′ TGAGGCGTAGGAGAAAGACG 3′ for sox21b. Each real time PCR reaction was performed in triplicate using β-actin as the control for normalization. Primers were diluted to.5 pmoles/μl in a forward and reverse primer mix, and 10.5 μl primer were mixed with 12.5 μl 2X SYBR® Green PCR Master Mix (Applied Biosystems) and 2 μl cDNA for each developmental stage containing a 25μl reaction volume in a 96 well format (Applied Biosystems). Reactions were performed using the ABI Prism 7900 HT Sequence Detection System with cycling parameters that were 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Raw data was analyzed and exported using the ABI prism SDS 2.0 software. Dissociation curves were also recorded to verify a single product was amplified for each primer set.
Raw fluorescent data was imported into the automated calculation workbook entitled Data Analysis for Real-Time PCR (DART-PCR) designed by Peirson, Butler, and Foster  that enables the rapid calculation of threshold cycle, amplification efficiency, and resulting R0 values. Results for each target gene were then normalized against β-actin to determine a fold expression value. Each reaction in triplicate was repeated in 3 experiments using independent RNA preparations.
Partial cDNAs for phlda3 and sox21a were amplified by RT-PCR and cloned into the pCR® II-TOPO® vector. The primers containing sequences 5′ GACGGGTATCTGGAGAAGAGG 3′ and 5′ CTGTTCCTGCCGGTCTGA 3′ were used for phlda3. Sox21 primer sequences were used previously as reported in . Antisense probes were transcribed from linearized vector using a MEGAscript™ Sp6 high yield transcription kit (Ambion Inc., Austin, TX) with each reaction containing a 3:1 ratio of Digoxigenin-11-UTP (Roche Diagnostics Corporation Indianapolis, IN) to UTP.
ASMO and MMMO embryos were prepared, prehybridized, and hybridized overnight at 70°C as previously reported . Embryos were washed to remove probe and blocked in blocking solution (2mg/mL BSA, 5% sheep serum, 1% DMSO in PBT, pH 7.4) for 4h and then incubated overnight at 4°C with Anti-Digoxigenin-AP Fab fragments (Roche Diagnostics Corporation Indianapolis, IN) diluted 1:4500 in blocking solution. Embryos were washed extensively to remove unbound antibody, equilibrated in NTMT buffer (.1M Tris pH 9.5,.1M NaCl,.05 M MgCl2,.1% Tween 20), and stained in NBT/BCIP solution (Roche Diagnostics Corporation). Reactions were stopped by PBT washes, dehydrated in methanol as done previously and stored until clearing in 2:1 benzyl benzoate/benzyl alcohol. Images were then under captured under bright field using a Nikon CoolPix 995 CCD camera mounted on a Nikon Optiphot microscope using 2X and 10X objectives.
To identify prospective genes affected by thrombin signaling in the zebrafish embryo, pooled total embryonic RNAs from three separate microinjection experiments at 20 hpf were used for probe labeling and microarray analysis using a 14K zebrafish array. Only ASMO embryos that exhibited posterior abnormalities with a reduction of the yolk sac extension as reported previously were used for RNA extraction along with MMMO embryos at the same stage (Fig. 1). Hybridizations were performed on duplicate slides using the same sample and were referred to as array 1 and 2. Following corrections for background intensity and normalization between filters, an overall intensity distribution for the ratio of expression for each representative gene was constructed by plotting log ratio intensity of ASMO versus MMMO. These plots revealed that the overall distributions for array 1 and 2 duplicates were similar, and verified that independent hybridizations gave comparable normalized intensity values (Fig. 2). There were relatively few genes expressed above the 2-fold ratio and below the 0.5-fold ratio as indicated by red or green colored outliers. A 2-fold change criterion was then used to obtain a gene list that excluded flagged data. The majority of genes affected by prothrombin knockdown were found with both of the hybridizations performed. According to both arrays, 35 total represented genes were upregulated with 27 of them being confirmed in duplicate, and 28 were downregulated with 23 in duplicate. All of these genes were classified according to cellular localization, function, or structure (Table 1). The greatest number of genes affected was related to metabolism and mitochondria (19% of total genes affected), while only a single ligand and transcription factor gene were affected in their expression. Cell structure and motility comprised 17% of genes affected, but contained different isoforms for similar genes such as keratin and parvalbumin. Some of the genes on the array such as for an unknown cysteine protease, the alpha globins, a putative membrane protein, the heat shock protein 90-beta, annexin max 3, and smd2 were spotted more than once which gave multiple values that were similar for each array.
Microarray findings were then validated by QRTPCR by selection of 5 upregulated and downregulated genes with priority given to greatest fold level change. Zebrafish genomic DNA sequence information for each of these genes was obtained by using the accession number provided with each spot identity. Primers were designed from zebrafish gene sequence reported on the array. Table 2 shows primer sequences used for microarray validation. Individual amplification curves demonstrated significant differences in gene expression in comparison to β-actin controls in which no change was detected between ASMO and MMMO embryos. Raw fluorescent data was used to analyze reaction efficiency and R0 values were generated to obtain a fold level change value relative to MMMO embryos after normalization using β-actin controls according to the DART procedure . QRTPCR results indicated a confirmation of trends in the changes of these selected genes with respect to their upregulation or downregulation as shown by each microarray (Fig. 3).
Interestingly, one particular gene reported on the microarray as bwr1c is also named IPL. BLAST results using mouse bwr1c/IPL sequence against the zebrafish genome gave greatest percent identity for another gene called phlda3. Both IPL and phlda3 belong to members of a pleckstrin homology-related gene family . We could distinguish that the gene affected as shown by microarray and QRTPCR was indeed phlda3 and not IPL in the zebrafish by synteny (Fig. 4A). The cluster of genes flanking phlda3 is arranged similar in human, but there is an inversion when compared to zebrafish and mouse. QRTPCR using phlda3 gene specific primers showed greater upregulation for phlda3 than the microarray (Fig. 3). To further confirm this result, semi-quantitative RTPCR using an alternative phlda3 primer set to amplify a larger sized product was performed to visualize product intensity by agarose gel electrophoresis compared to an another control that previously showed no difference in expression between ASMO and MMMO embryos. Results indicated that phlda3 is highly upregulated in knockdown embryos (Fig. 4B). This corroborated the fact that phlda3 is the gene that is altered in its expression shown by microarray and QRTPCR.
Similar to difficulties in discerning between phlda3 and IPL in the zebrafish, gene duplication was noted for sox21 in the zebrafish genome. The duplicate genes were designated as sox21a and sox21b . Since the primer set designed for confirmation by QRTPCR could amplify both sox21 isoforms, two more primer sets were designed specifically to amplify sox21a and sox21b by QRTPCR. Results revealed a greater downregulation of sox21b compared to sox21a when normalized to β-actin controls in reference to MMMO embryos, with both indicating an approximate 2-fold repression or greater (Fig. 4C). To further validate this data, two additional primer sets were designed to amplify a greater sized product that also gave results indicating a decrease in expression between ASMO and MMMO embryos for sox21a and sox21b, but with a greater reduction in sox21b (Fig. 4D). Control reactions using β-actin specific primers showed no difference in expression between the two groups.
The PCR products for phlda3 and sox21a were amplified (Figure 4B and 4D) and cloned for probe synthesis. In situ hybridization analysis revealed an expansion of phlda3 expression in the central nervous system particularly in the forebrain, midbrain, endoderm of the head, and the branchial arches (compare Fig. 5A,B and 5C,D). Interestingly, there also appeared to be abnormal upregulated expression in the tail region that was not observed in the mismatch control that showed a segmented pattern (compare Fig. 5B and 5D). It was also noted from phlda3 expression that furrow formation that occurred at the MHB appeared to be absent in the knockdown embryos along with the upregulation. A decrease in sox21a expression was found in the MHB of ASMO embryos compared to controls (compare Fig. 5E,F and 5G,H). Comparison of lateral views between ASMO and MMMO embryos showed that some MHB expression was lost in ASMO embryos (Fig. 5E and 5G). It was apparent from dorsal views that there was also a lack of furrow formation at the MHB along with the preservation of characteristic sox21a patterning, albeit with the reduction of expression at the MHB (Fig. 5F and 5H).
This study used a method to understand gene expression changes in prothrombin knockdown embryos using total RNA extracted from embryos and not from specific tissues or cell types. Elucidation of the greatest differences in gene expression was advantageous because it found potential genes controlled by thrombin in the entire embryo compared to approaches in which only tissue or cell type specific genes would be isolated. Since use of embryonic total RNA revealed a small number of relatively large changes in gene expression, minute changes in gene expression that may be vital to developmental processes may have been undetected. However, this method is currently used to find basic transcriptional programs in development induced by single factors with the use of mutant zebrafish embryos . This approach was also used for zebrafish embryos treated under hypoxic conditions as well as for retinoid deficient rat embryos [17, 18]. The use of total embryonic RNAs for microarray analysis also gave a manageable number of genes that allows microarray validation by QRTPCR possible for every gene identified. This technique may also be used to gain a broader understanding of the basis for embryonic lethality caused by prothrombin gene inactivation. The zebrafish system is beneficial for direct observation of prothrombin knockdown phenotype, which resembles the knockout phenotype  and facilitates analysis of gene expression by preparing total RNA from the affected embryos.
A total of 63 genes were revealed by both microarrays with 50 of these genes being confirmed in duplicate over a total of 14,000 genes represented. It is known that thrombin may upregulate collagen synthesis in platelets, fibroblasts, hepatic stellate cells, mesangial cells, and vascular smooth muscle. Thrombin signaling may further amplify signaling events as an autocrine mediator during vascular injury, cell signaling, or in the process of wound healing [19–22]. Collagen Ia2 downregulation as well changes in the expression of other structural components, such as claudin e and keratin, suggests a role for thrombin stimulation or regulation of extracellular events in embryonic growth.
After microarray validation, it is apparent that phlda3 (pleckstrin homology like domain family A, member 3) is the most significantly affected gene. It is approximately 4-fold upregulated by microarray analysis, and 11-fold upregulated according to QRTPCR. This discrepancy may be due to cross-hybridization of closely related family members such as with Ipl (Imprinted in Placenta and Liver) or TGAG51 (T-cell Death Associated Gene 51) which also contain pleckstrin homology (PH) domains, and indicates that precaution should be taken with the interpretation of oligo-based arrays if array construction did not take into account the presence of other highly related sequences used to represent each zebrafish gene. Multiple methods of confirmation such as synteny analysis between zebrafish and mammalian genes, and in situ hybridization provided validation in addition to spatial and temporal information.
Phlda3 is a very small protein; the central PH domain contains 100 amino acids out of a total size of 125 amino acids. It has moderate affinity for membrane phosphatidylinositol phophate binding, and PH-domain dependent accumulation at membrane ruffles . Phlda3 is expressed ubiquitously in mouse fetal tissues except for liver , while Ipl is specifically expressed in extraembryonic tissues, i.e., the visceral yolk sac endoderm, embryonic liver, and kidney [15, 24]. Knockout of Ipl results in overgrowth and expansion of the spongiotrophoblast layer while gene deletion of phlda3 had no effect . Although it is difficult to discern the function of phlda3 in embryonic development, it could possess a redundant function in mammals based upon knockout studies, but may have a specific role in zebrafish. Thrombin signaling could possibly regulate phlda3 to prevent G protein- induced, cytoskeletal and membrane shape responses in certain cells [15, 26, 27] since it has been shown previously that PH domain phosphorylation appears to inhibit Gβγ-activated PI3K by interacting with Gβγ [28–30]. However, phlda3 gene regulation by subsequent thrombin activation of its receptor in cells remains to be determined.
The only transcription factor identified in our studies was sox21. After validation by independent primers for sox21a and b, it is evident that both may be affected. This observation indicates that precaution should be taken in consideration of duplicated genes. Sox21 is a part of the group B Sox proteins, belongs to the HMG box superfamily, and has been described as a repressor of neuronal differentiation and causes neurite retraction . Sox21a expression controlled by potential thrombin activity could be a mechanism by which neurite outgrowth is regulated in embryonic development since thrombin plays a substantial role in neurite retraction [32–34], but this also requires further investigation.
The reduction of expression observed in the MHB of knockdown embryos provides a new avenue to be explored for the mechanism of thrombin action in the embryonic brain before the initiation of blood circulation. Thrombin is known to signal through PARs expressed on neuronal cells [2, 4, 35–38]. All of these data taken together with the known roles for phlda3, sox21, thrombin activity on neuronal cells, and their coexpression in the CNS suggest a potential model for thrombin activity in the developing vertebrate brain.
The microarray approach used here may provide insight into potential downstream transcriptional activity induced by thrombin signaling in early vertebrate embryogenesis in processes such as cell growth, differentiation, and migration. The genes elucidated here may possess unknown roles in response to thrombin activity that are unique during embryogenesis.
This study was supported by National Institutes of Health Grant to P.J.
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.