Characterization of zebrafish gp1ba mutant and modeling Bernard Soulier Syndrome

Abstract

Objective:

To model classical Bernard Soulier Syndrome in the zebrafish by targeting Gp1ba.

Results:

We obtained gp1ba mutant embryos from Zebrafish International Resource Center and grew them to adulthood. The tail clips from these fish were used to prepare DNA and sequenced to identify heterozygotes. They were then bred to obtain homozygotes. The mutation was confirmed by DNA sequencing as a termination codon UAA in place of AAA codon at position 886 in the gp1ba transcript. Thus, at the Pro-295, the Gp1ba protein could be terminated. The blood from gp1ba homozygous and heterozygous mutants showed decreased ristocetin-mediated agglutination in the whole blood agglutination assay. The gp1ba heterozygous and homozygous larvae were subjected to a laser-assisted arterial thrombosis assay, and the results showed the prolonged occlusion in the caudal artery. These results suggested that the gp1ba mutant had a bleeding phenotype. The blood smears from the adult gp1ba, heterozygous and homozygous mutants, showed macrothrombocytes, similar to the human GP1BA deficiency that showed giant platelets. The bleeding assay on these heterozygous and homozygous mutants showed greater bleeding than wildtype, confirming the above findings. Taken together, the characterization of gp1ba zebrafish mutant suggested an autosomal dominant mode of inheritance.

Conclusion:

The zebrafish gp1ba mutant models classical Bernard Soulier Syndrome and could be used for reversing this phenotype to identify novel factors by the genome-wide piggyback knockdown method.

Introduction

Von Willebrand Factor is a primary ligand for a glycoprotein membrane complex GP1B-IX-V ¹. This complex consists of four subunits, namely GP1BA, GP1BB, IX, and V ². The deficiency in any of the subunits of this complex except for V results in a rare autosomal recessive disorder called Bernard Soulier Syndrome ³. Also, a type A2 Bernard Soulier Syndrome has an autosomal dominant mode of inheritance ⁴. Previously, we created a Von Willebrand Factor mutation in zebrafish by CRISPR/Cas9 mutagenesis and modeled Von Willebrand Disease ⁵. To complement this work, we wanted to model Bernard Soulier Syndrome in zebrafish.

In this paper, we characterized a zebrafish gp1ba mutant with a termination codon at position 296 in the protein sequence. We demonstrated that adult gp1ba mutants showed profuse bleeding after mechanical injury, diminished ristocetin-mediated agglutination in whole blood agglutination assay, and the presence of macrothrombocytes in the blood smears. We did not observe any thrombocytopenia. In mutant larvae, we found prolonged time to occlusion in the arterial thrombosis assay. We also noted the Bernard Soulier Syndrome in zebrafish had a dominant mode of inheritance. Thus, the zebrafish gp1ba mutant models Bernard Soulier Syndrome and could be used to reverse this phenotype to discover the compensatory factors by whole genome knockdown studies.

Materials and Methods

gp1ba gene and Gp1ba

gp1ba gene (ENSDARG00000086846) and the cDNA (ENSDART0000123542) sequence were retrieved from ENSEMBL. The predicted translational product of the gp1ba gene, Gp1ba (ENSDARP00000109690) was compared with the human GP1BA using the publicly available MultAlin program.

gp1ba zebrafish mutant breeding and genotyping

gp1ba zebrafish heterozygote embryos from an N-ethyl-N-nitrosourea mutant, sa9915 were purchased from the ZFIN mutant repository (https://zfin.org/ZDB-ALT-130530-653). The embryos were collected and placed in the embryonic E3 medium (0.17 mM KCl, 0.33 CaCl₂.2H₂O, 5 mM NaCl, 0.33 mM MgSO₄, and 0.1% methylene blue) in small plastic containers. The larvae from 4 days post fertilization (dpf) to 8 dpf were fed with live paramecium twice daily. From 8 dpf, the larvae were fed with brine shrimp. The 1-month-old juvenile fish were transferred into the recirculating freshwater system at 28°C (82°F) and a pH 7.6 supplemented with Instant Ocean. They were grown to adulthood (2 to 3 months) and were kept under 14 hours light and 10 hours dark cycle and were fed with live brine shrimp and fish flakes twice in the day regularly ⁶.

The male and female fishes were kept in individual tanks, and they were genotyped as follows. The genomic DNA from the above individual male/female fishes was isolated from the tail clips ⁵. The tails were clipped by a pair of scissors and each tail was incubated in 50 µL DNA extraction buffer (10 mM Tris pH 8.5, 50 mM KCl, 5 mM EDTA, 0.45% NP-40, 0.01% gelatin, and 0.45% Tween 20) containing 0.5 µL proteinase K (20 mg/mL) for 12 hours at 55°C. Then proteinase K was inactivated by keeping this sample at 96°C for 10 minutes. The sample was centrifuged at 3000 rpm for 3 minutes, and the supernatant was transferred into fresh tubes. 4 µL of this supernatant was then amplified by polymerase chain reaction (PCR) containing 1 μL of 25 μM forward primer, 5’-TCAATTCCAAAGCCAATTCC-3’, 1 μL of 25 μM reverse primer, 5’-CGACTGTGGTGTGCAGATTT-3’, and 10 µL of 1-Drop PCR Mix (101Bio, Palo Alto, CA). The PCR products of size 332 bp were then resolved on 1.2% agarose gels. This DNA band was then excised and processed using EZNA gel extraction kit (Omega Biotek, Norcross, GA), and this gel-purified DNA was sent for sequencing to Genewiz, South Plainfield, NJ. The sequences were then analyzed using the software FINCH TV 1.5.0 (Geospiza, Inc., Seattle, WA) to genotype the fish.

From the heterozygotes identified by the above genotyping, one male and one female were kept in a breeding tank separated by a divider in the evening. The next day morning, the divider was removed after the lights came on, and the fish were kept undisturbed for about 30 minutes. The eggs were then collected, and the larvae were grown to adulthood and genotyped as described above. The homozygotes, heterozygotes and the wildtype littermates were used in the subsequent analysis as described below. All animal experiments were approved by the University of North Texas Institutional Animal Care and Use Committee.

Arterial laser-assisted thrombosis

Embryos from the homozygote x homozygote, homozygote x wildtype, wildtype x wildtype crosses were collected. Ten of the resulting 5 dpf larvae from each cross were anesthetized in an Eppendorf tube containing 0.5 mL of E3 medium and 10µL of 10 mM Tricane (MS222) for 1 minute. To this, 0.5 mL of 1.6% low melting agarose was added and the agarose mixture containing larvae was poured on a glass slide having a rubber gasket glued to the peripheral sides of the glass slide with petroleum jelly. The larvae were aligned using a pipette tip such that the larva lies on its lateral side. The slide carrying the larvae were placed under a Nikon Optiphot fluorescence microscope, and the larvae were focused using the 20X objective. Each larva was injured by using a pulsed nitrogen laser (445 nm) at 18 hits per cycle passed through coumarin dye via a fluorescence port (Micro Point Laser, Stanford Research Systems Inc., Sunnyvale, CA) around the fifth somite post anal pore on the caudal artery. The time taken to occlude the vessel completely from the time of laser injury was recorded as time to occlusion (TTO). TTO > 120 seconds was taken as 120 seconds ⁷.

Blood collection

Adult zebrafish was placed on its lateral side on a clean paper towel, and its head was covered with a wet laboratory tissue. The fish body was gently wiped with a laboratory tissue, and a lateral incision was made with a pair of dissection scissors between the dorsal and ventral fins and between second and fourth black stripes counted from the dorsal fin. 2 µL of blood oozing out was collected using a micropipette ⁸.

Bleeding time assay

Bleeding time was measured after mechanical injury. For mechanical injury, the adult zebrafish was laid on its lateral side, and the caudal artery was clipped as described above. To make sure the clip was uniform for all the fishes, we marked the ends of the dissection scissors with a marker pen approximately 2 mm above the scissors’ tips. The scissors’ tips were then inserted into the fish body until the marked line touches the skin in between the ventral and the dorsal fins at positions described above. The fish was allowed to bleed for 1 minute, and a photograph was taken. To quantify the bleeding, we used ImageJ software and separated red, blue, and green colors. The red color pixels were counted as an index of the area and were multiplied by the mean pixel intensity to yield the total red intensity representing the extent of bleeding ⁹.

Flow cytometry

Flow cytometry was used for thrombocyte analysis. In a 1.5-mL Eppendorf tube containing 0.5µL of 3.8% Sodium Citrate, 2 µL zebrafish blood was added and mixed gently to prevent blood clotting. For labeling thrombocytes with DiI, 1µL of 10mM DiI in dimethylformamide was added to 100µL of 1X PBS, and 5 µL of this diluted DiI was added to the blood collected above and mixed with finger tapping followed by 400 µL of 1X PBS. The blood cells were then analyzed by BD AccuriTM C6 Plus Flow Cytometer for 10000 events ¹⁰.

Blood smears

For the preparation of blood smears, 1 µL whole blood was added on a clean microscope slide. Another slide at an angle of 30° to 40° away from the drop of the blood was placed on the top, moved from the drop towards the edge of that slide, and air-dried completely. The blood cells were first fixed with methanol (Thermo Fisher Scientific, Waltham, MA), stained with Hematoxylin and Eosin stain (Biocare Medical, Pacheco, CA), and rinsed with distilled water. The blood cells were photographed under the Axio Imager microscope attached with Zeiss Axio Cam MRC camera using 40X objective and 2.0X zoom. Zen 3.4 ZenLite software was used to analyze the size of thrombocytes. The thrombocytes were identified by their unique morphology (scanty cytoplasm with a large nucleus almost covering the entire cell) and using the 20 µm scale bar ¹¹. The size of 101 thrombocytes of each type of fish was analysed by measuring the diameter from one end to the other end of the cell using Zen 3.4 ZenLite software, which was used to take images of the thrombocytes. The measurements were made by noting the longer diameter of the cell to maintain uniformity.

Whole blood agglutination assay

A whole blood agglutination assay was carried out as described previously ⁸. Briefly to a Nunc microtiter plate conical well with a total capacity of 10 µL, 4.5 µL 1X PBS was added, followed by the addition of either 5 µL of collagen (1.9 mg/ml) or ristocetin (3 mg/ml). After mixing them, 0.5 µL of whole blood was then introduced gently from the top, making sure that blood goes through the agonist solution. The samples were incubated at room temperature, and after 10 minutes the plate was tilted to 45° for 10 seconds. Photographs were taken to record the migration of blood down the inclined well wall. The photographs were normalized by resizing the images such that the diameter of all the wells was the same. No mobility or slight mobility of the blood in the inclined well was considered as a positive aggregation. The length of migration of the blood was measured by using the ruler in the PowerPoint ⁸.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 9.2.0 software. Groups were compared by using one-way ANOVA and selecting column data, and multiple comparisons feature. p ≤ 0.05 was considered statistically significant. Error bars represent standard deviation.

Results and discussion

In humans, the GP1BA gene consists of two exons and one intron, and the gene is located on chromosome 17 ¹². The zebrafish gene also contains two exons and one intron. However, the entire protein sequence is encoded by one exon in humans, whereas it is encoded by two exons in zebrafish, and the gene is located on chromosome 23. The human GPIBA is coded by 652 amino acids (NM_000173.7), while the zebrafish protein is 610 amino acids (ENSDARP00000109690). Comparison of the human GPIBA and the zebrafish Gp1ba revealed 38% identities and 58% positives between these two proteins at their amino-terminal regions. However, the overall identities were 19% (Figure 1).

Figure 1:

Image showing the MultAlin analysis of the zebrafish Gp1ba protein sequence compared with the human GP1BA protein sequence. The amino acids highlighted with the red color are the amino acids that are conserved (high consensus) between zebrafish and humans. Blue indicates low consensus sequences, and black indicates neutral according to the MultAlin program.

Since the deficiency of GPIBA in humans has been shown to cause Bernard Soulier Syndrome, we wanted to test whether such a deficiency in zebrafish causes a similar phenotype. To test this, we identified a gp1ba zebrafish mutant in the ZFIN database, which had a termination codon UAA (TAA in DNA) in place of AAA codon at position 886 in the transcript (ENSDART0000123542), and thus at the Pro-295, the protein could be terminated. The mutant was obtained from Zebrafish International Resource Center (ZIRC) as heterozygote and wildtype embryos. These embryos were grown to adult hood (F0), the DNA prepared from their tail clips, and sequenced. The wildtype (gp1ba^+/+) and heterozygote (gp1ba^+/−) sequences are shown in Figure 2. Interestingly, there were two wildtype alleles with SNPs T and C immediately 5’ to the AAA encoding Lys-296. These SNPs generate CCT and CCC that encode proline. Thus, the amino acid is not altered by the SNPs. We also found these SNPs and superimposing AAA and TAA corresponding to termination codon in the heterozygotes. The male and female gp1ba^+/− heterozygotes were crossed, and the progeny (F1) were genotyped again to obtain homozygotes (gp1ba^−/−), heterozygotes (gp1ba^+/−), and the wildtypes (gp1ba^+/+). The homozygote sequences are shown in Figure 2.

Figure 2:

Sequence analysis of gp1ba mutant. DNA obtained from tail clips from the F1 generation resulting from gp1ba^+/− (identified in F0 generation) crosses were amplified by PCR and resolved on agarose gel electrophoresis (Figure S1). The DNAs from these bands were eluted and were sequenced. gp1ba^+/+, gp1ba^+/−, and gp1ba^−/− sequence chromatograms are shown. The top two chromatograms show SNPs, T and C located immediately 5’ to the AAA sequence that is in the box and codes for Lys-296. All the heterozygotes carried one allele, which had the SNPs adjacent to AAA and TAA sequences.

To perform arterial laser thrombosis on gp1ba^+/+, gp1ba^+/− and gp1ba^−/− larvae, we set up three different crosses; gp1ba^+/+ x gp1ba^+/+; gp1ba^+/+ x gp1ba^−/−; and gp1ba^−/− x gp1ba^−/−. The progeny (F2) 5 dpf larvae from these crosses were subjected to arterial laser thrombosis, and the arterial TTO was measured. The gp1ba^+/+, gp1ba^+/− and gp1ba^−/− larvae yielded TTO of 75 seconds, >120 seconds, and >120 seconds, respectively (Figure 3). These results suggested that the phenotype is dominantly inherited.

Figure 3:

Comparison of TTO of the caudal artery after laser injury of 5 dpf larvae obtained from the F2 crosses of gp1ba^+/+ and gp1ba^+/+, gp1ba^+/+ and gp1ba^−/−, and gp1ba^−/− and gp1ba^−/− fish. Both gp1ba^+/− and gp1ba^−/− larvae show a prolonged TTO (in seconds) compared to gp1ba^+/+ control larvae by one-way ANOVA. Error bars represent mean ± SD; **** indicates a p < 0.0001. (N=10).

To test whether there is a defect in Von Willebrand Factor-mediated agglutination that also depends on its receptor GP1B-IX-V on the thrombocyte, we performed whole blood agglutination by a plate-tilt assay using gp1ba^+/+, gp1ba^+/− and gp1ba^−/− adult zebrafish blood in the presence of ristocetin. In this assay, we measured the length of migration of blood in a microtiter plate well from its origin as an indication of the extent of agglutination. The length of migration is inversely proportional to the function of the receptor. Our results showed that the length of migration of blood from the gp1ba^+/+, gp1ba^+/− and gp1ba^−/− adult fish were 0.89 cm, 1.06 cm, and 1.31 cm, respectively, suggesting the lack of agglutination in both the heterozygotes and homozygotes, which is consistent with the laser thrombosis data (Figure 4). Collagen was used as control, and as expected, there were no differences in the length of migration of blood among wildtypes, heterozygotes, and homozygotes (Figure 4).

Figure 4:

Plate tilt assay with ristocetin and collagen agonists on the blood obtained from gp1ba^+/+, gp1ba^+/− and gp1ba^−/− adult zebrafish. The photograph shows the migration of blood from the origin of the well (indicated by the white vertical line, the length of migration of blood from the origin of well is shown in cm in the presence of ristocetin (top) and collagen (bottom). The length of migration of blood from the origin of well (adjacent to the photographs) of gp1ba^+/+ fish was compared with gp1ba^+/− and gp1ba^−/− fish by one-way ANOVA showed significance with ristocetin and not with collagen. * and **** indicates p values < 0.05 and < 0.0001, respectively. ns represents no significance. Error bars represent standard deviation. (N=4).

To test whether the gp1ba deficient zebrafish have macrothrombocytes, we stained the blood smears from gp1ba^+/+, gp1ba^+/− and gp1ba^−/− fish with hematoxylin and eosin and found that the blood smears from gp1ba^+/− and gp1ba^−/− had relatively larger thrombocytes compared to the ones from the gp1ba^+/+ fish (Figure 5), although heterozygotes appear to be not as large as the ones from homozygotes. We have also measured the diameters of the thrombocytes and found that thrombocytes from gp1ba^+/− and gp1ba^−/− had 6.86 µm and 6.88 µm diameters, respectively compared to 5.65 µm thrombocytes from the gp1ba^+/+ fish (Figure 5).

Figure 5:

Images of hematoxylin and eosin-stained blood smears (top panels) generated from gp1ba^+/+, gp1ba^+/− and gp1ba^−/− zebrafish blood. The images were taken using an inbuilt scale of 20 µm provided by the microscope. The black arrows show the thrombocytes. The diameters of the thrombocytes were measured according to the scale from 101 thrombocytes each, from gp1ba^+/+, gp1ba^+/− and gp1ba^−/− zebrafish blood smears and compared (bottom panel) by one way ANOVA. **** indicates a p < 0.0001. (N=101).

To confirm whether there are any differences in bleeding of the gp1ba^+/+, gp1ba^+/− and gp1ba^−/− adult fish, we used these fish and performed a bleeding assay induced by a mechanical injury. The bleeding was measured as total red pixel intensities quantified as the number of red pixels in the photograph one minute after the injury. The results showed that red pixel intensities were 6.6 × 10⁵, 1.1 × 10⁶, and 2.2 × 10⁶, respectively in gp1ba^+/+, gp1ba^+/− and gp1ba^−/− fish (Figure 6). In this assay also, the heterozygotes and homozygotes showed equal red pixel intensities and were greater than the ones from the wildtype, suggesting a bleeding phenotype. These results are similar to the ones we observed in the laser thrombosis assays.

Figure 6:

Representative images of caudal vessel bleeding of the gp1ba^+/+, gp1ba^+/−, and gp1ba^−/− zebrafish. After allowing the fish to bleed for one minute, the total red pixel intensities were measured, and the intensities from gp1ba^+/+ fish were compared with those obtained from gp1ba^+/− and gp1ba^−/− fish by one-way ANOVA. * and *** indicates p values <0.05 and < 0.001. Error bars represent standard deviation. (N=10).

We also checked the number of thrombocytes by counting them after DiI labeling using flow cytometry because DiI, although it does not label all thrombocytes, it has been shown to label selectively only thrombocytes at a specific concentration ¹³. We found there were no differences in DiI-labeled thrombocyte counts in the blood, derived from gp1ba^+/+, gp1ba^+/− and gp1ba^−/− fish (Figure 7). Total thrombocytes could be counted separately by their size. However, occasionally other red cells and leukocytes that vesiculate in vitro could lead to errors in the total thrombocyte counts. Therefore, we used the DiI-labeled method, which measures only the thrombocyte population and is relatively more reliable.

Figure 7:

Dot plots showing the percentages of DiI labeled thrombocytes in whole blood of gp1ba^+/+, gp1ba^+/− and gp1ba^−/− fish, respectively. PE-A and SSC-A channels on the x-axis and y-axis show fluorescence and side scattering, respectively. Thrombocytes are shown in the Q2-UR (upper right quadrangle) gate and are colored red. The percentages of thrombocytes are shown in the respective gates. The large percentage of cells shown in Q2-UL (upper left quadrangle) gates represents other blood cells population. Gating was done according to the fluorescence intensities of DiI labeled thrombocytes. Approximately >10⁵ fluorescence intensity on the x-axis was considered positive thrombocyte fluorescence. gp1ba^+/+, gp1ba^+/− and gp1ba^−/− fish yielded similar percentages of DiI-labeled thrombocytes.

The results from the above data established that both gp1ba heterozygote and homozygote mutants had prolonged arterial TTO compared to wildtypes and consisted of macrothrombocytes. Also, there was a greater extent of bleeding in heterozygote and homozygote mutants than in the wildtypes. Likewise, in the ristocetin assay, there was a defective ristocetin-mediated agglutination of blood from the heterozygote and homozygote gp1ba mutants, compared to the wildtype. However, we did not find thrombocytopenia in either heterozygote or homozygote gp1ba mutants. Interestingly, platelet counts in Bernard Soulier Syndrome patients have been reported to considerably vary with a range from marginally low <30,000/µL to normal _˜200,000/µL of blood ³. Thus, our observation of lacking thrombocytopenia in the fish is consistent with the earlier observations in the patients. Therefore, our gp1ba mutants essentially mirrored the Bernard Soulier Syndrome phenotype. While our work was under progress, recently a publication reported that gpix mutation also produced Bernard Soulier Syndrome characteristics in zebrafish and the authors found that the fish had thrombocytopenia ¹⁴. Based on our findings and the other gpix report, it appears that there is a variation in the characteristics of Bernard Soulier Syndrome phenotype in zebrafish similar to what has been observed in human patients ¹⁴. Also, all the in vivo assays suggest a dominant mode of inheritance since both heterozygote and homozygotes have similar features in these assays. In one of the patients of type A1 Bernard Soulier Syndrome a nonsense mutation changing the TGG encoding Trp-343 to TGA corresponding to a termination codon ¹⁵. Our mutation changes the Lys-296 to the termination site. Thus, this nonsense mutation in zebrafish gp1ba is in the similar location relative to the human mutation; however, this human mutation results in an autosomal recessive disease. Interestingly, type A2 Bernard Soulier Syndrome has an autosomal dominant mode of inheritance ⁴. The above differences in lack of thrombocytopenia and the mode of inheritance seem to be the nature of Bernard Soulier Syndrome. Thus, Gp1ba is conserved in zebrafish and appears to have a similar hemostatic function as in humans, and the functional deficiency of Gp1ba in zebrafish seems to mimic the Bernard Soulier Syndrome patients ^16,17.

In conclusion, we have characterized a gp1ba mutant from ZIRC and obtained a phenotype that is similar to the classical Bernard Soulier Syndrome phenotype. This model, in combination with Von Willebrand Disease model could be used to develop therapeutic protocols by screening for novel drugs that could reverse the phenotype ^18,19.