Introduction

Hepatic stellate cells (HSCs) are mesenchymal cells in the liver that are critical for normal liver physiology and injury response (1). In their normal state within the liver, HSCs are quiescent and help maintain liver structure, regulate extracellular matrix (ECM) turnover, and store vitamin A. However, in response to liver injury, HSCs transition to an activated state through myofibroblast transdifferentiation. This activation triggers HSC proliferation and increased ECM secretion, resulting in scar formation at the injury site(s) and the production of cytokines and growth factors that promote the proliferation of other liver cell types. Prolonged HSC activation can lead to liver fibrosis and cirrhosis. Identifying mechanisms that impact HSC development can offer valuable insights into pathways involved in liver development, pathophysiology, and regeneration.

We and others previously identified podocalyxin-like (podxl) as a gene whose expression is highly enriched in adult HSCs in zebrafish (2, 3). In humans, PODXL belongs to the CD34 gene family that is comprised of PODXL, CD34, and ENDOGLYCAN (PODXL2)(4). It is a transmembrane protein recognized for its role in enhancing cell proliferation, invasion, and migration (5). Podxl null mice demonstrate severe defects in renal development accompanied by 100% perinatal mortality (6), and morpholino knockdown of podxl in zebrafish leads to disruption of the characteristic kidney podocyte architecture (7), supporting a conserved role for podxl in kidney development. However, the role of podxl in liver development has not been explored.

In the present study, we sought to define the role of podxl in zebrafish liver development using a combination of genetic approaches. This manuscript is based in part upon A.N.R.’s dissertation (8). We found that podxl knockdown using CRISPR/Cas9 or an ATG morpholino (7) resulted in a significant decrease in HSCs at 6 days post-fertilization (dpf). In contrast, podxl deletion mutants showed either no significant change in HSC number (three mutants) or a significant increase in HSC number (two mutants). Podxl mutants injected with podxl CRISPR/Cas9 knockdown, unlike wildtype control siblings, did not show a decrease in HSC number, supporting the hypothesis that decreased HSC number in podxl CRISPants is due to disruption of the podxl gene rather than to non-specific or off-target effects. Quantitative PCR analysis of podxl transcripts in adult mutant livers revealed mRNA and pre-mRNA levels similar to those of wildtype control siblings. CRISPR/Cas9 knockdown of the only other zebrafish podxl family member, endoglycan (endo), did not affect HSC number in wildtype zebrafish or podxl mutants. qPCR of endoglycan show similar expression levels between podxl mutants and wildtype control siblings. RNA sequencing of podxl mutants and wildtype control siblings showed no significant change in transcript levels of genes with sequence similarity to podxl, but it demonstrated upregulation of a complex network of ECM-related genes in response to podxl loss. Together, these data support a role for podxl in zebrafish HSC development and suggest that multiple genes are upregulated to compensate for genomic podxl loss.

Results

Knockdown of podxl in zebrafish results in fewer HSCs

As a first step to explore the role of podxl in zebrafish liver development, we knocked down podxl using CRISPR/Cas9 technology, injecting Cas9 protein and guide RNA targeting the first exon of podxl (sgRNA #1, S1 Table) into single-cell embryos (Fig 1). Knocking down genes in this manner using CRISPR/Cas9 (“CRISPant” larvae) typically results in mosaic animals with diverse indel mutations; mutagenesis rates of 75-99% have been reported in CRISPants (9), but some cells will have monoallelic and/or in-frame mutations that may not affect gene function. To visualize HSCs, we performed the knockdown in Tg(wt1b:eGFP) transgenic zebrafish that express eGFP in HSCs (10). As a control, we knocked down tyrosinase (tyr) in age-matched siblings. The tyr gene encodes an important enzyme in the melanin production pathway (11), and tyr knockdown is not predicted to impact liver development.

Figure 1.

We analyzed HSC number at 6 days post-fertilization (dpf) by performing confocal microscopy and manually counting all HSCs in each liver. We found that knockdown of podxl using CRISPR/Cas9 (CRISPant larvae) decreased the number of HSCs from 63 ± 19 (mean ± standard deviation) HSCs per liver in uninjected control siblings to 44 ± 11 HSCs per liver in podxl CRISPants (31% decrease; p < 0.05; Fig 1A, C, D). Knockdown of tyr had no significant impact on the number of HSCs (73 ± 20 HSCs per liver; Fig 1B, D). Knocking down podxl in a similar manner but targeting a different exon (exon #2, sgRNA #2, podxl KD #2) (S1 Table) decreased the number of HSCs by 30% (60 ± 20 HSCs per liver in uninjected control siblings versus 42 ± 7 HSCs per liver in podxl KD#2; p < 0.001; Fig 1E). Successful mutagenesis of podxl was confirmed using high-resolution melt analysis (HRMA) (S1 Fig), and tyr knockdown was confirmed by identifying decreased pigmentation using brightfield microscopy (S2 Fig).

To determine mutagenesis efficiency in our CRISPant larvae, we examined the podxl locus in DNA extracted from larvae that were injected with podxl sgRNA #1 and from uninjected control siblings. Using fluorescein-labeled primers and capillary electrophoresis, we determined the percent of allele mutation by calculating the fraction of mutant amplicon product that did not migrate like wildtype amplicon product, suggesting insertions, substitutions, and/or deletions (12). We found that 62% of podxl alleles in injected larvae showed mutations compared to 0% of uninjected controls (S3 Fig).

As a complementary approach to determine the effect of podxl knockdown on HSC development, we knocked down podxl using an ATG morpholino targeting podxl (S1 Table) (7). We injected each zebrafish embryo with 4.93 ng of morpholino, as using greater than 5 ng has been reported to decrease the specificity of this approach (13). We saw that podxl morphants had 47 ± 22 HSCs per liver, compared to 77 ± 21 HSCs per liver in siblings injected with control morpholino (56% decrease; p < 0.001; S4 Fig). The finding that multiple methods of podxl knockdown produced a similar decrease in HSC number supports the hypothesis that podxl is required for normal HSC development.

The podxl knockdown phenotype is not due to a developmental delay

We used two approaches to test the possibility that podxl knockdown decreased HSC number by delaying development. First, as liver size increases rapidly from 3 dpf to 5 dpf (14), a developmental delay could significantly decrease liver size at 6 dpf. We found that liver size was not significantly impacted by podxl CRISPant knockdown (Fig 2A) even when HSC count was significantly reduced (Fig 2B).

Figure 2.

Second, we examined HSC number in podxl CRISPant knockdown larvae at different time points. Some genetic manipulations that decrease liver size during larval development result, at least partially, from a transient developmental delay, such that mutant liver size “catches up” to wildtype liver size after hours or days (15). An example of this phenomenon is observed in wnt2bb mutant zebrafish, which have very small or absent livers at 50 hours post-fertilization; some animals developing normal functioning livers at later stages (15, 16).

As HSC numbers steadily increase from 3 to 8 dpf (2), a general developmental delay would be expected to decrease the number of HSCs at 6 dpf, rendering it similar to what is seen at earlier time points. If “catch-up” occurs, then differences in HSC number would be less pronounced at later time points. We knocked down podxl using sgRNA #1 and examined HSC number at 5 and 7 dpf in age-matched siblings. At 5 dpf we noted 67 ± 21 HSCs per liver in uninjected control larvae and 50 ± 16 HSCs per liver in podxl knockdown larvae (25% decrease; p < 0.05; Fig 2C). At 7 dpf we noted 78 ± 15 HSCs per liver in uninjected control larvae and 53 ± 18 HSCs per liver in podxl knockdown larvae (32% decrease; p < 0.001; Fig 2C).

The finding that there was not a more pronounced phenotype at 5 dpf than at 7 dpf suggests that podxl knockdown does not transiently delay development. This hypothesis is further supported by our finding that 7 dpf podxl knockdown larvae still tended to have fewer HSCs than 5 dpf uninjected larvae (p = 0.0502, ANOVA). Taken together, these results argue that there is not a significant developmental delay in the podxl CRISPant larvae.

The podxl knockdown phenotype is not specific to the Tg(wt1b:eGFP) reporter line

The Tg(wt1b:eGFP) transgenic reporter line used to visualize HSCs in the above-mentioned experiments expresses eGFP in HSCs and a few other cell types, including kidney cells (10). The transgenic reporter line Tg(hand2:eGFP) can also be used to visualize HSCs, along with lateral plate mesoderm and neural crest cells (2). To test the possibility that the effects of podxl knockdown are specific to the Tg(wt1b:eGFP) transgenic line, we knocked down podxl using podxl sgRNA #1 in Tg(hand2:eGFP) zebrafish. We found that HSC number at 6 dpf decreased from 84 ± 31 HSCs per liver in uninjected control siblings to 63 ± 26 HSCs per liver in podxl CRISPants (25% decrease; p<0.05; Fig 2D-F). This result indicates that podxl knockdown decreases HSC number independently of the reporter line used.

Generation of podxl mutants

To further understand the role of podxl in the developing zebrafish liver, we generated a series of podxl mutants, exploiting the Tg(wt1b:eGFP) reporter line to facilitate HSC visualization. Using CRISPR/Cas9 techniques, we made a mutant with a five base pair deletion, resulting in a predicted premature termination codon in the second exon (podxl ^Ex1,-5bpΔ) (Fig 3A, B and S5 Fig). We next made a mutant lacking the entire region between exon 1 and intron 7, leading to a predicted premature stop codon after 18 amino acids (podxl^Ex1(p)_Ex7Δ) (Fig 3C, S6 Fig and S7 Fig).

Fig 3.

Previous experiments examining genetic compensation in mutant zebrafish have shown that premature stop codons can trigger nonsense mediated decay (NMD), which produces degradation products that promote upregulation of gene(s) with similar sequence that compensate for the loss of the gene of interest (17, 18). This form of genetic compensation, termed transcriptional adaptation (17, 19), requires the degradation of mutant mRNA containing a premature termination codon (17, 20). To circumvent the possibility of genetic compensation induced by transcriptional adaptation, we made three additional podxl mutants. First, we made an in-frame mutant removing the region between introns 4 and 7 (podxl^Ex5_Ex7Δ) (Fig 3D and S8 Fig), corresponding to the transmembrane domain, which is critical for podxl localization and function (21, 22). The podxl^Ex5_Ex7Δ mRNA lacks a premature stop codon, so it is not predicted to trigger NMD and transcriptional adaptation. We also made two podxl mutants lacking the major transcriptional start site marked by H3K4me3 in several tissues including the liver (S9 Fig): podxl^{- 194_Ex7Δ} (Fig 3E, S10 Fig) and podxl^{-319_Ex1(p)Δ} (Fig 3F, S11 Fig). We predicted little podxl mRNA would be transcribed in these mutants, making NMD and transcriptional adaptation unlikely.

Analysis of HSC number in podxl mutants

We analyzed HSC number in podxl mutants by incrossing podxl^+/-heterozygotes and examining HSC number in podxl^-/- mutants and wild-type control siblings at 6 dpf. We observed no significant difference in HSC number between podxl ^Ex1,-5bpΔ mutants (68 ± 20) and wildtype siblings (62 ± 19) (Fig 3G), podxl^Ex1(p)_Ex7Δ mutants (82 ± 12) and wildtype siblings (76 ± 21) (Fig 3H), and podxl^Ex5_Ex7Δ mutants (56 ± 21) and wildtype siblings (56 ± 19) (Fig 3I)

Conversely, in the two promoter mutants, we observed a significant increase in the number of HSCs per liver. We observed a 46% increase in HSC number in the podxl^-194_Ex7Δ mutant (73 ± 11) compared to wildtype siblings (50 ± 13) (Fig 3J). In the podxl^{-319_Ex1(p)Δ} mutant we observed a 32% increase in HSC number (83 ± 20) compared to wildtype siblings (63 ± 18) (Fig 3K).

Podxl mutants are resistant to podxl CRISPant knockdown

The discrepant phenotypes in podxl knockdown larvae and podxl mutants raise two sets of possibilities. First, podxl is not important for HSC development, and the podxl knockdown phenotype is due to effects on another gene(s). Alternatively, podxl is important for HSC development, so acute podxl loss causes a decrease in HSCs, but other gene(s) compensate in the case of constitutive/germline podxl loss. We would expect that in the first scenario podxl CRISPR/Cas9 knockdown would decrease HSC number even in the absence of wildtype podxl, while in the second scenario podxl mutants would be resistant to podxl knockdown.

We distinguished between these two possibilities by knocking down podxl in a podxl^Ex1(p)_Ex7Δ heterozygous incross in the Tg(wt1b:eGFP) background. We chose this mutant for these experiments because it lacks the complete cut site for podxl sgRNA #1 (S12 Fig). We analyzed HSC number in podxl^-/- and podxl^+/+ 6-dpf larvae with and without podxl sgRNA #1 injection.

Consistent with our prior findings, the HSC numbers of podxl^+/+ larvae were significantly decreased in the group injected with podxl sgRNA compared to the uninjected controls. In contrast, we found no significant change in HSC number between the podxl^-/- groups with and without podxl KD injection (Fig 3L). These data support the hypothesis that podxl CRISPR/Cas9 knockdown requires an intact podxl gene to exert its effects on HSC development.

Maternal deposition of podxl does not compensate for podxl loss

Maternal expression and mRNA deposition of certain genes is sufficient to compensate for zygotic loss (23, 24). To address the possibility that maternally deposited mRNAs might compensate for podxl loss in mutants, we examined the mRNA-Seq models in the zebrafish ensemble database looking at both pre-zygotic and post-zygotic tracks. We found that podxl is not maternally expressed and expression is first detected at 75% epiboly. Additionally, we examined progeny from a podxl^Ex1(p)_Ex7Δ mutant incross. Homozygous mutant progeny showed no obvious developmental defects and had unremarkable HSC number and morphology (S13 Fig). Homozygous mutant progeny from the podxl^Ex1(p)_Ex7Δ mutant incross survived to adulthood and were fertile. These data suggest that normal HSC numbers are maintained in podxl mutants via a mechanism other than maternally deposited podxl mRNA.

Analysis of podxl transcript levels in podxl mutants

In some zebrafish mutants, transcriptional adaptation leads to upregulation of related genes, compensating for loss of the gene of interest and blunting the loss-of-function phenotype observed with transient gene knockdown adaptation (17, 19). When transcriptional adaptation occurs, pre-mRNA levels of the gene are close to normal, but mRNA levels are low.

To determine if transcriptional adaptation could be happening in podxl mutants, we performed qPCR to examine podxl mRNA and pre-mRNA in podxl mutant livers alongside wildtype control siblings. We assessed three different parts of the podxl mRNA: 5’ UTR, exon 7 to exon 8, and exon 8 (Fig 4). Additionally, we performed a qPCR on the following sections of podxl pre-mRNA: exon 2 to intron 2, intron 2 to exon 3, and intron 7 to exon 8 (Fig 5). Since some mutants lacked large portions of the podxl gene, using three different primer sets for each gene insured that at least one of the sets was predicted to have intact binding sites within each podxl mutant transcript (S2 table).

Fig 4.

Fig 5.

We found that podxl mRNA levels were not significantly different in podxl ^Ex1,-5bpΔ and podxl^Ex5_Ex7Δ mutants compared to wildtype sibling controls (Fig 4B and 4D). Podxl pre-mRNA levels were also not significantly different in these two mutants compared to controls (Fig 5B and 5D). These data suggest that podxl ^Ex1,-5bpΔand podxl^Ex5_Ex7Δ mutant pre-mRNA is transcribed normally and podxl ^Ex1,-5bpΔand podxl^Ex5_Ex7Δ mutant mRNA is not unstable. On the other hand, podxl mRNA levels were significantly reduced in podxl^Ex1(p)_Ex7Δ mutants (Fig 4C), supporting the hypothesis that mRNA transcription and/or mRNA stability may be decreased in these animals.

The promoter mutants podxl^-194_Ex7Δ and podxl^{-319_Ex1(p)Δ} showed very little podxl mRNA expression (Fig 4E and 4F), including the exon 8 region where the qPCR promoters are present in the mutants. The pre-mRNA levels of both promoter mutants were also significantly lower than those observed in wildtype control siblings (Fig 5E and 5F). This result supports the hypothesis that podxl transcription is reduced in podxl^-194_Ex7Δ and podxl^{-319_Ex1(p)Δ} mutants; podxl^-194_Ex7Δ and podxl^{-319_Ex1(p)Δ} mRNA may also be unstable.

Although at least two podxl mutants (podxl^Ex1(p)_Ex7Δ and podxl^-194_Ex7Δ) showed a significant reduction in podxl mRNA levels, at least some of this decrease might be related to decreased podxl transcription, given the decreased podxl pre-mRNA levels in podxl^-194_Ex7Δ and podxl^{-319_Ex1(p)Δ} . The finding that two podxl mutants (podxl ^Ex1,-5bpΔ and podxl^Ex5_Ex7Δ) had normal podxl mRNA levels argues that transcriptional adaptation triggered by NMD is not the sole mechanism by which podxl mutants compensate for podxl loss.

Upregulation of endoglycan does not compensate for podxl loss

In some cases of genetic compensation, genes in the same family compensate for the loss of one family member (20). In humans and mice, PODXL is part of the CD34 gene family that includes PODXL, CD34, and ENDOGLYCAN (4). Zebrafish have two CD34 family members: podxl and endoglycan (endo). To determine if endo compensates for podxl loss, we knocked it down using CRISPR/Cas9 in a podxl^Ex1(p)_Ex7Δ heterozygous incross (S1 Table). We confirmed cutting of endo using HRMA (S14 Fig). We found that there was no significant change in HSC number between the endo sgRNA injected podxl mutants and the uninjected control mutants (Fig 6A). This result suggests that endo is not compensating for the loss of podxl in podxl mutants.

Fig 6.

To further test the possibility that endo might compensate for podxl loss, we performed qPCR for endo on podxl mutant livers (S2 Table). Though we occasionally observed increased endo expression in some podxl mutants, overall there was no significant difference in endo expression between podxl mutants and wildtype sibling controls (Fig. 6B-F).

Response to liver injury is unaffected in podxl mutants

Some mutants appear unremarkable under normal physiologic conditions, but a phenotype is revealed in response to stress (25). To determine if podxl mutants might respond differently to liver injury, we subjected podxl^Ex1(p)_Ex7Δ and podxl^-194_Ex7Δ zebrafish larvae to incubation with 2% ethanol or to hepatocyte ablation with metronidazole/nitroreductase (26). In wildtype zebrafish, ethanol treatment increased HSC number as previously reported (2)(S16 Fig). Hepatocyte ablation decreased liver size as previously reported (26), and it also increased HSC density (S17 Fig). We found that HSC numbers and HSC density in podxl ^Ex1(p)_Ex7Δ and podxl ^-194_Ex7Δ mutants were similar to those of wildtype zebrafish in response to ethanol or hepatocyte ablation (S16 and S17 Fig), supporting the hypothesis that podxl is not required in HSCs in these contexts.

A network of extracellular matrix genes is upregulated in response to podxl loss

To characterize transcriptional changes in podxl mutant livers, we performed RNA-sequencing on adult livers dissected from three podxl^Ex1(p)_Ex7Δ mutants and wildtype control siblings. We identified 472 genes that were upregulated and 629 genes that were downregulated in podxl^Ex1(p)_Ex7Δ mutant livers (log2 fold change > 2 and adjusted p-value < 0.05). By performing GO Enrichment Analysis (27), we found that the upregulated genes were enriched for extracellular matrix, extracellular region, and peptidase activity genes (Table 1). Additionally, out of our top 100 most upregulated genes, 15 of those were genes with extracellular regions and three were extracellular matrix genes (S3 Table). Other significantly upregulated genes included thrombospondin 1a (thbs1a), an extracellular matrix protein (28), and ezrin a (ezra), a cytoskeleton linker protein that is known to interact with podxl (29). Several significantly upregulated genes have been previously reported to be enriched in zebrafish HSCs (2, 3, 30) (S4 Table). Together these data suggest that podxl mutants upregulate a network of extracellular matrix genes and other HSC-enriched genes to compensate for the loss of podxl.

Table 1.

We analyzed RNA sequencing data to further investigate the possibility of transcriptional adaptation in podxl mutants. If transcriptional adaptation were occurring, genes with similar sequences to podxl would be expected to be significantly upregulated in podxl mutants (17, 18). We used NCBI BLAST (31) to find genes that had either highly similar (S5 table) or somewhat similar (S6 table) sequences to the zebrafish podxl gene and analyzed the expression of these genes in podxl mutants compared to wildtype control siblings. We found that none of these genes with sequence similarity to podxl were significantly dysregulated, consistent with the hypothesis that transcriptional adaptation is not occurring in podxl^Ex1(p)_Ex7Δ mutants. Expression of the other zebrafish CD34 gene family member, endo, was not significantly different in podxl^Ex1(p)_Ex7Δ mutants compared to wildtype control siblings, being undetected in both groups.

We also examined RNA sequencing reads to confirm the podxl^Ex1(p)_Ex7Δ mutation. We found that podxl transcripts in the deleted region were not detected and podxl transcripts outside the deleted region were very low (S15 Fig), confirming the presence of the mutation and supporting the qPCR results.

Discussion

Here we report that podxl CRISPant knockdown in zebrafish significantly and robustly decreases HSC number, supporting a role for podxl in zebrafish liver development. We found that the decrease in HSC number seen in podxl CRISPant zebrafish was not due to a transient developmental delay or to an artifact of the transgenic reporter line. As podxl promotes migration and proliferation in colorectal cancer and spermatogonial stem cells (32, 33), it is possible that loss of podxl leads to decreased HSC migration into the liver and/or decreased proliferation of HSCs and/or HSC precursors.

We generated and characterized a series of podxl deletion mutants and found they had either no change in HSC number or a significant increase in HSC number, in sharp contrast to what we observed with acute podxl knockdown. Several lines of evidence argue that the discrepant phenotypes in podxl CRISPant knockdown zebrafish compared to podxl mutants are due to genetic compensation in podxl mutants rather than to an off-target effect of podxl knockdown. First, we noted a similar decrease in HSCs with podxl knockdown in eight independent CRISPant experiments using two different sgRNAs and performed by two different lab members. Second, podxl ATG-morpholino injection, which is expected to decrease Podxl levels by a distinct mechanism involving disruption of translation (7), also decreased HSC number. Third, podxl mutants were resistant to podxl knockdown.

Genetic compensation represents a highly conserved process that enables organisms such as fly, yeast, mice, and zebrafish to maintain their fitness despite losing an important gene (19, 34). Three major categories of genetic compensation have been described: redundant genes, protein feedback loops, and transcriptional adaptation (19). Redundant genes are expressed in the same tissues and have overlapping functions with the gene of interest, while protein feedback loops lead to the upregulation of genes in the same pathway in response to decreased protein levels (35, 36, 37). The finding that endo knockdown in podxl mutants did not impact HSC number argues against genetic compensation through redundant genes. The finding that HSC numbers were decreased in podxl knockdown zebrafish, which likely have decreased Podxl levels, suggests that protein feedback loops are not sufficient to compensate for podxl loss.

Transcriptional adaptation was first described in zebrafish, where Rossi et al. found that loss of the epidermal growth factor-like domain multiple 7 (egfl7) gene leads to severe vascular phenotypes in morphants though egfl7 mutants show no apparent phenotype. Further investigation revealed that this compensation is likely due to the upregulation of members of the emilin gene family, which share significant sequence similarity with egfl7 (17, 18). Like Rossi et al., we found that podxl mutants were resistant to podxl CRISPant knockdown, and three podxl mutants showed decreased mRNA levels, compatible with mRNA degradation. On the other hand, two podxl mutants showed mRNA levels that were not significantly decreased, suggesting that mRNA degradation is not necessary for genetic compensation in at least some types of podxl mutants. Additionally, while Rossi et al. found differential expression of only one gene in egfl7 mutants, our RNA sequencing data showed a plethora of differentially expressed genes in podxl mutants. Thus, it seems like genetic compensation in podxl mutants occurs through a mechanism that is distinct from the transcription adaptation that occurs in egfl7 mutants.

We observed an increase in HSC number in two podxl mutants, podxl^-194_Ex7Δ and podxl^{- 319_Ex1(p)Δ}, both of which disrupt exon 1 and the canonical promoter region of podxl. Genetic overcompensation has been reported in zebrafish with frameshift mutations in exon 1 of the BMP signaling-promoting gene marcksb: compared to wild-type control embryos, marcksb^ihb199 embryos exhibit increased Bmp2b fluorescence and increased p-Smad1/5/9 intensity, likely due to upregulation of MARCKS-family members and MARCKS-interacting protein Hsp70.3 (38). It is possible that disruption of the canonical promoter region in podxl^-194_Ex7Δ, podxl^{-319_Ex1(p)Δ}, and/or marcksb^ihb199 mutants leads to creation of an orphan enhancer that promotes transcription of nearby compensatory genes. In the human and mouse β-globin family, competition between promoters for a common enhancer, termed the locus control region (LCR), determines the developmental order and compensatory expression of related globin genes (39, 40). To further investigate the possibility that an orphan enhancer mediates the increase in HSC numbers observed in podxl^{- 194_Ex7Δ} and podxl^{-319_Ex1(p)Δ} mutants, it would be interesting to examine the long-range chromatin interactions of the residual promoter proximal DNA in the mutants by Hi-C (41) and/or by using a targeted approach such as 3C (42) aimed at the dysregulated genes we uncovered.

Egfl7 morphants display severe, highly penetrant vascular phenotypes (18). In contrast, the decrease in HSC number observed in podxl CRISPant zebrafish is not binary and instead falls on a spectrum, making statistical analysis necessary to define the phenotype(s) for each experimental group. Thus, although the podxl CRISPant phenotype is reproducible, its penetrance is variable, making it challenging to define a compensatory mechanism.

In a recent study of genetic compensation in slc25a46 mutant zebrafish, researchers found over a hundred dysregulated genes that did not have sequence similarity to slc25a46 (43). They did not observe mRNA degradation in slc25a46^238s mutants and speculated that the compensatory mechanism in their model may involve a genetic modifier or an orchestrated interaction among multiple genes within relevant networks. It is possible that a similar mechanism is at play in podxl^Ex1(p)_Ex7Δ mutants, as they also show dysregulation of hundreds of genes. Taken together, the findings in podxl and slc25a46 mutants provide support for a mechanism of genetic compensation that is distinct from redundant genes, protein feedback loops, and transcriptional adaptation.

Materials and methods

Zebrafish lines and husbandry

Zebrafish (Danio rerio) lines were maintained in compliance with the University of Utah Institutional Animal Care and Use Committee guidelines under standard conditions (44). In addition to male and female wildtype AB strain, the transgenic lines Tg(wt1b:eGFP), Tg(hand2:eGFP), and Tg(kdrl:mCherry) were also used. Embryos and larvae were cultured in egg water (2.33 gm Instant Ocean in 1 L Milli-Q water with 0.5 ml methylene blue) and stored in a 28.5°C incubator. Adult zebrafish were maintained on a system of recirculating water and fed powdered food, flakes, and brine shrimp. Animals were euthanized using tricaine methanesulfonate (0.03%) and/or immersion in ice water (rapid chilling).

SgRNA design and morpholino

Small guide RNAs (sgRNAs) were selected from the top targets identified by CHOPCHOP software (http://chopchop.cbu.uib.no/) with NGG PAM sites and no predicted off-targets. gRNAs were then ordered from IDT and annealed with tracerRNAs at a concentration of 3μM (S1 Table). Morpholino injections were performed consistent with published guidelines (13) with a well-characterized and validated ATG morpholino (7). The morpholino used was ordered from Gene Tools and each embryo was injected with approximately 4.93 ng of either standard control or podxl morpholino at the single-cell stage.

CRISPant knockdown experiments

To knockdown podxl, tyr, and endo, we injected single cell embryos with an injection mix that contained 0.25 µg/µL of Cas9 protein, 18 ng/µL podxl, tyr or endo crRNA, and 33.5 ng/µL tracrRNA. We assessed podxl and endo knockdown using high resolution melt analysis (HRMA). For podxl we used forward primer 5′–GACTGAACGCGGAGAATCTG–3′ and reverse primer 5′– CGTCGTATATGAAGGTAAACTCACC–3′. Podxl knockdown experiments were performed by two different lab members to validate the results. For endo we used forward primer 5′– GTGCCACGGGATCGAAA–3′ and reverse primer 5′– CCGTCCTCGCTGTCTTC–3′. For tyr, gene disruption was assessed by visually inspecting zebrafish larvae for loss of pigmentation.

Zebrafish embryo genomic DNA extraction

For initial screening and founder screening to check for the presence of podxl mutations, ten 2-day old embryos were pooled into a single well. They were digested for 2 hrs at 55 °C in 100 µl DNA digestion buffer (20 mM Tris pH 8, 50 mM KCl, 0.3% IGEPAL, 0.3% Tween20, 0.5 µg/μl Proteinase K). Proteinase K was then inactivated at 95 °C for 10 min. Samples were centrifuged at 2,500 rpm for 2 minutes; 1 µl was used in 12.5 µl PCR reactions.

Generating podxl mutants

We generated five podxl mutants using CRISPR/Cas9 technology. The putative podxl promoter region and major transcriptional start site were identified by epigenetic feature annotation (Supplementary Methods). We injected single-cell embryos with 0.25 µg/µL of Cas9 protein and 18 ng/µL podxl crRNA + 33.5 ng/µL tracrRNA (S6 Table). The injected (F0) zebrafish were screened for germline transmission by crossing with wildtype zebrafish and screening the F1 progeny for mutations using PCR (S8 Table). Temperature cycles were as follows: 98°C for 5′; 35 cycles of 95°C for 30′′, 58-68.3°C (S8 Table) for 30′′, 72°C for 35′′; 72°C for 5′; and 10°C hold. PCR samples were loaded into a 2% agarose gel in TAE.

Whole-mount Immunofluorescence

Larvae were fixed overnight in cold 4% PFA. After fixation, larvae were rinsed in cold PBS and the skin was removed to reveal the liver (45). To better visualize the eGFP expression in the Tg(wt1b:eGFP) and Tg(hand2:eGFP) zebrafish, larvae were blocked for at least 1 h with PBS+4% BSA+0.3% Triton X-100 (PBT) and then incubated with chick anti-GFP primary antibody (1:500, Aves Cat #GFP-1020, Lot #0511FP12) for at least 12 h followed by AlexaFluor goat anti-chick 488 secondary antibody (1:200, A11039, Lot #1937504) for at least 12 additional hours.

Confocal imaging and analysis of hepatic stellate cells

Larvae for all confocal imaging experiments were mounted on their back in 1% low-melt agarose plus SlowFade Diamond Antifade Mountant (Invitrogen) and cover-slipped. Image acquisition for all samples was done using an Olympus FV1000 confocal microscope using the same parameters (HV, gain, offset, etc.) for all zebrafish within an experiment. A 20x lens was used and 3.5-μm thick z stacks were collected imaging the entire liver. Images were blinded and randomized using R (45, 46) and the number of hepatic stellate cells was counted manually from Z-projections of all liver images using Fiji software (ImageJ) (47, 48).

Larval ethanol exposure and hepatocyte ablation

Two podxl mutants were selected for liver injury experiments; podxl^-194_Ex7Δ and podxl^Ex1(p)_Ex7Δ. Ethanol exposure was performed as previously described (2) on progeny of a podxl^+/- incross. Larvae were exposed to ethanol at 4 dpf by transferring them to 2% ethanol diluted in egg water. After 24 hours of incubating in ethanol, larvae were moved to fresh egg water at 5 dpf. Larvae were euthanized and fixed in PFA at 6 dpf. After genotyping, wildtype and mutant larvae were dissected to expose the liver and imaged by confocal microscopy.

Hepatocyte ablation was performed with a metronidazole (MTZ)-nitroreductase (NTR) ablation system (26). Larvae from an incross of podxl^+/- expressing Tg(wt1b:egfp), with one parent expressing Tg(fabp10:CFP-NTR), were exposed to either 10mM MTZ or DMSO from 3.5 dpf to 5 dpf. At 6 dpf larvae were euthanized and fixed in PFA. After genotyping, wildtype and mutant larvae were dissected to expose the liver and imaged by confocal microscopy.

RNA extraction

Total and pre-mRNA was isolated from adult zebrafish livers using the Qiagen RNeasy Mini kit. We euthanized adult zebrafish by ice-water immersion, dissected out their livers, added Qiagen RLT buffer, ground with a pestle, vortexed on ice for 20 minutes, and then followed the Qiagen RNeasy Mini kit protocol. DNAse I was used to remove all traces of DNA from the samples. RNA concentration and purity were quantified using NanoDrop.

Real-time quantitative PCR

cDNA was synthesized using SS3 Vilo Synthesis kit with OligoDt (Invitrogen) or Maxima First Strand cDNA Synthesis Kit (Thermofisher). Quantitative RT-PCR was performed using PerfeCTa Sybr Green FastMix (QuantaBio, Cat # 95072-012) with the primers listed in S8 Table. The PCR conditions were: 95°C for 3′; 45 cycles of 95°C for 15′′, 60°C for 45′′; melt curve analysis. The fold changes were calculated relative to Rpl13a expression (2, 49).

Three technical replicates (three wells) were performed for each sample and were used to generate standard deviations. Melt curves were examined for all wells, and only Ct values with distinct Tm peaks (valid melt curves) were included in the analysis. If one well lacked a valid melt curve, the other two wells were used for the analysis. If two or three wells lacked valid melt curves, that sample was considered not detected. P-values were derived using Welch’s t-test on the normalized mean expression for wildtype and mutant.

RNA sequencing

Total RNA was isolated from cell lines using DirectZol RNA Miniprep kit according to the manufacturer’s instructions (Zymo). The concentration and purity were quantified using NanoDrop. The quality of the preparation was considered adequate based on the lowest RIN value of 9.3. Library preparation, sequencing, and analysis were performed at the University of Utah High-Throughput Genomics Shared Resource using the Illumina TruSeq Stranded mRNA Library Prep (Illumina) and NovaSeq S4 Flow Cell 2 platform (Illumina) according to the manufacturer’s instructions.

CRISPR efficiency testing by fragment analysis

Injected and uninjected control larvae were euthanized and DNA was extracted using the protocol stated above. PCR was performed using FAM-labeled primers ordered from IDT (Forward primer 5′–/56-FAM/AGACGAAAAGCGGAACCGAG–3′; Reverse primer 5′–/56- FAM/CCTCGTCGTATATGAAGGTAAACTC–3′) and the KAPA2G HotStart PCR Kit (KAPA Biosystem, #KK5501). The PCR conditions were as follows: initial denaturation step at 95 °C for 3 min followed by 35 cycles at 95 C° for 12 sec (denaturation), 68 °C for 12 sec (extension), and 72 °C for 5 sec (elongation). The final elongation step was 72 °C for 10 min. Samples were analyzed as described previously (50) with support from the University of Utah DNA Sequencing Core Facility.

Quantifying larval liver size

Larvae were euthanized at 6 dpf using tricaine methanesulfonate (0.03%) and fixed in 4% paraformaldehyde (PFA) overnight at 4 °C with gentle rocking. Larvae were then rinsed out of PFA using cold 1X PBS and the skin surrounding the liver was removed with forceps (45). Larvae were mounted in 3% methyl cellulose and images of the left side of each liver were taken using a dissecting microscope on the lowest magnification. Images were blinded and liver size was quantified using FIJI/ImageJ (47).

Identifying podxl homologs, genes with sequence similarity to podxl, and HSC-enriched genes

To identify podxl homologs, we searched ensembl genome browser and UCSC genome browser for cd34 and endoglycan (podxl2). To look for genes that had sequence similarity to podxl, we used NCBI BLAS to query Danio rerio Reference RNA Sequences and either highly similar sequences or somewhat similar sequences. We examined all genes with a 5% query cover or higher.

We examined genes that were significantly upregulated in podxl mutants (log2fold change> 2, padj<0.05) alongside three publicly available datasets for HSC-enriched genes. Genes meeting criteria for HSC enrichment and/or upregulation in at least two of these four datasets were included in S4 Table. “Mutant” column indicates that the gene was upregulated with a log2fold change>2 and padj<0.05 in adult zebrafish podxl mutant livers compared to podxl wildtype livers (RNA sequencing experiment reported in this manuscript; data to be deposited in GEO upon manuscript acceptance). “Yin” column indicates that the gene was among the top 500 most upregulated (greatest log2fold change) in FACS-sorted HSCs from adult zebrafish livers compared to non-fluorescing other liver cell types (2). “Morrison” column indicates that the gene was upregulated with a log2fold change>2 and padj<0.05 in endothelial/HSC clusters upon single-cell RNA sequencing of wildtype zebrafish livers (30). “Spanjaard” column indicates that the gene was one of 50 genes denoted as characteristically expressed in HSCs in adult zebrafish (3).

Acknowledgements

We thank Jerry Kaplan for critical review of the manuscript, Director of Aquatics Carrie Barton and the entire Office of Comparative Medicine aquatic animal care team for outstanding zebrafish care and husbandry, the Cell Imaging Core at the University of Utah for use of equipment, and Xiang Wang for assistance in image acquisition. Research reported in this publication utilized the High-Throughput Genomics and Cancer Bioinformatics Shared Resource at Huntsman Cancer Institute at the University of Utah and was supported by the National Cancer Institute of the National Institutes of Health under Award Number P30CA042014. This work was supported by the American Cancer Society (RSG-22-014-01-CCB). K.J.E. was also funded by the National Cancer Institute (R01CA222570).

Additional information

Author contributions

A.N.R.: Data acquisition, analysis, and interpretation; N.M.M.: Data acquisition, analysis, and interpretation; S.M.K.: Data acquisition and analysis; J.H.: Data acquisition and analysis; M.T.: Data acquisition; A.S.: Data acquisition; S.A.B.: Data analysis and interpretation; K.J.E.: Conception, experimental design, funding, and data interpretation. The initial manuscript was drafted by A.N.R. and N.M.M. and reviewed critically for important intellectual content by all other authors.

Additional files

Supplementary Materials