osr1 couples intermediate mesoderm cell fate with temporal dynamics of vessel progenitor cell differentiation

Proper embryonic patterning depends on the establishment of progenitor territories with well-defined gene expression patterns (Briscoe and Small, 2015). For example, the medial-lateral axis of the vertebrate posterior mesoderm is divided into precise stripes of progenitor territories that give rise to various organs and cell types, including the kidneys, blood vessels and blood cells (Prummel et al., 2020). Kidney progenitors originate within a pair of bilateral territories called the intermediate mesoderm (IM) (Davidson et al., 2019; Dressler, 2009; Gerlach and Wingert, 2013). Across species, the dimensions of the IM are defined by the expression of conserved transcription factors, such as Lhx1/Lim1 and Pax2, that are required for its development (Carroll et al., 1999; Cirio et al., 2011; Torres et al., 1995; Tsang et al., 2000), but the mechanisms that establish boundaries between the IM and its neighboring territories remain poorly understood.

The zinc-finger transcription factor Osr1 is an intriguing candidate for playing a central role in IM boundary formation. Gene expression analyses and temporal fate mapping in amniotes demonstrated that Osr1 is expressed initially in the IM and the laterally adjacent mesoderm, which contains vessel progenitors, before its expression becomes restricted to kidney progenitors (James et al., 2006; Mugford et al., 2008). In zebrafish, osr1 is expressed in the posterior mesoderm, initially adjacent to the IM; later, a stripe of lateral vessel progenitors (LVPs) arises between the IM and the osr1-expressing territory (Mudumana et al., 2008; Perens et al., 2016). Mouse Osr1 knockout embryos have decreased Lhx1 and Pax2 expression during early stages of kidney development, thought to be due to increased apoptosis (Wang et al., 2005). In zebrafish, osr1 knockdown studies yielded varying conclusions: one study of osr1 morphants determined that osr1 is only required for maintenance of the pronephron lineage (Mudumana et al., 2008), while another indicated that osr1 was required for IM formation (Tena et al., 2007). Additionally, osr1 knockdown resulted in an expanded venous vasculature (Mudumana et al., 2008). As cell tracking experiments have shown that LVPs contribute to the cardinal vein (Kohli et al., 2013), it is interesting to consider whether osr1 may influence LVP development. In total, however, the functions of osr1 in the initial development of the IM and vessel progenitors remain unclear.

Considering the discrepancies often seen between morphants and mutants (Schulte-Merker and Stainier, 2014), we chose to augment previous osr1 morphant studies (Mudumana et al., 2008; Tena et al., 2007; Tomar et al., 2014) by analyzing a TALEN-generated osr1 mutation (Askary et al., 2017). The osr1 mutant phenotype demonstrated important roles of osr1 in promoting IM and pronephron differentiation and inhibiting premature LVP formation. Previously, we have found that osr1 and hand2, which encodes a bHLH transcription factor, are co-expressed in the most lateral territory of the posterior mesoderm and that hand2 promotes LVP development while inhibiting the lateral extent of IM formation (Perens et al., 2016). Each of these phenotypes was partially suppressed by mutation of osr1. Intriguingly, wild-type embryos displayed a striking reduction of osr1 expression immediately before LVP formation, and overexpression of osr1 inhibited LVP emergence while elevating IM formation. Together, our studies suggest a new model in which osr1 expression dynamics balance IM differentiation with the temporal emergence of neighboring vessel progenitors.

To enhance our understanding of osr1 function in the posterior mesoderm, we analyzed a TALEN-generated osr1 mutation (Askary et al., 2017). osr1^el593 is a 7 bp deletion leading to a frameshift; the predicted mutant Osr1 protein would contain its first 80 amino acids, followed by 45 missense amino acids, and would lack its zinc fingers (Fig. 1A). Thus, osr1^el593 is likely a strong loss-of-function allele.

Homozygous osr1 mutant embryos exhibited progressive pericardial and body edema (Fig. 1B,C) comparable with other zebrafish mutants with defects in pronephron development (Kroeger et al., 2017; Lun and Brand, 1998). Consistent with this, osr1 mutants displayed deficits in multiple pronephron segments. Most dramatically, markers of the glomerular precursors (wt1b, Fig. 1F) and glomerular precursors and neck region (pax2a, Fig. S1B) were absent. cdh17, which is normally expressed throughout the pronephron tubules (Fig. 1I), lacked expression at the proximal extent of the tubule, while the remaining tubule appeared thinner (Fig. 1J). Similarly, a marker of the proximal convoluted tubule segment (slc20a1a, Fig. 1M,N) was reduced in intensity, while a marker of the distal late segment (slc12a3, Fig. 1Q,R) revealed a slightly thinner expression pattern in the mutant. Thus, although previous morphant studies found that osr1 is required for only proximal tubule development (Mudumana et al., 2008), the osr1 mutant phenotype revealed that osr1 is required for development of the entire pronephron but has a higher impact on the development of the proximal territories.

The proximal deficiencies observed in osr1 mutants were reminiscent of the pronephron phenotypes previously shown to result from overexpression of hand2 (Perens et al., 2016). Additionally, pronephron size was increased in a hand2 null mutant (Perens et al., 2016) (Fig. 1G,K,O,S), and knockdown of osr1 function partially suppressed this hand2 mutant phenotype (Perens et al., 2016). Similarly, we found that mutation of osr1 partially suppressed the pronephron phenotypes in hand2 mutants (Fig. 1H,L,P,T; Fig. S1D). Confirmation of this genetic interaction during pronephron development raised the possibility that osr1, like hand2, regulates IM formation.

Previously, we found that hand2 constrains the size of the pronephron by repressing IM formation (Perens et al., 2016). We therefore investigated whether osr1 also regulates the extent of initial IM formation. We observed narrowed expression of lhx1a in the osr1 mutant IM, in comparison with the wild-type IM (Fig. 2A,B), and this aspect of the osr1 mutant phenotype could be rescued by injection of wild-type osr1 mRNA (Fig. S2). Quantification of the number of Pax2a⁺ cells revealed that the altered IM appearance in osr1 mutants reflected a significant decrease in the number of IM cells (Fig. 2I). Thus, although other studies reached varying conclusions regarding an early role for osr1 in IM formation (Drummond et al., 2020 preprint; Mudumana et al., 2008; Tena et al., 2007), our osr1 mutant phenotype indicated that osr1 is required for the initial generation of the full complement of IM cells. Furthermore, we found that loss of hand2 function partially suppressed the IM defects in osr1 mutants (Fig. 2D,H,I). Thus, osr1 and hand2 operate in antagonistic genetic pathways to regulate IM differentiation, raising the possibility that osr1, like hand2, may execute this function at the lateral border of the IM.

We investigated whether osr1, like hand2, might play a role in the development of the neighboring lateral vessel progenitors (LVPs). The LVPs normally arise at the lateral boundary between the IM and the hand2/osr1-expressing territory of the posterior mesoderm during a short, consistent time window, subsequent to the appearance of the earlier-arising vessel progenitors located medial to the IM (Kohli et al., 2013; Perens et al., 2016) (Fig. 3A,B,E). Surprisingly, we found that LVPs form prematurely in osr1 mutants. Specifically, while etv2-expressing LVPs rarely appeared prior to the 10-somite stage in wild-type embryos, some osr1 mutants exhibited etv2-expressing LVPs as early as the 6-somite stage, and most osr1 mutants possessed etv2-expressing LVPs by the 8-somite stage (Fig. 3C-E, Fig. S3). In contrast, the timing of the appearance of medial vessel progenitors was unaffected (Fig. S4). Thus, osr1 constrains vessel progenitor development by inhibiting the premature differentiation of LVPs at the lateral border of the IM.

Because osr1 and hand2 interact antagonistically during IM development, we examined whether the same genetic interaction occurs during vessel progenitor development. Notably, while hand2 mutants rarely form LVPs (Fig. 3H; 74% have no LVPs and 26% have one to three LVPs, n=31) (Perens et al., 2016), more etv2-expressing LVPs do form in hand2; osr1 double mutants (Fig. 3I; 22% have no LVPs, 56% have one to three LVPs and 22% have four to 10 LVPs, n=9). However, although hand2 mutants lack expression of flt4 and mrc1a within the cardinal vein (Perens et al., 2016), expression of these genes is not affected in osr1 mutants (Fig. S5E,H). Altogether, our data suggest that, as in the IM, osr1 and hand2 act in antagonistic genetic pathways to regulate LVP formation.

In addition to the IM and LVP phenotypes, we found that gata1-expressing blood progenitors were reduced throughout the posterior mesoderm in osr1 mutants (Fig. S6B). Unlike the IM and LVPs, however, blood progenitors did not seem altered in hand2 mutants (Fig. S6C), and the osr1 mutant phenotype did not appear to be suppressed by loss of hand2 (Fig. S6D). Thus, in addition to the antagonistic pathways through which osr1 and hand2 regulate IM and LVP formation, osr1 may function in additional genetic pathways that influence posterior mesoderm patterning.

Considering the dynamic nature of Osr1 expression in amniotes (James et al., 2006; Mugford et al., 2008), we surmised that osr1 expression in the zebrafish posterior mesoderm might also be dynamic. Indeed, unlike hand2, which has consistent expression in the posterior mesoderm from tailbud stage to the 10-somite stage (Fig. 3J-L), osr1 expression decreases dramatically during this same time period and before the emergence of the LVPs (Fig. 3M-O). Thus, decreased osr1 expression may be required for LVP emergence.

To test whether sustained osr1 expression would alter LVP development, we used the transgene Tg(hsp70:osr1-t2A-BFP) to overexpress osr1. Strikingly, induction of osr1 overexpression at tailbud stage could inhibit LVP formation (Fig. 4A-D, Fig. S7). Additionally, osr1 overexpression increased the formation of medial and proximal vessel progenitors (Fig. 4B,D, Figs S5C and S7), and caused increased and ectopic expression of vascular genes (Fig. S5F,I), suggesting that osr1 has distinct influences on different subsets of vessel progenitors. Interestingly, osr1 overexpression resulted in a range of severity for each of these vascular phenotypes, consistent with a dependence of vessel progenitor formation on precise osr1 expression levels (Fig. S7).

We next examined whether there was a concomitant change in the IM when LVP formation was suppressed by osr1 overexpression. Quantification of Pax2a⁺ cells showed a moderate, but significant, increase in the IM when osr1 is overexpressed (Fig. 4E-G). What might be the origin of these additional IM cells? In hand2 mutants that exhibit IM expansion and loss of LVPs, an increased number of Pax2a⁺ cells emerge within the hand2-expressing territory (Fig. S8C) (Perens et al., 2016). In contrast, in osr1-overexpressing embryos, Pax2a expression remained excluded from the hand2-expressing territory (Fig. S8B). Consistent with this difference, the increase in IM generated by osr1 overexpression was less than that generated by hand2 loss of function (Figs 2I and 4G) (Perens et al., 2016). Together, our findings suggest that elevated osr1 expression drives cells at the lateral IM border toward an IM fate, rather than a LVP fate. Like hand2 loss of function, osr1 overexpression can suppress LVP formation and increase IM production; however, unlike hand2 loss of function, elevated osr1 expression does not also convert the hand2-expressing lateral posterior mesoderm into Pax2a⁺ IM.

Because osr1 overexpression at tailbud was sufficient to increase IM formation, we wondered when osr1 function is normally required to promote IM and pronephron development. Other studies have implicated osr1 in mesendoderm development prior to gastrulation: notably, osr1 knockdown resulted in increased formation of endoderm progenitors in the early embryo, suggesting a possible influence of excess endoderm on IM development (Mudumana et al., 2008; Terashima et al., 2014). We found milder endoderm phenotypes in osr1 mutants (Fig. S9): we observed a trend toward a mild increase in endoderm progenitors at shield stage (Fig. S9A,B); however, unlike previous observations in osr1 morphants (Mudumana et al., 2008), we did not observe an increased amount of endoderm at the 18-somite stage (Fig. S9C,D). Because previous work suggested that osr1 acts during the earliest stages of endoderm differentiation to inhibit the formation of endoderm progenitors (Terashima et al., 2014), we chose to assess when induction of osr1 expression is able to rescue the osr1 mutant defects. Induction of osr1 expression at tailbud clearly rescued the osr1 mutant IM, podocyte and pronephron tubule defects (Fig. 4H-P). Thus, osr1 function after gastrulation is sufficient to regulate IM development, and osr1 function during earlier stages of mesendoderm development is not absolutely required for proper IM and pronephron formation. Conversely, induction of osr1 at the 10-somite stage failed to rescue the pronephron defects in osr1 mutants (Fig. 4P). Together, our analyses suggest a timeframe after the completion of gastrulation during which osr1 function is sufficient to promote the development of pronephron progenitors within the IM. Intriguingly, the timepoint at which osr1 induction was no longer able to rescue pronephron development coincides with the normal timing of LVP emergence.

Altogether, our studies provide new insights into the roles of osr1 in IM and vessel progenitor development. We show that osr1 is both necessary and sufficient to promote the initial differentiation of some, but not all, IM. In addition, we reveal context-dependent roles for osr1 in inhibiting vessel progenitor development, including an intriguing role in preventing the premature appearance of vessel progenitors at the lateral border of the IM. Finally, our findings suggest that the dynamic nature of osr1 expression in the posterior mesoderm is necessary for balancing the extent of IM formation with the timing of neighboring LVP emergence.

How might osr1 regulate both the IM and LVP lineages within the posterior mesoderm? Our findings indicate the presence of a unique territory at the boundary between the developing IM and the laterally adjacent mesoderm in which osr1 initially acts in opposition to hand2 to promote IM differentiation while inhibiting vessel progenitor differentiation; later, as osr1 expression recedes, IM differentiation ceases and vessel progenitors emerge. Conceptually, we envision that the dynamic levels of osr1 expression couple developmental timing with cell fate acquisition in order to set boundaries that delineate the extent of each progenitor territory. In future studies, it will be important to determine how directly or indirectly Osr1 and Hand2 regulate expression of the downstream genes that control IM and vessel identity, such as pax2a and etv2. Previous work suggested that Osr1 and its Drosophila homolog Odd function as transcriptional repressors (Goldstein et al., 2005; Tena et al., 2007). Considering the importance of reciprocal repressor interactions in establishing boundaries between neighboring progenitor territories in other developmental contexts (Briscoe and Small, 2015), it is interesting to speculate that, in the posterior mesoderm, a key function of Osr1 is to repress LVP formation while a primary role of Hand2 is to inhibit IM differentiation.

Our studies also suggest three subregions capable of contributing to the IM, arranged along the medial-lateral axis of zebrafish posterior mesoderm, with distinct genetic networks regulating IM formation in each area: a medial osr1-independent territory; a far lateral territory with latent IM-forming potential that is repressed by the sustained expression of hand2; and a boundary territory in between these in which a balance between hand2 and osr1 determines the precise amount and timing of IM and LVP formation. High-resolution lineage tracing will be necessary to delineate the precise fate map within the posterior mesoderm. Furthermore, it remains unknown whether hand2 and osr1 act cell-autonomously within the boundary territory to direct the fate of progenitors with the potential to contribute to the IM or LVP lineages. Considering the ectopic appearance of Pax2a within hand2-expressing cells in hand2 mutants (Fig. S8C; Perens et al., 2016), we hypothesize that hand2 acts cell-autonomously to regulate a decision between IM and LVP fates. Likewise, it is appealing to propose that osr1 acts in the same manner as hand2. In addition to being expressed in the same territory (Fig. 3J-O; Perens et al., 2016), osr1 and hand2 seem to function in the same timeframe: the stage after which induction of osr1 expression fails to rescue the osr1 mutant pronephron defects (Fig. 4P) coincides with the stage after which hand2 overexpression fails to inhibit pronephron development (Perens et al., 2016). Alternatively, instead of functioning autonomously within progenitor cells in the boundary territory, either hand2 or osr1 may influence IM and LVP cell fate decisions by controlling the production of diffusible signals that pattern the medial-lateral axis of the posterior mesoderm. Others have suggested that osr1 regulates the development of pronephron and vessel lineages non-autonomously, through its function during early endoderm development (Mudumana et al., 2008; Tomar et al., 2014). Our demonstration that induction of osr1 expression at tailbud can rescue the osr1 mutant phenotype argues against a mechanism in which osr1 regulates pronephron development by controlling the initial formation of endoderm progenitors; however, we cannot rule out a later role for osr1 in the endoderm. Ultimately, mosaic analysis will be necessary to determine where Osr1 and Hand2 function to shape the developmental potential of specific territories within the posterior mesoderm.

In the long term, an understanding of the impact of osr1 on medial-lateral patterning of the posterior mesoderm may have important implications for understanding congenital anomalies of the kidney and urinary tract (CAKUT), as OSR1 mutations have been associated with CAKUT phenotypes, including renal hypoplasia and vesicoureteral reflux (Fillion et al., 2017; Zhang et al., 2011). Additionally, because generation of IM is a key step in the production of vascularized kidney organoids (Takasato et al., 2015), in vivo analyses of osr1 function during IM and vessel progenitor formation have the potential to inform future refinements of relevant in vitro differentiation protocols.

We generated embryos by breeding wild-type zebrafish, zebrafish heterozygous for the osr1 mutant allele osr1^el593 (RRID: ZFIN_ZDB-ALT-171010-14) (Askary et al., 2017), zebrafish heterozygous for the hand2 mutant allele han^s6 (RRID: ZFIN_ZDB-GENO-071003-2) (Yelon et al., 2000), and zebrafish carrying Tg(hand2:egfp)^pd24 (RRID: ZFIN_ZDB-GENO-110128-35) (Kikuchi et al., 2011), Tg(hsp70:osr1-t2A-BFP)^sd63 or Tg(etv2:egfp)^ci1 (RRID: ZFIN_ZDB-GENO-110131-58) (Proulx et al., 2010). For induction of heat shock-regulated expression, embryos were placed at 37°C for 1 h and then returned to 28°C. All heat shocks were performed at tailbud, except for those described in Fig. 4P, which were performed at the 10-somite stage. Following heat shock, transgenic embryos were identified based on BFP fluorescence; embryos used for cell counting in Fig. 4E-G were confirmed to carry the transgene by PCR genotyping for the bfp-coding region, using the primers 5′-CTGGAAGGCAGAAACGACAT-3′ and 5′-TGCTAGGGAGGTCGCAGTAT-3′. Nontransgenic embryos were analyzed as controls. PCR genotyping was conducted as previously described for osr1^el593 mutants (Askary et al., 2017), for han^s6 mutants (Yelon et al., 2000) and for han^s6 mutants containing Tg(hand2:EGFP)^pd24 (Perens et al., 2016). For osr1^el593 mutants containing Tg(hsp70:osr1-t2A-BFP), we used a primer pair that would only amplify the endogenous osr1 locus, followed by digestion with EarI: 5′-AATGTTCTCTCTGTTTGTGTCTCC-3′ and 5′-AGGTTGGCAAAGTCAAAACG-3′. All zebrafish work followed protocols approved by the UCSD IACUC.

To generate transgenes for heat-activated overexpression of osr1, we first amplified the osr1-coding sequence from pCS2-osr1 (Mudumana et al., 2008), using the primers 5′-AAAAAAGCAGGCTGCCACCGATGGGTAGTAAGACGCTC-3′ and 5′-CTCCTCCGGACCCGCCGCCGTACTTTATCTTGGCTGGC-3′, and cloned the amplicon into the vector hsp70-BamHI-t2a-BFP at the BamHI restriction site. We employed standard protocols to create transgenic founders (Fisher et al., 2006). The F1 progeny of prospective founder fish were screened for BFP fluorescence following heat shock for 1 h at 37°C, and phenotypic analysis was performed on the F1 and F2 progeny of three separate founders carrying distinct integrations of Tg(hsp70:osr1-t2A-BFP). Similar phenotypes were observed in all three transgenic lines; data shown in Figs 4, S5, S7 and S8 depict results from the line Tg(hsp70:osr1-t2A-BFP)^sd63.

We synthesized capped mRNA from a pCS2-osr1 plasmid using the Ambion mMESSAGE mMACHINE kit, and injected 64 ng into embryos at the one-cell stage. To knock down sox32 function, we injected 3.4 ng of a previously characterized translation-blocking sox32 morpholino at the one-cell stage (Dickmeis et al., 2001).

Standard whole-mount in situ hybridization were performed as previously described (Thomas et al., 2008) using the following probes: atp1a1a.4 (ZDB-GENE-001212-4), cdh17 (ZDB-GENE-030910-3), etv2 (etsrp; ZDB-GENE- 050622-14), flk1 (kdrl; ZDB-GENE-000705-1), flt4 (ZDB-GENE-980526-326), foxa2 (ZDB-GENE-980526-404), gata1 (ZDB-GENE-980536-268), hand2 (ZDB-GENE- 000511-1), lhx1a (lim1; ZDB-GENE-980526-347), mrc1a (ZDB-GENE-090915-4), osr1 (ZDB-GENE- 070321-1), pax2a (ZDB-GENE-990415-8), slc12a3 (ZDB-GENE-030131-9505), slc20a1a (ZDB-GENE-040426-2217), sox17 (ZDB-GENE-991213-1) and wt1b (ZDB-GENE-050420-319).

Whole-mount immunofluorescence was performed as previously described (Cooke et al., 2005), using polyclonal antibodies against Pax2a at 1:100 dilution (Genetex, GTX128127) (RRID: AB_2630322) and against GFP at 1:250 dilution (Life Technologies, A10262) (RRID: AB_2534023), together with the secondary antibodies goat anti-chick Alexa Fluor 488 (Life Technologies, A11039) (RRID: AB_2534096) and goat anti-rabbit Alexa Fluor 594 (Life Technologies, A11012) (RRID: AB_10562717), both at 1:100 dilution. Samples were then placed in SlowFade Gold anti-fade reagent (Life Technologies) and mounted in 50% glycerol.

Bright-field images were captured with a Zeiss Axiocam on a Zeiss Axiozoom microscope and processed using Zeiss AxioVision. Confocal images were collected by a Leica SP5 or SP8 confocal laser-scanning microscope using a 10× dry objective and a slice thickness of 1 µm, and analyzed using Imaris software (Bitplane).

To count Pax2a⁺ cells, we flat-mounted and imaged embryos after dissecting away the yolk and the anterior region of the embryo. We examined a representative 250 µm long region in roughly the middle of the IM, selecting contiguous regions that were unaffected by dissection artifacts. When the quality of the dissection allowed both the right and left sides of the embryo to be counted, we counted Pax2a⁺ cells on both sides and represented the embryo by the average of the two values; otherwise, only one side was counted. In all cases, Pax2a⁺ cells were identified through examination of both three-dimensional reconstructions and individual optical sections. To differentiate Pax2a⁺ cells from background immunofluorescence, we used Imaris to decrease brightness until staining clearly outside of IM was no longer visible.

Statistical analysis was performed using Graphpad Prism 8 to conduct non-parametric Mann–Whitney U-tests when data involved a continuous variable. Fisher's exact test was used when data involved categorical variables. All results represent at least two independent experiments (technical replicates) in which multiple embryos, from multiple independent matings, were analyzed (biological replicates). For wild-type and mutant in situ hybridization results for which the phenotype was not quantified, phenotypes shown are representative examples from at least 10 embryos for wild-type and osr1 mutant phenotypes, and from at least five embryos for hand2 mutant and hand2; osr1 double mutant phenotypes. For wild-type, transgenic and mutant antibody staining results for which the phenotype was not quantified (Fig. 4C,D, Fig. S8), phenotypes shown are representative examples from at least five embryos.

We thank members of the Yelon lab for valuable discussions, A. Houk for providing the hsp70-BamHI-t2A-BFP plasmid, H. Kwan for generating the hsp70:osr1-t2A-BFP plasmid and T. Sanchez and the UCSD Animal Care Program for zebrafish care.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.198408