Anp32b Deficiency Suppresses Ocular Development by Repression of Pax6

The development of vertebrate eyes is a complex process, proceeding through the specification of the anterior neural plate, invagination of the optic vesicle, and formation of the lens and retina. At embryonic day 13.5 (E13.5), all basic ocular tissues, including the lens, cornea, retina, and optic nerve, have already formed. Lens fiber cells lose their organelles and form terminally differentiated fiber cells near the time of birth in mice. The cornea, retina, and optic nerve continue to differentiate until E15.5 or even E17.5 and become fully functional only after birth [1]. A common genetic network underlies in mammalian ocular development, which is characterized by reciprocal interactions and multiple signaling pathways [2, 3]. Over the past decade, researchers identified several master control genes such as PAX6, SOX2, and RAX, etc. in the initial stages of ocular development and elucidated that the disruption of these key factors leads to developmental ocular defects [2].

Developmental ocular disorders including microphthalmia and anophthalmia influence one in 18,500 live human births [2, 4]. Approximately 80 percent of microphthalmia cases are associated with mutations in OTX2, PAX6, RARB, RAX, SOX2, or VSX2, most of which encode transcription factors [2, 4, 5]. While the downstream regulation of these transcription factors is widely characterized in various developing tissues, the effect of upstream regulatory mechanisms on the fate of their transcriptional activity remains less clear.

ANP32B, belongs to the acidic leucine-rich nuclear phosphoprotein 32 kDa (ANP32) family, with a leucine-rich repeat domain (N-terminal) and a low-complexity acidic region (C-terminal) [6]. The ANP32 family is implicated in an extensive scope of physiological processes including mRNA transport, chromatin regulation, protein phosphatase inhibition, and apoptotic caspase regulation [7]. ANP32B is considered to be the most vital member for embryogenesis in the ANP32 family. Anp32a-deficient or Anp32e-deficient mice and the combined deletion of Anp32a and Anp32e do not exert any obvious impact on wellbeing in a congenic C57BL/6 background [8, 9]. However, Anp32b-deficient mice display strain-specific phenotypes, with a partial perinatal lethality in a mixed genetic background of 129:C57BL/6, and complete perinatal lethality in a pure C57BL/6 background. Moreover, because of defects in various organ systems, surviving Anp32b-deficient mice exhibit decreased viability [10]. By knocking out the Anp32b allele in the genetic background of BALB/c or FVB/N, BALB/c-derived Anp32b^−/− mice exhibited high survival rates approaching Mendelian ratios whereas FVB/N-congenic Anp32b^−/− mice were rarely viable [11]. Using BALB/c-derived Anp32b^−/− mice, phenotypes of organ developmental abnormalities were identified to be associated with Anp32b deficiency [12]. Moreover, we indicated that Anp32b has a vital effect on the maintenance of stem cells through suppressing the activity of p53 in normal and chronic myeloid leukemia mice and hinders B-acute lymphoblastic leukemia by enhancing PU.1 activity in hematopoietic-specific Anp32b-knockout mice [13, 14].

Based on examining the pure C57BL/6-derived Anp32b^−/− mice, the defects killing Anp32b^−/− mice occur perinatally, and Anp32b^−/− embryos at E17.5 exhibit sporadic physiological abnormalities including palate closure defects, or large hematomas in the liver [10]. By contrast, we identified that the deletion of Anp32b resulted in severe defects in ocular development at the embryonic stage. Hence, we assumed that Anp32b plays a vital role in ocular development and explored its molecular mechanisms. Our findings may provide a scientific basis for Anp32b as a viable therapeutic target for ocular developmental defects.

HEK293T (purchased from American Type Culture Collection (ATCC): CRL-3216), HCT116 (purchased from ATCC: CCL-247), and SRA01/04 (purchased from Riken Cell Bank: RCB1591) cells were kept in Dulbecco’s Modified Eagle Medium. HLE-B3 (purchased from ATCC: CRL-3603) cells were cultivated in Eagle’s Minimum Essential Medium.

Male BALB/c mice with the Anp32b-null allele were a gift from Prof. Tak Wah Mak. To generate pure C57BL/6-derived Anp32b^+/− mice, the Anp32b-null allele from BALB/C mice was backcrossed into C57BL/6 background mice for at least ten generations. Interbreeding of Anp32b^+/− mice produced Anp32b^−/− embryos (male and female).

The Servicebio company conducted hematoxylin and eosin (H&E) and immunohistochemical staining using embryos in 4% paraformaldehyde.

RNA sequencing was performed using BGI. The reference genome was GRCm38 with HISAT2 (v2.2.5). In addition, differentially expressed genes (DEGs) were conducted with q < 0.05 and |log₂FC|≥1 as a baseline.

To identify enriched pathways, gene ontology (GO) analysis of DEGs was conducted using DAVID bioinformatics resource (https://david.ncifcrf.gov/). Ingenuity pathway analysis was employed to identify “upstream transcriptional regulators” among the DEGs based on QIAGEN. For gene set enrichment analysis (GSEA), RNA-seq data were analyzed using gene sets from the MSigDB bioinformatics resource (https://www.gsea-msigdb.org/gsea/msigdb/).

Immunoprecipitation (IP) of Flag-tagged or endogenous proteins utilized to identify interacting proteins was performed in line with the previous description [13, 14]. Antibodies for IP are listed in online supplementary Table S4 (for all online suppl. material, see https://doi.org/10.1159/000542447).

Recombinant SUMO-tagged ANP32B and PAX6 proteins were purified using Ni-NTA µSphere affinity chromatography. Recombinant GST, GST-tagged ANP32B, and GST-tagged PAX6 were purified using GST Beads (Beyotime Biotechnology). GST, GST-tagged ANP32B, or GST-tagged PAX6 were separately incubated with His-PAX6 or ANP32B in an incubation buffer (0.1% BSA, 0.3% Tween, 1% PMSF in PBS), followed by pull-down using GST beads.

The sgRNA targeting ANP32B was cloned into a pLenti-U6-gRNA vector. Online supplementary Table S4 lists the specific target sequences. In addition, HCT116 cells were infected with a virus, and ANP32B-knockout clones were isolated. Following the manufacturer’s instructions, relative luciferase activity was identified using a Dual-Luciferase Reporter Assay System Kit (Promega, E1910).

The shRNA targeting human ANP32B or a control and overexpression Flag-PAX6 or empty vector lentiviral supernatant were applied, respectively, to infect lens epithelial cells, which included SRA01/04 and HLE-B3. After infection, cell proliferation was evaluated using a cell counting kit 8 (purchased from Dojindo) according to the manufacturer’s instructions.

A two-tailed unpaired Student’s t test was used to compare specific pairs within multiple groups or between two experimental groups. Error bars indicate the mean ± SD. Two-way ANOVA was employed to compare the proliferation curves. Prism 8 and SPSS 20.0 were applied to conduct all statistical analyses. Differences were considered statistically significant at p < 0.05.

Although the pure C57BL/6-derived Anp32b^−/− mice show a fully penetrant perinatal lethality, the physiological abnormalities of Anp32b^−/− embryos are largely unknown [10]. In this study, we utilized an Anp32b constitutive knockout mouse model in a C57BL/6 background (shown in online suppl. Fig. S1A) and demonstrated that Anp32b was not detected at either the RNA or protein level in embryonic tissues (shown in online suppl. Fig. S1B, C). Despite previous findings indicating perinatal lethality in constitutive Anp32b-knockout mice [10], Anp32b^−/− embryos at E13.5 and E17.5 were observed at an approximately Mendelian frequency (online suppl. Table S1). Surprisingly, over 40% of Anp32b^−/− embryos exhibited ocular developmental defects, representing a significant difference compared to Anp32b^+/+ and Anp32b^+/− embryos (shown in Fig. 1a; online suppl. Table S2). Further analysis revealed that Anp32b^−/− embryos with eye malformations exhibited either unilateral or bilateral eye defects, ranging from microphthalmia (reduced eye size) to anophthalmia (absence of an eye). The proportion of eye malformations accounted for approximately 40% both at E13.5 and E17.5 (shown in Fig. 1b; online suppl. Table S3).

We then examined the eye phenotypes at different stages of embryonic development. Although most of Anp32b^+/+ embryos exhibited normal ocular morphology at E13.5 and E17.5 as examined (shown in Fig. 1c A–D), the Anp32b^−/− embryos exhibited overt microphthalmia (shown in Fig. 1c E–H) or absent/rudimentary ocular tissue owing to arrested ocular development phenotypes (shown in Fig. 1c I–L). Interestingly, most Anp32b^−/− embryos with severely small eyes exhibited hypoplasia or even the absence of a lens, while retinal tissue was visible (shown in Fig. 1c I–L). Further structural changes in Anp32b-deficient embryonic eyes at E13.5 and E17.5 were characterized through morphological and histological analyses (shown in Fig. 1d). Most Anp32b^+/+ embryos developed all ocular tissues and did not reveal ocular abnormalities (shown in Fig. 1d A, B), while various pathological changes were observed in Anp32b^−/− embryos. The lens showed a size reduction (shown in Fig. 1d C, D) which was completely absent in severe cases (shown in Fig. 1d E, F). Numerous epithelial cells remained in the posterior part of the lens (shown in Fig. 1d C, black arrowhead), which elongated into fiber cells (shown in Fig. 1d A, black arrowhead) in the control group. Worse still, it seemed that the lens vesicle failed to separate from the ectoderm in the early stage, causing the lens to attach to the cornea or even completely disappear, and there were only optic cup-like structures left. Additionally, the size of the anterior chamber was notably reduced, and there was a slight increase in the density of blood vessels in the posterior chamber, which is known as persistent hyperplastic primary vitreous (shown in Fig. 1d E–F). Based on the obtained data, the deletion of Anp32b contributes to severe ocular developmental defects.

Anp32b deficiency significantly increased the occurrence of eye malformations at E13.5 and E17.5, indicating its vital role in ocular development. Therefore, we aimed to explore the mechanisms through which Anp32b functions in this process. RNA-seq was employed to assay alterations in the eye tissue transcriptome of E17.5 embryos induced by Anp32b deletion. Principle component analysis indicated significant clustering of biological replicates (shown in online suppl. Fig. S2). A total of 757 DEGs, comprising 501 upregulated and 256 downregulated genes, were determined in Anp32b^−/− when compared with Anp32b^+/+ eye tissues (shown in Fig. 2a). Among the DEGs of Anp32b^−/− embryos, several biological processes of GO terms related to ocular and lens morphogenesis were enriched, including “lens development in camera-type eye,” “visual perception,” and “eye development” (shown in Fig. 2b). In parallel, a molecular function term, “structural constituent of eye lens”, was found to be the top 1 in addition to transcriptional regulation in general (shown in Fig. 2b). Obviously, Anp32b showed consistently high expression in lens epithelial cells around the anterior pole and germinative zone of Anp32b^+/+ embryos with Anp32b^−/− embryos as a negative control (shown in online suppl. Fig. S3, black arrowhead). Considering that Anp32b mainly influences lens development and has been indicated to modulate the activity of transcription factors through binding them [13, 14], this study compared 199 upstream regulators obtained from ingenuity pathway analysis with 5,443 annotations in the top 1 GO pathway term (Go0002088: lens development in camera-type eye), finding an overlap of nine candidate transcriptional regulators (shown in Fig. 2c; online suppl. Table S5). We performed GSEA of candidate genes and found that Pax6 is a key regulator of lens development [15]. Its target genes were most significantly (p = 0.033) enriched in the transcriptome of Anp32b^−/− embryos and showed the highest enrichment score (NES = −1.363) (shown in Fig. 2d; online suppl. Fig. S4), indicating that Anp32b positively regulated the transcriptional activity of Pax6.

This study continued to verify the association between ANP32B and PAX6. Figure 3a, b showed that Flag-tagged ANP32B immunoprecipitated endogenous PAX6 in 293T cells, and vice versa. Using an anti-ANP32B antibody for endogenous co-IP, ANP32B could immunoprecipitate endogenous PAX6 in 293T cells (shown in Fig. 3c). Moreover, it demonstrated that either GST-ANP32B pulled down His-PAX6 or GST-PAX6 pulled down ANP32B through an in vitro GST pull-down assay (shown in Fig. 3d, e), which could support a direct interaction between ANP32B and PAX6. To map the domains required for the interactions, Flag-tagged PAX6 truncations, as shown in Figure 3f, were transfected into 293T cells and conducted co-IP with Flag-M2 beads. As shown in Figure 3g, ANP32B was pulled down by full-length and paired domain but not the 132–422aa fragment of PAX6. In addition, the 1–163aa fragment of ANP32B made no interactions with PAX6 (shown in Fig. 3h). According to the data obtained, ANP32B interacts with the paired domain of PAX6 through its C-terminal domain.

These observations contribute to the exploration of how Anp32b regulates the function of Pax6. As shown in Figure 3i, j, the absence of Anp32b did not alter the mRNA and protein levels of Pax6. Consistent with previous GSEA analysis (shown in Fig. 2d), the mRNA levels of well-known Pax6-target genes including Maf, Cryaa, Crybb1 [16], were downregulated in Anp32b-deficient eye tissues (shown in Fig. 3j). This indicated that Anp32b might exert a positive role in regulating the transcriptional activity of Pax6. Consistently, a luciferase assay driven by a specific PAX6-responsive element indicated that overexpression of ANP32B enhanced PAX6 transcriptional activity in a dose-dependent manner, whereas ANP32B knockdown diminished PAX6 transcriptional activity in HCT116 cells. Clearly, reexpression of ANP32B in ANP32B-knockout HCT116 cells completely rescued the defects in the transcriptional activity of PAX6 (shown in Fig. 3k, l). These data indicated that ANP32B did not influence the expression of PAX6 but acted as a transcriptional activator of PAX6.

Numerous studies have indicated a critically positive regulatory role for Pax6 in ocular development [15]. To demonstrate the requirement for Pax6 signaling in Anp32b-mediated ocular development, we first performed endogenous co-IP using anti-ANP32B antibody and observed that ANP32B immunoprecipitated endogenous PAX6 in SRA01/04 and HLE-B3 cells (shown in Fig. 4a; online suppl. Fig. S5A). SRA01/04 and HLE-B3 cells were stably co-transfected with either shNC, shANP32B#1, or shANP32B#2, along with an empty vector or Flag-PAX6 (shown in Fig. 4b; online suppl. Fig. S5B). As expected, cell proliferation in SRA01/04 and HLE-B3 cells was significantly inhibited by the knockdown of ANP32B, whereas it was promoted by PAX6 overexpression. Overexpression of PAX6 partially but significantly reversed the inhibition of proliferation observed in ANP32B knockdown cells, which was slightly inferior to that observed in ANP32B wild-type cells (shown in Fig. 4c; online suppl. Fig. S5C). Similarly, the mRNA levels of downregulated PAX6-activated genes such as MAF, CRYAA, and CRYBB1 [16] in ANP32B knockdown cells were rescued in PAX6 overexpression ANP32B knockdown cells (shown in Fig. 4d; online suppl. Fig. S5D). These data demonstrate that the regulation of ocular development by ANP32B is mediated through PAX6 signaling.

Anp32b is vital for mammalian development and knocking out Anp32b in a pure C57BL/6 background results in perinatal lethality [10]. During crossbreeding of Anp32b^+/− mice, it was unexpectedly observed that Anp32b^−/− embryos displayed a significantly higher incidence of ocular malformation than Anp32b^+/+ embryos. Our findings suggest that Anp32b deficiency mainly impairs lens formation during ocular development. The consistently high expression of Anp32b in the lens may be one reason why the deletion of Anp32b can specifically damage the lens. Hypoplasia of the optic cup and vesicle in the earlier embryonic stage can lead to failure of lens induction and retinal pigmentation [17]. Considering that we only detected ocular malformations at E17.5 and E13.5, whether Anp32b deficiency may affect the formation of the optic cup and vesicle in initial embryos needs further investigation.

As a histone chaperone, Anp32b interacts with different transcription factors and modulates their activity. This modulation depends on several factors, including tissue types, developmental stages, and downstream signaling pathways [13, 14, 18]. RNA-seq indicated that the expression of lens development-associated genes was significantly disrupted, especially Pax6-regulated genes. Interestingly, Anp32b not only physically interacts with Pax6 but also enhances its transcriptional activity. Pax6 exerts a key role in ocular development by orchestrating myriad processes [15]. In mice, heterozygous mutants in Pax6 lead to small eyes, but homozygous mutants have only remnants of ocular tissues and die shortly after birth because of nasal dysfunction [19]. As a master eye regulator, Pax6 has been reported to regulate many target genes involved in the formation of the lens and cornea, such as Maf, etc. [20]. Additionally, PAX6 is indispensable for the differentiation of lens epithelial cells [21]. In this study, a failure of lens epithelial cells to properly differentiate into lens fiber cells is observed in some Anp32b^−/− mice. Moreover, we indicated that overexpression of PAX6 promoted the proliferation of lens epithelial cells, consistent with its role in maintaining self-renewal properties and regeneration of ocular structures [22, 23]. In lens epithelial cells of which ANP32B was knocked down, overexpression of PAX6 partially but significantly reversed the inhibition of proliferation resulting from ANP32B knockdown. Exactly, most of the experiments were done by in vitro assay in cell cultures, the mechanisms of how ANP32B regulates PAX6 in vivo, specifically in the lens epithelium need more tests to clarify.

Microphthalmia or coloboma has complex etiology with chromosomal, monogenic, and environmental causes identified. SOX2 and PAX6 mutations may act by causing lens induction failure, while OTX2, CHX10, and RAX may result in retinal differentiation failure [24]. Here, nine candidate Anp32b-interacting proteins were identified in this study. In addition to Pax6, other candidate factors including Ctnnb1 [25], Tfap2a [26], Smad3 [27], Atf4 [28], and Hsf1 [29] have been previously recognized for their critical roles in ocular development. Whether these transcription factors are involved in ocular malformation mediated by Anp32b requires further investigation. Besides, infants with microphthalmia or coloboma have often other associated congenital anomalies, especially musculoskeletal, cardiac, and central nervous system anomalies [30]. Interestingly, Anp32b^−/− mice exhibit sporadic physiological abnormalities and occur perinatal lethality. Whether there is a common biological basis or mechanisms among these deserves further exploration.

In summary, for the first time, our study reveals the crucial regulatory role of Anp32b in orchestrating ocular development, particularly lens development. Furthermore, this study provides the first evidence that ANP32B interacts with PAX6 and improves its transcriptional activity, thereby promoting the proliferation of lens cells. Further exploration of the molecular mechanisms underlying Anp32b-mediated regulation of ocular development may potentially provide novel therapeutic strategies for treating ocular disorders.

All the animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at Shanghai Jiao Tong University School of Medicine (SJTU-SM) as the Reference No. (JUMC2023-121-A) of the ethical approval(s).

The authors have no conflicts of interest to declare.

This study was supported by National Natural Science Foundation (91853206, 82270156).

Yu-Sheng Wei and Hao-Ran Liu: conceptualization, methodology, software, validation, formal analysis, investigation, resources, data curation, writing – original draft, and visualization. Qian Yang: methodology, validation, investigation, and resources. Zhe Zhi: methodology and resources. Yun Yu: writing – review and editing, supervision, funding acquisition, and project administration.

Yu-Sheng Wei and Hao-Ran Liu contributed equally to this work.

The data supporting the findings are openly accessible in Gene Expression Omnibus databases (No: GSE269101).