RFX2 Is a Major Transcriptional Regulator of Spermiogenesis

Abstract

Spermatogenesis consists broadly of three phases: proliferation of diploid germ cells, meiosis, and finally extensive differentiation of the haploid cells into effective delivery vehicles for the paternal genome. Despite detailed characterization of many haploid developmental steps leading to sperm, only fragmentary information exists on the control of gene expression underlying these processes. Here we report that the RFX2 transcription factor is a master regulator of genes required for the haploid phase. A targeted mutation of Rfx2 was created in mice. Rfx2-/- mice are perfectly viable but show complete male sterility. Spermatogenesis appears to progress unperturbed through meiosis. However, haploid cells undergo a complete arrest in spermatid development just prior to spermatid elongation. Arrested cells show altered Golgi apparatus organization, leading to a deficit in the generation of a spreading acrosomal cap from proacrosomal vesicles. Arrested cells ultimately merge to form giant multinucleated cells released to the epididymis. Spermatids also completely fail to form the flagellar axoneme. RNA-Seq analysis and ChIP-Seq analysis identified 139 genes directly controlled by RFX2 during spermiogenesis. Gene ontology analysis revealed that genes required for cilium function are specifically enriched in down- and upregulated genes showing that RFX2 allows precise temporal expression of ciliary genes. Several genes required for cell adhesion and cytoskeleton remodeling are also downregulated. Comparison of RFX2-regulated genes with those controlled by other major transcriptional regulators of spermiogenesis showed that each controls independent gene sets. Altogether, these observations show that RFX2 plays a major and specific function in spermiogenesis.

Fig 1. Generation Rfx2 ^-/- mice.

The mouse Rfx2 gene is depicted approximately to scale from exon 6 onward (A). Below is shown the region of homology of the targeting vector, with the details of the floxed exon 7 enlarged. Locations of PCR primers are indicated above and their sequences are provided in S5 Table. ES cell clones were tested for targeted insertions using primer sets with one member lying outside the region of homology either up (B) or downstream (C) of Rfx2 sequences present in the targeting vector. Mice were genotyped by tail biopsy using primer-pairs specific to the WT, floxed or deleted alleles (D). Reverse transcriptase PCR using testis cDNA, and a primer set anchored in exons 7 and 8, confirmed that exon 7 was absent in homozygous Rfx2 ^-/- mice (E). Control amplification (Ctrl) detected a ubiquitously expressed component of the mitochondrial F1 ATP synthase complex (Atp5a1). Western blotting of testis protein extracts from mice of 30 or 104 days of age confirmed that full length RFX2 was not detected in homozygous Rfx2 ^-/- mice (E). Actin (ACT) was used as control.

Fig 2. Spermatogenesis is arrested in Rfx2 ^-/- mice.

(A-B) Testis sections from Rfx2 ^-/- and Rfx2 ^+/+ mice were stained with anti-RFX2 antibodies. RFX2 is expressed in pachytene spermatocytes (arrows) and early round spermatids (arrowheads). No RFX2 staining is observed in Rfx2 ^-/- sections. (C-D) H&E stained sections from 30 day old Rfx2 ^+/+ and Rfx2 ^-/- littermates are shown. (C) Seminiferous tubules from Rfx2 ^+/+ males show orderly development of germ cells of different stages including spermatocytes (black arrowhead, spc), round spermatids (black arrow, rsp), and late stage of spermatids with condensed nuclei (white arrow). (D) In striking contrast, Rfx2 ^-/- mice exhibit an arrest in differentiation of haploid cells at approximately step 7 of the round spermatid phase (black arrow), without evidence of flagellum formation or condensation of the nucleus. Spermatids start to fuse at this stage (white arrow) leading to the formation of giant multinucleated cells that subsequently develop highly condensed nuclei (white arrowhead). Spermatocytes are normally present (spc, black arrowhead) and undergo meiosis. (E-G) In 3 month-old Rfx2 ^+/+ mice, the cauda region of the epididymis is filled with mature sperm (E, arrow in G). (F and H) The cauda epididymis of a 3-month-old Rfx2 ^-/- mouse contains no mature sperm and only cell remnants (arrow).

Fig 3. The characteristic acrosome organelle fails to develop properly in Rfx2 ^-/- mice.

(A-H) Periodic acid shiff (PAS) / light green stained segments of testis sections from Rfx2 ^+/+ mice (A-D) and comparable sections from Rfx2 ^-/- mice (E-H). Developmental stages of the seminiferous tubules are noted on each panel. Insets show a 3-fold magnification of a spermatid from the respective section (box). Insets show in: (A, E) step 2 spermatids, (B, F) step 5 spermatids, (C, G) step 7 spermatids. (I-P) High power views of individual cells with the acrosomes stained by fluorescently-tagged peanut agglutinin. For Rfx2 ^+/+ sections, different step of acrosome development are shown (I: step 2, J-L: step 6–7 spermatids). For Rfx2 ^-/- sections, variations in acrosome morphology are shown for spermatids that are about to arrest (M-P). In the majority of cases the acrosome fails to spread normally over the anterior end of the nucleus and frequently occurs as unattached vesicles (arrows).

Fig 4. Ultrastructural defects and perturbed Golgi organization in Rfx2 ^-/- testes.

Transmission electron microscopy images of control (A-F) or Rfx2 ^-/- (G-S) spermatids. (A) The acrosome vesicle (arrow, 2-fold magnified inset) has fused with the nucleus and is beginning to flatten at step 4. (B) As development progresses toward the end of the cap phase, acrosomes have flattened and spread over approximately half of the nuclear surface in step 7 spermatids (arrow). (C-D) cytoplasmic bridges (between two arrowheads in insets showing 2-fold enlargements) are visible between two spermatids. The acrosomal cap is marked with arrows. (E) Developing axonemes (arrows, insets show two-fold enlargements of the axonemes) are prevalent among late stage round spermatids and can be seen clearly on an early elongating spermatid of step 9 (F). (G-S) For Rfx2 ^-/- sections various aberrant morphologies are shown for cells at about the point of developmental arrest. (G-J) Examples of aberrant acrosomes observed in Rfx2 ^-/- spermatids. (G) Multiple acrosomic vesicles are present without attachment to the nucleus (arrows) or (H) coexisting with some degree of an attached and flattened acrosome (asterisk). (I) An atypical acrosome (asterisk) has attached and spread but contains multiple cytoplasmic inclusions. (J) Another example of multiple unattached vesicles (arrows). (K) Cytoplasmic bridges (between two arrowheads, 2-fold magnification inset) are apparently normal until spermatids fuse, forming round huge multinucleated cells, and cytoplasmic bridges disappear. (L) Nucleus with a protruding aneurysm-like rupture (arrow). (M, N) Cells contain isolated centriole pairs that have not generated axonemes (inserts). In (M), arrow points to the acrosomal vesicle. (O, P, Q, R) Within multinucleated giant cells, clusters of centrioles occur (arrows), but without associated axonemes. (Q, R) A giant multinucleated cell contains a cluster of centrioles (box enlarged in R, arrows). Ciliary rootlets are visible (2-fold enlarged inset in P), the asterisk indicates the chromatoid body. (S) A large bundle of microtubules that could be an ectopic manchette (arrow). Sizes of scale bars are indicated in individual panels.

Fig 5. Testis sections from P30 Rfx2 ^+/+ or Rfx2 ^-/- mice stained for the cis-Golgi compartment (GM130 antibody, red), the acrosome (peanut agglutinin, green) or nuclei (Dapi, blue).

In WT spermatids, the Golgi compartment (asterisks) is concentrated at one pole of the nuclei above the forming acrosome (arrow) on the nucleus. In Rfx2 ^-/- testes, GM130 staining (asterisks) is observed adjacent to the nuclei and acrosome (arrow) and can be correctly orientated in a few situations (left panel), but the overall orientation of the Golgi and acrosome is most frequently disturbed with the cis-Golgi being (from left to right) between the acrosome and the nucleus, apposed to the Golgi and the nucleus, or surrounding the acrosome.

Fig 6. Perturbed gene expression in Rfx2 ^-/- testis.

(A) Plots of the cumulative number of downregulated genes as a function of their expression level in Rfx2 ^-/- relative to Rfx2 ^+/+ testis: (top) downregulated genes at P21 and P30; (bottom) genes upregulated at P21 and P30. (B) Venn diagram representing the overlap between each set of differentially expressed genes. (C) Differentially expressed genes were plotted according to their fold-change (log2) at P21 and P30. Dot size is inversely proportional to the p-value at P21. Black to grey scale is inversely proportional to the p-value at P30. (D) Motifs that are significantly enriched in the promoters of P30 downregulated or P21 upregulated genes are compared with the X motifs defined for RFX1 and RFX2 in JASPAR, and the X motif identified experimentally for RFX1 [8]. The motif identified in JASPAR for CREB1/CREM is also reported, to show the absence of homology with RFX binding motifs.

Fig 7. Developmental expression profiles of genes downregulated in Rfx2 ^-/- testis.

Expression patterns derived from [29], are shown for genes that are downregulated in Rfx2 ^-/- testis at P21 (left plots) or P30 (right plots). The plots show the expression patterns for all genes in the group (top) or for genes exhibiting expression patterns that are the most consistent with activation by RFX2 (lower). Gene totals are less than in Fig 6 because genes exhibiting unreliably low read values in the Laiho et al. data set were not included.

Fig 8. ChIP-seq analysis of RFX2 target genes in mouse testis.

(A) Distribution of RFX2 binding peaks in different genomic regions. (B) Fractions of peaks mapping in the different regions relative to the p-value of the peaks. Peaks are ranked according to increasing p-value. (C) Distribution of peak centers relative to the TSS for genes having peaks within promoter regions. Peaks are mostly centered near the TSS. (D) A motif that is significantly enriched in the promoters of Rfx2 downregulated target genes is compared with the X motifs defined for RFX1 and RFX2 in JASPAR. (E) Position of the de novo identified motif (motif discovery) and predicted RFX1 or RFX2 motifs relative to the TSS of RFX2 target genes. All three motifs cluster near the TSS. (F) Venn diagram showing the overlap between RFX2 targets (identified by ChIP-Seq) and differentially expressed genes at P21 and P30.

Fig 9. Developmental expression profiles of RFX2 target genes.

Expression patterns derived from [29], are shown for target genes that are downregulated in Rfx2 ^-/- testis at P21 (A) or P30 (B). The plots show the expression patterns for all genes in the group (top) or for genes exhibiting expression patterns that are the most consistent with activation by RFX2 (lower). Left panels show all genes in the group and right panels only cilia-related genes from each set.

Fig 10. Downregulated genes having known or suspected functions related to cilia/flagellum formation, microtubule function or the acrosome.

(A) Plot of the cumulative number of cilia-related genes as a function of their expression level in Rfx2 ^-/- relative to Rfx2 ^+/+ testis. Downregulated genes analyzed are included in the Syscilia gold standard list [55] or have at least 1 high-confidence citation in the CilDB database. (B) Bars show the expression levels (RPKM, reads per Kb per million) in Rfx2 ^+/+ (grey) and Rfx2 ^-/- (black) testis for downregulated genes included in the Syscilia Gold or Potential lists. RFX2 targets are indicated above the bars. Genes are ordered according to their expression level in WT mice. (C) Downregulated genes assigned to GO terms related to microtubule or the acrosome (GO function or compartment). Expression in Rfx2 ^-/- testis is expressed relative to Rfx2 ^+/+ testis. RFX2 targets are indicated above the bars.