Response to: X‐linked miR‐506 family miRNAs promote FMRP expression in mouse spermatogonia

Abstract

The topic of this rebuttal is a set of miRNAs expressed from a large gene cluster located in the Fragile‐X region of the X chromosome that we call the “Fragile‐X miRNA (Fx‐mir) cluster”. This miRNA cluster has several other names. Wang et al calls them the “miR‐506 family”, as some members of this cluster are related to this particular miRNA. Other names for this cluster from recent papers are “spermiR” ¹ and “XmiR” ².

In our Ramaiah et al ³ paper, we reported several lines of evidence that four members of the Fx‐mir cluster—miR‐741‐3p, ‐743b‐3p, ‐878‐5p, and ‐880‐3p—are highly expressed in Sertoli cells (SCs), where they target the adjacent gene Fmr1. The Wang et al rebuttal claims that most members of the Fx‐mir cluster and the Fmr1 gene are instead primarily expressed in germ cells (GCs). Below, we go over the evidence both for and against these claims.

Are Fx‐mir family members expressed in SCs?

To address the cell types that express Fx‐mir family members, we took three approaches: (i) genetic, (ii) cell fractionation, and (iii) in situ hybridization. For the genetic approach, we examined the expression of Fx‐mir miRNAs in the testes of two germ cell‐deficient mouse strains: (i) juvenile spermatogonial depletion (Jsd) and (ii) XXY, a mouse model of Klinefelter's syndrome. The former lacks all germ cells except for type‐A spermatogonia ⁴ and the latter lacks virtually all germ cells ^{5, 6}. Using TaqMan probes specific for the mature forms of Fx‐mir family members, we found that Fx‐mir miRNAs are expressed at normal or slightly elevated levels in Jsd mutant and XXY adult male mice testes relative to littermate control testes (Fig 1A, below; and fig 3B in Ramaiah et al ³ ). Since XXY testes completely lack GCs, this addresses Wang et al's concern that the residual GCs in Jsd‐mutant mice could be a major site of Fx‐mir expression. We believe the simplest interpretation of this result is that these Fx‐mir miRNAs are mainly expressed in somatic cells in the testis. We then examined their expression in purified testicular cell populations obtained from Drs. John McCarrey (U. Texas, San Antonio) and Marvin Meistrich (U. Texas, M.D. Anderson Cancer Center). We found that Fx‐mir miRNAs are expressed in the highly purified SC fraction (generated by elutriation from adult mouse testes ⁷) at a level 4‐ to 5‐fold higher than in total testes (Fig 1B, below; and fig 3C in Ramaiah et al ³). In contrast, Fx‐mir miRNAs are expressed at lower levels in most purified GC fractions, prepared as described ⁸, than in total testes (Fig 1B, below; and Fig EV4D in Ramaiah et al ³). Finally, in situ hybridization analysis showed that miR‐883a‐5p was expressed in a “spoke‐like” pattern in seminiferous tubules (Fig 1C), which is the pattern characteristic of factors expressed in the cytoplasm of SCs ⁹. Together, these data suggest that at least the particular Fx‐mir miRNAs we tested are primarily expressed in SCs.

Figure 1. Fx‐mir expression in testicular cells.

(A) TaqMan‐PCR analysis of the indicated microRNAs from three biological replicates. Germ cell transcripts (e.g., Dazl) were nearly undetectable in jsd and XXY testes (unpublished observations). Error bar, SEM. (B) TaqMan‐PCR analysis of RNA from the indicated purified testicular cell subsets. The result is from three biological replicates. Expression is shown relative to unfractionated adult testes, which was given a value of 1. Sg, spermatogonia; Sp, spermatocytes; Sd, spermatids, Pr, primitive; A, type‐A; B, type‐B; Pr, pre‐leptotene; LZ, leptotene‐zygotene; Pa, pachytene; J, juvenile; A, adult; R, round; E, elongated. Error bar, SEM. (C) In situ hybridization fluorescence analysis of mouse adult testes with miR‐883a‐5p and negative‐control (miR‐9) LNA probes. Arrows indicate miR‐883a‐5p in the Sertoli cell cytoplasm. Scale bar, 20 μm.

Past papers have also obtained evidence that Fx‐mir family members are expressed in SCs. For example, Ro et al observed that purified SCs express miR‐741, ‐883a (which they call “t25”), miR‐465‐3p (t28), ‐470, ‐471, ‐741‐3p (t27), and 883a‐3p (t24) ¹⁰. We note that this paper is from the same group as the Wang et al rebuttal. Similarly, another group—Panneerdoss et al ¹¹—reported that purified SCs express many Fx‐mir family members: miR‐880, ‐741, ‐465c‐3p, ‐470, ‐471, ‐201, ‐883a‐3p, ‐743a, ‐743b‐3p, and ‐871. The expression of many of these miRNAs was found to be regulated by androgen, consistent with SC expression, as SCs, not GCs, have androgen receptor. Finally, germ cell‐deficient mice have been shown to express normal testicular levels of at least two X‐miR1 family members—miR‐201c and ‐465—suggesting they are primarily expressed in SCs or other somatic cells ¹².

Are Fx‐mir family members expressed in GCs?

As indicated in the rebuttal from Wang et al, several papers have reported evidence that some or all Fx‐mir family members are expressed in GCs. Indeed, we noted this in the Discussion section of our Ramaiah et al paper: “It has been previously reported that Fx‐mir miRNAs are expressed in germ cell‐enriched fractions ^{10, 13, 14}, a finding we reproduced, but we found that expression in the germ cell fractions was much lower than in the total testis fraction” ³.

Wang et al cite five papers that they claim provide evidence that Fx‐mir “miRNAs are predominantly expressed in germ cells”. While we agree that these 5 papers provide evidence for GC expression of Fx‐mir family members—including from enriched cell populations ^{1, 13, 15} and from in situ hybridization ^{16, 17}—we do not believe that any of these papers demonstrate that GCs are the main site of Fx‐mir expression in the testis. The paper that comes closest to demonstrating this point—Song et al ¹³—is from the same group as the Wang et al rebuttal. This paper, which provides a comprehensive analysis of the expression of X‐linked miRNAs reports evidence that several Fx‐mir family members are expressed at similar levels in total testes as in purified germ cell subsets. This suggests that these Fx‐mir family members are expressed primarily in GCs, but there are several caveats with this conclusion. First, expression levels were analyzed by Q‐PCR, but were reported using a semi‐quantitative numbering system ¹⁰, and thus it is not possible to draw firm conclusions about relative expression levels. Second, rather than using a TaqMan assay (as is typically done to distinguish mature miRNAs from their precursor/primary transcripts), Song et al used a SYBR green‐based qRT‐PCR approach that has been shown to yield inconsistent results due to sequence‐specific differences in primer annealing ¹⁸. Finally, because SCs and GCs are in direct contact with each other in vivo, it is common for purified GC preparations to have significant SC contamination (as indicated in our Ramaiah et al paper ³), thereby further obscuring the ability to draw firm conclusions.

In their rebuttal letter, Wang et al provide new RNA‐seq data showing that several Fx‐mir family members are expressed at higher levels in enriched spermatogonia than in enriched SCs. There are several possible explanations for the discrepancy between these results and our findings. First, as described above, it is common for GC‐ and SC‐fractions to co‐purify and thus the enriched spermatogonia fractions assayed by Wang et al could have considerable SC contamination. Second, individual Fx‐mir miRNAs clearly have different expression patterns and thus it would not be surprising if some were predominantly expressed in GCs and others were predominantly expressed in SCs. Indeed, Wang et al's heatmap shows that several Fx‐mir family members appear to not be expressed higher in spermatogonia than SCs (their data are presented on a Z‐scale and thus absolute relative expression is not clear). Third, strain differences may be responsible, as Wang et al did their analysis on outbred (CD1) mice, and not inbred (C57/B6) mice, as analyzed in our study ³ (of note, Tan et al ¹⁵ also examined Fx‐mir expression in purified SCs from CD1 mice). Finally, Wang et al purified SCs from postnatal day 8 (P8) mice, a time point likely to express low levels of many Fx‐mir miRNAs in SCs, as we found that all four Fx‐mir miRNAs we examined in our Ramaiah et al ³ study were expressed >5‐fold lower level in P5 mice than in P15 and older postnatal mice.

We conclude that there is abundant evidence that most (if not all) Fx‐mir family members are expressed in SCs, with some of these miRNAs very likely being expressed at higher level in SCs than in GCs. Likewise, there is clear evidence (mainly from in situ hybridization) that many Fx‐mir miRNAs are also expressed in GCs, but the level of their expression in GCs is not clear.

Where is FMRP expressed?

In our Ramaiah et al ³ paper, we provide several lines of evidence from transfected cell lines that many Fx‐mir miRNAs regulate the immediately adjacent gene: Fmr1. As the Wang et al rebuttal points out, in order for Fx‐mir miRNAs to regulate Fmr1 in an endogenous setting, the Fx‐mir miRNAs and Fmr1 need to be expressed in the same cell type. In Ramaiah et al, we cited two papers that we claimed report the protein product of Fmr1—FMRP—is expressed in SCs ^{19, 20}, but Wang et al indicated that these two papers “actually reported that FMRP was mainly expressed in spermatogonia and hardly detectable in the SCs in either human or mouse testes”. We agree that one of these papers—Devys et al ¹⁹—does not provide evidence that FMRP is expressed in SCs, and we apologize for citing this paper incorrectly. However, we note that this paper indicates a lack of FMRP signal in what they call a “Sertoli cell”, when actually it is an SC nuclei. This negative signal would be expected given that FMRP is known to mainly be in the cytoplasm ²⁰. The cytoplasm of SCs extends all the way from the basement membrane of the seminiferous tubule to its lumen, which greatly contrasts with SC nuclei, which are of similar size as a whole GC. SCs require being large, as they are bound by many germ cells that line up in a developmentally orchestrated manner along the length of the SC. However, the large volume of the SC cytoplasm greatly dilutes molecules, even when expressed in high amounts in a single SC, making it challenging to observe the expression of mRNAs and proteins by visualization methods such as immunohistochemistry, immunofluorescence, and in situ hybridization.

The other reference we cited—Bakker et al ²⁰—shows there is clear FMRP antibody staining in the cytoplasm of SCs in postnatal mouse testes. This signal is no longer detectable in adult mouse testes, which may indicate developmental regulation of FMRP or it may indicate insufficient sensitivity to observe the signal at this stage. The notion that FMRP protein is expressed in SCs is also supported by another paper—Hinds et al ²¹—which showed that the FMR1 signal detected by in situ hybridization in the seminiferous tubule is maintained in germ cell‐deficient W/W ^v mice ²². We note that the more intense FMR1 signal observed by Hinds et al ²¹ at the basement membrane leads to the intriguing possibility that FMRP is expressed in a polar manner in SCs to serve a functional role in the basal compartment.

The Wang et al rebuttal also brought up Tamanini et al ²³, a paper which Wang et al claim demonstrates “that FMRP is only expressed in spermatogonia in human testes”. We respectfully disagree with this conclusion. In addition to spermatogonia staining, there is a modest signal throughout the seminiferous tubule, consistent with SC cytoplasm staining. This signal is not apparent in the FMR1 patient testes, indicative of a signal specific for FMRP. As explained above, given the extremely large size of the SC cytoplasm, modest staining can actually indicate high expression.

Finally, the Wang et al rebuttal cited two recent single‐cell RNA‐seq (scRNA‐seq) studies ^{24, 25} as further evidence that Fmr1 is only expressed in spermatogonia in human and mouse testes. To independently assess this assertion, we re‐analyzed the scRNA‐seq datasets from these two studies and then examined the degree of co‐expression of Fmr1 with germ cell and SC markers. In adult mice, we found that Fmr1 is co‐expressed with both germ cell markers that we tested (Dazl and Ddx4) and both SC markers that we tested (Sox9 and Rhox8) (Table 1), indicative of Fmr1 expression in both spermatogonia and SCs. Because scRNA‐seq is an insensitive technique, the values in Table 1 are likely to be a conservative estimate of the percentage of these two cell types expressing Fmr1. An even greater proportion of both SCs and spermatogonia expressed Fmr1 in postnatal testes, based on analysis of our own scRNA‐seq dataset from postnatal day 7 (P7) testes (Table 1). We also re‐analyzed scRNA‐seq datasets from two studies examining adult human testes ^{24, 26}, and, again, observed that a proportion of both SCs and spermatogonia expressed FMR1 (Table 1). We conclude that both mouse Fmr1 and human FMR1 are heterogeneously expressed in both SCs and spermatogonia.

Table 1.

Expression of Fmr1 in Sertoli cells and spermatogonia

Testes source	Data source	Fmr1 + Sox9 + /Sox9 +	Fmr1 + Rhox8 + /Rhox8 +	Fmr1 + Dazl + /Dazl +	Fmr1 + Ddx4 + /Ddx4 +
Mouse postnatal day 7	Our unpublished data	42%	40%	59%	63%
Mouse adult	Hermann et al 2018 24	13%	36%	22%	7%
Mouse adult	Green et al 2018 25	19%	14%	28%	10%
		FMR1 + SOX9 + /SOX9 +	FMR1 + WT1 + /WT1 +	FMR1 + DAZL + /DAZL +	FMR1 + DDX4 + /DDX4 +
Human adult	Guo et al 2018 26	15%	20%	19%	13%
Human adult	Hermann et al 2018 24	8%	14%	12%	11%

The percentage of cells co‐expressing the indicated gene markers was done by re‐analyzing published scRNA‐seq datasets using the same quality control and clustering method as described ³¹, except that we used a more stringent filtering cut‐off (mitochondrial content < 0.2%; nGene > 1,000). Analysis of mouse P7 testes was done similarly. Sox9 and Rhox8 are mouse Sertoli cell markers; Dazl and Ddx4 are mouse germ cell markers; SOX9 and WT1 are human Sertoli cell markers; DAZL and DDX4 are human germ cell markers.

The finding that FMR1 is expressed in SCs is consistent with the fact that Fragile‐X Syndrome patients often have abnormally large testes (macro‐orchidism). First reported in 1975 ²⁷, at least 16 subsequent studies have reported that Fragile‐X patients have macro‐orchidism (at a frequency of between 11% and 91% of all patients examined in each study) ²⁸. While the underlying mechanism is not known, a simple explanation is that an excess of SCs is responsible given that it is well established that the number of SCs is a major factor that dictates the number of germ cells in the seminiferous tubule and, consequently, testes weight ²⁹. As direct evidence, loss of Fmr1 in mice causes hyper‐proliferation of SCs accompanied by increased testes weight ³⁰.

What is the functional role of Fx‐mir miRNAs?

To investigate the biological role of the Fmr1 gene cluster, Wang et al generated a knockout mouse that lacks most Fx‐mir family members (18 out of 21 Fx‐mir genes). They made the interesting discovery that FMRP levels were downregulated in the testes of these Fx‐mir cluster‐null mice, and claim that, “Therefore, these miRNAs do not repress, but rather promote FMRP expression”. This conclusion does not contradict our findings in Ramaiah et al ³, as we reported that, while many Fx‐mir miRNAs repress Fmr1 expression (consistent with the typical repressive action of most miRNAs), one Fx‐mir family member—mir‐883a‐5p—strongly promoted Fmr1 expression (and another Fx‐mir member—miR‐881‐3p—exerted a trend toward positive regulation of Fmr1). More importantly, one cannot conclude (as Wang et al did) that several miRNAs have a given function based on knocking all of them out simultaneously. Instead, each miRNA would need to be knocked out individually to dissect the role of each. As an added note, we believe the phenotypic results reported by Wang et al are quite important, and thus, it is critical they provide a little more detail (as allowed within space constraints), such as testes weight, sperm count, and testes morphology, as well as key controls (e.g., analysis of the expression of the 3 miRNAs they did not delete).

Conclusion

We conclude that there is considerable evidence that many Fx‐mir family members, as well as Fmr1, are expressed in both SCs and GCs. We thank Wang et al for bringing forward their criticisms, so that this potentially wider role of Fx‐mir miRNAs—in regulating FMRP expression in both germ and somatic cells in the testis—could be highlighted.

EMBO Reports (2020) 21: e49354