Motivated by observations in C. elegans that double-stranded RNA expression in neurons can result in siRNA activity in the germline (Devanapally et al., 2015), we set out to explore the potential for neuronal Cre expression in mice to drive cell non-autonomous Cre activity elsewhere in the body. In two series of experiments, Cre expression was induced in the mouse brain via stereotaxic delivery of AAV constructs engineered to drive Cre expression in transduced cells (Figure 2A). In an initial series of experiments, we utilized an AAV2-retro-Ef1a-Cre construct introduced into the prefrontal cortex (Figure 2—figure supplement 1), while follow-up studies targeted the lateral somatosensory cortex, caudate putamen, and nucleus accumbens using an AAV9-pCMV-Cre construct (Figure 2B). Remarkably, examination of the epididymis in both cases revealed robust Tomato expression in the cauda (distal) epididymis and in the vas deferens (Figure 2C). Tomato expression was reproducibly observed in all epididymis samples isolated from ten AAV-transduced animals, including three animals injected in the prefrontal cortex (Figure 2—figure supplement 1) as well as seven animals injected in the lateral somatosensory cortex, caudate putamen, and nucleus accumbens (Figure 2C, Figure 2—figure supplement 2). As in the case of the Defb41^iCre driver, closer examination of our imaging data showed that Tomato expression was confined to the epididymal epithelium (inset, Figure 2C, Figure 2—figure supplement 2), with no detectable expression in the sperm filling the epididymal lumen (see below).

An exciting potential explanation for the ability of neuronal Cre expression to drive recombination in the epididymis would be that Cre RNA or protein synthesized in neurons is somehow trafficked to the male reproductive tract. However, the use of viral transduction to induce Cre expression leaves open the possibility that viral particles entering the circulation could plausibly exhibit unanticipated tropism for the epididymis.

We therefore turned to an orthogonal method for driving Cre expression in neuronal subpopulations by crossing Ai14D mice to various driver lines expressing Cre under the control of cell type-specific promoters (Figure 3A). We illustrate this approach using Slc32a1^Cre mice that express Cre under the control of the vesicular GABA transporter (Slc32a1, or Vgat) promoter, which drives Cre expression in inhibitory GABAergic neurons in the CNS. Consistent with expectations, we observed widespread Tomato expression in the brain of double transgenic animals, with no detectable expression in the liver, kidney, intestine, or testis (Figure 3B). Remarkably, similar to the reporter expression observed following viral induction of neuronal Cre expression (Figure 2), Cre expression from the Slc32a1 transgene also resulted in robust epididymal Tomato expression in Ai14D; Slc32a1^Cre double transgenic animals (Figure 3C). Closer inspection of double transgenic epididymis by microscopy revealed Tomato expression exclusively in the principal cells of the epididymal epithelium (Figure 3D), and we further confirmed the presence of Tomato-positive principal cells by FACS (Figure 3—figure supplement 1). Intriguingly, we observed patchy Tomato expression in any given histological cross section, suggesting either that a subset of principal cells are capable of receiving Cre shipments, or that Cre transfer is extremely inefficient and only a subset of cells are stochastically labeled.

Finally, we sought to determine whether Cre expression in reporter animals resulted in epididymal Tomato signal as a result of Cre itself being shipped from neurons to the epididymis, vs. Cre-induced Tomato produced in neurons being shipped to the epididymis. We therefore assayed recombination at the LoxP-STOP cassette in the epididymis, finding recombination at the reporter locus in genomic DNA isolated from the epididymis (Figure 3E). These findings strongly support the hypothesis that Cre activity (RNA or protein) is the relevant molecular signal trafficked from the CNS to the epididymis.

Together, our findings suggest the possibility that Cre RNA or protein synthesized in the CNS is transported to the male reproductive tract, presumably through the circulation (see below). Importantly, the two experimental schemes used to drive Cre expression in the CNS have distinct and unrelated potential artifacts. In the case of AAV transduction, the injected virus could conceivably enter the circulation and might have an unknown tropism for the epididymis. However, this concern does not apply to the use of transgenic constructs to drive Cre expression. Conversely, it is well known that purportedly tissue-specific Cre drivers are often less cell type-specific than expected (Song and Palmiter, 2018; Stifter and Greter, 2020), and the male reproductive tract – the testis in particular – is well known to express an unusually high fraction of the genome. However, this cannot explain our results using viral transduction to drive Cre expression in the CNS. We further test the hypothesis that Cre is trafficked via the circulation to the male reproductive tract below.

Our findings thus far suggest the intriguing possibility that the epididymis might be a privileged recipient of neuronally derived molecular cargo. To expand our survey of neuronal Cre lines beyond our initial studies using Slc32a1^Cre, we crossed the Ai14D line to a number of additional CNS Cre drivers: Fev (serotonergic neurons), Gad2 (GABAergic neurons), Syn1 (pan-neuronal), Nestin (pan-neuronal), Drd1a (dopaminoceptive neurons), ChAT (cholinergic neurons), Dat1 (dopaminergic neurons), and Gfap (glia). We also explored two non-neuronal Cre drivers, crossing the Ai14D animal to Alb (liver) and Adipoq (adipose tissue) Cre lines. We note that the majority of these genes are either undetectable or expressed at extremely low levels (~1 ppm) in the epididymis as assayed either by bulk or single cell RNA-Seq (Rinaldi et al., 2020).

For each of these double transgenic lines, we surveyed a range of tissues, typically including brain, intestine, liver, lung, kidney, testis, seminal vesicle/prostate, and the epididymis. Images for key examples are documented in Figures 4–5 and Figure 4—figure supplements 1–4. Overall, we find that the connection between Cre expression in neurons and reporter expression in the epididymis is not a one-to-one mapping, but rather many-to-many: not only do some non-neuronal Cre lines drive reporter expression in the epididymis, but many Cre lines also drive reporter expression in additional tissues such as the intestine or seminal vesicle. Findings of particular interest are detailed below.

Most notably, we find that a wide range of neuronal Cre lines drive Tomato expression in the epididymis. Focusing first on the putative ‘pan-neuronal’ Cre lines, driven by the Synapsin or Nestin promoters, we found extensive Tomato staining throughout the epididymis in both double transgenic lines (Figure 4—figure supplement 2). However, we noted that Tomato expression in the Syn^Cre animals was not limited to the epididymal epithelium and instead filled the epididymis lumen (Figure 4—figure supplement 2A), suggestive of Tomato-positive sperm. Indeed, offspring of the Ai14D; Syn^Cre double transgenics exhibited Tomato expression throughout the body, consistent with prior reports of Syn^Cre driving germline recombination (Luo et al., 2020). Similarly, the Nestin Cre driver has also been reported to drive germline recombination (Betz et al., 1996; McLeod et al., 2020), and we did obtain Tomato-positive offspring of Ai14D; Nes^Cre double transgenics. That said, in contrast to the Ai14D; Syn^Cre animals we found no detectable labeling of sperm in Ai14D; Nes^Cre animals, with Tomato expression in the epididymis largely confined to the epididymal epithelium (Figure 4—figure supplement 2B). This suggests that Nes^Cre-mediated recombination occurs relatively late during spermatogenesis in germ cells, with recombination in the epididymal epithelium occurring independently – probably induced by the same CNS→epididymis trafficking seen for several other neuronal Cre drivers (Figure 3 and below).

Turning from pan-neuronal promoters to more cell type-specific CNS Cre drivers, we consistently observed Tomato expression in the epididymis when Cre was expressed from Slc32a1, Fev, Gad2, and Drd1a promoters (Figures 3–4). For two of these Cre drivers – Fev and Slc32a1 – we find that Tomato expression was strongest in the cauda epididymis and vas deferens, similar to our findings obtained using AAV to drive Cre expression in the CNS. In contrast, both Ai14D; Gad2^Cre and Ai14D; Drd1a^Cre double transgenic animals exhibited patchy epithelial staining throughout the epididymis (Figure 4). Intriguingly, off-target epididymal recombination was not unique to neuronal Cre drivers, as we also observed robust Tomato expression and reporter locus recombination in animals carrying the adipose-specific Adipoq^Cre (Figure 5). In contrast, reporter animals carrying the liver-specific Alb^Cre transgene exhibited barely detectable Tomato expression in the epididymis with few Tomato-positive cells detected throughout the entire tissue – at most one or two cells per longitudinal section (Figure 4—figure supplement 1) – indicating that not all Cre transgenes drive extensive recombination in the epididymis.

Thus, although our initial studies suggested a privileged molecular trafficking pathway from the CNS to the male reproductive tract (Figures 2—4), the robust Tomato expression driven by the adipose-specific Adipoq Cre driver (Figure 5) suggested the possibility of additional avenues for unanticipated inter-tissue Cre trafficking. Not only did non-neuronal Cre drivers direct recombination in the epididymis, but multiple neuronal Cre drivers exhibited off-target activities in tissues beyond the epididymis. Most notably, within the male reproductive tract we observed robust Tomato expression in the reproductive accessory glands of the Nes, Fev, and Gad2 lines (Figure 4—figure supplements 3–4).

Our data demonstrates that a wide range of putatively tissue-specific Cre lines drive recombination in the epididymis of Ai14D animals. Given the extraordinary nature of this finding, we were concerned that our results might reflect some idiosyncracy of the LoxP-flanked Tomato reporter locus in the Ai14D strain. We first explored publicly available annotations for Cre drivers from MGI (Mouse Genome Informatics: https://www.informatics.jax.org, data retrieved repeatedly from ~2018 through October 2022), which includes images from a variety of Cre drivers crossed to a LacZ reporter line. Of the lines that drive recombination in Ai14D in our study, there is also clear evidence for recombination in the epididymis epithelium for Syn1^Cre, along with ‘ambiguous’ expression annotated for Nes^Cre and Gad2^Cre. However, closer examination of the image provided for Nes^Cre also revealed clear evidence for reporter expression in the epididymis epithelium; no image was available for Gad2^Cre. These reporter assays thus are consistent with our findings using the Ai14D reporter animal. Conversely, Adipoq, Slc32a1, Drd1, and Fev were all annotated as ‘absent’ in the epididymis epithelium (at postnatal days 7 and 56). However, given that epididymal activity was annotated as ‘ambiguous’ for Nes^Cre despite clear microscopic evidence for reporter expression, we were concerned about false negative annotations for these other Cre drivers, for which either no images were available for scrutiny, or the available image was limited to a very narrow region of the epididymis. Nonetheless, the lack of annotated recombination in this dataset, for Cre drivers which exhibit robust activity in our system, suggested the possibility that some reporter loci could be relatively resistant to Cre activity – by virtue of the target locus being packaged in relatively inaccessible chromatin in the epididymis, for instance – compared to the Tomato reporter locus present in the Ai14D animal.

To extend our findings to other target loci, we therefore obtained double transgenic animals carrying the Adipoq Cre driver along with a conditional allele of Dicer1 with LoxP sites flanking exons 21 and 22 (Figure 5—figure supplement 1A). PCR genotyping confirmed the expected robust recombination in adipose tissue, while tail tip DNA was used as a negative control (Figure 5—figure supplement 1B). Using this assay, we document recombination of this locus specifically in the cauda epididymis, but not in the testis or other parts of the epididymis (Figure 5—figure supplement 1B). These data thus extend our results to a second target locus, demonstrating that off-target recombination in the epididymis is not unique to the Ai14D reporter.

Together, our findings reveal that the epididymis is a surprisingly common target of a variety of Cre drivers that are not generally thought to be expressed in the male reproductive tract. While the observations made using double transgenic lines could result from leaky expression of the promoters in question – whether in adulthood or during early development of the epididymis – this hypothesis would not explain the results observed following AAV-mediated Cre expression in the brain (Figure 2). Nonetheless, we sought to definitively test the hypothesis that Cre activity, whether protein or RNA, synthesized in distant tissues can make its way to the male reproductive tract through the circulation.

As an initial test of this hypothesis, we generated parabiotic animal pairs (Gibney et al., 2012), in which two animals are surgically joined (hereafter represented as animal1::animal2) to establish a conjoined circulatory system (Figure 6A, Figure 6—figure supplement 1A–C). Here, parabioses were established between a Cre driver line (focusing on the Slc32a1 and Nestin Cre lines) and an Ai14D reporter animal. Eight to nine weeks after establishing the parabiosis, we sacrificed the animal pairs and harvested tissues for analysis. We detected moderate Tomato expression in the epididymis of the Ai14D recipient animal in a parabiotic Ai14D:: Slc32a1^Cre pair (Figure 6B, Figure 6—figure supplement 1B), along with a faint band supporting recombination at the reporter locus in one recipient (Figure 6C). However, in two other Ai14D animals we found no evidence for recombination by PCR (Figure 6—figure supplement 1C), motivating more extensive characterization of the potential for Cre transfer through the circulation.

Given the technical and logistical challenges involved in generating large numbers of parabiotic animal pairs, we attempted to transfer any potential circulating Cre activity by injecting serum or circulating exosomes/extracellular vesicles (EVs) (enriched via tangential flow filtration [TFF] Heinemann et al., 2014), from various Cre lines into Ai14D reporter animals (Figure 6D). In an initial experiment, we obtained serum from a Slc32a1^Cre animal and transferred it into the tail vein of an Ai14D male. This procedure was repeated every other day, for a total of three transfers over 5 days. As shown in Figure 6E–F, this resulted in epididymal LoxP recombination and Tomato expression in animals receiving Slc32a1^Cre serum, while recipients of serum from FVB or Alb^Cre animals did not express Tomato or exhibit recombination at the reporter locus.

Follow-up studies were, unfortunately, somewhat ambiguous (Figure 6—figure supplement 1D–F). We initially set out to leverage the serum transfer approach to fractionate serum and identify the circulating material responsible for driving recombination in recipient animals. We found no Tomato expression in an Ai14D male that had received TFF-enriched exosomes from nine Slc32a1^Cre animals spread across 3 injection days (Figure 6—figure supplement 1E, top right panel). Across many subsequent experiments, we used a range of donor animals carrying various Cre drivers and varied our injection scheme to focus on whole serum, concentrated exosomes, or combinations of the two. Ultimately, across over 20 recipient animals we documented clear evidence for recombination in six animals, with positive results in recipients of Slc32a1^Cre serum, Slc32a1^Cre serum and exosomes, Gad2^Cre serum and exosomes, a mixture of Gad2^Cre and Drd1^Cre serum and exosomes, Nes^Cre serum and exosomes, and Adipoq^Cre serum and exosomes (Figure 6—figure supplement 1E–F). However, similar or nearly identical injection schemes often resulted in no detectable Tomato expression or recombination. Although we attempted to account for variables ranging from the recipient animal’s age, to injection time during the day, we have not been able to identify variables that reliably distinguish injection schema that lead to recombination from those that fail.

Taken together, our data support the hypothesis that off-target Cre activity in the epididymis is likely mediated by Cre trafficking through the circulation, although the difficulty in reliably replicating this transfer impedes further mechanistic follow-up.

The fact that putatively tissue-specific Cre drivers can often drive recombination in unanticipated tissues has long been known. This is commonly thought to reflect ‘leaky’ expression of a given Cre transgene, potentially owing to the removal of a given promoter from its genomic context (and thus from long-range regulatory elements) in many Cre drivers. In addition, while many promoters chosen as cell type-specific Cre drivers are highly expressed in the cell type of interest, they are often also expressed at lower levels in other cell types throughout the body where the Cre transgene therefore has the potential to drive recombination.

That said, our data showing off-target Cre activity in the epididymis are unlikely to result from leaky expression of the Cre drivers studied here, for multiple reasons. First of all, we find a wide range of Cre transgenes drive recombination in the epididymis, and most of the neuron-specific genes in question are not detectably expressed in the adult epididymis (Johnston et al., 2005; Rinaldi et al., 2020). It remains plausible that these genes could be expressed in the epididymis only during early development, before becoming undetectable in the adult epididymis. However, our observation that induction of Cre expression in the brain via AAV-mediated transduction leads to reporter activity in the epididymis epithelium (Figure 2, Figure 2—figure supplements 1–2) cannot be explained by leaky transgene expression. Conversely, the possibility that unanticipated tropism of the AAV constructs used here might result in viral transduction of the epididymis would not explain any of the results obtained using double transgenics (Figures 3—5). Together, these findings strongly suggest that Cre activity synthesized in neurons is transported to the epididymis.

The delivery of Cre from neurons to the epididymis could occur via a number of routes, including release of Cre-containing vesicles by peripheral neurons that innervate the cauda epididymis. We consider this to be unlikely for several reasons. First, we show that adipose-specific Cre expression reliably drives recombination in the epididymis (Figure 5), arguing against Cre transfer requiring neuronal termini. Second, induction of Cre expression specifically in one hemisphere of the brain via stereotactic AAV injections was shown to lead to Tomato expression in both the ipsilateral and contralateral epididymis (Figure 2), arguing against short-range neuronal delivery (not to mention the challenges in considering how Cre might move from the CNS to peripheral nerves).

Indeed, our follow-up studies using parabioses and serum transfers support the hypothesis that Cre activity synthesized in a variety of tissues can access the circulation and subsequently be trafficked to the male reproductive tract. Although these efforts to transfer Cre activity from a Cre-expressing animal to a reporter animal were inefficient, the fact that recombination was ever observed in these studies is best explained by the trafficking hypothesis. We speculate that the circulating Cre activity consists of Cre mRNA, rather than protein, and we anticipate that the RNA is present either in exosomes or in ribonucleoprotein complexes protected from serum nucleases. Indeed, we can document Cre mRNA by PCR in the circulation of multiple Cre lines (Figure 6—figure supplement 2), and there is abundant precedent for RNA transmission between different tissues. We cannot presently determine whether the transferred Cre activity is contained in exosomes, microvesicles, ribonucleoprotein complexes, or other forms. Again, precedent argues for exosomes as the carriers of inter-tissue molecular information (Valadi et al., 2007), but given the inefficiency of Cre transfer in serum and exosome injections (Figure 6, Figure 6—figure supplement 1), a detailed fractionation effort would require an inordinate number of animals to be used, and must be reserved for future studies.

Discovery of a pathway by which neurons (and some other tissues) potentially traffic RNAs (or proteins) to the male reproductive tract would have exciting implications. Most intriguingly, we speculate that environmental conditions experienced by a male could be interpreted by the CNS, with salient features of the environment then communicated to the male reproductive tract via this trafficking system. Given the emerging role for the epididymis in remodeling multiple aspects of maturing sperm (Gervasi and Visconti, 2017; Tamessar et al., 2021; Zhou et al., 2018), an appealing hypothesis would be that receipt of CNS-derived information – whether RNAs (mRNAs, lincRNAs, sncRNAs, etc.), proteins, etc. – could then influence epididymal physiology and thus alter the post-testicular maturation of germ cells in the male reproductive tract.

Interestingly, at least two prior studies in the paternal effects literature have suggested the possibility that circulating information may play a role in reprogramming the sperm epigenome. First, in their studies of intergenerational effects of carbon tetrachloride, Zeybel et al. reported that repeated serum transfers from CCl₄-treated animals to a recipient were sufficient to reproduce H3K27me3 changes in the sperm of the recipient animal (Zeybel et al., 2012). More recently, Mansuy and colleagues reported that transferring serum from males subject to early life stress to a recipient male was sufficient to drive altered glucose responses in the F1 offspring of the recipient animal (van Steenwyk et al., 2020), again suggesting that circulating factors may play a role in reprogramming the sperm epigenome. Although neither of these studies implicated circulating RNAs in modulating sperm biology – Zeybel et al. did not speculate about the nature of the soluble factor(s), while van Steenwyk et al. focused on lipids with the potential to activate various nuclear hormone receptors – they highlight the potential for circulating factors to mediate communication between distant tissues and the germline.

This possibility of inter-tissue RNA shipments between distant tissues and the germline is particularly intriguing considering the role of systemic RNA trafficking in worms and plants. Most relevant to our study, small RNAs produced in C. elegans neurons have been documented to silence genes in the germline and induce transgenerational silencing (Devanapally et al., 2015). Intriguingly, a prior report in mouse suggested that microRNAs expressed in the brain could be delivered to sperm, and subsequently to the next generation (O’Brien et al., 2020). However, with the exception of Syn1 and Nestin (Figure 4—figure supplement 2), we do not find any evidence for neuronal Cre trafficking directly to sperm cells, arguing against the idea that the circulating Cre activity is stably delivered to sperm. Even in the case of Syn1 and Nestin Cre drivers, recombination in the germline likely occurs prior to the last stages of spermatogenesis in the testis – potentially in spermatogonial stem cells or even earlier in development – and not during epididymal transit. Our data instead would place the epididymis as a mediator between neuronal information processing and programming of the sperm epigenome, with CNS-derived information potentially driving altered epididymal physiology and consequent remodeling of sperm small RNAs, proteins, or surface lipids.

An alternative explanation of our findings would be that neuronally derived molecular information does not have any physiological consequences upon uptake by the epididymis. In such a ‘trash can’ model, the epididymis could simply be unusually receptive to circulating exosomes or RNPs, with the unique sensitivity of Cre-Lox systems revealing the delivery of Cre molecules at abundances that would likely be meaningless for a typical cellular sncRNA, mRNA, or protein. We disfavor this model, in part because different Cre drivers result in recombination in different parts of the epididymis, arguing for some level of specificity between different tissue-derived exosomes (for instance) and target cell populations in the epididymis.

Nonetheless, without any evidence for physiological functions for this trafficking system we cannot formally rule out the trash can hypothesis at present; we envision two potential approaches to this question in the future. One possibility would be to subject male mice to some stress paradigm or other exposure, then identify changes in transcription or small RNA production in the epididymis. Subsequently, males could be subject to the exposure in question, with circulating serum/exosomes then isolated and injected into an unexposed recipient. In this scenario, if circulating material is responsible for programming epididymal gene expression, the recipient animal would be expected to exhibit the molecular changes seen in directly exposed animals. While this approach could be very powerful, our experience with serum/exosome transfers has been that exosome transfer results in very inefficient transfer of Cre activity between animals, with exosomes from at least nine donor animals required for patchy and low-penetrance reporter activity (Figure 6, Figure 6—figure supplement 1). This suggests that serum from scores of animals might be required for a single recipient to recapitulate the epididymal changes driven by direct exposures.

An alternative approach would be to identify the molecular basis for epididymal receptivity to neuronal information. For instance, if the circulating carrier of neuronal information is indeed present in exosomes, tetraspanins or other cell surface molecules are presumably responsible for tissue-specific targeting of exosomes to the epididymis. Identification of the relevant exosome receptor could then be used in knockout studies to prevent receipt of neuronal shipments, and changes in epididymal responses to various organismal stressors in wild type vs. knockout animals could then be used to identify potential effects of neuronal RNA/protein shipments on epididymal physiology.

Here, we have demonstrated that Cre activity synthesized in neurons and in adipose tissue consistently directs recombination in the epididymis – with some lines also having other off-target tissues. These findings add to the literature on the dangers of Cre-Lox genetics in conditional knockout studies, identifying additional off-target tissues for a number of Cre drivers. In addition, our data suggest that this poor tissue specificity could be caused in part by mobilization of Cre activity through the circulation, implying a surprising mechanism behind some of these off-target effects. Finally, the possibility that this pathway plays a role in sensing of environmental conditions in modulation of the sperm epigenome represents an exciting hypothesis for future studies in paternal effect epigenetics.

All animal use was approved by the Institutional Animal Care and Use Committees of UMass Chan Medical School and Harvard Faculty of Arts and Sciences, under protocol PROTO202100029 to Dr Oliver Rando and protocol 29014 awarded to Dr Amy Wagers. Double transgenic Adipoq^Cre; Dicer1^flx/flx animals were a gift from Bruna Brandao and C Ronald Kahn.

The transgenic mice utilized in this study were maintained by filial mating, giving preference toward homozygous carriers whenever possible. To avoid introduction of genetic confounders the breeding colonies were maintained for up to three generations, after which animals were either backcrossed to their background strain or the colony was culled, and new breeding pairs acquired.

Transgenic animals were typically sacrificed for imaging at 10–12 weeks of age, unless otherwise noted. AAV injections were carried out between 8 and 12 weeks of age. For parabiosis experiments, animals underwent surgery to establish conjoined circulation between 21 and 28 days after birth. Finally, for serum/exosome transfer experiments, recipient animals were 10–15 weeks, while donors of plasma were typically older than 15 weeks of age as this afforded more material for injections.

Below is a list with the official names for all the mice strains and the genotype abbreviations used throughout this text, as well as the strain number listed at the Jackson Laboratory.

All genotypes were determined using Transnetyx genotyping services.

An initial set of three animals were injected with rAAV2-retro-Eif1a-Cre. Subsequent experiments used an AAV9-based construct carrying pCMV-CRE: AAV9.CB-PI-Cre 5.03×10¹²GC/ml; VCAV-04562. Virus was a gift from the Gao lab.

Stereotaxic injections were performed to deliver AAV constructs into different regions of the CNS of adult Ai14D males. Adult male mice (8–12 weeks of age) were anesthetized with a mixture of 100 mg/kg ketamine and 10 mg/kg xylazine, administered via intraperitoneal injection. Prior to surgery, the top of the skull was shaved and disinfected. Surgeries were performed using aseptic technique with the aid of a stereotaxic frame (Stoelting Co.). Mice were placed into the stereotaxic frame and the skull was exposed by making a small incision with a scalpel blade. Using bregma and lambda as landmarks, the skull was then leveled along the coronal and sagittal planes. Two small drill holes were made in the skull that allowed injections to target the somatosensory cortex, caudate putamen, and nucleus accumbens. Microinjections were made using a Hamilton 10 μl neurosyringe (1701RN; Hamilton) and a microsyringe pump (Stoelting Co); 0.5 μl of virus was delivered through the syringe and at a constant flow rate of 30 nl/min. After injection, the needle was left unmoved for 10 min before being slowly retracted. The incision was then closed and held together with glue.

Male mice of similar age from two different strains – one ‘donor’ animal expressing Cre, and the Ai14D Tom^flx/flx ‘recipient’ – were surgically connected to allow the sharing of their circulatory systems. Surgical preparations were performed on the shaved skin, where incision was made along the opposing flanks of each mouse to be joined. A skin flap was made without damaging the underlying peritoneal lining, and the mice placed side-by-side in a prone position. To better secure the connection, the animals’ joints (elbows and knees) were held together using suture. The dorsal and ventral skin flaps from the different mice were then pinched together and sutured. Parabiotic pairs were maintained for 8 weeks after surgery and therefore expected to have maintained cross-circulation for at least 2 weeks (Wright et al., 2001).

Single cell suspension of the epididymis were performed as previously described (Rinaldi et al., 2020). Briefly, epididymis samples were separated into caput corpus cauda and vas and enzymatically dissociated. Cells were resuspended in HSS and transferred to a 1.5 ml Eppendorf tube for staining with viability dye and nuclear stain. Afterward the cell suspension was once more filtered while being transferred to a 5 ml polypropylene FACS tube, placed in ice, and protected from light until flow cytometry data acquisition. Data acquisition was performed in Cytec or LSRII with PMT voltages adjusted for each experimental day using single stained cells from homozygous Ai14D mice as well as from ubiquitously Cre-expressing Ai14D mice. Fluorescence minus one was also made. No compensation was necessary.

Recombination PCR was done using 2× MyTaq Red Mix (Meridian Biosciences, cat. BIO-25044) per manufacturer’s instructions. Primers (Supplementary file 1) were diluted to final concentrations of 0.33 μM for Rec7, 0.67 μM for Rec11 and SeqRec, and 0.5 μM for Dcr1 KO. Two-hundred ng DNA was used per reaction. PCR products were run on 2% agarose gel.

Slightly modified from the described steps on the Qiagen RNA&DNA extraction protocol. Namely, mandatory use of the antifoam reagent and adoption of a fixed 1 ml volume of lysis buffer to 3 mm³ of tissue. The best lysis was obtained using bead beater with the pre-programmed settings for kidney mouse tissue repeated three times and starting every cycle on frozen tissue/lysis buffer.