Sexually dimorphic peripheral sensory neurons regulate copulation duration and persistence in male Drosophila

Peripheral sensory neurons are the gateway to the environment across species. In Drosophila, olfactory and gustatory senses are required to initiate courtship, as well as for the escalation of courtship patterns that lead to copulation. To be successful, copulation must last long enough to ensure the transfer of sperm and seminal fluid that ultimately leads to fertilization. The peripheral sensory information required to regulate copulation duration is unclear. Here, we employed genetic manipulations that allow driving gene expression in the male genitalia as a tool to uncover the role of these genitalia specific neurons in copulation. The fly genitalia contain sex-specific bristle hairs innervated by mechanosensory neurons. To date, the role of the sensory information collected by these peripheral neurons in male copulatory behavior is unknown. We confirmed that these MSNs are cholinergic and co-express both fru and dsx. We found that the sensory information received by the peripheral sensory neurons from the front legs (GRNs) and mechanosensory neurons (MSNs) at the male genitalia contribute to the regulation of copulation duration. Moreover, our results show that their function is required for copulation persistence, which ensures copulation is undisrupted in the presence of environmental stress before sperm transfer is complete.

form weaker connections to the intromittent organ and can be easily separated by dissection. The periphalilic structures are subdivided into four regions: the epandrial ventral lobe (formerly called: lateral plate) (EVL), the surstylus (formerly called: clasper) (SUR), the epandrial posterior lobe (formerly called posterior lobe) (EPL), and the cercus (formerly called: anal plate) (CER) 21 (Fig. 2A). Genital coupling, as revealed by a high-resolution electron microscopic time sequence analysis, involves the active movement of these periphalic structures 22 . The surstylus bends medially and is hidden from view. Within 10 min of copulation, the cercus aligns with the female oviscape and achieves genital coupling. Each periphallic region contains an array of stereotypic, species and sexspecific bristles, and each bristle is innervated by a bipolar mechanosensory neuron 23 . The neural implication of this sensory information collected during copulation is unclear.
Copulation must last long enough to ensure successful sperm transfer, which does not occur until five minutes after copulation starts 24,25 . The regulation of copulation time in Drosophila is a complicated process that does not simply depend on the volume or the transfer of sperm and seminal fluid, since mutants defective in their synthesis still have normal copulation duration 24 . Both neuronal and non-neuronal factors can influence mating time. Neuronal regulators include a small cohort of fru neurons that co-express the transcription factor engrail 26 and a subset of 4-5 fru neurons that innervate the male reproductive tissues 27,28 . The latter neurons control copulation duration in response to sperm transfer 27 . Non-neuronal factors include mutations in the circadian clock gene period, which exhibit longer copulation duration 29 , and environmental factors such as the gut microbiome 30,31 , the social environment of the male [32][33][34][35] , and environmental stressors 24 . Copulation duration reflects the reproductive investment of the males 32 . Sustaining copulation with a suboptimal mate is a loss of opportunity to gain a better mate. Therefore, males must respond to extrinsic and intrinsic factors by adjusting copulation duration to maximize reproductive success 33 . During the first 5 min before sperm transfer occurs, the benefit of maintaining copulation outweighs any risk because terminating copulation will lead to no fertilization. Copulation persistence describes the maintenance of copulation in the presence of stressful stimuli, which peaks within the first 5 min of copulation and decreases over time 24 . This is an important cost-benefit analysis that ensures species survival. The neural circuit for this cost-benefit analysis includes 8 dsx/GABA neurons in the ventral cord that decrease copulation persistence. With opposite action, ventral cord DA neurons increase copulation persistence 24 . However, the neural input required to initiate copulation persistence is currently unknown. In this study, we show that sensory information received by the peripheral sensory neurons from the front legs (GRNs) and mechanosensory neurons (MSNs) at the male genitalia contribute to the regulation of copulation duration and their function is critical for the maintenance of copulation in the presence of stressful stimuli before sperm transfer is complete.

Discovery of fruitless (fru) neurons that regulate copulation duration.
To uncover the neuronal regulators of copulation duration, we utilized an intersectional genetic approach, involving both the GAL4/ UAS 36 and FLP/FRT 7 expression systems. The details of the intersectional genetic system and how we generated a FLP enhancer trap screen had been described elsewhere 28,37,38 . Briefly, the target gene is downstream of a DNA sequence that contains the binding site (UAS) for the transcription factor GAL4, followed by a DNA sequence that contains a stop codon flanked by the FLP recombinase recognition site FRT (UAS-FRT-stop-FRT-target gene). A cell must possess an active promoter for FLP (to remove the stop codon by recombination) and an active promoter for GAL4 (to bind to UAS) to activate expression of the target gene. As an additional tool to restrict gene expression, tsh GAL80 is used to inhibit GAL4 expression specifically in the ventral cord. From here forward, the genetic nomenclature will list all transgenes that are responsible to drive expression of the target gene (e.g. "X GAL4 ,FLP # > target gene" denotes a fly that carries a GAL4 driver controlled by the promoter of X, and the FLP line # from the enhancer trap screen, expressing the target gene). For the relevant neuronal populations, we expressed either a membrane tagged GFP (UAS-FRT-stop-FRT-mCD8::GFP) to visualize the arborizations of the neurons or the neuronal silencer tetanus toxin (UAS-FRT-stop-FRT-TNT) to block neuronal activity. Previously, we identified a cluster of ~ 5 fru 5HT/DA neurons (fru-sAbg-1) in the male abdominal ganglion that innervate various male reproductive tissues and regulates copulation duration 28 . In that study, we observed a reduction in copulation duration when we combined FLP335 with either fru-GAL, that targets most fruitless neurons (Fru GAL4 , FLP 335 , tsh GAL80 > TNT), with Th-GAL4, that targets dopaminergic neurons (Th GAL4 , FLP 335 , tsh GAL80 > TNT), or with 5HT-GAL4, that targets serotonergic neurons (5HT GAL4 , FLP 335 , tsh GAL80 > TNT) 28 . However, neither silencing the subsets of dopaminergic nor serotonergic neurons can recapitulate the copulation phenotype observed when we targeted the broader fruitless circuit. The result suggested that additional fru neurons contributed to the copulation duration phenotype 28 . Indeed, the expression pattern for Fru Gal4 , FLP 335 , tsh GAL80 > mCD8::GFP males extended beyond the accessory glands and seminal vesicles all the way to the genitalia (Fig. 1A). Upon further examination, we were able to trace GFP expression to the male genitalia.
To further confirm the genitalia neurons are implicated in copulation duration control, we screened our enhancer trap FLP library 28 to search for other FLP lines that affected copulatory behaviors. We identified another FLP line that showed expression at the genitalia and also reduced copulation duration when it is combined with fru-GAL4 to express TNT in males (Fru GAL4 , FLP 386 , tsh GAL80 > TNT) (control:Inactive TNT/silenced:Active TNT) (Fig. 1D). The experimental males (Fru GAL4 , FLP 386 , tsh GAL80 > mCD8::GFP) showed similar expression in fru ORNs, JONs, and GRNs compared to Fru GAL4 , FLP 335 , tsh GAL80 > mCD8::GFP males (Fig. 1B). However, the prominent fru-sAbg-1 neurons in the abdominal ganglion that innervate the male reproductive tissue were noticeably missing, leaving only the sexually dimorphic arborizations (ANN1) in the abdominal ganglion originating from the MSNs in the genitalia. The expression pattern in the peripheral tissues that extend out from the ventral cord of Fru GAL4 ,FLP 386 ,tsh GAL80 > mCD8::GFP males was highly restricted, showing GFP expression only in the male genitalia (Fig. 1B, video of confocal stacks). Although we still observed expression for the fru LAN1 arborations originating from the GRNs, it is less consistent compared to Fru GAL4 ,FLP 335 , tsh GAL80 > mCD8::GFP.  Table 1). The copulation duration of Fru GAL4 , FLP 386 , tsh GAL80 > TNT was shortened by ~ 26% compared to the control (Fig. 1D). This difference is comparable to that observed in Fru GAL4 , FLP 335 , tsh GAL80 > TNT males. Of the three groups of fru neurons targeted by both FLP lines, we hypothesized that  www.nature.com/scientificreports/ ANN1 from fru MSNs in the genitalia were the most likely to contribute to the copulation duration phenotype since we showed previously that silencing fru ORNs and GRNs did not affect copulation duration 28 . Recently, it was discovered that females sing in copula and that this song influences the reproductive success of the male 39 . However, the study showed that copulation duration was unaffected by the absence of female singing 39 .

Fru MSN neurons co-express doublesex (dsx).
To rule out the contributions of other neuronal populations to the copulation duration phenotype, we investigated other genetic combinations that target genitalia neurons more specifically. Previous research indicated that the genitalia neurons express dsx 11,12,40 . Indeed, we confirmed that the fru-MSN neurons are also dsx positive (Fig. 3A). Replacing fru-GAL4 with dsx-GAL4 in our genetic combination with FLP 335 (Dsx GAL4 , FLP 335 > mCD8::GFP) eliminated all expression in the brain (Fig. 3C).
In the VNC, Dsx GAL4 , FLP 335 > mCD8::GFP showed consistent expression only in the sexually dimorphic arbors originating from the foreleg GRNs and the genitalia neurons ( Fig. 3B,D-F). Using this highly restrictive genetic combination, we silenced these neurons by TNT expression and evaluated the post-copulatory behaviors. Consistent with our hypothesis, copulation duration was still reduced at the same level for Dsx GAL4 ,FLP 335 > TNT compared to the control (Fig. 3G). These results ruled out the involvement of CNS neurons in the copulation duration phenotype and indicate that that the fru neurons responsible for the phenotype are also dsx positive. The characteristic male-specific midline crossing of the LAN1 arbors that originated from GRNs on the front legs is regulated by both fru and dsx 41 . These GRNs co-express the ion channel ppk25 that is critical for pheromone detection during early courtship steps 17 . To investigate whether the copulation duration phenotype requires female pheromone perception, we paired Dsx GAL4 ,FLP 335 > TNT males with females whose pheromone-producing cells were ablated genetically via expression of the pro-apoptotic gene hid under an oenocyte-specific promoter 42 . Graphical evaluation of the four groups created by two levels of male (control:Inactive TNT/silenced:Active TNT) and two levels of female (control:oe+/no pheromones:oe−) suggests that there is an interaction between the effect of male (control/silenced neurons) and female (control/no pheromone) on copulation duration (Fig. 3H). Assumptions were not met for analysis with a two-way factorial ANOVA. An overall effect of the male factor can not be evaluated; in the presence of interaction the effect of male must be evaluated separately for each female group. Overall, a difference in copulation time in the four unique combinations was observed (one-way Kruskal Wallis ANOVA [nonparametric data (unequal variance and non-normal distribution)], p ≤ 0.0001). Experimental males with both GRNs and MSNs silenced (Active TNT) exhibited shortened copulation duration compared to the control (Inactive TNT) males, irrespective of whether they were paired with oenocyte-less females (oe−; Dunns' test, p = 0.0289) or their genetic controls (oe+; Dunn's test, p = 0.0088) (Fig. 3H). However, the copulation duration of the control males (TNT-in) is significantly shorter when paired with oe− females compared to control oe+ females (Fig. 3H). This suggests that female pheromone perception also contributes to the shortened copulation duration. www.nature.com/scientificreports/ Adding the transgene cha-Gal80, which inhibits GAL4 in cholinergic neurons, eliminated ANN1 expression that originated from the genitalia MSNs (compare to Fig. 1A,B). Only fru-Abg-1 and other neurons in the abdominal ganglion showed expression. Scale Bar = 50 µm. www.nature.com/scientificreports/  www.nature.com/scientificreports/ Sensory information detected by peripheral senseory neurons is required for copulation persistence. Next, we asked the biological relevance of a shortened copulation duration as a result of silencing fru/dsx GRNs and MSNs. In a productive copulation pairing, duration must be long enough for the transfer of sperm and accessory gland fluid. In the three genetic combinations presented here, the median copulation duration for the experimental males is 10-14 min, which is longer than the minimal time (~ 8 min) necessary for sperm transfer 26,27 . To evaluate if sperm is successfully transferred from the test male to the target female, we quantified the number of copulation pairings that resulted in fertilization after a single copulation event between a test male and a virgin Canton-S female. Compared to inactive TNT controls, a significantly lower percentage of males where both copulation duration regulating neural clusters (sAbg-1 and MSNs) are silenced (Fru GAL4 , FLP 335 , tsh GAL80 > TNT males) could fertilize the virgin females. In contrast, males in which sAbg-1 are not manipulated but MSNs are silenced (Fru GAL4 , FLP 386 , tsh GAL80 > TNT and Dsx GAL4 , FLP 335 > TNT males) have relatively normal levels of fertilization rates compared to their respective inactive TNT controls (Fig. 4A). Therefore, post-copulatory fertility is unaffected by the shortened copulation duration due to the silencing of fru/dsx GRNs and MSNs. If fertility is unaffected, we wondered if peak copulation persistence is still possible without sensory input from GRNs and MSNs. To evaluate if copulation persistence is affected by the silencing of the fru/dsx GRNs and MSNs, we applied heat shock as the stress stimulus to copulating flies when copulation persistence is at its peak. We heat-shocked pairs of flies 5 min after the onset of copulation and quantified the number of pairs that disengaged (Fig. 4B). While a large number of pairs in which the fru/dsx GRNs and MSNs were silenced by our three genetic combinations (Fru GAL4 , FLP 335 , tsh GAL80 > TNT; Fru GAL4 , FLP 386 ,tsh GAL80 > TNT; Dsx GAL4 , FLP 335 > TNT) terminated copulation, all pairs with the corresponding control males remained copulating (Fig. 4C). These results indicate that the fru/dsx GRNs and MSNs are important in maintaining copulation persistence in response to environmental stress before sperm transfer is complete.

Discussion
Different neuronal populations have been identified to control three distinct aspects of male copulatory behaviors: genital coupling, copulation duration, and sperm transfer 27,28,40 . Here, we have characterized the fru/dsx-MSNs at the genitalia. We also identified novel function of the peripheral sensory neurons at the front legs (GRNs) and genitalia (MSNs) in regulating copulation duration and maintaining copulation persistence in the presence of environmental stressors.
Retrograde labeling of various genitalia MSNs have shown that most of these neurons project only to the abdominal ganglion except for one neuron from the surstylus (clasper) that projects all the way to the suboesophageal region of the brain 40 . Since we did not observe any arborizations in the brain, the copulation phenotype www.nature.com/scientificreports/ appears to be regulated by a subset of the fru/dsx − MSNs at the surstylus (claspers) and epandrial ventral lobe (lateral plate) that project specifically to the abdominal ganglion. In addition, the retrograde labeling experiment showed that the axonal terminals of the genitalia neurons juxtapose the dendrites of the abdominal dsx glutamatergic (dsx/vGlut-Abg) and GABAergic (dsx/GABA-Abg) neurons 40 . Moreover, artificial mechanical stimulation of the genitalia with a minuten pin activated both dsx/vGlut-Abg and dsx/GABA-Abg neurons 40 . Therefore, we can infer that fru/dsx-MSNs neurons make functional synaptic connections to both dsx/vGlut and dsx/GABA neurons in the abdominal ganglion (Fig. 5). The shortened copulation duration phenotype when fru/dsx-MSNs are silenced is a result of dsx/vGlut-Abg and/or dsx/GABA-Abg not receiving sensory feedback signals from the genitalia. dsx/vGlut-Abg are all motor neurons that innervate the phallic and periphallic musculature responsible for genital attachment 40 . On the other hand, dsx/GABA-Abg is comprised of a heterogeneous population of interneurons with different functions. While some of these interneurons inhibit the activity of dsx/vGlut-Abg Figure 5. Proposed model of how sensory information from the MSNs is incorporated into the neural circuit that governs copulation persistence. Genital coupling requires the activation of the dsx/vGlut motor neurons that innervate the copulatory muscles. The resulting muscular tension is proportional to the activity level of the dsx/vGlut motor neurons, which are antagonistically regulated by dsx/GABA interneurons. These copulation regulating neurons are functionally connected to the axonal terminals of sensory neurons from the genitalia bristles 40 . We propose that a suboptimal sensory input will reduce the total muscular tension due to less activation of the dsx/vGlut motor neurons or more activation of the dsx/GABA interneurons. Copulation persistence ensures the maintenance of copulation before sperm transfer in the presence of stressful stimulus and is regulated by opposing actions of 8 dsx/GABA and the dopaminergic system in the ventral cord 24 . This persistence requires optimal sensory input from the genitalia bristles. www.nature.com/scientificreports/ to terminate copulation 40 , others have different functional roles, such as regulating copulation persistence 24 . Therefore, sensory signals received from the fru/dsx-MSNs can influence the relative activity level of dsx/vGlut-Abg and dsx/GABA-Abg neurons with opposing actions and modulate the overall tension of the copulatory muscles. In the absence of sensory information, the baseline muscle strength is sufficient to maintain genital coupling long enough for sperm transfer since males with fru/dsx-MSNs silenced has normal fertility (Fig. 4A). However, our results show that the sensory information encoding the male's correct engagement in copulation is necessary to achieve peak copulation persistence before sperm is transferred (Figs. 4B, 5). Sensory information-provided by fru/dsx-MSNs-might be a way to measure the quality of the copulation. Genital coupling will lead to bending of the bristle hairs and activate the fru/dsx-MSNs. Which bristle hair gets stimulated and the strength of the stimulation will depend on whether genital coupling is established and the morphology of the female genitalia. For example, an abnormal amount of pressure received by a bristle hair could send less signal to activate the copulatory muscle innervated by dsx-vGlut-Abg. Similarly, a wrong set of bristles bent could activate the dsx-GABA-Abg neurons that inhibit the dsx-vGlut-Abg. A suboptimal activation decreases the total muscular tension during genital coupling. Copulation persistence is a result of the fly's ability to analyze the tradeoff of maintaining copulation during exposure to an environmental stressor. Sensory information that signals the quality of the copulation provides a critical factor in this assessment.
Our results showing males' perception of female pheromone contributes to the regulation of copulation time is surprising because our previous work showed that genetically silencing olfactory receptor or gustatory receptor independently did not result in shortening of copulation duration 28 . Since females defective in pheromone synthesis would lack both volatile (detected by olfactory receptor) and non-volatile (detected by gustatory receptor), it is possible that males use olfactory or gustatory cues to evaluate female fitness. In the absence of both sensory cues, the males assess that the female has poor fitness for reproductive success and decide to shorten the copulation duration in order to select for an alternative mate. Non-volatile pheromones are cuticular hydrocarbons (CHCs) synthesized from long-chain fatty acids. Therefore, abnormal distribution of CHCs could reflect metabolic deficiency. In line with this explanation, diet has been shown to influence copulation duration by altering the gut microbiome, whose metabolism changes the cuticular hydrocarbon profiles that could affect the amount of sex pheromones present 30,31 .
As is observed in many species, copulation duration reflects the reproductive investment of the males 33 . Although sperm transfer occurs in the first 5 min of copulation 24,25 , mating lasts much longer to an average of 20 min in Drosophila melanogaster 22 . Males can adjust their mating time in response to their social-sexual environment. Males exposed to other males prior to mating copulate longer compared to socially isolated males 32 . The prolonged mating time resulted in females that laid more eggs and reduced the likelihood of remating 32 . Prolonged mating time is necessary for the transfer of seminal proteins that ensures sperm competitiveness of the mating male against other rivals, particularly for heterospecific males that can cause reproductive interference 43 . Synthesis of seminal fluid is energetically costly to the male and therefore, the allocation of seminal fluid must be strategized. Sustaining copulation with a suboptimal mate is a loss of opportunity to gain a better mate. The evaluation of an optimal copulation time requires a neural circuit that processes sensory information from the environment and triggers the motor program for copulation. Males require at least two sensory cues from sound, smell, or touch to recognize the presence of rivals and respond by increasing mating time 44 . With our genetic tools to specifically silence the front leg fru-dsx-gustatory neurons and genitalia mechanosensory neurons, it will be interesting to test whether these males will lose the ability to perceive the presence of rivals and respond with a lengthened copulation duration.
Our work revealed the importance of sensory information collected by both front leg dsx/fru GRNs and genitalia MSNs in regulating copulation duration. To further dissect the functional role of gustation and mechanosensation, new genetic strategies to target gene expression to either GRNs and MSNs is necessary. The recently published transcriptomic atlas for both the ventral nerve cord 45 and the male genitalia 46 provides new possibilities to develop new tools that will allow further investigation of the neural mechanism of how sensory information is encoded and processed in the CNS.

Materials and methods
Fly strains. The following strains were used in this study: fru-GAL4 7 , dsx-GAL4 11 , UAS > stop > TNTin, UAS > stop > TNT 7 , UAS > stop > mCD8::GFP 9 , FLP 335 and FLP 386 were generated as described in previous study 28 , cha-GAL80 47 , tsh-GAL80 from Julie Simpson, and Canton-S strain from the Bloomington Stock Center, Bloomington, IN. Oenocyte less (oe−) flies and their controls were generated as previously described 42 . Immunohistochemistry. Dissection of 3-7 day old adult flies and immunohistochemistry of the adult nervous system were carried out as described previously 48 with some modifications in primary and secondary antibodies. Dissection of male reproductive organs was performed on Sylgard plate covered with 1 × PBS. To obtain male genitalia, the lower half of the abdomen were dissected and fixed in 4% PFA at room temperature for 20-30 min. After which the PFA was replaced with 1xPBS. The genitalia were cut out with a pair of microscissors to ensure a flat surface for mounting. Dissected samples were fixed and proceeded with immunostaining as described previously 48 . The following primary and secondary antibodies were used in this study: rat polyclonal anti-mCD8 (1:100; Caltag, Burlingame, CA), mouse anti-GFP (1:500, Life technologies), mouse anti-nc82 (1:20; Hybridoma Bank) 49  www.nature.com/scientificreports/ Microscopy. Images of the ventral cord, the reproductive tissues, and genitalia were acquired using an Olympus Fluoview FV1000 confocal microscope with a 20X objective. ImageJ was used to stitch the overlapping images together. All other images were acquired using a Zeiss LSM 700 confocal microscope. For the genitalia, lambda stacks from 490 to 600 nm at 20 nm intervals were acquired. The autofluorescence signals were unmixed from the 488 nm signals using the linear unmixing algorithm in the Zen Black software.
Behavioral assays. Husbandry. Flies were raised on standard cornmeal medium and kept on a 12 h:12 h day:night cycle at 25 °C in ambient relative humidity. Each newly eclosed adult was collected and aged for 3-7 days in an isolation vial (16 × 100-mm) supplied with ~ 2 ml of fly food. Virgin, wildtype Canton-S females were aged in groups of 20-40 for 3-7 days. All behavioral experiments were performed at 25 °C with ~ 50% humidity during the first 3 h. after lights on.
Copulation assay. Copulation assays were performed in 12-well plates (Thermo Scientific BioLite Multidish). A square glass plate covering four wells was used as the lid. Assays were performed in 4 wells with ~ 5 ml standard fly food to maintain humidity in each well. An experimental male was paired with a virgin female and the courtship behavior was videotaped for at least an hour. If a pair did not copulate within an hour, it was considered unsuccessful. Copulation duration was calculated from the beginning of genital coupling until the male was dismounted from the female.
Fertility assay. Freshly hatched males were isolated into a glass tube (16 × 100 mm) with 2 ml of food and kept for 4-5 days at standard conditions (25 °C, 50% RH). A 6-7 day old CS virgin female was introduced into each tube using an aspirator. After transferring the vials back to standard conditions, the flies were allowed to interact for 1 h and observed every 10 min for successful copulation. Tubes with pairs that did not copulate were discarded. For the rest of the tubes, males were removed by quick anaesthetization and females were allowed to lay eggs for 24 h and then discarded. Fertility was recorded after 7 days by observing the presence of progeny.
Persistence assay. A single experimental male was paired with a wildtype CS virgin female in a glass tube (16 × 100 mm) with or without food. If copulated in a vial with food, the pairs were transferred to an empty vial before heat shock to ensure efficient heat transfer. Each pair that successfully copulated (within 5 min) was subjected to heat stress by submersing the empty glass vial (16 × 100 mm) in a 37 °C water bath for 30 s. Frequency of copulation termination was recorded.
Statistics analysis. Prism version 9.3.1 (https:// www. graph pad. com/ scien tific-softw are/ prism/) was used for all statistical analysis. For Fig. 3H, presence of interaction between factors was assessed visually (data from groups at different levels of one variable showed differing observed mean values and distributions for separate levels of another variable). Results from the two way factorial design of this study (control vs. silenced males x control vs. pheromones-less females) were assessed for fit to a two way factorial ANOVA (normality and homogeneous variance of residuals). For nonparametric data, a one way Kruskal-Wallis ANOVA (nonparametric) was fit to combinations created of levels for each of the two factors (control vs. silenced males x control vs pheromone-less females) when interaction was present. Post hoc pairwise comparisons, controlling for multiple comparisons, were performed using Dunn's test for multiple comparisons. www.nature.com/scientificreports/