A Novel Neuroprotective Role of Phosphatase of Regenerating Liver-1 against CO₂ Stimulation in Drosophila

Summary

Neuroprotection is essential for the maintenance of normal physiological functions in the nervous system. This is especially true under stress conditions. Here, we demonstrate a novel protective function of PRL-1 against CO₂ stimulation in Drosophila. In the absence of PRL-1, flies exhibit a permanent held-up wing phenotype upon CO₂ exposure. Knockdown of the CO₂ olfactory receptor, Gr21a, suppresses the phenotype. Our genetic data indicate that the wing phenotype is due to a neural dysfunction. PRL-1 physically interacts with Uex and controls Uex expression levels. Knockdown of Uex alone leads to a similar wing held-up phenotype to that of PRL-1 mutants. Uex acts downstream of PRL-1. Elevated Uex levels in PRL-1 mutants prevent the CO₂-induced phenotype. PRL-1 and Uex are required for a wide range of neurons to maintain neuroprotective functions. Expression of human homologs of PRL-1 could rescue the phenotype in Drosophila, suggesting a similar function in humans.

Graphical Abstract

Highlights

• •PRL-1 functions to protect the nervous system against olfactory CO₂ stimulation
• •PRL-1 physically interacts with Uex and controls Uex expression levels
• •PRLs may retain a similar neuroprotective function in humans

Molecular Genetics; Molecular Neuroscience

Introduction

The mammalian phosphatase of regenerating liver (PRL) family contains three members, PRL-1, PRL-2, and PRL-3 (Diamond et al., 1994, Zeng et al., 1998). Human PRLs have been implicated in multiple cancers (Bessette et al., 2008, Campbell and Zhang, 2014, Saha et al., 2001, Al-Aidaroos and Zeng, 2010). The PRL expression patterns in several animal models have been characterized. In Drosophila, amphioxus, and zebrafish, PRL family members have been detected in many tissues including those of the central nervous system (CNS) during embryonic development (Pagarigan et al., 2013, Lin et al., 2013). In mice, mPRL-2 is expressed ubiquitously in the hippocampal pyramidal neurons, ependymal cells, and cone and rod photoreceptor cells (Gungabeesoon et al., 2016). An early study of rat brains demonstrated that the expression of PRL in neurons and oligodendrocytes was enhanced in the cerebral cortex following transient forebrain ischemia (Takano et al., 1996). In Drosophila, PRL-1 is the only homolog of mammalian PRLs. The exact physiological functions of the PRLs remain largely unknown.

CO₂-evoked behavioral responses in many winged insects are important for food foraging, reproduction, and survival (Guerenstein et al., 2004, McMeniman et al., 2014, Stange and Stowe, 1999). CO₂ as a natural gas is odorless for humans, although CO₂-responsive neurons do exist (Shusterman and Avila, 2003, Ji et al., 2007). Drosophila is highly sensitive to CO₂, and the sensing of CO₂ is usually accompanied by immediate physiological and behavioral responses. These responses have been previously studied mainly in terms of anesthetic and toxic effects under high concentrations of CO₂ (Dijken et al., 1977, Badre et al., 2005). When using standard CO₂ anesthetic protocol in fly work, wt adult flies respond with a brief loss of motion and activity and subsequently develop a held-up wing phenotype under extended exposure to a high concentration of CO₂. These flies recover normal wing function and return to normal activity levels upon the resumption of normal atmospheric levels of CO₂.

CO₂ also acts as an unfavorable stress odorant eliciting avoidance behavior in Drosophila (Suh et al., 2004). It has been reported that such avoidance behavior is mediated by the CO₂ receptors Gr21a and Gr63a that function together as a heterodimer (Scott et al., 2001, Jones et al., 2007). These chemosensory receptors are specifically expressed in CO₂-responsive neurons harbored in the third segment of the antennae (Scott et al., 2001, Jones et al., 2007). In the sensilla of the antenna, olfactory receptor neurons (ORNs) send their axonal projections into the antennal lobe (AL) to form glomeruli. It is these glomeruli that act as the primary olfactory center of the brain (Couto et al., 2005). The stereotyped V-glomerulus in the most ventral AL is responsive to CO₂ stimuli. It then conveys the signals to the mushroom body (MB) as a higher processing center (Sachse et al., 2007, Vosshall and Stocker, 2007). Eventually, the nervous system translates the processed olfactory signals into animal behavior.

In this study we found that in the absence of PRL-1, the adult flies treated with high concentrations of CO₂ exhibited a permanent wing held-up phenotype that failed to recover in the ambient environment. The deprivation of the CO₂ chemosensory receptor protein Gr21a or the overexpression of PRL-1 in the nervous system was able to suppress the wing phenotype in mutant animals. In addition, we found that PRL-1 interacted with Unextended wing (uex) and regulated its expression. The down-regulation of Uex alone resulted in the same wing held-up phenotype and elevated Uex levels in PRL-1 mutants to prevent the wing phenotype induced by CO₂. Expression of human homologs of PRL-1 could rescue the phenotype in Drosophila. Our data demonstrate a novel function of PRL-1 in preventing neural dysfunction from CO₂ insult and shed light on the understanding of hPRL functions in human diseases.

Results

Absence of PRL-1 Leads to CO₂-Induced Wing Phenotype

To search for potential physiological functions of the hPRLs, we took advantage of the simplicity of the Drosophila genome. Two independent mutant isolates of PRL-1 were generated with the CRISPR/Cas9 method (Bassett et al., 2013, Bassett and Liu, 2014). Both lines turned out to be frameshift mutations and caused a stop codon-terminating translation after 35 (PRL-1¹⁶) or 49 (PRL-1^44F) amino acid residues (Figure 1A). These two alleles were viable and developed into morphologically normal adults. Western blot analyses with the antibody against full-length amino acid sequence of PRL-1 exhibited an obvious band of 20 kDa in wt flies, but not in the mutants (Figure 1B), suggesting that mutant lines were loss-of-function alleles. As both isolates were null alleles, we only employed the PRL-1¹⁶ mutant (referred to as PRL-1^−/− or PRL-1 mutant) in this study.

Figure 1.

Absence of PRL-1 Leads to CO₂-Induced Wing Phenotype

(A) Two independent frameshift mutant alleles, PRL-1¹⁶ and PRl-1^44F, were generated with the CRISPR/Cas9 method. The targets of PRL-1 guide RNA (gRNA) were located in the first exon and nearby intron separately. Target 1 produced a site mutation, which added one base pair into the gRNA sequence. Target 2 produced a much longer genomic DNA deletion, which removed 51 bp near the gRNA sequence. Both of these targets caused a stop code in the following sequence, terminating translation after the 35th (PRL-1¹⁶) or 49th (PRL-1^44F) amino acid residue.

(B) Western blot analyses of mutant flies. Lysates were probed with anti-PRL-1 and anti-GAPDH antibodies.

(C and D) PRL-1 mutant flies exhibited a permanent wing held-up phenotype upon CO₂ stimulation (D), compared to wt flies under the same condition (C). Scale bar, 25 μm.

(E) Statistical calculations of fly populations with wing held-up phenotype. About 90% of 3-day-old male PRL-1 mutant animals displayed a permanent wing held-up phenotype upon CO₂ exposure within 24 h (orange). A low background of spontaneous wing phenotype was also detected in the mutant flies in the ambient environment (green).

Data are expressed as mean ± SD. See also Figure S1.

It appeared that PRL-1 was not critical for animal development and survival. One of the possibilities was that PRL-1 was involved primarily in stress responses. We interrogated mutant adult flies with various stress stimuli, including ultraviolet, X-ray, cold, heat, starvation, and CO₂. Interestingly, only high concentration of CO₂ treatment (a pulse of 5 L/min CO₂ for 20 s in a vial with a plug) caused a vertical held-up wing phenotype in PRL-1 mutant flies within 24 h after CO₂ exposure (Figure 1D), whereas wt control flies did not show any such response (Figure 1C). Further study found that young flies, 3 days after eclosion, were most responsive to such CO₂ exposure, with about 90% of such flies displaying the held-up wing phenotype (Figures 1E and S1A). We also detected a low background of spontaneous held-up wing phenotype in the PRL-1 mutant flies in the ambient environment (Figure 1E). It appeared that the male animals displayed the more prominent wing phenotype than the females (Figure S1B). We therefore only focused on male responses in the following experiments. We monitored the wing held-up phenotype in PRL-1 mutant flies induced by CO₂ exposure over time and found that this wing phenotype was permanent (Figure 1E). The lifespan of the mutant flies with this wing phenotype remained the same as that of the wt animals, despite their inability to fly (Figure S1C).

Wing Phenotype Is Rescued by PRL-1 Expression in the Nervous System

It was clear that the wing held-up wing phenotype could be either caused by a defect of the nervous system or a dysfunction of the wing muscles in the absence of PRL-1. To address this issue, we employed genetic approaches to identify the tissue in which PRL-1 expression could rescue the wing phenotype. PRL-1 expression in the mutant background was driven by an array of tissue-specific GAL4 lines. Our data showed that only the pan-neuronal expression (elav-GAL4) of PRL-1 completely prevented the wing phenotype induced by CO₂ exposure (Figure 2A). Neither the muscle-specific (Mhc-GAL4) nor the glial (repo-GAL4) expression of PRL-1 had any rescue effect (Figure 2A). Examination of the indirect flight muscles by phalloidin staining and transmission electrical microscopy showed no obvious differences between the wt and the mutants (Figures S2A and 2B). Based on these observations we conclude that PRL-1 plays an important protective role in the nervous system. The held-up wing phenotype induced by CO₂ exposure was caused by a defective neuronal function occurring in the absence of PRL-1.

Figure 2.

Statistical Data Showed Protective Role of PRL-1 in Nervous System against CO₂ Insult

(A) Neuronal-specific expression of PRL-1 (elav-GAL4) in the mutants completely rescued the wing held-up phenotype induced by CO₂, but expression in other tissues such as muscles (Mhc-GAL4) or glial cells (repo-GAL4) did not result in any rescue.

(B) The response of PRL-1 mutant flies to alternative anesthetics, such as carbon dioxide (CO₂), ether, and nitrogen (N₂). Permanent wing phenotype was only induced by CO₂.

(C) In the PRL-1 mutant background, knockdown of CO₂ sensory receptor protein Gr21a driven by Gr63a-GAL4 prevented the wing phenotype.

(D) Within the transition time (about 10 min) when flies were shifted from non-permissive temperature (29°C) back to the permissive temperature (25°C), the vast majority of flies ectopically expressed shi^ts1 in the nervous system (elav-GAL4>UAS-shi^ts1) and exhibited a transient wing held-up phenotype. A similar wing phenotype was also observed when shi^ts1was specifically expressed in motor neurons (D42-GAL4>UAS-shi^ts1).

Data are expressed as mean ± SD. *p < 0.05, ***p < 0.001. See also Figures S2 and S4.

We also explored the possibility whether human homologs of PRLs could rescue the wing phenotype. Two human homologs, hPRL-1 and hPRL-2, were tested. Either of them fulfilled Drosophila PRL-1 function and was able to effectively rescue the wing phenotype (Figure S1D). This result implies that hPRLs may have a similar role in humans.

CO₂ Sensory Circuitry Is Required for the Wing Phenotype

Our data indicated that PRL-1 plays a protective role in the nervous system. However, the permanent held-up wing phenotype could simply be due to the depletion of O₂ during the CO₂ exposure. To test this possibility, N₂ was used to anesthetize mutant flies, and no held-up wing phenotype was observed (Figure 2B), suggesting that the wing phenotype was not caused by the lack of O₂. In addition, ether was used to test whether anesthesia alone could cause the wing phenotype, and, again, such a positive correlation did not occur (Figure 2B).

Analysis of CO₂-evoked avoidance behavior using T-maze assays (Kwon et al., 2007, Suh et al., 2004) revealed no significant differences between PRL-1 mutants and the wt animals (Figure S4A). This indicates that the olfactory sensing of CO₂ remains active in the PRL-1 mutants. To further exclude the possibility that it was pH changes in body fluid, occurring through the trachea via exposure to the high concentration of CO₂ that may have led to the wing phenotype, we blocked CO₂ sensory circuitry in mutant flies. As most of the double mutant flies bearing either Gr21a or Gr63a with the PRL-1 mutation died at the pupal stage, we employed RNA interference (RNAi) method to specifically knock down Gr21a in the CO₂-responsive neurons. PRL-1 mutant flies carrying Gr63a-GAL4>Gr21a-RNAi no longer displayed any wing held-up wing phenotype upon CO₂ exposure (Figure 2C). This observation indicates that it is the signals generated by CO₂-responsive neurons that trigger the characteristic wing phenotype in the PRL-1 mutants. It is therefore unlikely that any pH change in body fluid has played a role in this event.

Held-up Wing Phenotype Is due to Neural Dysfunction

To further confirm that it was neural dysfunction that caused wing phenotype, we took advantage of a temperature-sensitive shi^ts1 allele. In this, at a non-permissive temperature (29°C), the synaptic vesicle recycling is halted and neuronal transmission is blocked (Kosaka and Ikeda, 2010, Kitamoto, 2001). As expected, at 29°C the flies (elav-GAL4>UAS-shi^ts1) were paralyzed. When the flies were shifted back from non-permissive temperature to the permissive temperature (29°C–25°C), the vast majority of flies with shi^ts1 in the nervous system exhibited a transient held-up wing phenotype within the recovery time (about 10 min) (Figures 2D and S4B), which was reminiscent of that observed in PRL-1 mutant flies induced by CO₂ exposure (Figure 1D). We also observed a similar wing phenotype when shi^ts1was expressed in motor neurons (D42-GAL4>UAS-shi^ts1) (Figure 2D). The same held-up wing phenotype also appeared during the temperature shift from 25°C to 29°C.

It is reasonable to assume that the synaptic vesicle recycling in elav-GAL4>UAS-shi^ts1 flies is not fully functional and that neurotransmission is affected within the transition time from 29°C to 25°C. Therefore the held-up wing phenotype is most likely due to neural dysfunction. Our data suggest that it was the neurotransmission defects in a wide range of neurons, including motor neurons, that were responsible for the wing phenotype. Based on these observations, we conclude that the CO₂-induced held-up wing phenotype in PRL-1 mutant flies is due to neural dysfunction.

PRL-1 Is Expressed in the CO₂ Sensory Neural Circuitry

PRL-1 expression was detected in the adult head by western blot (Figure S4C). We stained the adult brains with anti-PRL-1 antibodies. PRL-1 was detected in the AL including the V-glomeruli (Figures S3B–S3B″). As a control, these PRL-1 signals were not detectable in the PRL-1 mutant brains (Figures S3C–S3C″). To visualize the PRL-1 expressing neurons, we generated PRL-1-GAL4 transgenic flies with a 6.1-kb genomic DNA fragment immediately 5′ of the ATG codon of the PRL-1 gene. The GFP signals were robust in the head, the third segment of antennae, and the maxillary palp (Figures 3A and 3A′). Confocal images showed an obvious distribution of PRL-1 in the basiconic sensilla (Figures 3B and 3C), which harbors the cell bodies of ORNs, including those of CO₂-responsive neurons (Scott et al., 2001, Suh et al., 2004). The strong GFP signals were observed in the V-glomeruli (arrows, Figures 3D and 3D′), as well as the MB (arrow heads, Figures 3D–3E′). The signals of EGFP-PRL-1 driven by PRL-1-GAL4 and anti-PRL-1 antibody staining were both observed in the CO₂ neural circuitry (Figures 3D–3D′ and S3A–3A″).

Figure 3.

PRL-1 Is Expressed in the CO₂ Sensory Circuitry

Images of the expression pattern of EGFP-PRL-1 driven by PRL-GAL4.

(A and A′) GFP fluorescence signals in the adult head.

(B and C) Confocal scanning of the antennae (B) and maxillary palps (C) shows the expression of PRL-1 in the basiconic sensilla.

(D–E′) Immunofluorescence staining reveals the spatial expression of UAS-EGFP-PRL-1 driven by PRL-GAL4 in the anterior (D and D′) and posterior parts (E and E′) of the brain. The brains were double labeled with anti-GFP (green) and anti-nc82 (red). The distribution of PRL-GAL4 was strong in the V-glomeruli (arrows) and mushroom body (arrowheads).

Scale bar, 30 μm in (B–E′) and 39 μm in (A′). See also Figure S3.

CO₂ Triggers Hyperactive Ca²⁺ Activity in the Antennal Lobe of the PRL-1 Mutant Brain

To analyze the PRL-1 function in the brain, we estimated the neuronal activity with calcium-sensitive fluorescent protein (GCaMP) (Jones et al., 2007, Wang et al., 2003). It is known that the neurons in the V-glomeruli of the AL respond to CO₂ stimuli (Jones et al., 2007). We examined the transient Ca²⁺signals triggered by CO₂ in the AL region of the brain in living flies. Upon 20-s CO₂ exposure, the GCaMP signals in wt and mutant brains were recorded and analyzed (Figure 4A). The magnitude of GCaMP signal changes in PRL-1 mutant brains was about 2-fold of those of the controls (average peak ΔF/F of PRL-1 mutants: 1.21 ± 0.51; average peak ΔF/F of wt: 0.55 ± 0.46) (Figures 4B and 4C). The overexpression of a UAS-PRL-1 transgene driven by elav-GAL4 in the mutants could restore CO₂-evoked Ca²⁺ sensitivity (Figures 4D and 4E). This result suggests that, in the absence of PRL-1, the neurons in the AL, including those of the V-glomeruli, are hyperactive upon CO₂ stimulation. This phenomenon may unveil neural dysfunction in the nervous system of the mutant.

Figure 4.

CO₂ Triggers Hyperactive Ca²⁺ Activity in the Antennal Lobe of the PRL-1 Mutant Brain

(A) Calcium activity in the antennal lobes of control, PRL-1 mutant, or PRL-1-rescued flies with CO₂ stimulation. Representative images of GCaMP6.0 fluorescence were obtained from living imaging at the baseline and after CO₂ stimuli (%ΔF/F).

(B and C) The magnitude of GCaMP signal changes in PRL-1 mutant brains was about 2-fold that of the controls (average peak ΔF/F of PRL-1 mutants is 1.21 ± 0.51; average peak ΔF/F of WT is 0.55 ± 0.46) (B), also shown in the bar graph quantitation (C).

(D and E) The overexpression of a UAS-PRL-1 transgene driven by elav-GAL4 in the mutants could restore CO₂-evoked Ca²⁺ sensitivity (D), also shown in the bar graph quantitation (E).

Statistics for (C) and (E): Data are expressed as mean ± SD. Two-tailed Student's t test with *p < 0.05, ns, not significant. n = 6 for each group.

Uex Knockdown Results in an Identical Held-up Wing Phenotype

We next asked if any genes downstream of PRL-1 are involved in the wing held-up phenotype. It was reported that cyclin M/ancient conserved domain proteins (CNNMs) interacted with hPRL-1 or hPRL-2 during cancer metastasis (Funato et al., 2014, Hardy et al., 2015). We examined whether PRL-1 also interacted with Uex, the only Drosophila homolog of human CNNMs. S2 cells were transfected with PRL-1 tagged with 3xHA at the N terminus. Cell lysates were subjected to immunoprecipitation with an anti-hemagglutinin antibody and analyzed by western blot with the anti-Uex. In this assay, endogenous Uex was co-immunoprecipitated by PRL-1 (Figure 5A). We performed a biotin pull-down experiment to confirm the interaction between PRL-1 and Uex (Figure S5A). A glutathione S-transferase (GST)-pull-down assay validated the direct interaction between PRL-1 and Uex (Figure 5B). The co-localization of PRL-1 and Uex on the plasma membrane of S2 cells was also detected by double-immunofluorescence staining with the antibodies against PRL-1 and Uex (Figure 5D).

Figure 5.

Uex Knockdown Results in an Identical Wing Held-up Phenotype

(A) Coimmunoprecipitation analysis with transient transfection of either empty or hemagglutinin (HA)-tagged PRL-1 in S2 cells showed that Uex was co-immunoprecipitated with PRL-1.

(B) In vitro glutathione S-transferase (GST) pull-down assay with purified GST-tagged PRL-1 confirmed a direct interaction between Uex and PRL-1.

(C) Western blot with brain extracts at day 5 revealed that Uex expression was noticeably reduced in the uex knockdown samples. Concurrent overexpression of a full-length uex transgene in the uex RNAi background elevated total Uex expression.

(D) S2 cells were subjected to immunofluorescence staining with anti-PRL-1 (red) and anti-Uex (green), and localization of Uex and PRL-1 on cell membranes was observed (arrows).

(E–G) Loss of Uex function (elav/+; UAS-uex-IR/+) in the fly nervous system results in abnormal wing posture (F), compared to the control (E).

Wing held-up phenotype was restored with concurrent UAS-uex expression (elav-GAL4/+; UAS-uex-IR/UAS-uex) (C and G). Scale bar, 100 μm in (D) and 37 μm in (G).

See also Figures S4 and S5.

Given the abnormal wing phenotype induced by CO₂ in PRL-1 mutant flies, and a direct interaction between PRL-1 and Uex, this prompted us to examine the uex phenotype. As uex mutations generated with CRISPR/Cas9 method were larval lethal (Figure S5B), we employed an alternative RNAi approach. An array of different tissue-specific GAL4 drivers was used to evaluate the Uex knockdown phenotype (Figure S4D). As compared to the control (Figure 5E), only knockdown of uex in the nervous system (elav > uex-IR) led to the held-up wing phenotype after eclosion (Figure 5F), which was identical to the phenotype observed in the PRL-1 mutant (Figure 1D). Western blot with brain extracts at day 5 after eclosion revealed that Uex expression was noticeably reduced in the uex knockdown samples (Figure 5C). Concurrent overexpression of a full-length uex transgene in the uex RNAi background elevated total Uex expression (Figure 5C) and rescued the wing phenotype (Figure 5G).

Uex Functions Downstream of PRL-1

To clarify a potential PRL-1/Uex interrelationship, we examined Uex expression in PRL-1 mutant flies. In the wt, the Uex protein was detected as a doublet with a major band of 94 kDa and a faint band of 90 kDa (Figure 6A). We assume that the lower-molecular-weight band was the depredated product of the 94-KDa Uex protein. The expression levels of the Uex protein were significantly decreased in the absence of PRL-1 (Figure 6B). The lower-molecular-weight band was obvious in the newly hatched PRL-1 mutant flies (day 1 after eclosion). However, this lower band almost disappeared in older animals (day 3) (Figures 6A and 6B). It appeared that in the absence of PRL-1, approximately one-third of the Uex protein started to degrade from day 1. At day 3, the lower band was almost undetectable but the full-length Uex band remained unchanged. This observation may explain the reason why 3-day-old mutant flies were more responsive to CO₂ exposure (Figure S1A).

Figure 6.

Uex Functions Downstream of PRL-1

(A) Western blot analysis of protein levels of Uex in day-1 and day-3-old controls, PRL-1 mutants, and Uex knockdown flies. Uex was detected as a doublet in the brains, with a major 94-kDa band and a fairly weak band of 90 kDa. The lower band could be a degraded protein product.

(B) The statistical analysis of the western blot results from triple experiments, Data are expressed as means ±SD. *p < 0.05.

(C) Conditionally induced overexpression of uex by RU486 with elav-Gal4 GeneSwitch was observed in the adult nervous system. The expression of Uex was induced with 0.5 mg/mL RU486 in experimental groups for 2 days, and all flies were exposed with 20 s CO₂ on day 3. Within the next 24 h, the percentages of wing held-up flies were analyzed.

(D) The elevated Uex expression drastically prevented CO₂-induced wing phenotype, and the initial wing held-up phenotype (4 day) was completely reversed over the next few days. Data are expressed as means ± SD. ***p < 0.001. n = 20 for all groups.

(E) The diagram depicts the proposed protective functions of PRL-1/Uex in the nervous system in response to CO₂ stimulation.

Based on these data, we propose that in wt animals, the neuroprotective role of PRL-1 against CO₂ challenge depends on Uex. We reason that although Uex levels are down-regulated in the absence of PRL-1, this Uex level is higher than that of uex RNAi knockdown (Figure 6A, elav > uexIR) and the wing phenotype is not observed in newly hatched adults (Figure 1A). Upon CO₂ exposure, the compromised Uex levels in the nervous system could not antagonize the effect of the stimuli, resulting in the held-up wing phenotype.

In this scenario, the ectopic expression of uex in PRL-1 mutants should block the CO₂-induced wing phenotype. A conditional RU486-dependent elav-GAL4 (GeneSwitch) (Osterwalder et al., 2001) was employed to induce tissue-specific expression of uex in PRL-1 mutants. As RU486 was fed to the newly hatched flies, the Uex protein was elevated in the brains (day 5, Figure 6D). Consistent with our compromised Uex hypothesis, the increased Uex expression comprehensively prevented the CO₂-induced wing phenotype with only about 20% of the flies showing the phenotype, when compared with the 70% of unfed flies (Figure 6C). Interestingly, any initial held-up wing phenotype (day 4) observed in RU486-fed flies was completely reversed over the following few days (Figure 6C). This experiment clearly demonstrates that Uex acts downstream of PRL-1 and protects the nervous system against CO₂ challenge. Ectopic expression of uex in the nervous system not only prevents the wing phenotype development but also reverses the phenotype in individual mutant flies.

Discussion

In this study we demonstrate that in Drosophila adult flies, PRL-1 functions in the nervous system and prevents CO₂-induced neural defects manifested by a held-up wing phenotype. Our genetic rescue data strongly indicate that it is defects in the nervous system that cause the CO₂-induced wing phenotype in PRL-1 mutant flies. No obvious defects in muscles were observed (Figure S2), and ectopic expression of PRL-1 in muscles alone could not rescue the phenotype. The CO₂-induced wing phenotype was triggered initially by the signals from the CO₂ sensory neurons. Specific knockdown of CO₂ receptor protein Gr21a in these neurons fully prevented the wing phenotype in the PRL-1 mutants.

The holding-up of wings in the fly is a behavioral output signal usually indicating avoidance or acceptance of a stimulus. Olfactory CO₂ detection via the receptors Gr21a and Gr63a in the CO₂-reseponsive neurons mediates avoidance (Jones et al., 2007), whereas E409 neurons, a population of taste neurons, have been identified to mediate taste acceptance in flies that are attracted to CO₂ in solution (Fischler et al., 2007). This indicates that compartmentalization of CO₂ detection may allow Drosophila to distinguish local versus global CO₂ levels and finely regulate behavior. In our study, the PRL-1 mutant flies retained normal responses to anesthesia including N₂, volatile ether, and CO₂. However, without PRL-1 protection, the processing of olfactory CO₂ stimulation was rendered defective, resulting in a permanent holding up of wings. Many other gene mutations such as Apterous (Weihe et al., 2001) and Beadex (Kairamkonda and Nongthomba, 2014, Biryukova et al., 2009) could produce held-up curled wings. Parkin and pink mutants also exhibit held-up or drooped wing phenotype due to muscle defects (Yang et al., 2006, Fernandes and Rao, 2011). Such a hold up differs in nature to the PRL-1 case. Here we report that the occurrence of a held-up wing phenotype is caused by gene disruption, which might regulate neuronal homeostasis, and this demonstrates that olfactory CO₂ stimulation is associated with the risk of neurological dysfunction for which PRL-1 provides defense.

Neural expression of shi^ts1 in wt background provides a valuable clue to understand the rationale behind the held-up wing phenotype. Within the recovery time (about 10 min) when flies were shifted back from the non-permissive temperature (29°C) to the permissive temperature (25°C), they exhibited a held-up wing phenotype, which was reminiscent of that observed in PRL-1 mutant flies induced by CO₂ exposure, except that in this case it was transient rather than permanent. As the nervous system is only partially functional during the period of recovery, we conclude that CO₂-induced held-up wing phenotype in PRL-1 mutant flies is most likely due to neural dysfunction. Our data showed that expression of shi^ts1 in motor neurons (D42-GAL4) also induced wing hold-up phenotype, although with a lower penetrance. This could simply be due to the lower-level expression of shi^ts1 in motor neurons.

There are three members of the mammalian PRL family. Drosophila PRL-1 shares high similarities (74%–76%) to all three mPRLs (Zeng et al., 2000, Lin et al., 2013). Bai et al. recently reported that PRL1/PRL2 double knockout mice were embryonic lethal. However, PRL1^−/−/PRL2^+/− and PRL1^+/−/PRL2^−/− mice are viable, suggesting that there is a functional redundancy between PRL1 and PRL2 (Bai et al., 2016). Mice deficient for PRL3 were grossly normal (Zimmerman et al., 2013). Our study reveals that the PRL-1 mutant flies are viable and fertile, even when they occurred with held-up wings, which negated flight for their entire lifespan. Using molecular mapping, we found that PRL-1 was enriched in the V-glomeruli of the AL and the MB of the Drosophila brain. We demonstrate that PRL-1 functions to protect against olfactory CO₂ stimulation. Our study suggests that Drosophila PRL-1 might not be critically required for survival, but essential for the maintenance of the neural homeostasis under stress conditions.

In mammals, PRL-2 regulates intracellular magnesium levels by forming a functional heterodimer with the magnesium transporter CNNM3 (Hardy et al., 2015). However, a substrate-trapping assay revealed that the mutation of catalytic cysteine to serine, or the mutation of aspartic acid to alanine in the WPD motif of PRL-2, did not lead to increased complex formation but to a strong reduction in the binding between the two proteins. This suggests that a catalytically active form of PRL-2 is still crucial for its association with CNNM3. We have also obtained a similar result by using substrate trapping mutants in analyzing the binding of Drosophila PRL-1 to Uex and have confirmed that Uex is not a typical phosphorylated substrate for PRL-1. The physiological substrate of PRl-1 is still unknown. It is possible that Drosophila PRL1 acts both as a trigger of Uex for a particular neuronal pathway and as a lipid phosphatase to maintain an active conformation for additional functions, for example, to control magnesium homoeostasis through the PRL-1/Uex complex.

The CBS pair domain of the magnesium transporter MgtE acts as a magnesium sensor and regulates the gating of the activity of the magnesium-transporting pore (Hattori et al., 2007). To confirm that Uex protein does indeed bind PRL-1 through its CBS domain, we designed a guide RNA targeted to the CBS domain using CRISPR/Cas9 method. We got many mutants, but most of them were lethal. Only one of them was homozygous viable, named uex⁷⁻⁷⁻¹, which caused two amino acids to be turned to one amino acid in the CBS domain (Figure S5C). The disrupted Uex protein extracted from this single mutant line exhibited decreased binding to PRL-1, as revealed by a GST pull-down assay (Figure S5D).

In our study, loss of PRL-1 clearly decreased the expression of Uex. Direct knockdown of Uex resulted in the same wing phenotype as observed in the PRL-1 mutants, whereas abnormal wing posture in PRL-1 mutants could be restored by rescuing Uex expression, particularly in the nervous system. However, we found that the loss of Uex causes fly lethality. In the mouse model, knockout of PRL-1 or PRL-2 only affects the related CNNMs protein. In this case, because the CNNM family has four members, the partial degradation of only one CNNM member is not enough to cause lethality. However, double mutants of PRL-1 and PRL-2 are clearly enough to decrease CNNMs' protein expression, which then causes the lethality of the mouse. Mg²⁺ acts as a physiological Ca²⁺ antagonist for blocking the excitatory N-methyl-D-aspartate receptors in the CNS (Zito and Scheuss, 2009, Iseri and French, 1984) and has therefore been suggested as a possible means of resolving muscle rigidity and spasms in cases of tetanus (Ceneviva et al., 2003). In humans, mutations in CNNM2 cause seizures and mental retardation in patients with hypomagnesemia (Arjona et al., 2014). CNNM4 can regulate Ca²⁺ influx during sperm capacitation (Yamazaki et al., 2016). Although we were unable to measure the Mg²⁺homeostasis status in the PRL-1 mutants and uex-IR flies, enhanced Ca²⁺ activities were induced in the PRL-1 mutants. It would be possible that, if the cations, either magnesium or calcium, were added to the flies, this would affect the CNS homeostasis in Drosophila.

We have achieved a complete rescue in the Drosophila PRL-1 wing phenotype by using either hPRL-1 or hPRL-2 transgenic flies. This may imply that human PRL phosphatases are poised to function in a way similar to that we have shown for neuroprotection in Drosophila. Human PRL-3 has been demonstrated to dephosphorylate lipids and to affect phosphatidylinositol 3-kinase (PI3K) signaling (Wang et al., 2007). Drosophila PRL-1 is also thought to affect phosphoinositide-dependent PI3K-PTEN signaling loop, leading to the spatially restricted synapse formation (Urwyler et al., 2019). For an unknown reason we have found it a technical difficulty to produce hPRL-3 transgenic flies for the rescue experiment. Weather PRL1 in Drosophila acts as a lipid PTP (protein tyrosine phosphatase) in CO₂ neural circuits remains to be illustrated.

In conclusion, we have identified a novel neural protective function of PRL-1/Uex (Figure 6E). In the absence of PRL-1, Uex expression levels are down-regulated. Upon CO₂ exposure, the receptors in the CO₂ sensory neuron send signals to the nervous system, triggering behavioral responses. The AL region of the brain in PRL-1 mutants exhibits hypersensitive Ca²⁺ responses to CO₂ exposure. This hypersensitivity combined with low levels of Uex leads to neural dysfunction, resulting in the held-up wing phenotype. Although primarily recognized for PRL's oncogenic properties in mammals, here we highlight its neuroprotective role in the nervous system, particularly in relation to the CO₂ sensory motor pathway in Drosophila. Our study implies that PRLs may retain a similar neuroprotective function in humans. It also comes to our attention that the phenomena of neurological dysfunction induced by CO₂ insult in PRL-1 mutants resembles the post-traumatic stress disorder in humans, in which transient severe unfavorable stimulating factors cause ongoing neurological dysfunction. Further investigations are needed to confirm the correlation.

Limitations of the Study

Although we have revealed a Prl-1-Uex complex-based neuroprotective mechanism in which Prl-1 protects against nervous system insult related to olfactory CO2 stimulation, any human neuroprotective mechanisms related to the issue of CO₂ toxicity, particularly those relating to olfactory pathways, have yet to be elucidated. It is true that in human brain disorders such as Parkinson and Alzheimer diseases, there is profound olfactory disorder in odor threshold detection, odor memory, or odor identification often occurring before disease onset. These are often associated with aspects of limb dysfunction. However, the reasons and mechanisms of such still remain unknown. The potential role of hPRL-1 in this process requires further study.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: September 27, 2019