Introduction

Drosophila researchers routinely immobilize flies with CO₂ without regard to the physiological consequences. CO₂ anesthesia, however, can have adverse effects: it impairs fertility, suppresses the immune system, and it can negatively impact climbing and flight behavior for hours or days [1, 2]. We demonstrated that CO₂ exposure enhances specific morphogenetic defects in flies with null mutations in Imaginal disc growth factors (Idgfs), a family of six genes that encode Drosophila chitinase-like proteins with orthologs in humans [3]. In addition, CO₂ exposure induces a loss of cortical actin during oogenesis and embryonic development in both wild-type and Idgf-null flies [3]. Wild-type flies proceed to develop normally, but Idgf-null flies do not, suggesting a possible role for Idgfs in protecting against adverse effects of CO₂. The mechanisms for how Idgfs mediate this protection are unknown.

How does elevated CO₂ induce these morphogenetic and cytoskeletal defects? One way could be by disrupting pH homeostasis (CO₂ reacts with water to form carbonic acid [4]). The actin cytoskeleton is extremely sensitive to pH, and pH changes can induce drastic changes in actin organization [reviewed in 5]. Dynamic remodeling of the cytoskeleton drives cell protrusion, cell migration, and tissue morphogenesis and relies on the activity of pH-sensitive proteins such as cofilin [6–8], profilin [9], and Talin [10]. For example, actin polymerization at the leading edge of migrating cells increases with a pH_i >7.2 in mammalian cells [11]; decreasing pH_i or increasing extracellular pH (pH_e) inhibits cell migration [6, 12–14]. Cancer cells reverse their intracellular-to-extracellular pH gradient (pH_e < pH_i) compared to normal cells (pH_e > pH_i), thereby promoting cell protrusion, migration, and metastasis [5, 10, 11, 15, 16]. Furthermore, pH_i dynamics regulate follicle stem cell differentiation in the Drosophila ovary and in mouse embryonic stem cells. [17]. During Drosophila oogenesis, stage-specific changes in pH_i regulate follicle cell development by modifying the cytoskeleton and cell polarity [18, 19].

Few, if any studies have addressed the real-time response of intracellular pH to CO₂ in whole intact Drosophila. The duration and magnitude of large perturbations in pH could severely impact pH-sensitive processes such as cytoskeletal organization and morphogenesis and could be factors in whether, or how quickly, the system can recover from such perturbations.

Here, we sought to characterize the nature of the pH response to elevated CO₂ in living, intact, adult Drosophila using a pH sensor (tpHusion) and imaging directly through the abdominal cuticle. tpHusion is a genetically encoded, ubiquitously expressed pH sensor [20] consisting of a tubulin promoter, two fluorophores—pH-sensitive superecliptic pHluorin and pH-insensitive FusionRed—and a short HRas sequence to tether the sensor to the intracellular membrane (Fig 1A). Whereas the intensity of pHluorin varies positively with the pH, FusionRed is insensitive. A change in the ratio of pHluorin-to-FusionRed intensities indicates a change in intracellular pH (Fig 1B). We imaged pHluorin and FusionRed intensities in the ovarian follicular epithelium directly through the ventral abdominal cuticle (Fig 1D, 1D’ and 1D”) using “Bellymount” (Fig 1C), a noninvasive method for imaging live, intact Drosophila [21].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. pH sensor construct and live-imaging setup.

(A) The tpHusion construct consists of a fusion protein of superecliptic pHluorin (SEpHluorin) and FusionRed cloned under control of a tubulin promoter (alpha-Tub84B), which induces ubiquitous expression. An HRas sequence is included to tether the protein to the plasma membrane [20]. (B) Intracellular configuration of tpHusion. SEpHluorin fluoresces more brightly at high pH than at low pH. FusionRed is relatively insensitive to pH. (C) Imaging setup: a CO₂ chamber consists of a glass-bottom Petri dish with a CO₂ inlet and outlet. CO₂ flow rate was ~3.5 l/m (see Methods). The ventral abdomen of the living fly is glued to the glass-bottom coverslip with UV glue and slightly flattened using a small piece of cut coverslip secured with a dab of vacuum grease at each corner. (D) Schematic showing orientation of ovaries within a fly. (D’) A pair of ovaries and an ovariole pulled out from an ovary. Egg chambers develop in assembly-line fashion from anterior to posterior along each ovariole. (D”) Enlarged view of the egg chamber indicated by the rectangle in D’. Fluorescence intensity was measured within a subset of main body follicle cells, excluding dorsal appendage primordia (indicated by the two curved lines on the dorsal side of the egg chamber). The dorsal appendage primordia develop into tubes that form the dorsal appendages. (D‴) Laid egg with dorsal appendages indicated (not to scale).

https://doi.org/10.1371/journal.pone.0302240.g001

Results

We exposed control (y w;; tpHusion) and Idgf-null (w¹¹¹⁸ Idgf^4Δ; Idgf^{(1Δ dsRed, 2–3Δ, 6Δ, 5Δ)}; tpHusion) flies to a one-minute pulse of 100% CO₂, followed by a recovery period. To sustain a CO₂ flow rate of ~3.5 l/m into the CO₂ chamber, we used a Flowbuddy^™ with a manual on/off switch and controlled the timing using the elapsed time displayed by the Leica software during image acquisition (see Methods). We monitored the change in pH_i (inferred from the ratio of pHluorin/FusionRed intensities) in the main-body follicle cells in mid-stage egg chambers (Fig 1D, 1D’ and 1D”). The pHluorin fluorescence intensity dropped sharply and nearly disappeared for both genotypes within 30 seconds while the FusionRed intensity remained relatively constant (examples in Fig 2A, 2A’, 2B and 2B’, S1 and S2 Videos). The ratios, normalized to 1.00 at the beginning of the CO₂ pulse, showed at least a two-fold decrease (Fig 2A” and 2B”). Fig 2C shows the normalized ratio plots for the control (n = 11) and Idgf-null (n = 11) flies, and Fig 2C’ shows the averaged ratio profiles for each genotype. The initial response (0–2 minutes) was remarkably similar in both genotypes, but the response began to diverge during the recovery phase.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. pHluorin/FusionRed CO₂ response and recovery.

(A, B) Examples of still images extracted from videos (S1 and S2 Videos) of control (y w;; tpHusion) and Idgf-null (w¹¹¹⁸ Idgf^4Δ; Idgf^{(1Δ dsRed, 2–3Δ, 6Δ, 5Δ)}; tpHusion) egg chambers in live, intact, adult Drosophila. Dotted lines indicate ROIs measured within main-body follicle cells. The brightness of the still images was increased for better visualization. Identical adjustments were applied to all images within a series. The videos were not adjusted. (A’, B’) 10-minute time lapse of pHluorin and FusionRed intensities from the same samples shown in panels A and B. (A”, B”) pHluorin/FusionRed intensity ratios normalized to 1.0 just prior to the 1-minute CO₂ pulse and showing the decrease and return to 1.0. Note: The images in panel B were scanned as 512x512-pixel images with an acquisition time of less than five seconds. The images in panel A were scanned as 1024x1024-pixel so that at the same scan speed, the images took double the acquisition time of those in panel B. The longer acquisition captured a small decrease in the pH within the 1-minute frame immediately after the CO₂ was turned on. This decrease is evident in the second image in panel A and in the graphs in A’ and A”, showing a slight drop in the ratio in the frame starting at 1-minute. All of the wild-type videos (S1 and S3–S12 Videos) and Idgf-null videos S13–S20 Videos were scanned as 1024x1024 images. Idgf-null videos S2, S21 and S22 Videos were scanned as 512x512 images. (C) Normalized 10-minute time-lapse of pHluorin/FusionRed ratios for 11 control and 11 Idgf-null flies. Each line represents the profile for one fly. (C’) Averaged normalized pHluorin/FusionRed ratios—averages of the profiles of the 11 control and 11 Idgf-null flies shown in panel C. Error bars represent 95% confidence limits. (D) Recovery time for pHluorin/FusionRed ratios. Each dot represents the amount of time between when the CO₂ was turned off to when the pHluorin/FusionRed ratio returned to 1.0. Error bars represent 95% confidence limits.

https://doi.org/10.1371/journal.pone.0302240.g002

After removal of the CO₂, the pH returned to normal within a few minutes. The recovery time was measured starting at the time that the CO₂ was turned off (at 2 minutes) until the pHluorin/FusionRed ratio returned to a value of 1.0. The intracellular pH in Idgf-null flies recovered faster than in the control (3.2 minutes in Idgf-null vs. 4.7 minutes in control, p = 0.0029, Student’s t-test) (Fig 2D). Note that the recovery time includes the time for the CO₂ to clear from the CO₂ chamber, so the start of normoxia is slightly delayed.

Discussion

We found that pH_i in Drosophila ovarian cells drops sharply, in just a few seconds, in response to a one-minute pulse of 100% CO₂, and then recovers gradually over several minutes after removal of the CO₂. Our findings give a real-time picture of how pH_i responds in a living, intact animal, rather than in dissected or fixed samples. Dissected tissues are removed from their normal microenvironment inside the animal and can potentially be influenced by the pH of the buffer. The time it takes to dissect and fix tissues is problematic for capturing a rapid and dynamic pH response to CO₂ exposure, and the tissue pH can be influenced by the pH of the buffer or mounting media (personal observation).

Despite the ephemeral nature of the CO₂ pulse and the resulting drop in pH, the cytoskeletal and morphogenetic consequences are relatively long-term. The loss of cortical actin we previously observed in both wild-type and Idgf-null flies was apparent in embryos and in ovaries dissected and fixed 30 minutes after a single, one-minute, 100% CO₂ pulse. CO₂ exposure enhanced dorsal appendage defects in Idgf-null flies—dorsal appendages are eggshell structures produced by the follicular epithelium when two patches of cells form tubes and secrete eggshell proteins into the tube lumens. Thus, the dorsal appendages provide a readout for proper tube formation (Fig 1D” and 1D”‘). The defects in Idgf-null dorsal appendages peaked at 8–10 hours after the CO₂ pulse and returned to the basal level by 27 hours [3]. The wild-type flies developed normally despite the loss of cortical actin apparent at 30 minutes post CO₂ exposure.

Considering the sensitivity of the cytoskeleton to changes in pH_i, we hypothesize that these cytoskeletal and morphogenetic defects are a direct result of the pH_i perturbation, and the Idgf-null flies are less robust to perturbations in pH_i than wild-type flies. Evidently a slower recovery to pH homeostasis in control versus Idgf-null ovarian cells is either not a factor in determining normal dorsal appendage development or could actually be beneficial. We propose that, rather than buffering against the pH_i change itself, Idgfs protect against pH_i-induced cytoskeletal modifications by indirectly modulating the cytoskeleton, which is sensitive to pH. Future studies exploring the nature of the pH-induced cytoskeletal changes (e.g., monitoring endogenously expressed fluorescent markers for key proteins such as Idgfs, actin, and actin-modifying proteins) could shed light on this unexpected result.

The rapid drop in pH_i was nearly identical in the two genotypes, yet the pH_i in ovarian cells recovered faster in Idgf-null flies than in control flies. The rapid drop in pH_i likely results from passive diffusion of dissolved CO₂ down a concentration gradient across the cell membranes and a reaction with water to form carbonic acid, which dissociates into hydrogen ions (H⁺) and bicarbonate ions (HCO₃^-). The relatively slower return to homeostasis after removal of the CO₂ depends on transport of ions across the cell membranes and is regulated by carbonic anhydrases and several membrane pumps and transporters [reviewed in 5, 17–19] as well as pH-gated channels [22, 23]. Changes in expression or activity of these proteins could impact the rate at which pH returns to a pre-CO₂ level, but whether Idgfs influence these activities is unknown. Future studies defining the molecular mechanisms of Idgfs should give insight into this process and explain why pH_i recovers more rapidly in flies lacking Idgfs.

We previously tested how different CO₂ exposure regimes affected dorsal appendage development, including a single 1-minute pulse of 100% CO₂, which increased defects in Idgf-null eggs but not in control eggs; a 1-minute pulse of 100% CO₂ every 12 hours for 3.5 days, which increased defects in Idgf-null eggs but not in control eggs; and continuous 20% CO₂ for up to eight days, which produced no increase in defects in either genotype [3]. Note that CO₂ makes up about 0.04% of the ambient atmosphere [24]. Similar experiments monitoring the pH response to different exposure regimes will determine whether the cells can manage to maintain pH homeostasis under lower (e.g., ≤ 20%) CO₂ concentrations and for how long, whether the pH recovery profile is sensitive to the length of the CO₂ stimulus, and whether prior exposure to CO₂ at different concentrations alters the sensitivity of the pH response. A study in mice showed that, after an initial decline in arterial pH, after three days of exposure to 10% CO₂ pH returned to normal as a result of an increase in HCO₃^- due to renal compensation [25]. Could Malpighian tubules, the Drosophila renal organ in insects, play a similar adaptive role in maintaining pH at lower concentrations of CO₂? Our previous finding that dorsal appendage defects are enhanced in Idgf-null flies when exposed to a single 1-minute pulse of 100% CO₂ but not after continuous exposure of 20% CO₂ for several days suggests the possibility that some sort of adaptation could be occurring at lower CO₂ concentrations.

Drosophila sense CO₂ through co-expression of gustatory receptors (Gr21a and Gr63a) in the antennae; these receptors activate specific neurons that affect behavior, i.e., attraction to CO₂ at low concentrations and avoidance at high concentrations [26]. Also, flies sense and avoid acidity through stimulation of different antennal nerves that express Ionotropic Receptor IR64a. CO₂ dissolved in the fluid inside the antennae can form carbonic acid and activate these acid-sensitive neurons [27]. It would be interesting to test whether the pH response is in any way altered in flies lacking these receptors. We predict, however, that the follicle-cell response would not differ in flies lacking these receptors. Such a response would depend on some sort of neural stimulus that would alter follicle cell physiology. Even given that the flies see CO₂ for a whole minute, this scenario seems unlikely. Our hypothesis is that CO₂ simply enters through the spiracles and into the tracheal system, then diffuses into the hemolymph and across cell membranes of the muscle sheath and follicle cells of the ovary. Thus, CO₂ affects pH by entering cells passively rather than by stimulating a G-protein coupled receptor.

One of the challenges in this study was that the proximity of specific egg chambers to the coverslip was variable from sample to sample. Within samples, peristaltic movements of the muscles surrounding the egg chambers produced sometimes dramatic movements of the egg chambers. This movement required adjustment to the imaging parameters (gain, laser intensity, and pinhole) from sample to sample. Longer imaging times for some samples with higher laser intensity could impact the results due to photobleaching after 8–10 minutes. We found that if we could image the samples with low intensities and lower resolution with faster scan times, we were able to avoid significant photobleaching during a 10-minute interval. Variation in the imaging parameters did not affect the results because each pHluorin/FusionRed ratio was normalized to the frame just prior to the beginning of the CO₂ pulse so that the beginning ratio just before the CO₂ pulse had a value of 1.0 for each profile. Therefore, all ratio curves were directly comparable.

Another limitation of this study is that because we used living, breathing, animals, we could not create a calibration curve to relate our pHluorin/FusionRed ratios to a known pH_i within the animal. Our objective, however, was to monitor the change in pH_i, rather than the absolute pH_i.

Understanding pH dynamics in a tractable model organism such as Drosophila can provide a better understanding of pH alterations in human pathologies such as cancer [11], neurodegenerative disorders [28], and several other diseases [29]. In this study we have characterized the intracellular pH dynamics in response to a CO₂ pulse in a live intact animal using the Bellymount method and a genetically encoded, ubiquitously expressed pH sensor. These tools provide opportunities to explore the effects of pH dysregulation in fly ovaries as well as a variety of other tissues [20, 21].

Methods

Supporting information

Acknowledgments

The authors thank Hugo Stocker for the tpHusion fly stock and Todd Nystul for helpful advice on ratiometric imaging. They thank Nathaniel Peters at the University of Washington Keck Center for technical support and advice on imaging.