Numerous cells die during embryogenesis. To ensure that these effete cells do not interfere with development, they must be cleared by engulfment. Much has been learned about the process of cell engulfment from studies in Caenorhabditis elegans that relied on DIC microscopy to visualize un-engulfed cells (Mangahas and Zhou, 2005). More recent studies in C. elegans utilized time-lapse imaging to identify novel cell engulfment and corpse-processing genes and to establish their temporal relationships (Yu et al., 2006, 2008; Venegas and Zhou, 2007; Zou et al., 2009). Mammalian cell culture has been critical to our understanding of cell engulfment in mammals (Erwig and Henson, 2007). However, there is a need to study cell engulfment in vivo, particularly in the context of animals that undergo regulative development. This report describes a new reagent for time-lapse monitoring of cell engulfment in Drosophila embryos, tissues, and cells. In Drosophila embryos, epithelial cells are mostly engulfed by neighboring cells, while cells dying within internal organs, such as the brain, are engulfed by macrophages (Pazdera et al., 1998; Mergliano and Minden, 2003; Robertson et al., 2003; Sears et al., 2003). The molecular connection between cell death and cell engulfment within the context of the living Drosophila embryo is poorly understood. Therefore, it is important to be able to monitor cell death and cell engulfment events in real time.
Until recently, injection of the vital dye, acridine orange (AO), was the primary method for monitoring apoptosis in live Drosophila embryos. Newly developed genetically expressed reporters of apoptosis include: Apoliner and SCAT, which detect caspase 3 activation (Takemoto et al., 2003, 2007; Bardet et al., 2008), and CIETDY, which detects the activity of the initiator Caspase 8, DED (Mazzalupo and Cooley, 2006). These reporters eliminate the need for AO and allow one to observe cell death in different tissues and developmental stages. The engulfment of dead cells in Drosophila embryos has been monitored by injecting VGAL, a reporter for cell engulfment that becomes fluorescent when the cytoplasm of a dying cell is mixed with the lysosomal compartment of the engulfing cell (Mergliano and Minden, 2003). Since VGAL is membrane impermeable, it has to be loaded into cells by injection into syncytial stage embryos, where all cells receive equal amounts of VGAL. Thus, VGAL can only be used to visualize the engulfment of embryonic cells. Moreover, VGAL cannot be used to monitor the engulfment of specific cell types since it is uniformly loaded into all cells. More importantly, because VGAL can only be introduced by injection, it is impractical to use VGAL in large screens for mutations that affect cell engulfment. Recently, a pH-sensitive fluorescent dye, pHrodo, has been used to assess cell engulfment by FACS analysis (Miksa et al., 2009). To circumvent the limitations of dyes such as VGAL, we designed a genetically encoded reporter for cell engulfment that will allow one to study the clearance of apoptotic corpses in live animals.
A hallmark of apoptotic cell engulfment is the fusion of the phagosome with the lysosomal compartment (Odaka and Mizuochi, 1999). We have taken advantage of the acidic environment of the lysosome and employed the pH-sensitive ratiometric derivative of GFP, pHluorin, to report the movement of dying cells' contents into the acidic milieu of the engulfing cells' lysosome. To aid in the visualization of pHluorin in healthy cells, we chose to tether it to the cell cortex via the Moesin actin-binding domain. The cortical localization of pHluorin was hypothesized to reveal morphological changes in apoptotic cells. This pHluorin::moesin actin-binding-domain chimera is referred to as pHMA. Here we show that when pHMA is expressed in cultured Drosophila cells, its ratiometric signal reports that the pH of healthy cells drops slightly from 7.4–7.0 to 6.8–6.6 prior to engulfment, and then drops rapidly to as low as 5 after cell engulfment. As predicted, the number of acidified pHMA bodies increases in embryos with increased cell death and is not present in embryos lacking cell death. Time-lapse analysis of pHMA-expressing embryos reveals a sequence of characteristic morphological changes in apoptotic, epithelial cells prior to and after their engulfment. Thus, pHMA serves as a robust engulfment reporter that will be a useful reagent to study the dynamics of apoptosis and to screen for new mutations that affect apoptotic cell engulfment.
RESULTS AND DISCUSSION
Design of pHMA
To generate a dynamic, genetically encoded reporter for cell engulfment, we relied on pHluorin and the actin-binding domain of Moesin. pHluorin is a pH-sensitive GFP derivative that emits in the usual GFP spectrum, but has a bimodal excitation spectrum; it is maximally excited at 475 nm in acidic environments, and at 395 nm in neutral surroundings (Miesenbock et al., 1998). Therefore, local changes in pH can be detected by taking a ratio of emission intensities at excitations of 470 and 410 nm (R470/410). pHluorin has been used to determine the acidification of intracellular compartments and neurotransmitter uptake (Li et al., 2005). To localize pHluorin cortically, we engineered a fusion of pHluorin and the actin-binding domain of Moesin to generate pHMA. The actin-binding domain of Moesin was chosen because it has been successfully used in the Drosophila embryo in a fusion known as GMA, which consists of GFP fused to the 140–amino acid C-terminal actin-binding domain of Moesin (Edwards et al., 1997; Dutta et al., 2002). GMA reveals the cortical actin network and allows one to visualize cell geometry and morphogenetic events (Edwards et al., 1997). GMA has also been used to examine the role of apoptosis in the squamous epithelial cells of the Drosophila amnioserosa during dorsal closure (Toyama et al., 2008). To allow for tissue-specific and inducible expression in Drosophila, pHMA was placed downstream of the UAS promoter (Brand and Perrimon, 1993). pHMA was also placed under the control of polyubiquitin promoter (Ubi) (Davis et al., 1995) promoter for uniform expression (Fig. 1A).
We hypothesized that pHMA would reveal the cell-shape changes that take place during apoptosis. Some studies suggest that apoptotic cells become slightly acidified prior to engulfment (Barry and Eastman, 1992). pHMA will be useful for determining if this is also the case for Drosophila cells. When an apoptotic cell is engulfed, the cortically localized actin, along with pHMA, will be internalized into the apoptotic phagosome of the engulfing cell where they undergo further acidification as they first fuse with the endocytic compartment and eventually fuse with the lysosomes of the engulfing cell (Fig. 1B).
pHMA Expression in S2 Cells Reveals pH Changes of Living, Dying, and Engulfed Cells
As proof of principle, pHMA was expressed in Drosophila cultured cells and its localization and fluorescence characteristics were monitored over the life-cycle of the cells. S2 and S2R+ cells, which are hemocyte-like, phagocytic cells derived from late-stage Drosophila embryos, were used for this analysis (Stroschein-Stevenson et al., 2006). To establish the relationship between intracellular pH and pHMA's fluorescence response, pHMA-expressing cells were fixed and permeabilized and the pH was adjusted with phosphate buffers ranging from pH 4.5 to 8.5, in 0.5-pH intervals. The ratio of pHMA emission when excited by 470 and 410 nm light (R470/410) was determined at each pH interval (Fig. 2A). Nonlinear regression analysis was used to generate a standard curve of pHMA R470/410 relative to pH (r2 = 0.998). The resulting model was used to approximate the pH of individual living, dying, and engulfed cells observed during time-lapse recordings (Fig. 2B). Time-lapse recordings of S2 cells that were transfected with Ubi-pHMA for two days were made over 24-hr periods with images taken at 10-min intervals. Over 400 S2 cells transiently expressing pHMA were analyzed. During the 24-hr imaging period, 48 pHMA-expressing cells (11%) were engulfed, while 30 (7%) suddenly disappeared, suggesting a necrotic demise. The number of engulfed pHMA-positive cells did not appreciably accumulate from the beginning 31/418 (7%) to the end of the recording period 40/443 (9%), indicating that the corpses were completely processed over time, and that the exposure conditions of the 10-min time-lapse cycle did not induce photo-toxic apoptosis. The pH of cells that were morphologically healthy stayed between 7.0–7.4 (cell a: Fig. 2B–D, C′, D′; see Supp. Movie 1, which is available online). Most, but not all, of these cells had cortically localized pHMA. Healthy cells displayed variable morphology that ranged from highly crenated to relatively round with long processes. Cells that were eventually engulfed or ruptured displayed decreased motility, rounded up, and the expressed pHMA frequently transitioned from cortical to cytoplasmic. These changes were accompanied by slight acidification to 6.8–6.6. Acidification of cytoplasm has previously been described as a component of apoptosis (Barry and Eastman, 1992). For example, in U937 cells the pH has been observed to vary from 7.2 in normal to 6.8 in pre-apoptotic to 5.7 in apoptotic cells (Nilsson et al., 2003, 2006). About 40% of the slightly acidified cells rapidly and completely lost their pHMA signal (cell b: Fig. 2B, E–G, E′–G′; Supp. Movie 2); this loss is presumed to result from the loss of membrane integrity since the non-fluorescent, poorly refractile cell corpses were still visible after the loss of pHMA signal, indicating necrosis. The frequency of cell rupture was greatly increased by adding cell death inducing agents such as 10 μM actinomycin D (data not shown). Cells that were engulfed, acidified over 30–60 min to a much greater degree, pH <6, than pre-engulfment. Most engulfed corpses remained acidified at pH 5 or below for several hours, while the phago-lysosome decreased in size over time (cell c: Fig. 2B, H–K, H′–K′; Supp. Movie 3). The pH of the phago-lysosomes appeared to rebound; most cell corpses partially re-neutralized (cells d and e: Fig. 2B, L–O, L′–O′; Supp. Movie 4). The biological significance of this phagosome pH recovery is not clear at this point, but will require further examination. All of the dying cells were engulfed whole. Over the 24-hr recording period, we observed approximately 140 partial engulfment events that possibly involved the pruning of cell processes from healthy cells or the engulfment of detached cell fragments. As will be described in the in vivo analysis of pHMA, this wholesale engulfment of dying cells by S2 cells differs from the piecemeal engulfment of epithelial cells seen in live embryos. This may be due to the experimental conditions or to the hemocytic nature of S2 cells. These data clearly show that pHMA-expressing S2 cells provide a useful platform for studying cell engulfment in vitro.
The In Vivo Pattern of pHMA Acidification Mirrors Known Apoptosis and Engulfment Reporters
To confirm that pHMA can serve as an in vivo cell engulfment marker in Drosophila embryos, the pattern of pHMA acidification was compared to the known injectable cell death and engulfment reporters, acridine orange (AO) and a fluorogenic marker, VGAL, respectively (Fig. 3A) (Namba et al., 1997; Pazdera et al., 1998; Mergliano and Minden, 2003). Cell death and engulfment initiate during embryonic stage 12, when retraction of the germ band begins (Schock and Perrimon, 2002). As in mammals, Drosophila embryonic-cell engulfment is performed by both professional phagocytic cells, macrophages, and neighboring cells in the epithelium (Abrams et al., 1993; Pazdera et al., 1998; Mergliano and Minden, 2003). Time-lapse analysis has shown that epithelial cell death and engulfment occurs in a characteristic segmented pattern during germ band retraction and dorsal closure, stages 12–14 (Fig. 3A, arrows and Supp. Movie 5) (Abrams et al., 1993; Pazdera et al., 1998). Epithelial cell engulfment is performed mostly by neighboring cells rather than dedicated phagocytes (Pazdera et al., 1998). The apoptotic AO signal appears first, followed by the cell engulfment signal from VGAL (Fig. 3A; Mergliano and Minden, 2003). Phagocytosis of apoptotic cells is piecemeal as evidenced by the limited overlap between AO- and VGAL-positive vesicles (Fig. 3A, inset). This indicates that different phagocytic compartments contain variable amounts of apoptotic nuclear and cytoplasmic material. Time-lapse analysis allows one to distinguish between AO and VGAL signals emanating from epithelial cells and macrophages based on their size and motility. Epithelial cells have smaller and less motile AO and VGAL bodies (Supp. Movie 5).
To generate strong, uniform pHMA expression, UAS-pHMA embryos were injected with purified GAL4VP16 protein during the syncytial blastoderm stage (Fig. 3B) (Cambridge et al., 1997). Since pHMA is based on the original GFP allele, not the enhanced GFP derivative, it requires about 2 hr to mature to its fluorescent form (Heim et al., 1995). Thus, pHMA that is expressed in the cellular blastoderm, stage 5, begins to fluoresce around stage 12. The majority of the experiments reported here were done on a wide-field, deconvolution microscope system. Relatively low, 20× magnification recordings were used to assess the pattern of pHMA expression and acidification. This magnification was not sufficient to visualize the cortical localization of pHMA. Acidified pHMA is more strongly excited by 470 nm light, which showed a stronger increase in fluorescence of both diffuse and punctate structures during germ band retraction (Fig. 3B, 470-nm row). Excitation at 410 nm, which preferentially excites pHMA at neutral pH, displayed a gradual and slightly mottled increase in fluorescence over time (Fig. 3B, 410-nm row). This diffuse fluorescence seen both in the 410- and 470-nm excitation wavelengths may be due to variation in cell density or expressivity. Ratioing the 470- and 410-nm images, which biases toward acidified pHMA, cancelled out the diffuse signal and revealed a punctate pattern that closely resembled the spatial and temporal patterns of the AO apoptotic pattern and VGAL engulfment pattern (Fig. 3B, R470/410 row). Notice that the segmental pattern of pHMA R470/410 is similar to the segmental patterns of AO and VGAL (Fig. 3A and B, arrows; Supp. Movie 6). Further, the size of AO-positive, VGAL-positive, and acidic pHMA-positive vesicles is also similar (Fig. 3A and B, insets). These data show that pHMA behaves as expected in wild-type embryos consistent with the engulfment of apoptotic cells.
Increasing Cell Death Increases the Number of Acidified pHMA Bodies
To validate the use of pHMA as an in vivo engulfment reporter, we tested if pHMA is able to indicate both increases and decreases in cell engulfment. To determine if pHMA can detect increases in cell engulfment, apoptosis was induced in a subset of embryonic cells by over-expressing UAS-reaper (UAS-rpr) and UAS-Wrinkled (UAS-W), along with UAS-pHMA, under the control of engrailed-GAL4. At 60× magnification, the cell cortices of wild-type, engrailed-positive cells that express pHMA are visible when excited with both 410- and 470-nm light (Fig. 4A–D; Supp. Movie 7). It is important to note that the background threshold setting for these ratio images was set to that of the non-engrailed-expressing cells. Cell cortices appear as mid-grey, while acidified pHMA vesicles appear as white (Fig. 4A–D, arrows). R470/410 images of engrailed cells expressing rpr, W, and pHMA reveal that pHMA appears in diffuse, irregular stripes with numerous bright vesicles (Fig. 4E–H and Supp. Movie 8). The number of acidified pHMA-containing bodies in engrailed cells expressing rpr and W was much greater than the number observed in wild-type engrailed cells. The acidified pHMA bodies in rpr, W-expressing cells were similar in morphology to vesicles seen in wild-type embryos; however, they were slightly larger, indicating that the removal of this larger than usual population of dying cells may be perturbed. There was also a significant increase in the number of macrophages that contained acidified-pHMA bodies in embryos with increased apoptosis (data not shown). Only a few of the rpr, W, pHMA cells retained their normal cortical pHMA localization (Fig. 4E–H, arrowheads). Since the rpr, W, and pHMA genes are turned on simultaneously, most of the affected cells are dead or dying by the time pHMA becomes strongly fluorescent.
Loss of Cell Death Abrogates Acidified pHMA Bodies
To assess if pHMA can detect decreases in cell death, we utilized the H99 mutation, which is missing the pro-apoptotic genes rpr, W, and grim (White et al., 1994). H99 homozygous embryos exhibit virtually no apoptosis (White et al., 1994). Embryos heterozygous for this mutation develop normally and do not have any cell death defects. Ratio images of UAS-pHMA; H99/TM6B embryos injected with GAL4VP16 showed that these heterozygous embryos had wild-type patterns of acidified pHMA signal in the epithelium and macrophages (Fig. 5, left column, arrows and arrowhead). UAS-pHMA;H99/H99 embryos injected with GAL4VP16 had normal levels of pHMA expression in the individual pHMA channels (data not shown), but the R470/410 ratio images showed an almost total lack of acidified pHMA in epithelial cells and macrophages (Fig. 5, right column). The lack of pHMA signal in H99 homozygous embryos supports the hypothesis that pHMA will detect events that are associated with apoptosis. Taken together, these data clearly demonstrate that pHMA will be useful for detecting mutations and treatments that alter the normal pattern of cell death and engulfment.
Autophagy Does Not Result in Acidification of pHMA
Since pHMA is acidified in the lysosome after cell engulfment, it is pertinent to ask if pHMA also detects autophagy, a process that also moves cytoplasmic material into the lysosome. In Drosophila embryos, autophagy is prevalent during dorsal closure, particularly in the amnioserosa. During dorsal closure, this entire extra-embryonic epithelial tissue is removed via apoptosis and autophagy (Toyama et al., 2008). Interestingly, in H99 mutants, the amnioserosa is eventually removed solely by autophagy that is readily detectable from late dorsal closure onward (Mohseni et al., 2009). We monitored pHMA fluorescence in H99 embryos throughout dorsal closure, and detected no acidified pHMA. In embryos past 50% dorsal closure, the acidification of pHMA in amnioserosa cells is clearly visible in H99/TM6B embryos, which have normal levels of apoptosis and autophagy (Fig. 6A and B, left column). In similarly staged homozygous H99 embryos, pHMA acidification was not observed, indicating that pHMA only indicates apoptotic cell engulfment, not autophagy (Fig. 6A and B, right column and Supp. Movie 9). One possible reason that pHMA does not indicate autophagy may be due to its cortical localization and that autophagy primarily targets non-cortical, cytoplasmic components, such as mitochondria. Other internalization events, such as endocytosis and multi-vesicular body formation, rarely introduce cortical cytoplasm into the lysosomal compartment. If such processes do exist, the flux of cortical cytoplasm into the lysosome is below our limit of detection since healthy, pHMA-expressing S2 cells do not show any internal, acidified pHMA. Thus, pHMA appears to be specific for monitoring changes associated within apoptotic cells and the consequences of their engulfment.
pHMA Reveals Morphological Changes Prior to and During Apoptotic Cell Engulfment
Apoptosis and cell engulfment are highly dynamic processes. Understanding the cell morphology changes associated with these processes in the context of the living tissue will be very helpful in determining the timeline of cell elimination. Since pHMA is visible in both healthy and apoptotic cells, we are able to track the morphological changes prior to engulfment by running time-lapse movies in reverse. To study these morphological changes, we recorded high resolution (60× magnification), high time sampling (30-sec interval) time-lapse movies of wild-type, UAS-pHMA embryos injected with GAL4VP16. Under these extreme recording conditions, we found that the 410-nm light used to detect neutral pH pHMA had phototoxic effects; on the other hand, embryos were unaffected by the 470-nm light used under these time-lapse conditions to visualize acidified pHMA. Imaging with 410-nm light at lower magnification and longer time-lapse intervals (>2 min) had no effect on embryo development. The lower resolution recordings revealed that engulfed corpses appear as bright vesicles. To trace the morphological changes that give rise to these bright spots, the time-lapse recordings were played in reverse from the time a pHMA spot first appeared to when it converts back into an intact, normal-sized cell. Since these high spatial- and temporal-resolution recordings were made only with 470-nm excitation light, this section only comments on changes in overall pHMA signal. It is not known if these changes are related to changes in acidification or concentration.
About 50 min prior to its engulfment, apoptotic cells appeared to “drop” basally by 2–10 μm, but remained within the epithelial cell layer (Fig. 7A, −50 min, red shaded cell; for a marked version, see Supp. Fig. 1). This cell dropping was marked by an accumulation of cortical pHMA around the dropping cell and an increase in diffuse cytoplasmic pHMA (Fig 7A, −40 min). It is not certain if this cortical pHMA accumulation is in the dying cell or the neighboring cells. These observations are consistent with studies in chick embryos describing a ring of actin and myosin within apoptotic cells that were extruded and within the surrounding cells (Rosenblatt et al., 2001). Once the dying cells had begun to drop, they exhibited membrane blebbing, as suggested by the irregular shape of the dying cell and partial spatial overlap with the engulfing cell (Fig. 7A, −15 to −5 min, arrows). The blebbing cells accumulated pHMA internally, depleting pHMA from the cortical surfaces of the blebs (Fig. 7A). The first signs of internalization of apoptotic cell fragments by neighboring cells occurred 5 to 15 min prior to the formation of a discrete vesicle, or phagosome, within the engulfing cell (Fig. 7A, −15 to 0 min, arrows). Initially, pHMA associated with the phagosome appeared both cortically and as internal foci (Fig. 7A, 0 to 30 min, arrows). At this resolution, it was not possible to determine if the cortical pHMA surrounding the phagosome was on the cytoplasmic surface or internal to the phagosome. Forty minutes after engulfment, the pHMA appeared diffuse and uniformly distributed throughout the phagosome, suggesting that the phagosome had completely fused with the lysosome and that breakdown of the engulfed corpse had begun. pHMA persisted in the engulfed vesicle for many hours after engulfment. The remainder of the apoptotic corpse that was not engulfed by the cell shown in Figure 7A was eventually removed by other neighbor cells and later by a macrophage (data not shown).
pHMA Acidification In Vivo Occurs After Cell Corpse Engulfment
To determine when phagosomes of engulfed cell corpses become acidified, the high-magnification time-lapse recordings were repeated at 2-min intervals with 410- and 470-nm illumination, conditions that allowed wide-field imaging for several hours without adverse effects. Between the period of apoptotic cell dropping and the final removal of apoptotic corpses, the fluorescence intensity of pHMA increased both in the corpses and the newly engulfed vesicles (Figs. 7A and 8A, asterisks and arrow), but showed little change in the ratio of acidified pHMA, indicating this increase in fluorescence was mostly likely due to cell shrinkage rather than acidification (Fig. 8A, R470/410). The signal from yolk auto-fluorescence interferes with our ability to make fine R470/410 ratio calculations so that subtle pH changes prior to engulfment went undetected. The engulfed bodies appeared to become strongly acidified about 30–60 min after cell engulfment (Fig. 8A, R470/410, arrows). The time of acidification correlates with the transition from localized pHMA within engulfed vesicles to uniform distribution of pHMA within the phagosomes, indicating that it takes this time for the phagosome to fuse with lysosomes and begin breaking down the contents of the phagosome. Our analysis of pHMA in S2 cells showed that pHMA acidification generally took 30–60 min for the apoptotic cell pH to drop from 6.8 to below 6, the pH we consider to be strongly acidified.
It is known that wild-type multimeric GFP-like proteins can survive in the lysosomal environment (Katayama et al., 2008). The diffused pHMA signal persisted for several hours within pHMA-positive vesicles. During their maturation, engulfed vesicles became smaller and showed a decrease in the pHMA signal (Fig. 8A, 100 min). While it is difficult to follow individual apoptotic cells and their remnants for long periods of time due to the embryo's large-scale morphogenetic movements, we have observed acidified pHMA vesicles persist for more than 5 hr (data not shown). The same longevity of engulfed pHMA was seen in S2 cells.
The delay between apoptotic cell engulfment and acidification of the resulting phagosome is not unexpected and has been observed in other situations. The timeline of cell engulfment and apoptotic phagosome maturation in the epithelium of Drosophila embryos (Fig. 7B) is comparable to the engulfment schedule observed in other systems. In C. elegans, the majority of corpses that appear refractile with DIC microscopy disappear within 30 min of their appearance (Yu et al., 2006). Time-lapse fluorescence imaging studies in C. elegans have shown that lysosomes are recruited to the phagosomal surface within 10 min of engulfment, that lysosomal fusion occurs 20 min later, and that phagosomal degradation takes place 50–70 min after engulfment (Yu et al., 2006, 2008; Zhou and Yu, 2008). Mammalian cell culture systems show more rapid acidification of phagosomes. Murine peritoneal and bone marrow macrophages acidify phagosomes to a steady state of pH 5 within 15 min of engulfment (Underhill, 2005; Yates and Russell, 2005; Haggie and Verkman, 2007). Another study observed phagosome acidification rates by non-professional and professional phagocytes, using LysoTracker, with apoptotic cells and various opsonized beads to be on the order of 60–90 min (Erwig et al., 2006). These studies show that different cell types process engulfed material, such as apoptotic corpses, bacteria, and beads, at different rates. The engulfment schedule for Drosophila embryonic epithelial cells appears to be within known limits.
Timeline of pHMA Acidification With VGAL and AO Signaling
Prior to the development of pHMA, VGAL was used to monitor cell engulfment events. To determine the correlation between VGAL and pHMA acidification, VGAL and GAL4VP16 were injected into UAS-pHMA embryos and time-lapse recordings of VGAL fluorescence and pHMA ratios were made. Figure 8B shows a vesicle about 40 min after engulfment where the pHMA was strongly acidified. At this time, there was very little VGAL signal. Over the next 90 min, the pHMA signal collapsed into a smaller volume and decreased in intensity. There was a concomitant increase in VGAL fluorescence, indicating that the VGAL signal matures more slowly than pHMA acidification. This lag between acidification and VGAL maturation is not surprising since proton diffusion is very rapid compared to the enzymatic degradation of VGAL to its fluorescent form by lysosomal β-galactosidase (Minden, 1996). Like pHMA, the VGAL signal persisted for long periods of time in the developing embryo. VGAL fluorescence persists much longer than pHMA because its fluorescent moiety, resorufin, is not bio-degradable.
AO has been prominently used in Drosophila to identify apoptotic corpses. Our previous study determined that there is a 15–45-min delay between the initial appearance of AO and subsequent VGAL fluorescence (Mergliano and Minden, 2003). It is desirable to also determine the temporal relationship between pHMA acidification and AO signaling. However, since pHMA and AO have overlapping fluorescence emission wavelengths, it was not possible to record pHMA acidification and AO fluorescence simultaneously. We surmise that if VGAL fluorescence peaks about one to two hours after the initial appearance of engulfment vesicles and that AO precedes VGAL by 15–45 minutes, then one can conclude that AO fluorescence appears well after initial engulfment events. This estimate contradicts previous claims that AO fluoresces in un-engulfed apoptotic cells (Abrams et al., 1993). It is important to note that this earlier study used single time-point analysis of AO-stained embryos and focused on interstitial cell bodies that may have been either macrophages or sloughed-off brain cells. Our time-lapse analysis of the engulfment of apoptotic epithelial cells by neighboring epithelia is more akin to studies in C. elegans where AO stains corpses after they are engulfed (Kinchen et al., 2008). All engulfment in C. elegans is mediated by neighboring epithelia, not macrophages. Mutations in C. elegans that prevent phagosomal maturation or acidification lack AO staining. A timeline of the events surrounding the engulfment of a dying cell is diagrammed in Figure 7B. This timeline was assembled from several different time-lapse experiments as described in the preceding sections. This timeline will be helpful when analyzing mutations or treatments that affect different aspects of apoptosis and engulfment.
pHMA Reveals Changes in Larval Tissues
To demonstrate the broad utility of pHMA, it was expressed in larval wing imaginal discs and salivary glands (Fig. 9). The wing imaginal disc is a model system to study and identify apoptotic signals (Umemori et al., 2009) as well as organ size regulation and cell competition (de la Cova et al., 2004; Moreno and Basler, 2004; Li and Baker, 2007; Martin et al., 2009). In wild-type discs, apoptosis occurs at low levels and is interspersed throughout the disc (Milan et al., 1997). The absence of fast-moving macrophages and the amenability to genetic manipulation that allows to induce or suppress apoptosis and cell engulfment (Li and Baker, 2007) make an imaginal disc an attractive model for studying cell engulfment.
en-Gal4 was used to drive the expression of UAS-pHMA in the posterior compartment of the developing wing imaginal disc (Fig. 9A and B). Time-lapse imaging of the wing pouch of the dissected disc reveals relatively uniform pHMA expression across the posterior compartment. A few cells displayed increased pHMA concentration, which was represented by an increase in both 410- and 470-nm fluorescence signal, and R470/410 was unchanged. Soon after the initial increase in neutral pH pHMA, the R470/410 signal increased, indicating acidification. The pHMA signal from the presumably engulfed apoptotic cell was later extinguished (Fig. 9A′, B′, A″, B″, and Supp. Movie 10). The magnification used for these recordings was not sufficient to resolve individual cells and cell fragments. These time-lapse data show that pHMA will be a useful reporter for cell death and engulfment in the wing disc. The dynamics of the pHMA response appears to be similar between embryos and wing discs. Like the embryo epidermis, the engulfment of dying wing epithelia appears to be neighbor dependent.
Another Drosophila tissue that has served as a model for cell death is larval salivary gland. Salivary glands from wall-crawling Ubi-pHMA larvae and white pre-pupae were dissected and mounted for time-lapse microscopy. The pHMA signal in these late larvae/early pupae is mostly cortical (Fig. 9C and D). There appears to be some fibrillar and vesicular accumulation of pHMA. Under the culture conditions used in this experiment, the signal from 470-nm excitation is slightly higher than the 410-nm signal in the salivary gland relative to the attached fat body, suggesting that the salivary gland is acidified relative to the adjacent fat body. Further studies that reproduce the anterior-to- posterior wave of ecdysone-induced caspase activation (Takemoto et al., 2007) will be instrumental in relating caspase activity to pHMA acidification and engulfment in the salivary gland.
Autophagy is prevalent during histolysis of the salivary gland (Lee and Baehrecke, 2001), yet it has also been shown that histolysis is mediated by an apoptosome-dependent process that is parallel to or downstream of autophagy (Akdemir et al., 2006). While pHMA does not appear to report autophagy in embryos, further detailed studies of pHMA in salivary glands undergoing histolysis will be interesting for determining the relative contributions of autophagy and engulfment to the histolysis of the salivary glands.
The expression of pHMA, a fusion of pHluorin to actin-binding domain of Moesin, in cultured Drosophila cells and embryos has proved to be a useful marker for apoptotic cell engulfment. In healthy cultured cells, pHMA resides in the cell cortex and reveals the cell's dynamic morphological behavior. It also reports on the healthy cell's neutral pH. When cultured cells begin to die, the cells become rounded and quiescent and the cytoplasmic pH drops to about 6.7. Some of these apoptotic cells burst and lose their pHMA signal, while others are engulfed. Thirty to sixty minutes after engulfment, the engulfed pHMA reports the acidification of the phagosome to about pH 5. The acidified pHMA signal persists for several hours, but the pH of the phagosome tends to partially neutralize over time. Expression of pHMA in embryos shows similar changes over the life cycle of the cell, the main difference being that apoptotic cultured cells are engulfed as a single entity by one phagocytic cell, while dying epithelial cells are removed in a piecemeal fashion by multiple neighboring cells, as well as macrophages. pHMA faithfully reports both increases and decreases in embryonic cell death and discriminates between apoptotic cell engulfment and autophagy. Overall, pHMA provides a tool for monitoring several different events during apoptosis in a characteristic time sequence. These features make pHMA a useful reagent for screening for new mutations that affect embryonic apoptosis and cell engulfment. The timeline, which shows that morphological changes occur as early as 50 min prior to engulfment and require several hours to completely degrade the cell corpse, will be a useful tool for the detailed characterization of existing and yet-to-be-discovered mutations. Likewise, expression of pHMA in cultured cells will allow one to screen for treatments that affect apoptosis and engulfment in vitro. It will also be interesting to place the other in vivo apoptosis reporters within this timeline to see when they are activated relative to the changes indicated by pHMA. Finally, we anticipate that pHMA will also be a useful reporter for cell death and engulfment in other organisms.
DNA Constructs and Fly Stocks
UAS-pHMA was constructed by replacing the GFP portion of pUAST-GFP::Moesin (GMA) (Bloor and Kiehart, 2001; Dutta et al., 2002), a plasmid provided by D. P. Kiehart (Duke University, Durham, NC), with pHluorin, which was provided by G. Miesenböck (Yale University, New Haven, CT). A 727-bp fragment of pHluorin was subcloned into UAS-GMA vector, to replace GFP between EcoRI and HindIII sites. To generate the Ubi-pHMA reporter, we replaced the GFP fragment in the pCaSpeR4 Ubi-GFP plasmid (Davis et al., 1995) with the pHluorin::moesin fusion. Transgenic Drosophila were generated using standard procedures (Baek and Ambrosio, 1994) by BestGene Inc. (Chino Hills, CA). Transgenic stocks with strongest expression were recombined to generate stocks that have two copies of UAS-pHMA (2xUAS-pHMA) on either the second or third chromosomes. Progeny of recombined third chromosome insertion, Cyo/Bl;2xUAS-pHMA/TM6B, were used for Gal4VP16 injections. Engrailed-Gal4 and H99/TM3 (Bloomington Stock Center, Bloomington, IN) were used to generate en-Gal4/Cyo;UAS-pHMA/TM6B and 2xUAS-pHMA; H99/TM6B, respectively. UAS-rpr, UAS-W on X was provided by J.R. Nambu (University of Massachusetts, Amherst, MA).
S2 Cell Culture, Transfection, and Fixation
S2 and S2R+ cells (Drosophila Genomics Resource Center, Bloomington, IN) were grown in Modified Schneider's Drosophila Medium (Lonza, Basel, Switzerland; 04-351Q) supplemented with 10% heat inactivated Fetal Bovine Serum (Lonza 14-503E) and pen/strep (HyClone, Logan, UT; SV30010). One million cells per well, in a 6-well plate, were transiently transfected with Ubi-pHMA, using Effectene transfection Reagent and adherent cells protocol (Qiagen, Valencia, CA). For each transfection, 1 μg of DNA, 8 μl of Enhancer, and 25 μl of Effectene were used. Using this approach about 1/10 of cells expressed pHMA within two days of transfection. Cells retained pHMA for up to two weeks.
S2R+ cells were fixed with 4% paraformaldehyde in PBS for 20 min at 4°C with gentle rocking. The fixed cells were pelleted, resuspended in PBS, and permeabilized with 30 μg/ml digitonin in PBS for 5 min at room temperature. The pH was adjusted to 4.5–8.5, in 0.5-U increments, by diluting the cell suspension in two volumes of 100 mM sodium phosphate buffer.
pHMA Fluorescence Versus pH Standard Curve Construction
Fixed and pH-adjusted S2R+ cells were imaged with 410- and 470-nm excitation. For each pH value, twenty fields of cells, each containing an average of 5 cells per field, were recorded. Three focal planes 2 μm apart were imaged by wide-field microscopy with a 40× lens for each field of view. A MATLAB script was created to calculate the R470/470 ratios for each field of view. The exposure was adjusted to avoid pixel saturation. The median value for each image was used as background. To calculate the ratio, the background value at the corresponding excitation wavelength was subtracted from each image, and R470/410 of pixel values that were greater than three times the background but less than saturation were computed. The average R470/410 and standard deviation was calculated for each set of images. The data points were plotted using GraphPad software and fitted with a nonlinear curve pH = 6.379-log10(((0.6173–2.077)/ (R470/410-2.077))-1), r2 = 0.9981.
Live S2 and S2R+ cells were imaged using microscope exposure settings that were identical to fixed cells. These cells were exposed with 410 and 470 nm at 10-min intervals. Areas that contained individual cells were cropped. The average of the background was calculated for an entire movie before cropping. This background was used for all cropped images in the movie. The pH of individual cells was calculated at each time point using the computed R470/410 and the above equation.
Dechorionated embryos were prepared for injection as previously described (Li et al., 1999). Stage-3 embryos were injected with AO (0.5 mg/ml), VGAL (∼1.8 mg/ml), or purified Gal4VP16 protein (0.2 mg/ml) as previously described (Mergliano and Minden, 2003). Post-injection, embryos were incubated at 25°C for 5–6 hr or at 15°C for 18–20 hr prior to time-lapse imaging.
Salivary Gland and Wing Imaginal Disc Preparation
The wing discs were collected from wall-crawling larvae, while the salivary glands were collected from either wall-crawling larvae or white pre-pupae. Wing imaginal discs and larval salivary glands were dissected in supplemented Schneider's medium that was also used for S2 cell culture. The discs were mounted in Schneider's medium on the gas-permeable membranes of petriPERM dishes. The samples were surrounded with halocarbon oil to limit desiccation and covered with a No. 1 coverslip. Time-lapse recordings were made with a wide-field microscope system.
Wide-field, time-lapse images were collected on a Delta Vision imaging system controlled by softWoRx software (Applied Precision, Issaquah, WA) configured around an Olympus IX70 inverted microscope, using a filtered Xenon white light source. Images of VGAL (resorufin-β-galactoside-polyethylene glycol 1,900) and AO images were obtained with a standard “quad” dichroic filter set, using FITC and TRITC emission filters, respectively. The 410- and 470-nm excitation wavelength images were collected using a custom dichotic filter set (Chroma Technology, Rockingham, VT); the images were collected with a FITC emission filter. Confocal images were collected with a Carl Zeiss (Thornwood, NY) LSM 510 Meta/ UV DuoScan Spectral Confocal Microscope mounted on an inverted Axio Observer Z1 microscope frame equipped with a 405-nm laser diode (30 mW); 458-, 477-, 488-, and 514-nm Argon laser (30 mW all lines); and 561-nm DPSS laser (15 mW). We found that the light intensity of the wide-field microscope system was less phototoxic than the confocal microscope. The confocal was used for the pHMA plus VGAL experiments because this system was capable of near instantaneous line-by-line automatic switching of each excitation laser via an accusto-optical tunable filter (AOTF) in conjunction with a quad-line dichroic mirror.
Processing of pHMA-Expressing Embryo Images
Wide-field ratio images were generated by dividing individual pixel values and rescaling the output minimum-maximum. Images of embryos acquired with a 20× lens were processed using softWoRx software to subtract local background, which included auto-fluorescence that was generated by the yolk during 410-nm excitation. To approximate the background, the images were blurred using a mean 2D Filter with kernel size of 45 pixels. The reference images were smoothed with a median 2D Filter with a three pixel kernel size. The background-subtracted images were generated by subtracting blurred images from the reference images. A constant value was added to the background-subtracted images to eliminate negative pixel intensities. The resulting images were then divided, 470-nm excitation by 410-nm excitation, to generate a ratio image. The images in Figure 7 were processed to enhance phagocytic vesicles in ImageJ using an FFT bandpass filter set to “Filter Large Structures Down” to 15 pixels and “Filter Small Structures Up” to 0 pixels. The products were subsequently smoothed to suppress graininess. In Figure 9, to determine R470/410 only for the posterior region of wing disc that expressed pHMA, the ratio images were calculated using the MATLAB script that was generated for S2 cell analysis. Ratios for points with signal at 470-nm excitation that was greater than four times the background level was used. To remove noise, the ratio images of the wing imaginal disc in Figure 9 were median filtered with a radius of two pixels.
We thank Olga Strachna for technical assistance and Brooke McCartney and the Minden laboratory for their critical reading of the manuscript. We are indebted to Dr. Bino John for help with MATLAB programming and high throughput image analysis. This work was supported by a grant from the National Institutes of Health (1R21 HD049554) to J.S.M. and an American Cancer Society Postdoctoral Fellowship (PF-07-118-01-DDC) to E.F. J.F. also acknowledges financial support through the NIH National Technology Center for Networks and Pathways (5U54 RR022241) funded through the NIH Roadmap for Medical Research.