Phosphatidylserine (PS) appears on the outer membrane leaflet of cells undergoing programmed cell death and marks those cells for clearance by macrophages. Macrophages secrete lactadherin, a PS-binding protein, which tethers apoptotic cells to macrophage integrins.
We utilized fluorescein-labeled lactadherin together with the benchmark PS Probe, annexin V, to detect PS exposure by flow cytometry and confocal microscopy. Immortalized leukemia cells were treated with etoposide, and the kinetics and topology of PS exposure were followed over the course of apoptosis.
Costaining etoposide-treated leukemoid cells with lactadherin and annexin V indicated progressive PS exposure with dim, intermediate, and bright staining. Confocal microscopy revealed localized plasma membrane staining, then diffuse dim staining by lactadherin prior to bright generalized staining with both proteins. Annexin V was primarily localized to internal cell bodies at early stages but stained the plasma membrane at the late stage. Calibration studies suggested a PS content ≲2.5%–8% for the membrane domains that stained with lactadherin but not annexin V.
Phosphatidylserine (PS) appears on the outer membrane leaflet of cells undergoing programmed cell death (1–3) and marks those cells for clearance by macrophages (1). Transient exposure of PS on viable cells accompanies sperm capacitation (4), phagocytosis of apoptotic cells (5), myotube formation (6), and neutrophil (7) and platelet stimulation (8). On stimulated platelets, exposed PS supports procoagulant reactions, and on lymphocytes PS externalization appears mechanistically coupled to the function of ion channels and the multidrug resistance protein in lymphocytes (9). However, in spite of the essential relationship of PS exposure to cellular function, the topology and kinetics of PS exposure remain only partially characterized.
PS exposure is frequently detected through binding of fluorescence-labeled annexin V (3). The extent to which annexin V binds to a membrane is a complex function related to the free annexin V concentration, membrane PS content, phosphatidylethanolamine content, and ambient Ca++ concentration (10–15). Indeed, recent reports indicate that the cooperative interaction of annexin V on the plasma membrane drives a mechanism whereby invagination of the plasma membrane is followed by internal vesiculation (16). The complexity of annexin V binding and the internalization of annexin V by stressed cells leaves uncertainty about the extent to which annexin V fluorescence represents plasma membrane PS exposure in early apoptosis.
Lactadherin is also a PS-binding protein and has a domain structure that includes two epithelial growth factor-like domains and two lectin-type domains that are homologous with the PS-binding domains of blood clotting factor VIII (17). Lactadherin exhibits stereoselective interaction with phospho-L-serine of PS (18, 19). In contrast to annexin V, binding appears proportional to PS content and is independent of Ca++ concentration and membrane phosphatidylethanolamine (PE) content (20). Lactadherin is secreted by mammary epithelium (21), epididymal epithelium (22), vascular cells (23), and stimulated macrophages (24). Macrophage-derived lactadherin binds to PS on the membranes of dying cells and mediates interaction with the αvβ3 integrin of macrophages, thus enabling or enhancing phagocytosis of the dying or dead cells (24). Thus, the appearance of membrane domains that bind lactadherin relates to the timing and nature of interactions between dying cells and macrophages. The contrasting membrane binding properties of lactadherin and annexin V suggested that they could be utilized as complimentary probes of membrane PS content.
The present study was conducted to validate the use of lactadherin to detect PS and to determine whether the contrasting PS-binding properties of lactadherin and annexin V could be exploited to devise a method to detect the extent of local and global PS exposure. Accordingly, synthetic membranes were utilized to optimize complimentary conditions for lactadherin and annexin V binding and to calibrate the assay as to PS content. The cell death program of immortalized leukemic cells was initiated by etoposide exposure and PS externalization was monitored from the initial cell stress until the apoptosis program was completed.
The human erythrocytic leukemia cell line, K-562, the promyelocytic leukemia line, HL-60, were from American Type Culture Collection (Rockville, MD), Iscove's Modified Dulbecco's Medium (IMDM) was from Cambrex Bio Science Walkersville (Walkersville, MD). TACS TdT Kit (Apoptosis Detection Kit, Catalog Number: TA4625) was from R&D Systems (Minneapolis, MN). Etoposide and Annexin V-Cy 3.18 were from Sigma. Annexin V-fluorescein and Annexin V-Cy5.5 were from BD Biosciences Pharmingen (San Jose, CA). FluoReporter fluorescein isothiocyanate (FITC) protein labeling kit and Alexa Fluor 647 protein labeling kit were from Molecular Probes (Eugene, OR). Bovine brain PS, egg yolk PE, and phosphatidylcholine (PC) were from Avanti Polar Lipids (Alabaster, AL).
Lactadherin and Annexin V
Bovine lactadherin, free of lipid and protein contaminants, was purified as previously described (25). SDS-PAGE with silver staining revealed only two bands of Mr 47 and 52 kDa corresponding to the two glycosylation variants. Aliquots of 1 mg/ml in phosphate-buffered saline (PBS-150 mM NaCl, 10 mM Na2HPO4, pH 7.4) were flash-frozen in liquid nitrogen and stored at 80°C, and each aliquot was subjected to less than three cycles of flash-freezing and rapid thawing. Annexin V was handled as described for lactadherin. Lactadherin was stained with fluorescein isothiocyanate as previously described (19). Annexin V (Sigma) was coupled to Alexa Fluor647 according to the manufacturer's instructions.
Cell Staining for Apoptosis
Terminal deoxynucleotidyl transferase apoptosis staining was performed utilizing standard techniques described in the kit. PI was utilized as described (3).
Flow Cytometry Assay for PS Exposure
Membranes supported by glass microspheres were prepared as previously described (26). Cell samples were analyzed by a FACSCalibur flow cytometer (Becton Dickinson, San Diego, CA) equipped with CellQuest software with all channels in log mode. Events were triggered based upon forward light scatter and a gate was placed on the forward scatter-side scatter plot that excluded cell debris but included all cells. The samples were excited utilizing both the 496 and 633 nm laser sources. For PS exposure studies, cell suspensions were adjusted to ∼1 × 106 cells/ml. A 0.5 ml of cell suspension was transferred to a microcentrifuge tube; 10 μl FITC-lactadherin and/or Alexa Fluor 647-annexin V was added to the cells to give a final concentration of 2 nM. Fluorescein emission intensity was evaluated via the 530 ± 30 nm band pass filter set (Fl1) for fluorescein and 660 ± 20 nm band pass (Fl4) filter for Alexa 647. Mean log fluorescence intensity was converted to linear fluorescence intensity utilizing the geometric mean values of the single or major population for graphs in Figures 1 and 2.
Samples were incubated with the indicated concentrations of lactadherin and/or annexin V for 10 min at room temperature in the dark. Cells were centrifuged at 200g for 1 min at room temperature, supernatant removed and the cell pellet gently resuspended in 0.5 ml Tyrodes buffer with 1.5 mM CaCl2 and analyzed immediately. Samples were imaged using BioRad MRC 1024ES multiphoton/confocal imaging system (BioRad, Hercules, CA). The samples were excited with 488, 568, and/or 647 emission lines of a krypton-argon laser, and narrow band pass filters were used to restrict emission wavelength overlap. A Zeiss Axiovert S100 inverted microscope equipped with a high quality water immersion 40×/1.2 NA, C-apochroma objective was used to image the cells in an epifluorescence mode. The images were collected using internal detectors at a pixel resolution of 0.240–0.484 μm with a Kalman 3 collection filter, and were reconstructed using the BioRad LaserSharp and/or MetaMorph (MetaMorph Imaging Series, Universal Imaging Corp., West Chester, PA) software (27).
Calibration of Lactadherin/Annexin V Assay
We hypothesized that the PS-binding properties of lactadherin would enable detection of early PS exposure on dying cells. Preliminary binding studies that related PS content to binding of lactadherin and annexin V were performed with synthetic membranes supported by 2-μm glass microspheres (26) (Fig. 1). The Ca++ concentration was 1.5 mM near the upper limit of ionized Ca++ in blood plasma (28). Because the rate of PE exposure during early apoptosis has not been measured, we evaluated binding to synthetic membranes with minimal and maximal PE content. First, we assumed that PE would not be exposed together with PS. The PS content was varied from 0 to 12% in synthetic membranes with 2% PE (Fig. 1A). The quantity of lactadherin bound increased with PS content and approached a plateau at 8% PS. In contrast, annexin V binding was not detected until the PS content was 8%, after which it increased rapidly. Second, we assumed that the rate of PE efflux would exceed the rate of PS efflux. Synthetic membranes were synthesized with a PE:PS ratio of 4:1 (Fig. 1B). Again, lactadherin binding was proportional to PS content whereas the relationship between PS and annexin V binding was sigmoidal. The presence of excess PE reduced the PS threshold for annexin V to ∼2.5% PS (Fig. 1B). These results indicate that lactadherin can be expected to bind to cell membranes with PS content ≥0.5%, while annexin V is predicted to bind when the PS content exceeds 2.5%–8%, depending on the PE content. Thus, cell membranes that stain with lactadherin, but not annexin V are likely to have membrane PS content <8% and possibly <2.5%.
Similar results were obtained with annexin V conjugated to three different fluorophores, supplied by three vendors. In addition, a conjugate of annexin V-Alexa Fluor 647 prepared in our laboratory yielded the same pattern of binding. The Alexa Fluor 647 conjugate was utilized for the experiments displayed in Figures 1C and 2–5.
Competition binding experiments were performed to determine whether lactadherin and annexin V could compete with each other for the same binding sites (Figs. 1C and 1D). Unlabeled annexin V competed for only about 40% of the lactadherin binding sites on membranes with 4% PS and 15% PE. By comparison, unlabeled lactadherin competed for ∼70% of annexin V binding sites on these membranes. The results suggest that the binding sites for lactadherin and annexin V overlap but are not identical. They indicate, further, that the presence of either protein at concentrations less than 8 nM would not diminish binding of the other by more than 20%. The arrows indicate the degree of displacement predicted for each protein for experiments depicted in Figures 3–5.
Lactadherin Binding to Apoptotic Cells
We wished to confirm that lactadherin could detect PS exposure on apoptotic cells. Therefore, K-562 and HL-60 cells were treated with etoposide for 48 h and evaluated with conventional stains as well as lactadherin (Fig. 2).
Staining with hematoxylin and eosin confirmed that some K-562 cells had condensed and fragmented nuclei (Fig. 2A). Positive TUNEL staining confirmed that the DNA was fragmented in these cells (Fig. 2B). Costaining cells with lactadherin and propidium iodide (PI) (Fig. 2C) confirmed that PI permeable cells expose sufficient PS to enable diffuse binding of lactadherin. The diffuse blebs that appear transiently in the course of apoptosis (29) were identified on a small number of cells (Figs. 2D and 2E). These vesicles stained diffusely with lactadherin. Therefore, K-562 cells expose sufficient PS to bind lactadherin as they undergo programmed cell death.
The engagement of PS-binding motifs in staining by lactadherin and annexin V was evaluated in competition binding experiments (Figs. 2F and 2G). Phospholipid vesicles containing 8% PS competed with etoposide-treated cells for both lactadherin and annexin V binding. These vesicles competed for more than 95% of lactadherin binding but only about 85% of annexin V binding. This difference was rationalized by assuming that a larger fraction of annexin V was internalized by the cells (16). The results confirm that the PS-binding motifs of both lactadherin and annexin V participate in binding to the etoposide-treated apoptotic cells. Addition of EDTA at twice the Ca++ concentration reduced staining by annexin V but did not decrease lactadherin binding as detected by flow cytometry or by microscopy (not shown). These results confirmed the Ca++-dependent PS binding of annexin V and the Ca++-independent PS binding of lactadherin. Furthermore, they exclude the possibility that lactadherin binding was mediated by interaction with the αvβ3 or αvβ5 integrins of the dying cells.
Experiments were conducted to determine whether lactadherin could detect PS exposure before cells exhibit increased permeability to PI (Fig. 3, upper panel). Twenty-four hours after initiation of etoposide exposure, three cell populations were evident, indicating the progressive pathway of PS exposure and PI permeability. Approximately 25% of cells remained in the left lower quadrant (solid arrow), but PS exposure was greater than that of untreated cells (open arrow). Approximately 25% were positive for lactadherin staining but negative for PI (open arrowhead). Another 25% exposed the same quantity of PS but were also stained with PI (upper open arrowhead). Finally, a population of cells stained positively for PI and stained very strongly for lactadherin (solid arrowhead). After 48 h the majority of cells stained maximally with both PI and lactadherin. Samples studied at intermediate time points confirmed that most cells progressed through the two intermediate stages identified after 24 h (not shown). These findings indicate that PS exposure begins prior to PI permeability, but that complete PS exposure is delayed for many hours after PI permeability.
Lactadherin staining of both K-562 and HL-60 cells were monitored over 72 h to determine the comparative rates of PS exposure (Fig. 3, lower panel). The results were consistent with progressive PS exposure after treatment with etoposide for both cell types. The arrows and arrowheads on the HL-60 histogram depict the modal fluorescence intensities for untreated cells as well as minimally, intermediate, and maximally responding populations (Fig. 3, 24 h). The elapsed time between etoposide exposure and PS exposure differed for HL-60 cells vs. K-562 cells. HL-60 cells stained within 6 h of etoposide (not shown) while K-562 cells did not stain until at least 12 h had elapsed since etoposide exposure was initiated. Lactadherin staining was maximal for HL-60 cells by 48 h but required at least 72 h treatment for K-562 cells. The mean progression of PS exposure for eight separate experiments is depicted in the lower right panel. The gates utilized to detect the fraction of cells with each quantity of PS exposure are in the K-562 panel. These findings indicate that PS exposure is progressive over more than 24 h for both K-562 and HL-60 cells.
Costaining with Lactadherin and Annexin V
We utilized flow cytometry to evaluate simultaneous staining by lactadherin and annexin V (Fig. 4, upper panel). Dot plots indicate that most cells proceed through a pathway in which staining by lactadherin is increased to a greater degree than by annexin V. For example, prior to etoposide exposure, only 0.6% of K562 cells stained with lactadherin. After 24 h exposure to etoposide, the fraction of the population in the “lactadherin positive” region increased to 32.6%. After 72 h of etoposide exposure, 84.1% were strongly positive for both annexin V and lactadherin. Comparative staining with lactadherin and PI vs. Annexin V and PI confirmed that the two proteins have distinct staining patterns prior to PI staining (not shown). Lactadherin stained the majority of cells with a continuous intensity from negative to highly positive. In contrast, staining with annexin V showed most cells with either minimal staining or bright staining immediately prior to permeability to PI. Very few cells demonstrated intermediate levels of annexin V staining, indicating that most cells had not exposed sufficient PS to reach the annexin V threshold (Fig. 1). Confocal microscopy revealed the cellular features that display PS exposure after 24 h of etoposide exposure (Fig. 4, lower panels). Many cells had minimally changed gross architecture (black arrow) while a similar number were condensed (arrowhead). The normal-sized cells had patches and small appendages that stained with lactadherin (white arrows), but stained weakly or not at all with annexin V. The condensed cells had diffused lactadherin staining, detected as rings, but annexin V staining was punctate. Most cells exhibited staining of internal cell bodies by annexin V. In fact, the internalized bodies were the primary location of annexin V staining rather than the cell surface. These results indicate that an early response to etoposide treatment is localized PS exposure and that the quantity of PS exposure generally remains below the annexin V threshold.
Annexin V and lactadherin staining studies were also performed with Ca++ increased from 1.5 to 3 mM (Fig. 4), similar to the conditions generally employed to probe apoptotic cells for PS exposure with annexin V. The results showed an equivalent pattern of staining with lactadherin and annexin V. The blue background reflects superimposition of the phase contrast with fluorescence images. An enlarged view of a vesiculating cell, with 3 mM Ca++, highlights the preferential binding of lactadherin to membrane vesicles and the preferential localization of annexin V to internal bodies. Experiments were performed with increasing annexin V concentrations, as other investigators have sometimes utilized concentrations much higher than the apparent dissociation constant of <5 nM (30). These conditions led to increased staining of both etoposide and control cells to approximately the same degree. The results suggest that use of annexin V for selective staining of exposed PS is probably most effectively performed with annexin V concentrations <6 nM.
After 48 h of exposure to etoposide, the majority of K-562 cells exposed sufficient PS to stain with both lactadherin and annexin V (Fig. 5). Some cells had fragmented nuclei and irregular contours, changes that are consistent with completed apoptosis (Fig. 5A, arrowhead). Most cells were round with condensed, but regular nuclei, consistent with the ongoing cell death program (Fig. 5A, arrows). Most of the condensed cells stained diffusely with lactadherin (Fig. 5C, white arrows). Other cells, with a lesser degree of condensation, had scattered membrane appendages that stained with lactadherin, similar to dominant cell morphology observed at 24 h. Annexin V stained internal bodies of the cells with intact nuclei (Fig. 5B) confirming that significant annexin V is internalized over the 10 min time course of these experiments (16). Annexin V also stained discrete patches on the plasma membranes of these cells. A composite image (Fig. 5D) contrasts cell features that stain preferentially with lactadherin (green) annexin V (red) and highlights regions that stained with both lactadherin and annexin V (yellow).
The contrasting pattern of staining with lactadherin vs. annexin V is highlighted by the profile analysis of costained cells (Figs. 5E–5G). A composite image at higher magnification demonstrates that some internal bodies stain for both lactadherin and annexin V, though predominantly for annexin V. Color intensity profiles of the same cells illustrated that lactadherin diffusely stained the plasma membrane of cells progressing through the cell death program and that the intensity increased with progressive apoptosis (Fig. 5F). Annexin V stained focal regions of the plasma membrane on these cells and internal bodies (Fig. 5G). Thus, these results indicate that dying leukemoid cells expose PS in three stages. PS exposure on most of the cell membrane remains below the annexin V threshold until the final stage of apoptosis.
We have utilized lactadherin and annexin V to identify PS-containing membrane regions on leukemoid cells undergoing programmed cell death. The results indicate that PS exposure is first localized and later generalized. Within 12–14 h after exposure to etoposide, lactadherin binds to PS on ruffled regions of the plasma membrane, membrane-bound projections, and small vesicles. Next, PS is exposed diffusely over the cell surface, initially at a low level and increasing slowly over ∼24 h. Subsequently, PS is exposed briefly on extensive membrane blebs, and finally the cell corpse exposes PS over a large fraction of the cell body. Selective staining of cell appendages by lactadherin during the first 24 h and dim diffuse staining of the plasma membrane during the next period indicates that PS exposure remains regulated and the quantity of PS exposed remains below the annexin V threshold. At later time points, annexin V was evident in discrete patches on the same cell features stained diffusely by lactadherin, identifying localized domains with PS content apparently above the annexin V threshold of at least 2.5%. These results indicate that cells have a mechanism for regulating local PS exposure that remains intact until the final phase of apoptosis.
The results in this report are in agreement with several prior reports. Prior investigators have reported PS-dependent binding of lactadherin (24) and annexin V (3) to apoptotic cells, though the studies of lactadherin binding have been very limited and the cell features recognized by the two proteins have not been compared. The preferential internalization of annexin V by cells undergoing apoptosis has previously been reported (16). A prior report indicated that high concentrations of lactadherin compete with annexin V for binding sites and that high concentrations of annexin V exhibit partial competition with lactadherin for binding to apoptotic cells (24). These findings are consistent with the limited degree of competition between these two proteins at the concentrations utilized for these studies. The progression of PS exposure following PI staining has also been previously observed, though not emphasized (see Figure 5 in Ref.2). This study has not emphasized the impact of focal membrane curvature or focal segregation of lipids into microdomains or rafts. Future studies into the role of plasma membrane curvature and lipid segregation will likely lead to refinement of the insight into the selective binding sites for lactadherin (19) and annexin V (10).
In these studies, we utilized Ca++ concentrations that are near the physiologic plasma range of 0.94–1.33 mM (28), and concentrations of lactadherin and annexin V that may be readily achieved in the systemic circulation of animals. Our motivation was twofold. First, we wished to develop a semiquantitative assay for PS exposure rather than an exhaustive study of the impact of Ca++ and annexin V concentration on the extent to which annexin V stains cells membranes. Second, we wished to develop an assay that might be suitable for in vivo studies to determine the extent of PS exposure. Indeed, we have initiated in vivo studies costaining sites of vascular injury with lactadherin and annexin V. Our preliminary results with mice confirm that this approach can be utilized to detect limited PS exposure on platelets during thrombus formation (lactadherin positive) vs. extensive PS exposure of endothelial cells at the site of injury (annexin V and lactadherin positive) (Shi J, Pipe SW, Heegaard CW, Rasmussen JT, Gilbert GE, unpublished observation.)
Lactadherin shares homology with coagulation factor VIII and factor V and both of these proteins bind to PS-containing membranes in a Ca++ independent manner. It is plausible that either of these proteins could also serve as PS-binding detection probes, similar to lactadherin. Indeed, some binding of both proteins to cell microparticles and membranes has been interpreted as indicative of PS exposure (7, 31, 32). Lactadherin offers three advantages over factor VIII and factor V. First, it binds to all of the membrane sites recognized by either factor V or factor VIII (20), while each of these proteins recognize only a subset of PS-containing sites recognized by lactadherin or by the other proteins (26). Second, lactadherin is resistant to harsh conditions and some proteases (33) while factor V and factor VIII are both known for their susceptibility to proteolytic degradation. Third, factor VIII and factor V both have putative protein receptors on some cell membranes (34). Thus, binding of these proteins to membranes may be more complex to interpret.
Our data indicate that PS exposure occurs in three stages during the cell death program and that fluorescence-labeled lactadherin enables detection of the early PS exposure. In the first stage, PS is exposed on scattered vesicles, membrane-bound projections, and ruffled regions. Lactadherin staining is detectable by flow cytometry as a shift in green fluorescence that continues to overlap with intrinsic fluorescence (Figs. 2 and 5). In the second stage, diffuse PS exposure is evident over the plasma membrane. The intensity of PS exposure increases as the cytoplasm condenses and the plasma membrane becomes permeable to PI. In the third stage the plasma membrane phospholipid asymmetry collapses and the cells expose sufficient PS to enable intense staining by both lactadherin and annexin V. The long interval between exposure of PS on localized cell features and the completion of cell death suggests that limited exposure of PS occurs prior to irreversible commitment to the cell death program as previously suggested (35). The slow, progressive PS exposure also suggests that cells have a mechanism for graded PS exposure that is distinct from the Ca++-dependent “scramblase” that mediates collapse of phospholipid asymmetry in the final stage of apoptosis (36).
PS exposure during early stages of the cell death program raises questions about the physiologic function of this phenomenon. We have observed a similar pattern of PS exposure on two different leukemic cell lines, suggesting that this pattern of PS exposure is common in leukemia cells and leukocytes. Further studies will be necessary to determine whether a similar pattern of progressive PS exposure occurs on adherent cells in the course of apoptosis. Lactadherin is secreted by macrophages (24), dendritic cells (37), smooth muscle cells (38), and endothelial cells (23) and thus is likely to be present in the local milieu of some stressed cells. The lactadherin that binds to PS on stressed cells might then bridge the gap to macrophages or other cells expressing the αvβ3 and/or αvβ5 integrins. Normal leukocytes that are undergoing apoptosis are cleared by hepatic macrophages (39), so it seems reasonable to speculate that stressed or dying leukemic cells may also be cleared by these cells. Alternatively, exposed PS on a stressed or dying leukemic cell probably contributes to uncontrolled blood coagulation reactions that sometimes complicate acute leukemia. Lactadherin may play a role in limiting access of blood coagulation proteins to PS on these cells (20). Exposed PS directly affects ion channel activity and membrane transporters in viable lymphocytes (9). It is plausible that PS exposure could alter focal activity of ion channels or other membrane proteins in the ruffled edges or vesicular bulges on stressed leukemic cells. Thus, focal PS may lead to altered local function of the leukemic cells as well as altered interaction with proteins and cells in the environment of these cells.
The authors would like to thank Patricia Price for excellent technical assistance and Dr. Hemant Thatte for both teaching and assisting in confocal microscopy.