Classification of apoptosis and necrosis by morphological differences has been widely used for decades. However, this usefulness of this method has been seriously questioned in recent years, mainly due to a lack of functional and biochemical evidence to interpret the morphology changes. To address this matter, we devised genetic manipulations in Drosophila to study pyknosis, a process of nuclear shrinkage and chromatin condensation that occurs in apoptosis and necrosis. By following the progression of necrotic pyknosis, we surprisingly observed a transient state of chromatin detachment from the nuclear envelope, followed by the nuclear envelope completely collapsing onto chromatin. This phenomenon led us to discover that phosphorylation of barrier-to-autointegration factor (BAF) mediates this initial separation of nuclear envelope from chromatin. Functionally, inhibition of BAF phosphorylation suppressed necrosis in both Drosophila and human cells, suggesting that necrotic pyknosis is conserved in the propagation of necrosis. In contrast, during apoptotic pyknosis the chromatin did not detach from the nuclear envelope and inhibition of BAF phosphorylation had no effect on apoptotic pyknosis and apoptosis. Our research provides the first genetic evidence supporting a morphological classification of apoptosis and necrosis through different forms of pyknosis.
Morphological differences observed during cell death have been widely used to classify apoptosis and necrosis for a long time. These include apoptotic features such as cell shrinkage, membrane blebbing, nuclear condensation and apoptotic body formation (Kerr et al., 1972), and necrotic features such as cell swelling, plasma membrane rupture, intracellular vacuolization and nuclear chromatin clumping (Raffray and Cohen, 1997). However, many exceptions have been discovered and classification of cell death by morphology has been controversial (Raffray and Cohen, 1997). Recently, the Nomenclature Committee on Cell Death (NCCD) has recommended using biochemical markers for cell death classification instead of morphology (Galluzzi et al., 2015, 2012). The reasons include that morphology might not be linked to a functional aspect of cell death, a given morphology might be triggered by heterogeneous insults, and intermediate morphology shared by both apoptosis and necrosis might exist (Galluzzi et al., 2015, 2012; Raffray and Cohen, 1997).
Pyknosis has been considered as an irreversible condensation of chromatin and the nucleus. It commonly occurs in both apoptotic and necrotic cell death. For apoptosis, the nucleus usually undergoes condensation, chromatin marginalization and fragmentation into a few large and regular chromatin clumps, which are eventually packed into apoptotic bodies (Kerr et al., 1972; Niquet et al., 2003). In contrast, nuclei in necrotic cells condense into to smaller chromatin clumps with irregular and dispersed morphologies, which might be dissolved later (Bortul et al., 2001; Fujikawa et al., 2000; Hardingham et al., 2002; Niquet et al., 2003). Therefore, based on the morphology of nuclear fragmentation, pyknosis can be divided into nucleolytic pyknosis (mainly occurring in apoptosis) and anucleolytic pyknosis (mainly occurring in necrosis) (Burgoyne, 1999). Although pyknosis has been widely considered as a marker of cell death in vitro and in vivo (Colbourne et al., 1999; Fujikawa et al., 1999, 2010; Ji et al., 2013; Niquet et al., 2003; Sohn et al., 1998), it is unclear whether pyknosis is a regulated process.
Here, we studied the morphological changes visible in pyknosis during the execution of apoptosis and necrosis in a temporal manner. We found that necrotic pyknosis occurred in a distinct pattern compared to apoptotic pyknosis, and phosphorylation of BAF plays a key role only in necrotic pyknosis.
RESULTS AND DISCUSSION
Necrotic pyknosis shows a different morphology to apoptotic pyknosis
To study necrotic pyknosis, we used a previously established necrosis model in Drosophila. By allowing transient expression of a leaky cation channel, the glutamate receptor 1 Lurcher mutant (GluR1Lc), Ca2+ is overloaded into cells, which results in necrosis and fly lethality (Liu et al., 2014). This fly model contains the Gal4 driver Appl-Gal4 (expressed in neurons and a few epithelial cells in the larval anal pad), and the UAS expression constructs UAS-GluR1Lc and tub-Gal80ts, and we denote it the AG model (Appl>GluR1Lc, tub-Gal80ts) (Liu et al., 2014). The AG flies are healthy at 18°C due to inhibition of Gal4 function by Gal80ts. Upon shifting flies to 30°C, Gal80ts is deactivated and necrosis is initiated by GluR1Lc expression (Liu et al., 2014). To label the nuclear envelope and the chromatin, the fluorescent reporters UAS-koi.GFP and His2Av-mRFP1 (expressing His2Av–mRFP fusion protein in all cells) were used, respectively.
After being at 30°C for 22 h, the nuclei of control flies [Appl-Gal4;tub-Gal80ts, labeled as wild type (WT); Fig. 1Aa] showed normal morphology, with chromatin occupying the whole nucleus, and the interaction of chromatin with the nuclear envelope was clearly visible. However, after 18 to 20 h at 30°C, the chromatin in the epithelial cells of AG flies was dramatically condensed, whereas the nuclear envelope only slightly shrank, leading to the detachment of chromatin from the nuclear envelope (Fig. 1Ab,c). Later, both the nuclear envelope and the chromatin were further compacted and eventually collapsed together (Fig. 1Ad). The morphology is consistent with the features of anucleolytic pyknosis (Fujikawa et al., 2010; Sohn et al., 1998).
We also studied apoptotic pyknosis in vivo by transient expression of reaper (rpr) (Appl>rpr, tub-Gal80ts, denoted AR), which induces classical apoptosis (White et al., 1996). We found that the nuclear envelope and chromatin physically shrank together, with the nucleus eventually fragmenting into regular clumps (Fig. 1Ba–d).
Phosphorylation of BAF specifically mediates necrotic pyknosis
Our data suggest that necrotic pyknosis is likely to be initiated by the detachment of chromatin from the nuclear envelope. Previous studies have shown that barrier-to-autointegration factor (BAF) plays a key role in chromatin tethering to the nuclear envelope through its interaction with LAP2, emerin and MAN1 (LEM)-domain-containing proteins and double-stranded DNA (dsDNA) in a sequence non-specific manner (Umland et al., 2000; Zheng et al., 2000).
Based on the function of BAF, it is possible that the dissociation of chromatin from the nuclear envelope is induced by deregulation of BAF during necrosis. To examine BAF localization in necrosis, we generated a GFP-tagged BAF transgene (UAS-GFP-BAF). It has been reported that an N-terminal GFP tag does not affect BAF activity (Nichols et al., 2006; Shimi et al., 2004). In wild-type cells (Appl-Gal4;tub-Gal80ts) after 30°C for 18 h, GFP–BAF was distributed on chromatin and interacted with the nuclear envelope (Fig. 1Ca). In the AG flies, most GFP–BAF protein was also localized on the compacted chromatin (Fig. 1Cb), suggesting that BAF localization is not drastically altered during necrosis. However, GFP–BAF reduced its distribution in the chromatin and formed ring-like structures in apoptosis (Fig. 1Cc). This phenomenon is consistent with a previous report stating that BAF disassembled and disappeared from the nucleus during apoptosis (Furukawa et al., 2007).
The N-terminal phosphorylation of BAF has been reported to cause the detachment of chromatin from the nuclear envelope during karyosome formation in the meiosis of the oocyte in Drosophila (Lancaster et al., 2007). The BAF N-terminus has three potential phosphorylation sites, including serine 2, threonine 4 and serine 5. To test the importance of BAF phosphorylation in necrotic pyknosis, we generated transgenic flies expressing BAF proteins with these three potential phosphorylation sites mutated to alanine (the non-phosphorylatable form, denoted BAF-3A) or aspartic acid (the phospho-mimic form, denoted BAF-3D). These mutation sites of BAF are illustrated in Fig. 1D. When these mutant flies were crossed to AG flies to induce necrosis, GFP–BAF-3A localization appeared to be similar to the pattern of wild-type BAF (GFP–BAF-WT) under normal conditions (Fig. 1Cd), whereas GFP–BAF-3D showed a similar dissociation phenotype to that of the GFP–BAF-WT in the AG background (Fig. 1Ce). These data suggest that chromatin disassociation from nuclear envelope during necrosis is likely inhibited by GFP–BAF-3A expression. To further confirm the role of BAF-3A, we quantified the number of cells with early stage necrotic pyknosis features in cells with the chromatin and nuclear envelope labeled by His2Av–mRFP1 and koi.GFP. In the AG flies, 58% of cells displayed the early morphological features of necrotic pyknosis (Fig. 1Ea,b,F). Strikingly, the early stage morphology defect of necrotic pyknosis was completely abolished by the BAF-3A expression (Fig. 1Ec,F). Similarly, expression of BAF-3A rescued the late-stage morphology of necrotic pyknosis (Fig. 1Ga–c,H). For apoptotic pyknosis, BAF-3A expression had no inhibitory effect (Fig. 1Gd,e,H), indicating different molecule(s) might be required.
Anucleolytic pyknosis plays a functional role in necrotic cell death
Although anucleolytic pyknosis has been considered to be a marker of necrosis, its function in cell death is unclear. To address this question, we investigated the survival of AG flies under different transgenic BAF mutant backgrounds. Transient overexpression of BAF-3A or BAF-3D in neurons had no effect on fly survival (Fig. 2A). Upon induction of necrosis, the survival rate of adult AG flies decreased to 38.4% (Fig. 2A). Overexpression of BAF-WT resulted in a similar survival rate (Fig. 2A). Strikingly, the overexpression of BAF-3A increased the survival rate to 70.0%, whereas the expression of BAF-3D reduced the survival rate to 11.7% (Fig. 2A). At the cellular level, BAF-3A suppressed the necrotic morphology, whereas BAF-3D enhanced it, and BAF-WT had no effect (Fig. 2B,C). Taken together, BAF phosphorylation alone is not sufficient to induce cell death, but it is necessary for necrotic pyknosis and the propagation of cell death. In contrast, the expression of BAF-3A or BAF-3D had no effect on fly models of apoptosis, including the eye defects of GMR-Hid (Grether et al., 1995) and GMR>eiger (Hid and Eiger induce caspase-mediated and JNK-mediated apoptosis, respectively; Moreno et al., 2002) (Fig. 2D).
Owing to a lack of antibody that can directly detect the phosphorylation of Drosophila BAF, we assessed BAF phosphorylation by immunoprecipitating GFP–BAF proteins with an anti-GFP antibody followed by western blotting with an antibody against phosphorylated threonine (pThr). The result showed that BAF phosphorylation indeed increased in the AG flies, and the expression of BAF-3A abolished the phosphorylation (Fig. 2E). The subtle change in BAF phosphorylation is likely due to only ∼1% of neurons undergoing necrosis, which is sufficient to cause fly lethality in the AG flies (Liu et al., 2014). No signal could be detected when an antibody against phosphorylated serine was applied. This result suggests that threonine 4 in the N-terminal of BAF is the functional site that is phosphorylated during necrotic pyknosis. To test this hypothesis, we generated transgenes to express BAF proteins with both threonine 4 and serine 5 mutated to alanine (BAF-2A) or threonine 4 alone mutated to alanine (BAF-1A) (Fig. 1D). We found that both BAF-2A and BAF-1A could rescue the lethality of the AG flies, similar to BAF-3A (Fig. 2A).
BAF phosphorylation regulates necrotic pyknosis in mammalian cells
Because of the conserved function of BAF in chromatin anchoring in metazoans (Segura-Totten and Wilson, 2004), we asked whether BAF phosphorylation regulates necrotic pyknosis in mammalian cells. To induce necrosis, human neuroblastoma SH-SY5Y cells were treated with Ca2+ ionophore to overload Ca2+. Ca2+ ionophore treatment led to nuclear shrinkage into propidium-iodide-positive, small, round and bright chromatin clumps within 30 min (Fig. 3Aa–b′). However, we did not observe the transient detachment of chromatin from the nuclear envelope in SH-SY5Y cells and in several other cultured mammalian cells (Fig. 3Ac; Fig. S1). This variation might be due to the difference between in vivo and in vitro systems. In tissues, the cellular contacts play an important role in setting the threshold of cell death (Raffray and Cohen, 1997). In fact, the intermediate state of the nuclear envelope being detached from chromatin has been observed in the retinoblastoma patients (Buchi et al., 1994). In mammalian cells, the nucleus is highly compacted and the dissociation of nuclear envelope from chromatin might be less obvious at the cell morphological level. To study the interaction between the nuclear envelope and chromatin in mammalian cells, the DNA that tethered on the inner nuclear membrane was precipitated by an anti-lamin-B1 antibody. Then, the precipitated DNA was quantified by an agarose gel and a quantitative PCR (qPCR) assay. The result showed that the amount of lamin-B1-bound DNA was significantly decreased during necrosis in the SY5Y cells (Fig. 3B,C). This result indicates that the dissociation of chromatin and nuclear envelope might also take place during necrosis in the mammalian cells.
The potential phosphorylation sites of human BAF (BANF1) include threonine 2, threonine 3 and serine 4 (Fig. 3D). Because the site equivalent to the Drosophila BAF threonine 4 was unclear (Lancaster et al., 2007), we mutated all three sites (Fig. 3D, BANF1-3A), and expressed the mutant constructs in SH-SY5Y cells. The result showed that expression of BANF1-3A (a non-phosphorylatable mutant) abolished the reduction of lamin-B1-bound DNA during necrosis (Fig. 3B,C). To examine the functional role of BANF1 phosphorylation on necrosis, we performed Ca2+ ionophore experiments. Ca2+ ionophore treatment induced ∼30.4% propidium-iodide-positive nuclei (Fig. 3E,F). However, expression of BANF1-3A reduced the proportion of cells that died to 14.4% (Fig. 3Ec,d,F). BANF1-3A also blocked the ATP depletion upon Ca2+ ionophore treatment (Fig. 3G).
To examine BANF1 phosphorylation, BANF1–Flag proteins were precipitated with an anti-Flag antibody and assessed with an anti-pThr antibody. The result showed that the phosphorylation of BANF1-WT–Flag was not detectable under normal conditions but it was greatly increased upon treatment with Ca2+ ionophore (Fig. 3H). Importantly, there was no detectable phosphorylation of BANF1 when BANF1-3A–Flag was expressed (Fig. 3H). This result indicates that BAF phosphorylation is a conserved event of necrosis.
Our study identifies BAF phosphorylation as a specific biochemical marker for necrotic pyknosis in certain types of cells, suggesting that classical apoptotic and necrotic pyknosis might be distinct processes at the molecular level. A schematic model of necrotic pyknosis is shown in Fig. 4. For classification of cell death, identification of more biochemical markers that directly regulate morphology should greatly improve the uncertainty in categorizing more complicated modes of cell death (Raffray and Cohen, 1997). In fact, several regulators of apoptosis and necrosis have been identified, including caspase-activated DNase (CAD, also known as DFFB), endonuclease G and DNase I in nuclear fragmentation (Enari et al., 1998; Li et al., 2001; Liu et al., 1997; Oliveri et al., 2001), and phospholipase A2 in necrotic pyknosis (Shinzawa and Tsujimoto, 2003). Here, we provide an example of morphological regulation of necrotic pyknosis by a biochemical event.
MATERIALS AND METHODS
Drosophila stocks and maintenance
Flies were raised on standard cornmeal medium. The stocks were kindly provided by colleagues including: UAS-eiger (Lei Xue, School of Life Science and Technology, Tongji University, China), GMR-hid, GMR-Gal4 (Andreas Bergmann, Department of Cancer Biology, University of Massachusetts Medical School, Worcester, MA) and UAS-IAP1 (Denise Montell, Molecular, Cellular, and Developmental Biology, University of California, Santa Barbara, CA). We generated the following lines from the w1118 background by P-element insertion: UAS-GFP-BAF-WT, UAS-GFP-BAF-3A, UAS-GFP-BAF-3D, UAS-BAF-WT-Flag, UAS-BAF-3A-Flag, UAS-BAF-3D-Flag, UAS-GFP-BAF-2A and UAS-GFP-BAF-1A. UAS-koi.GFP (BL#26266) and H2Av-mRFP1 (BL#23650) were obtained from the Bloomington Drosophila Stock Center.
Protein extraction and immunoprecipitation
All cell lines were obtained from the China Infrastructure of Cell Line Resources and tested for contamination. Fly heads and human SH-SY5Y cells treated with Ca2+ ionophore (A23187; Tocris Bioscience #1234) were lysed in buffer (20 mM Tris-HCl pH 7.5, 100 mM NaCl, 0.5% NP40) supplemented with protease inhibitor cocktail (Roche) and phosphatase inhibitors (1 mM Na3VO4, 50 mM NaF, 30 mM glycerophosphate and 0.5 mM EDTA). After sonication and centrifugation, the supernatant was incubated with normal IgG (1:1000, Santa Cruz Biotechnology, sc-2025) and protein-A/G–agarose beads (Thermo Scientific Pierce, #20421). After centrifugation, the supernatant was incubated with anti-GFP antibody (1:200, Abcam ab1218, clone 9F9.F9) or anti-Flag antibody (1:1000, Sigma-Aldrich F3165, clone M2) for 2 h. Then, pre-washed beads were added and incubated overnight. After washing with lysis buffer six times, the pellet was boiled in SDS loading buffer. Tricine-SDS-PAGE was used to separate the small-molecular-mass proteins. BAF phosphorylation was detected with an anti-pThr antibody (1:500, Cell Signaling #9381).
Nuclear morphology and cell survival in cultured cells
Antibody against the nuclear pore complex (NPC) (1:1000, Abcam ab24609, clone Mab414) was used for immunostaining. SH-SY5Y cells were incubated with 20 mM Ca2+ ionophore for the indicated times. The ATP level was determined by a CellTiter-Glo® Luminescent Cell Viability Assay kit (Promega). The relative ATP level (%)=ATP [with Ca2+ ionophore]/ATP [without Ca2+ ionophore]×100.
Chromatin immunoprecipitation and quantitative PCR
After Ca2+ ionophore or DMSO treatment for 30 min, chromatin immunoprecipitation was performed as described previously (Boyd et al., 1998). Protein A+G agarose, salmon sperm DNA (Merck Millipore #16-201) and anti-lamin-B1 antibody (1:100, ab16048, Abcam) were used. The same primers were added 5' and 3' to sample DNA of different lengths through multiple annealing and looping-based amplification cycles (MALBAC) with a 27-nucleotide sequence followed by 8 variable nucleotides (N) as a primer (5′-GTGAGTGATGGTTGAGGTAGTGTGGAGNNNNNNNN-3′) (Zong et al., 2012), and then quantified by qPCR using the 27-nucleotide sequence (5′-GTGAGTGATGGTTGAGGTAGTGTGGAG-3′) as the primer. The DNA precipitated by the anti-lamin B1 antibody was quantified as a relative value to the IgG precipitation (the level of the IgG chromatin immunoprecipitation was set at 1).
Student's t-tests and one-way ANOVA analysis with post hoc Tukey algorithm were performed based on the hypothesis of normal distribution and the variances within or between the groups. All data were collected and analyzed without any preference.
L.H., K.L., X.J. and L.L. designed the experiments; K.L. and S.M. performed the initial Drosophila study; L.H. performed the Drosophila and mammalian cell study; Y.L. generated the Drosophila transgenes; and L.H., K.L., X.J. and L.L. wrote the manuscript.
This work is supported by grants provided to L.L. by the Ministry of Science and Technology of the People's Republic of China [grant number 2013CB530700]; and the National Natural Science Foundation of China for Distinguished Young Scholars [grant number 81325007 to X.J.]
The authors declare no competing or financial interests.