Promyelocytic leukemia bodies tether to early endosomes during mitosis

Abstract

During mitosis the nuclear envelope breaks down, leading to potential interactions between cytoplasmic and nuclear components. PML bodies are nuclear structures with tumor suppressor and antiviral functions. Early endosomes, on the other hand, are cytoplasmic vesicles involved in transport and growth factor signaling. Here we demonstrate that PML bodies form stable interactions with early endosomes immediately following entry into mitosis. The 2 compartments remain stably associated throughout mitosis and dissociate in the cytoplasm of newly divided daughter cells. We also show that a minor subset of PML bodies becomes anchored to the mitotic spindle poles during cell division. The study demonstrates a stable mitosis-specific interaction between a cytoplasmic and a nuclear compartment.

Introduction

PML bodies are distinct cellular compartments that are formed by the tumor suppressor protein promyelocytic leukemia (PML). They are mostly detected at the interchromatin space within the nucleus as 5–20 doughnut-shaped structures ranging in size from 0.1 to 1.0 µm in diameter. These compartments recruit and release a number of different proteins with diverse functions, and they respond dynamically to cell stress, viral infections, and growth factors.^1,2

Several studies have demonstrated a close relationship between nuclear PML bodies and the genome. For example, studies using electron microscopy have revealed direct contact between PML bodies and chromosomal DNA.³ In addition, these structures have been detected in close proximity to actively transcribed genes, centromeric DNA, telomeric DNA, sites of DNA damage, and sites of viral DNA replication.^4-9

During mitosis, PML bodies are released from the genome and enter peripheral regions of the cell.^10-12 Mitotic PML bodies, which have been designated mitotic assemblies of PML proteins (MAPPs),¹¹ are morphologically and biochemically distinct from the interphase nuclear PML bodies.^10,11 For example, MAPPs are generally larger and have a more amorphous morphology compared with their interphase counterparts and are formed through aggregation of nuclear PML bodies.^10,11 In addition, MAPPs lack several of the proteins (including Sp100, Daxx, and SUMO) that are found to be associated with nuclear PML bodies during interphase.^10,11,13 At exit from mitosis, MAPPs remain in the cytoplasm of newly divided cells, where they complex with components of the nuclear pore complexes (NPCs) to form cytoplasmic assemblies of PML and nucleoporins (CyPNs).¹² Following completion of cell division, the CyPNs disassemble and PML recycle to the nucleus to stimulate regeneration of progeny nuclear PML bodies.^10-13

The endocytic pathway is the main vesicle transport route from the cell surface to lysosomes or recycling endosomes.¹⁴ Previous studies have suggested that PML functionally interacts with endosomal vesicles through cytoplasmic PML splice variants that lack a functional nuclear localization signal (NLS).^15,16 Indeed, mutational analysis of nuclear PML isoforms revealed effective endosomal PML targeting when NLS was mutated.¹⁷ In the present study we use immunofluorescence microscopy and single-particle tracking to demonstrate a stable and specific interaction between PML bodies and early endosomes during mitosis. The results suggest that the endocytic pathway may be targeted by the full plethora of PML isoforms during this phase of the cell cycle.

Results

PML bodies tether to the surface of early endosomes during mitosis

To investigate a potential interaction between PML and endosomes in mitosis, we immunofluorescently labeled HaCaT cells using antibodies against PML in combination with antibodies against the early endosome resident protein EEA1.¹⁸ Analysis of labeled cells by confocal microscopy revealed that 65 to 85 percent of all MAPPs within a cell associated closely with early endosomes at all stages of mitosis (Fig. 1A and B). Despite the high degree of proximity observed between the 2 compartments, we detected limited colocalization between EEA1 and PML, indicating that these structures persisted as separate, but closely associated, entities after entry into mitosis (Fig. 1A; Fig.  S1). To verify this, we analyzed the interface between MAPPs and early endosomes by sub-diffraction-limit imaging using gated stimulated emission depletion (gSTED) microscopy. At this level of resolution (< 100 nm) the signals emitted by the 2 fluorophores were observed to be non-overlapping, indicating that the 2 structures are neighboring compartments that contact each other’s surfaces (Fig. 1C). We observed a significant reduction in the number of early endosome–MAPP interactions in the cytoplasm of early G₁-phase cells (10–20%) compared with mitotic cells (65–85%) (Fig. 1A and B). Thus, dissociation of the 2 compartments appears to occur concomitantly with mitotic exit, activation of nuclear import, and transformation of MAPPs into CyPNs.

Figure 1. Interaction between PML bodies and endosomal vesicles in mitotic cells. (A and B) Immunofluorescence (IF) analysis of fixed HaCaT cells. (A) Antibodies against PML (green) and EEA1 (red) were used for detection of PML bodies and early endosomes (upper panels). Antibodies against PML (green) and Lamp1 (red) were used for detection of PML bodies and late endosomes/lysosomes (lower panels). Images represent projections of 3 to 4 confocal Z-sections. DAPI staining is shown in blue. Scale bar: 10 µm. (B) Quantification of the percentage of endosome-interacting PML bodies per cell. Bars represent the average of 3 independent experiments ± standard deviation (SD) (**P < 0.01; ***P < 0.001). In each experiment, more than 100 cells were analyzed. (C) Gated stimulated emission depletion (gSTED) microscopy of a PML body–early endosome interaction in a fixed, mitotic HaCaT cell. The leftmost image shows a single confocal Z-section through the cell. Scale bar: 1 µm. (D) IF using antibodies against transferrin receptor (TfR, green) and PML (red). The cell surface and the chromatin region are outlined by white lines. Scale bar: 10 µm.

We also investigated potential interactions between MAPPs and late endosomes using anti-PML antibodies in combination with antibodies against the late endosome marker Lamp1. We observed significantly fewer interactions between MAPPs and late endosomes (10–20%) compared with MAPPs and early endosomes (65–85%) (Fig. 1A and B). In addition, we did not observe significant changes in the number of late endosomes proximal PML bodies as cells progressed from mitosis to G₁ phase (Fig. 1A and B). Thus, PML bodies become specifically targeted to early endosomes during mitosis.

Several plasma membrane surface receptors use early endosomes for intracellular transport. To investigate if PML associates with endosomes that contain plasma membrane receptor proteins during mitosis, we studied mitotic HaCaT cells immunofluorescently labeled using antibodies against PML and the transferrin receptor. As expected, we detected clustering of PML bodies to transferrin receptor-positive domains in cells undergoing mitosis (Fig. 1D). Thus, PML has a large potential of interacting with endocytosed plasma membrane anchored receptor proteins during cell division.

The interactions between PML bodies and early endosomes are stable throughout mitosis

To further investigate the dynamics of PML bodies and early endosomes during mitosis, we generated HaCaT cells that stably express EYFP-tagged PML1 in combination with mCherry-tagged Rab5 or 2xFYVE, protein components that are known to target early endosomes.^19,20 In agreement with the data obtained by using immunofluorescence labeling of fixed cells (Fig. 1), close proximity between PML bodies and early endosomes was also observed in living cells expressing PML-body and early-endosome markers (Fig. 2A–C; Videos S1–3). To track individual PML bodies during mitosis, we collected a single Z-section for the EYFP and mCherry channels with 2 seconds intervals between frames. This experiment revealed stable association between PML bodies and early endosomes throughout the mitotic phase (Fig. 2A and B; Videos S1 and 2). We observed separation of the 2 compartments at the mitosis-to-G₁ transition, confirming that the interactions between these 2 compartments are destabilized on completion of cell division (Fig. 2B; Video S2).

Figure 2. Tracking of PML bodies and early endosomes in mitotic cells .(A and B) Tracking of a single PML body in HaCaT cells expressing EYFP-PML1 and mCherry-Rab5 (A) or EYFP-PML1 and mCherry-2xFYVE (B). Images were captured at intervals of 2 s (every second time point [every fourth second] is displayed). The tracked area is indicated by the white rectangle. Movements in the Y-direction are visualized while movements in the X-direction are kept constant (see also Videos S1 and 2). (C) Imaging of EYFP-PML1 and mCherry-Rab5 during entry into mitosis. The nuclear boundary is indicated by white stippled lines in the first image. Arrows indicate formation of PML body–early endosome interactions at the time points when they are first detected (see also Video S3).

By examining cells at entry into mitosis, we observed de novo formation of small PML1-containing foci at the cell periphery immediately after breakdown of the nuclear membrane (Fig. 2C, white arrows; Video S3). Formation of these small foci, which occurred close to the plasma membrane, was generally detected 2–4 min prior to association between pre-existing PML bodies and early endosomes (Fig. 2C; Video S3). Together, these experiments demonstrate rapid formation of stable interactions between PML bodies and early endosomes at entry into mitosis and disruption of this contact at exit from mitosis.

A small subset of PML bodies becomes anchored to mitotic spindle poles during cell division

During the analysis of PML bodies by high-resolution tracking in living cells, we noticed a small sub-population of bodies that becomes anchored to the mitotic spindle. This was evident since the movements of these bodies were completely synchronized with the movements displayed by chromatin and mitotic spindle, whereas the majority of mitotic PML bodies showed movements that were independent of these structures (Fig. 3A; Video S4). Interestingly, these mitotic bodies were frequently seen at or near subcellular locations expected to contain the poles of the mitotic spindle (Fig. 3A; Video S4). In addition, they were consistently observed to dissociate at the cytoplasmic face of the nuclear membrane at completion of mitosis (Fig. 3A; Video S4).

Figure 3. Interactions between PML bodies and mitotic spindle. (A) Live cell imaging of HaCaT cells expressing EYFP-PML1. Individual PML bodies in mitotic cells were tracked by a single confocal scan at a frame rate of 2.5 s per interval. Every 48th frame is displayed. The orientation of the metaphase plate and the cleavage furrow is indicated by the red, dotted line. Arrows points to PML body anchored to mitotic spindle (see also Video S4). (B) Tracking of spindle pole-attached PML bodies in HaCaT cells stably expressing EYFP-PML1 (green) and mCherry-Centrin1 (red). Images were captured at intervals of 2 s (every second time point [every fourth second] is displayed). The tracked area is indicated by the white rectangle. Movements in the Y-direction are visualized while movements in the X-direction are kept constant (see also Video S5). (C) IF showing PML (red) and microtubules (green) in mitotic cells. Panels represent merged projections of multiple confocal stacks. (D) Quantification of the interaction between PML bodies and spindle pole(s). For each experiment more than 50 cells were analyzed. Data represent the average of 3 independent experiments ± SD (E) HaCaT, HeLa, and U2OS cells in the mitotic phase were immunofluorescently labeled using antibodies against PML (green) and pericentrin (red). Panels represent merged projections of multiple confocal Z-sections. PML bodies and centrosomes in close proximity are indicated by white rectangles.

To verify that PML bodies have the ability to become anchored to spindle poles during mitosis, we generated HaCaT cells that stably express EYFP-PML1 together with an mCherry version of the centrosome marker protein centrin1. Tracking of PML bodies and centrosomes in mitotic cells at 2-second intervals between frames for periods of several minutes revealed a persistent association between these 2 structures (Fig. 3B; Video S5).

To investigate the spindle pole-anchored PML bodies further, we analyzed the spatial organization of PML bodies and microtubules in fixed HaCaT cells using antibodies against PML and α-tubulin. We detected PML bodies at one of the 2 spindle poles in 33% of the cells examined, while 6% of the cells had detectable clusters of PML at both poles (Fig. 3C and D). In the remaining cases (61% of the cells examined) we did not detect PML bodies at any of the 2 poles.

Finally, we analyzed the localization of spindle pole-attached PML bodies in relation to the centrosome marker protein pericentrin in HaCaT, U2OS, and HeLa cells. We detected PML bodies in close proximity to mitotic centrosomes in all 3 cell lines (Fig. 3E). Notably, PML was generally observed close to, but was rarely seen to overlap extensively with, centrosomes (Fig. 3E). In addition, we did not detect association between centrosomes and PML bodies in interphase cells (data not shown).

Discussion

In the present paper we demonstrate that the cytoplasmic compartment early endosome and the nuclear compartment PML body join to become stably associated during the mitotic phase of the cell cycle. At completion of mitosis the 2 compartments separate concomitant with activation of endosome trafficking and PML body nuclear import (Fig. 4). Notably, PML bodies are also subjected to distinct biochemical alterations (including desumoylation, loss of nuclear proteins, and nucleoporin recruitment) at entry into and exit from mitosis.^11,12,21,22 Thus, the interactions between PML bodies and early endosomes may be strictly regulated by cell cycle-dependent control mechanisms.

Figure 4. Schematics of PML body and early endosome dynamics during mitosis. PML bodies and early endosomes are kept separate by the nuclear membrane during interphase and tether on mitotic entry. The 2 compartments separate in the cytoplasm of newly divided cells concomitant with activation of PML nuclear import. MAPP, mitotic assemblies of PML protein; CyPN, cytoplasmic assemblies of PML and nucleoporins.

Although PML is predominantly detected within nuclear PML bodies, several studies have indicated a role of this protein in cytoplasmic activities, such as transport of calcium and endosome-mediated signaling.^15,16 These functions may be partly mediated by cytoplasmic PML splice variants that lack a functional nuclear localization signal. Indeed, ectopic expression of cytoplasmic PML has been shown to support PML-mediated functions in calcium homeostasis and TGF-β signaling,^15,16 and nuclear PML isoforms carrying mutations in the nuclear localization signal have been found to be effectively targeted to early and late endosomes of interphase cells.¹⁷ However, endogenous cytoplasmic PML isoforms are expressed at low levels, and they can potentially become transported to the nucleus through heterodimerization with nuclear import-competent PML isoforms. The present study suggests that cytoplasmic proteins reseeding at the surface of early endosomes may be contacted by the full complement of PML isoforms at cell division.

Endocytosis and transport of endocytic vesicles are generally thought to be less active in mitotic cells compared with interphase cells.^23-25 It is possible that PML bodies and early endosomes interact during mitosis to facilitate component transfer between the 2 compartments. The effect of such protein exchange could potentially become executed after completion of cell division, when endocytic transport and signaling is more active. Alternatively, the tethering of PML bodies to early endosomes during cell division may represent a mechanism that controls faithful distribution of PML bodies to newly divided daughter cells. In agreement with this, mitotic endosomes have previously been shown to control segregation of plasma membrane-anchored surface proteins such as planar cell polarity proteins²⁶ and the notch–delta signaling complex.²⁷

Although the majority of MAPPs appeared to become anchored to early endosomes, our live cell particle tracking experiments also revealed a minor sub-population of MAPPs that becomes attached to structures near the poles of the mitotic spindle. In a previous study the PMLIII isoform was shown to target centrosomes in order to regulate their replication during cell division.²⁸ A subsequent study, however, failed to detect colocalization between PML and centrosome proteins in interphase cells.²⁹ The present study shows that a sub-population of MAPPs becomes anchored to the poles of mitotic spindle close to centrosomes, and that these interactions are disrupted as cells progress from mitosis to interphase.

This work shows that dividing cells may take advantage of the opportunity to bring cytoplasmic and nuclear compartments into close proximity following breakdown of the nuclear membrane at entry into mitosis. This concept may have important implications for nucleo–cytoplasmic trafficking, growth regulation, and cell fate determination.

Materials and Methods

Cells and lentivirus infection

The immortalized human keratinocyte cell line HaCaT³⁰ was obtained from CLS cell line services. Cells were grown in Iscove Modified Dulbecco medium (MedProbe) supplemented with 10% fetal calf serum (Fisher Scientific).

Lentivirus-based vectors expressing PML1 was provided by Roger Everett.³¹ The plasmid PGK-H2BmCherry³² was acquired from Addgene. Lentivirus constructs expressing mCherry-Rab5 and mCherry-2xFYVE under control of the PGK promoter were generated by fusing the open reading frame of the 2xFYVE probe or Rab5 with mCherry and transferring the resulting construct into pLenti-PGK-Puro-Dest (#19068, Addgene) using Gateway recombination. A lentivirus construct expressing mCherry-centrin1 was generated by fusing the open reading frame of centrin1 (provided by Sebastian Patzke) to mCherry and transferring the resulting construct to a gateway-enabled version of pCDH-EF1a-IRES-Puro (SBI).

Production of infectious lentiviral particles and stable transduction of HaCaT cells were performed as previously described.²²

Immunofluorescence and analysis of sub-compartment interactions

HaCaT cells grown on glass coverslips were fixed using 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS). Anti-α-tubulin-labeled cells were fixed in 100% ice-cold methanol. Fixed cells were processed for immunolabeling as previously described.³³ The following antibodies were used: rabbit anti-PML (H-238; Santa Cruz Biotechnology), mouse anti-PML (PG-M3, Santa Cruz Biotechnology), mouse anti-α-tubulin (DM1A; Sigma), rabbit anti-pericentrin (Covance), mouse anti-EEA1 (14/EEA1; BD transduction Laboratories), rabbit anti-transferrin receptor (CBL47; Millipore), and rabbit anti-Lamp1 (H4A3; Developmental Studies, Hybridoma Bank). To quantify PML body/endosome or PML body/mitotic spindle-interactions, we first acquired images using a Zeiss LSM 510 laser confocal microscopy system equipped with a Zeiss Axiovert Observer Z1 inverted microscope and a 40× oil immersion lens. For each of the cells analyzed 6–10 Z-scans spanning the entire cell were generated. Each Z-section was manually inspected, and in the case of samples labeled by EEA1 and PML antibodies, interactions were scored if the fluorescence signal emitted by the 2 fluorochromes were observed to be overlapping or completely adjacent (no visual space in between). In the case of samples labeled by α-tubulin and PML antibodies, attachment to spindle poles were scored if the PML signal localized at the mitotic spindle near the centrosome.

Gated stimulated emission depletion (gSTED) microscopy

HaCaT cells grown on glass coverslips were fixed in 4% PFA in PBS. After fixation and permeabilization in 0.5% Triton X-100, cells were incubated in blocking buffer (0.5% bovine serum albumin in PBS) for 15 min at room temperature, followed by incubation with primary antibodies in blocking buffer at 4 °C overnight. The primary antibodies used were rabbit anti-PML (NB100-59787; Novus Biologicals) and mouse anti-EEA1 (BD610456; BD transduction Laboratories). The next day, cells were incubated with secondary antibodies goat anti-mouse Chromeo505 (Active Motif) and goat anti-rabbit biotin-XX Fab (B21078; Life Technologies) in blocking buffer for 1 h at 37 °C, followed by incubation with V500-streptavidin (BD Horizon^™) in blocking buffer for 30 min at 37 °C. Lastly, samples were mounted in ProLong Gold anti-fade mounting media (Molecular probes^®; Life Technologies). STED images were acquired using a Leica TCS SP8 gated STED microscope and a 100× oil immersion lens. Images were deconvolved using the Huygens Professional software (Scientific Volume Imaging).

Live cell imaging and particle tracking

Cells used for live cell microscopy were grown in normal growth medium in 35-mm glass-bottom petri dishes from MatTek Corporation. To track single PML bodies, early endosomes, and centrosomes, we used a Zeiss 510 laser scanning confocal microscope system equipped with a Zeiss Axiovert Observer Z1 inverted microscope and a 40× oil immersion objective built into an incubation chamber maintaining 37 °C and 5% CO₂. Sequential Z-scans of the EYFP and mCherry channels were acquired by collecting a single Z-plane at intervals between 1 and 4 seconds. The Z-position was adjusted manually during imaging to keep tracked particles in focus. For whole-cell imaging, Z-stacks consisting of 3 Z-scans were collected at 2-minute intervals between frames.