Cytoskeletal influences on nuclear shape in granulocytic HL-60 cells
© Olins and Olins; licensee BioMed Central Ltd. 2004
Received: 11 April 2004
Accepted: 19 August 2004
Published: 19 August 2004
Skip to main content
© Olins and Olins; licensee BioMed Central Ltd. 2004
Received: 11 April 2004
Accepted: 19 August 2004
Published: 19 August 2004
During granulopoiesis in the bone marrow, the nucleus differentiates from ovoid to lobulated shape. Addition of retinoic acid (RA) to leukemic HL-60 cells induces development of lobulated nuclei, furnishing a convenient model system for nuclear differentiation during granulopoiesis. Previous studies from our laboratory have implicated nuclear envelope composition as playing important roles in nuclear shape changes. Specifically noted were: 1) a paucity of lamins A/C and B1 in the undifferentiated and RA treated cell forms; 2) an elevation of lamin B receptor (LBR) during induced granulopoiesis.
The present study demonstrates that perturbation of cytoskeletal elements influences nuclear differentiation of HL-60 cells. Because of cytotoxicity from prolonged exposure to cytoskeleton-modifying drugs, most studies were performed with a Bcl-2 overexpressing HL-60 subline. We have found that: 1) nocodazole prevents RA induction of lobulation; 2) taxol induces lobulation and micronuclear formation, even in the absence of RA; 3) cytochalasin D does not inhibit RA induced nuclear lobulation, and prolonged exposure induces nuclear shape changes in the absence of RA.
The present results, in the context of earlier data and models, suggest a mechanism for granulocytic nuclear lobulation. Our current hypothesis is that the nuclear shape change involves factors that increase the flexibility of the nuclear envelope (reduced lamin content), augment connections to the underlying heterochromatin (increased levels of LBR) and promote distortions imposed by the cytoskeleton (microtubule motors creating tension in the nuclear envelope).
Granulopoiesis, the differentiation of peripheral blood granulocytes, involves dramatic nuclear and cytoplasmic structural changes . Committed bone marrow progenitor cells possess ovoid-shaped nuclei with prominent nucleoli and a paucity of heterochromatin. The mature terminally differentiated human neutrophil (polymorphonuclear granulocyte) exhibits a distinctly lobulated (segmented) nucleus with shrunken nucleoli and extensive peripheral heterochromatin. The mature neutrophil is released into the bloodstream, where it circulates as a round unpolarized cell. Responding to chemotactic agents produced by infection and tissue damage, the circulating neutrophil changes cell shape, converting to a rapidly migrating polarized cell. Mature granulocytes have a limited lifespan, succumbing to apoptosis within a few days following release into the bloodstream.
Several established tissue culture cell lines have been investigated as model systems for understanding the events and mechanisms of granulopoiesis . Previous studies from our laboratory have employed the HL-60 cell system to examine nuclear lobulation and cytoskeletal polarization [3–5]. HL-60 cells exhibit granulocytic differentiation in response to added retinoic acid (RA) , eventually undergoing apoptotic death . Our studies on RA induced granulocytic differentiation of HL-60 cells implicated two major factors in the nuclear lobulation process: 1) very low cellular levels of lamins A/C and B1; 2) a significant increase in cellular levels of lamin B receptor (LBR). Ultrastructural studies suggested that during nuclear differentiation, nuclear envelope surface area was increased . In addition to nuclear lobulation, these studies demonstrated the formation of extensive nuclear envelope outgrowths, denoted "nuclear e nvelope-l imited c hromatin s heets" or ELCS. The suspected role of LBR was confirmed in subsequent studies [8, 9], which demonstrated that a genetic deficiency of LBR correlates with hypolobulated granulocyte nuclei in the human Pelger-Huet anomaly  and the murine Ichthyosis mutation . LBR is an integral membrane protein of the nuclear envelope inner membrane, with putative interactions to lamin B, chromatin and HP1α . The mechanistic relationships between changes in nuclear envelope composition and granulocytic nuclear lobulation are currently unknown.
The position of the interphase nucleus within the cell appears to be regulated by microtubules and associated dynein  and actin and associated spectrin-like proteins [12–15]. However, only fragmentary data exists examining the relationship, if any, between cytoskeletal elements and nuclear shape. Absence of intermediate filaments (vimentin) has been correlated with nuclear envelope folds or invaginations . Employing HL-60 cells, evidence has been published that neither microtubules (MTs) nor the actin microfilament system are essential for the establishment of nuclear lobulation . Our laboratory chose to investigate these conclusions concerning HL-60 cells in greater detail. Employing the same cell subline (HL-60/S4), we demonstrated that brief (i.e., 2 and 4 hour) treatments of undifferentiated or granulocytic (RA treated) cells with various cytoskeletal modifying chemicals had no obvious effects upon nuclear shape, although cell shape was strongly affected . However, prolonged (i.e., 2 day) exposure of undifferentiated HL-60/S4 cells to nocodazole (NC, disrupts MTs) or taxol (TX, stabilizes MTs), but not cytochalasin D (CD, disrupts actin microfilaments), resulted in rapid apoptotic death. For this reason, the present study emphasizes the prolonged exposure of cytoskeleton-modifying chemicals on Bcl-2 overexpressing HL-60 cells [18, 19]. These cells are more refractory to undergoing apoptosis, but can still be induced with RA to exhibit granulocytic differentiation. Employing HL-60-bcl-2 cells , we demonstrate that the integrity of the MTs system, but not the actin microfilament system, is essential for the nuclear lobulation process during in vitro granulopoiesis. Prolonged exposure of HL-60-bcl-2 cells to CD or TX does lead to perturbations of nuclear shape, independently of RA induced nuclear differentiation.
The previous publication  also examined the role of MTs on nuclear lobulation in HL-60/S4 cells. They exposed RA treated cells to 0.1 μM (0.03 μg/ml) NC for up to two days, concluding that the integrity of MTs was not important to the formation granulocytic nuclear lobes. We repeated this experiment, including a separate comparison of HL-60/S4 cells exposed to 0.1 μM TX. The results clearly indicated ~100% apoptotic cell death by day 2 with either NC or TX, preventing any firm conclusion of whether MT integrity is important for nuclear lobulation. We concluded that exploration of this issue would be better accomplished using HL-60-bcl-2 cells, which are more refractory to undergoing rapid apoptosis.
Prolonged exposure of HL-60 cells to various cytoskeleton-modifying chemicals (i.e., nocodazole and taxol) are very harmful to cell viability, inducing rapid apoptosis. The present study was conducted primarily on a Bcl-2 overexpressing subline of HL-60 cells, which is more refractory to apoptosis and exhibits the chemically induced differentiation properties of the parent cell line . Lethal effects of NC or TX are delayed in HL-60-bcl-2 cells, compared to HL-60/S4 cells, allowing a window of time for determining the effects of these MT modifying chemicals on nuclear shape and nuclear differentiation.
There are two additional differences between HL-60-bcl-2 and HL-60 cells observed in this study. The first difference: ~10% of the undifferentiated HL-60-bcl-2 cells exhibit micronuclei, compared to ~0.8% of HL-60 cells . Micronuclei are generally regarded as the products of abnormal mitoses, where the enclosed chromosomes or chromosome fragments fail to congress at the mitotic plate, but are still surrounded by a post-mitotic nuclear envelope [27–30]. In normal human cells growing in culture (e.g., lymphocytes), they occur in less than 0.5% of the cells . They can be induced in cells by treatment with a variety of DNA breakage conditions (e.g., irradiation) or spindle disrupting agents (e.g., colchicine). Micronuclei in HL-60 cells are described as representing amplified acentric euchromatic genes (such as c-myc) and appear to form dynamically during S phase . We observed that micronuclei in HL-60-bcl-2 cells exhibited immunostaining of centromeres, heterochromatin and nucleolar antigens (Figure 7), which suggests differences from the earlier interpretation of the nature of micronuclei in HL-60 cells.
Especially puzzling was our observation that the majority of micronuclei possessed comparable amounts of lamin B, but reduced amounts of LBR, in comparison to the companion main nucleus. A recent study employing MCF-7 cells observed micronuclei containing lamins A/C and B1 and a relative deficiency of LBR . The same study demonstrated that when significant numbers of micronuclei were induced by prolonged exposure to the spindle disrupting chemical curcumin, the resulting nuclear envelopes contained LBR. Current views on the sequence of protein additions to post-mitotically reformed nuclei agree that LBR enters the nascent nuclear envelope well before lamin B [34, 35]. It is possible that untreated HL-60-bcl-2 and MCF-7 micronuclear envelopes form later than the main nucleus, after the cytoplasmic pool of LBR is exhausted, whereas curcumin "induced" micronuclei form at about the same time as main nuclei.
The second difference: RA treated HL-60-bcl-2 cells exhibit a small population (up to 20%) with multilobed nuclei, which we almost never observed with HL-60/S4 cells. There may be some clinical significance to this latter observation. A small percent of granulocyte nuclei with 5 or more nuclear lobes in human blood smears is considered diagnostic for megaloblastic anemia (vitamin B12 or folic acid deficiency) . The present data with RA treated HL-60-bcl-2 cells suggests that delayed or dysfunctional apoptosis might play a role in these human diseases. It is of interest that neutrophil nuclear multilobulation (hypersegmentation) has been described in two circumstances that delay apoptosis: 1) Glucocorticoid administration to patients induces hypersegmentation , and in vitro glucocorticoid treatment of neutrophils prolongs their survival [37–39]. 2) Granulocyte colony-stimulating factor (G-CSF) administration to rats induces hypersegmentation in mature neutrophils , and in vitro treatment of neutrophils with G-CSF prolongs their survival [38, 41].
The most important present observation is the suppression of RA induced nuclear lobulation in HL-60-bcl-2 cells by simultaneous exposure to NC (Figure 9). This observation implies that MTs must be intact during nuclear differentiation. Furthermore, we have observed (data not shown) that HL-60-bcl-2 cells can be made 0.1 μM NC on day 2 or 4 after addition of RA, still exhibiting inhibition of nuclear lobulation. This observation suggests that the requirement for intact MTs is not an early event in the nuclear differentiation process. When RA and NC treated HL-60-bcl-2 cells were examined by confocal immunofluorescent staining with anti-lamin B, the ovoid nuclei revealed extensive "wrinkling" of the nuclear envelope (Figures 10 and 11). We suggest that this "wrinkling" reflects expansion (growth) of the nuclear envelope in the absence of nuclear lobulation.
The present study demonstrates a lack of requirement of an intact actin microfilament system for RA induced granulocytic nuclear differentiation (Figure 1). This conclusion is based upon the observation that incubation of HL-60/S4 and HL-60-bcl-2 cells with 1 μM CD in the presence of RA does not inhibit nuclear lobulation or the formation of ELCS. Our results agree with an earlier study . In addition, we extended the incubation (with CD and RA) to times where nuclear differentiation is more definitive. Our data also demonstrated that prolonged incubation with CD alone produces significant nuclear envelope folding, which is best appreciated by confocal immunostaining of the nuclei (Figure 8). The mechanism for the nuclear shape changes in undifferentiated cell nuclei exposed to prolonged incubation with CD is presently unknown. There is increasing evidence that actin microfilament interactions with the nuclear envelope exist, mediated via actin binding spectrin-like proteins, variously named Syne 1 and 2/ nesprin-1 and 2/ ANC1/ NUANCE [12–15]. However, a counter argument for the relevance of this actin interacting system to nuclear differentiation in the HL-60 cell system can be made. Nesprin-1 binds directly to lamin A and emerin ; but undifferentiated and granulocytic HL-60 cells possess negligible amounts of lamin A, with emerin primarily localized in the cytoplasm . Furthermore, published  and unpublished data (A. Karakesisoglu, A. Olins and D. Olins) indicate that HL-60 cells, undifferentiated or RA treated, possess only trace amounts of NUANCE.
The finding that TX treatment of cells has dramatic consequences to interphase nuclear structure has been reported before. Exposure of human carcinoma cells (Ishikawa and HeLa) to 0.01–0.1 μM TX for up 20 hours, followed by incubation in drug-free media for up to 72 hours, led to nuclear envelope "unraveling" and clustering of nuclear pores . The authors observed lobulated and micronuclei, much as observed here. In the case of HL-60-bcl-2, the nuclear structural changes superficially mimic the effects of RA treatment, but occur much faster (Figure 12). The bundling of MTs in HL-60 cells in response to exposure to 1.0 μM TX for 24 hours has also been reported , but no mention was made of the nuclear envelope changes. Furthermore, the authors noted that at 0.1 μM TX (or greater) the HL-60 cells showed clear apoptosis by 24 hours. In a later paper by the same group , increased expression of Bcl-2 or Bcl-xL yielded HL-60 cells with considerably greater resistance to TX induced apoptosis. It seems to us that the dramatic cell and nuclear changes that we observe have no obvious relationship to normal granulocytic nuclear lobulation, but underscore an effect of MT integrity upon nuclear shape. The formation of micronuclei, in particular, suggests that TX may exert its effects on nuclear shape by interfering with normal mitotic chromosome distribution.
We suggest that nuclear envelope deformability is an important factor and depends upon the amount of underlying lamins: the less lamin protein, the more pliable the nuclear envelope. Granulocytic forms of HL-60 exhibit deficiency of lamins A/C and B1, whereas both types of lamins are present in monocyte/macrophage forms . The absence of lamins A/C in normal granulocytes and their presence in macrophages has been previously noted . The present data on granulocytic differentiation in HL-60-bcl-2 demonstrates that nuclear lobulation correlates with low levels of lamins A/C and B1 coupled with a rise in LBR levels (Figure 4).
The pivotal role of LBR in determining granulocytic nuclear lobulation was demonstrated with the human Pelger-Huet anomaly and murine Ichthyosis mutations [8, 9]. These studies demonstrated that LBR functions in a dose-dependent manner: homozygous mutants present a more severe phenotype and lower amounts of LBR, than in the heterozygous state. The influence of LBR on granulocytic nuclear lobulation is consistent with its known properties . LBR is embedded within the nuclear envelope inner membrane via 8 transmembrane domains (the C-terminal ~400 aa), and associated with lamin B, chromatin and HP1α in the N-terminus (~200 aa). In the absence of sufficient LBR, nuclear lobulation is prevented and the normally peripheral heterochromatin is redistributed into a more centrally condensed form [8, 9].
A role for MT integrity is implicit in our present observation that exposure of HL-60-bcl-2 cells to NC prevents nuclear lobulation during exposure of the cells to RA. Direct interaction between MTs and the interphase nuclear envelope in mammalian cells have not documented. However, a recent model for mitotic nuclear envelope breakdown [47, 48] can be adapted to the situation of nuclear lobulation. The nuclear envelope breakdown model proposes that cytoplasmic dynein attaches MTs to the nuclear envelope, pulling the envelope towards the centrosomal region. The excess envelope near to the centrosome produces nuclear invaginations; the tension on the non-growing nuclear envelope generates tears and the mixing of nuclear and cytoplasmic materials. If we assume that the nuclear envelope is still growing in the case of RA differentiating HL-60 cells, invaginations and lobulations might be expected to accumulate within the intact nuclear envelope. The present study demonstrates proximity of the centrosome to nuclear lobulation (Figure 15). But as yet, there is no direct evidence for cytoplasmic dynein playing a role in granulocytic nuclear differentiation.
Our present data suggests that an intact actin microfilament system does not play a major role in granulocytic nuclear lobulation (Figure 1). The best described mechanism of actin interacting with the nuclear envelope involves spectrin-like proteins, which may bridge cytoplasmic actin to nuclear envelope proteins, such as lamin A and emerin [12–15]. But undifferentiated and granulocytic HL-60 cells possess very little lamin A/C and emerin is primarily cytoplasmic , suggesting that this bridging system may not be functional in these cell forms.
In SW-13 cells the absence of intermediate filaments (vimentin) has been correlated with nuclear envelope folds or invaginations . We have observed a decrease in vimentin during differentiation of granulocytic HL-60/S4 cells [3–5], suggesting that reduced vimentin concentrations may contribute to granulocytic nuclear lobulation.
The proposed model for granulocytic nuclear lobulation yields several testable predictions: 1) Expression of lamins A/C and B1 in HL-60 cells should strengthen the nuclear envelope, minimizing nuclear lobulation following exposure of the cells to RA; 2) "Knock-down" or expression of a "dominant negative" form of LBR in HL-60 cells would be expected to suppress granulocytic lobulation; 3) Overexpression of dynamitin in HL-60 cells should inhibit dynein activity , preventing nuclear lobulation following RA induced differentiation. A number of these experiments are already in progress.
Employing Bcl-2 overexpressing HL-60 cells, which are more refractory to induced apoptosis than the parent cell line, we demonstrated that disruption of the MTs by nocodazole prevented nuclear lobulation in response to RA treatment. These results implicate the necessity of an intact MT system for granulocytic nuclear shape. Cytochalasin D, on the other hand, did not suppress RA induced nuclear lobulation. Combined with the decreasing levels of intermediate filaments (vimentin) during differentiating granulocytic forms of HL-60 cells, the role of the MT system appears to be quite central to the nuclear shape change. Recent models on the role of a MT bound motor (dynein) in facilitating mitotic nuclear envelope breakdown suggest that similar tension forces on the differentiating granulocyte nucleus, combined with continued influx of nuclear envelope components, could explain nuclear invaginations and lobulation.
Two HL-60 cell sublines were employed in this study: HL-60/S4 , which achieves maximum nuclear lobulation in 4 days following addition of RA; HL-60-bcl-2 , which achieves maximum lobulation in ~7 days, comparable to the parent cell line. Cultivation conditions were exactly as described previously [3, 4].
The following chemicals were purchased from Sigma-Aldrich (St. Louis, MO): all-trans retinoic acid (RA), nocodazole (NC), taxol (TX), cytochalasin D (CD). Stock solutions of these chemicals were stored at -20°C as described earlier [3, 4].
Guinea pig antisera were the generous gifts of two Ph. D. students in the laboratory of H. Herrmann (German Cancer Research Center, Heidelberg): anti-LBR and anti-emerin, from C. Dreger; anti-lamin A and anti-lamin B1, from J. Schumacher. Goat anti-lamin B was obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Rabbit anti-4 × dimethyl H3K9 (lysine 9 on histone 3) was generously provided by T. Jenuwein (Biocenter, Vienna). Human auto-antisera anti-centromere (CREST) and anti-nucleolus (antigens unknown) were purchased from Antibodies Inc. (Davis, CA). Rabbit anti-calreticulin was from Calbiochem (San Diego, CA). Mouse monoclonal antibodies against α-tubulin, γ-tubulin and Golgi p58 were all purchased from Sigma-Aldrich. Mouse monoclonal anti-vimentin (3B4) was a gift of H. Herrmann and has been described before . FITC-, Cy3-, Cy5- and HRP-conjugated donkey secondary antibodies were all purchased from Jackson ImmunoResearch Laboratory, Inc. (West Grove, PA). TO-PRO-3 and SlowFade were obtained from Molecular Probes, Inc. (Eugene, OR).
For Wright-Giemsa staining, cells were cytospun onto ethanol-cleaned microscope slides, fixed in room temperature methanol for 15 min, air-dried and stained as described earlier . For analysis of the percentage of cells in the various nuclear morphology categories (Figures 3, 9 and 12), approximately 150 cells were observed and classified in each experiment.
In the majority of immunostaining experiments the procedure followed that described previously , with some modifications, as follows: 1) Microscope slides were soaked overnight in 1/1 ethanol/ether and freshly coated with poly-L-lysine (MW ~150–300,000; Sigma-Aldrich), just before centrifugation of the cells. 2) No coverslip was used during antibody incubations, to minimize loss of cells. 3) Prior to the application of primary antibodies, slides were incubated with 5% normal donkey serum (Jackson ImmunoResearch Laboratory) in PBS for 15–30 min at 37°C in a moist chamber. 4) In the cases of monoclonal anti-γ-tubulin and anti-vimentin, where antigenicity appeared to be destroyed by HCHO fixation, slides were fixed in methanol (-20°C, 10 min), acetone (-20°C, 1 min) followed by three washes in PBS (5 min each). The cells were not excessively flattened by either procedure, maintaining sufficient 3-D structure to justify stereo viewing. Confocal images were collected on a Zeiss 510 Meta. Stereo images were obtained as ± 15° projections through the stack of confocal images.
SDS total cell extracts, obtained from undifferentiated and RA treated HL-60-bcl-2 cells over a three-week period (simultaneously with preparation of slides for Wright-Giemsa staining), were analyzed by immunoblotting as previously described . No attempt was made to separate viable cells from debris prior to SDS extraction. Comparable amounts of total cell protein were loaded into each lane of the SDS-PAGE, as judged by Ponceau S staining of the PVDF membrane after protein transfer. Several different ECL exposures were collected on X-ray film and subsequently scanned with a Bio Rad Chemi Doc for densitometric analyses.
nuclear envelope-limited chromatin sheets
This work was supported by funds from the Department of Biology, Bowdoin College. We wish to express our gratitude to members of the Bowdoin faculty who have encouraged our research, especially W. Steinhart, B. Kohorn and D. Page. We also wish to express our appreciation to our colleagues (H. Herrmann and P. Lichter) at the German Cancer Research Center for stimulation and enlightened criticism.
This article is published under license to BioMed Central Ltd. This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.