CML PMNL do not show classical morphological responses

Unstimulated normal PMNL were round (Figure ​(Figure1a).1a). After fMLP stimulation for 0.5 min, 90% of PMNL showed either blebbing or classical oriented cells with lamellipodia and uropod (Figure ​(Figure1b).1b). At 5 min, the cells became elongated and later they rounded up. Unstimulated CML PMNL were round (Figure ​(Figure1c).1c). At early time points of fMLP stimulation, in about 45% of samples, 50% cells showed fine peripheral projections (Figure ​(Figure1d).1d). Classical lamellipodia and uropod formation was rare. With increasing time the cells rounded up. Thus, CML PMNL exhibited lack of morphological responses towards fMLP.

Open in a separate window

Figure 1

Morphology of PMNL: Phase contrast micrographs showing morphology of a) unstimulated and b) fMLP stimulated (0.5 min) normal PMNL. Arrows indicate classical morphology of oriented cells showing lamellipodia and uropods. c) Unstimulated and d) fMLP stimulated (0.5 min) CML PMNL.

Actin expression is lower and stimulation of actin polymerization is absent in CML PMNL

Western blot analysis would estimate expression of total actin i.e., G plus F actin in the cells. In normal PMNL, at early time points of fMLP stimulation, 60% of the normal samples showed a drop in total actin as compared to that in unstimulated PMNL (Figure ​(Figure2a).2a). The total actin reduced significantly by 20% and 36% after 5 and 45 min of fMLP stimulation, respectively (Table ​(Table1a).1a). In CML PMNL, 50% of the samples showed a slight drop in total actin during early time points of fMLP stimulation (Figure ​(Figure2a).2a). The average total actin dropped significantly (15%) at 0.5 and 45 min of fMLP stimulation (Table ​(Table1a).1a). Though unstimulated and fMLP-stimulated normal PMNL showed higher levels of total actin as compared to the levels in the respective CML PMNL, this difference was not statistically significant.

Open in a separate window

Figure 2

Time kinetics of GTPase and actin stimulation: (a) Western blot for expression of actin and GTPases - ras, rac1 and rhoA: lane 1- unstimulated control, lane 2- fMLP 0.5 min, lane 3- fMLP 5 min, lane 4- fMLP 10 min, lane 5- fMLP 30 min and lane 6- fMLP 45 min in normal (n = 14) and CML PMNL (n = 22); For ras and rac1 both the patterns of expression, i.e., 21 kd alone or 21 kd along with 25 kd are depicted. (b-e) Representative FCM plots for expression of F-actin and GTPases- Ras, rac1 and rhoA in normal (n = 14) and CML PMNL (n = 22): i) isotype, s) secondary control, a) unstimulated, b) fMLP 0.5 min, c) fMLP 5 min, d) fMLP 10 min, e) fMLP 30 min, and f) fMLP 45 min.

Flow cytometry (FCM) analysis showed that on fMLP stimulation for 0.5 min, F-actin in normal PMNL increased significantly (range 1.1 - 3.5 fold) (Figure ​(Figure2b).2b). This was followed by a drop and then a second increase. Since the time frame for this second response varied from sample to sample, the average of the median fluorescence did not show considerable change (Table ​(Table1b).1b). In CML, only 25% samples showed an increase in F-actin after 0.5 min of stimulation that was comparable to that in normal PMNL (Figure ​(Figure2b).2b). Overall, F-actin levels were steady in fMLP stimulated CML PMNL (Table ​(Table1b).1b). Thus, CML PMNL showed defective stimulation of actin polymerization.

Organization of F-actin is altered in CML PMNL

Unstimulated normal PMNL showed F-actin as weak cytoplasmic and bright peripheral fluorescence. At 0.5 min of stimulation, an increase in F-actin along with cell polarization was seen. F-actin was concentrated in blebs or lamellipodia and uropods (Figure ​(Figure3a).3a). With increasing time, the pattern of F-actin distribution was similar to that seen in the unstimulated cells. Changes in F-actin levels observed by laser scanning confocal microscope (LCM) were similar to those seen using FCM. In unstimulated CML PMNL, F-actin was seen as weak cytoplasmic and slightly brighter peripheral fluorescence. After fMLP treatment, though morphological changes were observed in some cells, F-actin picture remained unaltered (Figure ​(Figure3a3a).

Open in a separate window

Figure 3

(a) F-actin and rhoA localization: Middle Z planes of LCM images showing transmission images (upper panel) and distribution of F-actin (middle panel) and rhoA (lower panel) in unstimulated normal and CML PMNL and those stimulated with fMLP (0.5 min); U - Uropod, L - Lamellipodia, P - Projection. (b) Proposed model for interactions between GTPases and F-actin: (i) Unstimulated and (ii) fMLP stimulated normal PMNL; (iii) unstimulated and (iv) fMLP stimulated CML PMNL. Bold font indicates crucial regulatory molecules.

Actin provides structural framework and defines cell shape and polarity. Its dynamic properties provide the driving force for the cells to move and divide. Changes in its expression could alter G to F polymerization dynamics. Defects in actin could be at the level of expression or polymerization. Alterations within actin could be via mutations in actin, changes in upstream regulatory signaling proteins or actin binding proteins. Altered actin expression and polymerization is known to be associated with cancer [5]. In normal and CML PMNL, basal levels of total actin and F-actin were not significantly different. On fMLP treatment, total actin levels decreased in both, but normal PMNL showed significant increase in F-actin, indicating that the total actin expression remained above the critical levels and hence actin could polymerize. Tarachandani et al have shown that in CML PMNL, actin expression, though lower, was sufficient for polymerization [6]. Hence, lack of stimulation of actin polymerization and altered F-actin architecture in CML PMNL could be due to defects in signalling leading to defective actin polymerization. The signalling molecules - rhoGTPases, play an important role in the spatial and temporal organization of actin. Ras, the major target of bcr-abl, activates rhoGTPases. In the present studies, on fMLP stimulation CML PMNL showed filopodia-like but thin projections, suggesting involvement of rhoA. This finding was different from that in normal PMNL where lamellipodia formation indicative of rac signaling were seen [7]. Since rhoA and rac govern dynamics and spatial organization of F-actin differently, changes in actin expression and F-actin localization, leading to defective functions [4,8] in CML PMNL, could be due to altered rhoGTPases regulation.

About 70% of the bcr-abl protein localizes to the cytoskeleton. This localization is important in cellular transformation. An actin-binding domain of the bcr-abl kinase enhances its leukemogenicity [9]. Analogously, a mutation in the F-actin binding domain of c-abl reduces its ability to transform fibroblasts [10]. Constitutively active bcr-abl alters actin dependent cell adhesion and motility by phosphorylating actin-binding proteins [11]. Another mechanism that alters actin functioning in the neoplastic state targets upstream regulators of actin binding proteins. Ras - the key signalling molecule in the actin polymerization pathway, is also a major target of bcr-abl. Ras regulates cell proliferation pathways that in turn regulate gene expression, and morphological (or motility) pathways controlling the actin cytoskeleton [12]. Therefore, the GTPases - ras, rho A and rac1 were studied.

Ras expression in CML PMNL is refractory to fMLP treatment

In the majority of normal and CML PMNL, a sharp 21 kd band was seen for ras in the Western blots (Figure ​(Figure2a).2a). But 43% of normal and 32% of CML samples showed an additional band at 25 kd at all the time points, indicating existence of differential isoprenylation of ras. For densitometry analysis, both the bands were considered collectively.

On fMLP stimulation, 50% normal samples showed increase in ras at early time points. With increasing time, this percentage increased to 79% resulting in a significant increase in ras (about 1.5 fold) at 30 and 45 min (Table ​(Table1a).1a). On fMLP stimulation, CML PMNL showed very little or no change in ras expression (Figure ​(Figure2a).2a). Normal and CML PMNL expressed similar basal levels of ras. But after stimulation for 30 min, normal PMNL showed significantly higher ras levels as compared to the corresponding CML PMNL.

In FCM estimation, normal PMNL showed about two-fold increase in ras (range 1.35-4.2) after fMLP stimulation (Figure ​(Figure2c).2c). The increase was significant at 5 and 10 min of fMLP stimulation (Table ​(Table1b).1b). In CML PMNL, the majority of the samples did not show any response to fMLP. Only at 45 min of stimulation, 36% of the samples showed increase in ras resulting in a statistically significant increase (1.2 fold) in average ras expression (Figure ​(Figure2c,2c, Table ​Table11b).

Basal ras levels in normal PMNL were slightly higher than in CML PMNL. But an extremely delayed and low response of CML PMNL to fMLP resulted in significantly higher ras levels in normal PMNL, stimulated for 5 and 10 min than the respective CML PMNL. Hammond et al. have suggested that intracellular signalling could occur through modulation of the oscillations in response to stimulus. Cancer can result from changes in the oscillation frequencies, amplitudes and phasings of signaling molecules [13]. In Dictyostelium discoideum ras activation stimulates a small amount of preexisting, membrane-associated PI3K, causing F-actin polymerization [14]. Thus, defective ras dynamics might lead to defective actin polymerization. The present findings reveal that on fMLP stimulation, ras levels increased only in normal PMNL, indicating defects in signals regulating ras expression in CML PMNL.

Intracellular localization of ras in normal and CML PMNL is comparable

Unstimulated and stimulated normal and CML PMNL showed diffused cytoplasmic staining for ras. In normal, bleb formation, observed at 0.5 min of stimulation, was associated with an increase in ras. The fluorescence intensity increased with increase in stimulation time, indicating an increase in ras. Only 50% CML PMNL that showed morphological changes on stimulation showed slight increase in ras expression. Overall change was not noted in the levels of ras in response to stimulation. There was no distinct difference in the localization of ras in normal and CML PMNL (Additional file 1: Figure S1 & Additional file 2: Figure S2). Since only 10% of the total GTPase undergoes a transition from inactive to active state, limitations of the used LCM technology and lack of specific probe could have made these alterations undetectable. Thus, at present, defects in stimulation of actin polymerization could be partly attributed to alterations in dynamics of ras expression.

The morphological pathway is branched into PI3K-dependent and PI3K-independent pathways. The PI3K-dependent pathway also depends on protein kinase C (PKC) ξ and Akt/PKB, and controls 70-80% of F-actin. PKCξ is involved in pseudopodia formation and oxidative burst [12,15]. In fMLP stimulated PMNL, translocation of PKCξ to the plasma membrane started at 1 min, but peripheral actin polymerization was observed by 30 seconds [4]. This difference in time kinetics suggests that PKCξ may not have direct role in spatial distribution of F-actin in PMNL. The PI3K-independent pathway depends on rhoGTPases, ROCK, src kinases and NADPH, and is modulated by cAMP [16]. Ras and its associated rhoGTPases - rac, rho and cdc42, play an important role in the spatial and temporal organization of actin, and regulate cell adhesion and motility. Cdc42 is required for cell polarity and rac1 for protrusive activity [17]. Basal rhoA activity is necessary to maintain cell adhesion. This pro-adherent effect of rhoA probably counterbalances the effects of rac1 and Cdc42 as rac1 must inhibit rhoA to exert its activity toward myosin [18]. Since rhoGTPases cross-activate each other, balanced control of this activation determines outcomes like cell polarization, directional motility and substrate adhesion. As mentioned earlier, CML PMNL showed defects in cell polarization, adhesion, motility, pinocytosis, etc., and it was suggested that defective actin polymerization might have contributed to these defects [19]. In CML, defective actin polymerization may result in early egress of PMNL and immature myeloid cells from the bone marrow. Therefore, to understand defective actin polymerization in CML PMNL further, expression of GTPases - rac1, and rhoA, was examined.

Different rac1 isoforms are stimulated in normal and CML PMNL

In the Western blots 50% of normal and 59% CML samples showed a single band of rac1 at 21 kd, at all the time points studied (Figure ​(Figure2a).2a). But the remaining samples showed two bands, at 21 kd and 25 kd. The 25 kd band could be of the rac1b protein, as the molecular weight of recombinant rac1b isolated from E. coli is reported to be higher than 21 kd [20]. Therefore, for densitometry analysis, both the bands were considered collectively.

In Western blot studies, about 60% of normal and CML samples showed increase in rac1 levels at early time points of fMLP stimulation. Increase in rac1 expression was followed by a drop at later time points of fMLP stimulation. In normal, this increase was not significant., but in CML PMNL, increase at 5, 10 and 30 min of fMLP stimulation was statistically significant (Table ​(Table1a).1a). Rac1 levels were comparable in normal and CML PMNL, under unstimulated and stimulated conditions. But the major responder bands in normal and CML were 25 kd and 21 kd, respectively. The 25 kd band could be of rac1b. Rac1b acts like a fast cycling GTPase [21] and induces formation of lamellipodia in NIH3T3 [22]. Similarly in normal PMNL too, rac1b could be responsible for actin polymerization in lamellipodia. Though unstimulated CML PMNL showed higher levels of total rac1, and these levels increased further in response to stimulation, lower response of rac1b might have resulted in the absence of lamellipodia in CML PMNL leading to the absence of chemotaxis.

In FCM studies, only 50% of the normal samples showed increase in rac1 (range 1.3-2.3) at early time point, specifically at 0.5 min of fMLP stimulation and then showed a second increase (Figure ​(Figure2d).2d). Of the remaining, 30% samples showed a drop and 20% samples showed delayed increase in rac1 levels. Hence, the increase in the average median channel for rac1 on stimulation was statistically insignificant (Table ​(Table1b).1b). In CML PMNL, majority of the samples showed a drop (range 10- 60%) in rac1 levels on stimulation at early time points of stimulation followed by a partial recovery. At later time points, only 21% of the samples showed a true increase in rac1 levels (Figure ​(Figure2d).2d). Thus, CML PMNL showed a significant drop in rac1 levels after 0.5, 5, 10 and 45 min of stimulation (Table ​(Table1b).1b). FCM studies showed higher expression of rac1 in unstimulated CML PMNL than that in normal PMNL. On fMLP stimulation, the rac1 levels increased in normal PMNL and dropped in CML PMNL. Hence, significant difference between rac1 levels of both was seen at 45 min of fMLP stimulation (Table ​(Table11b).

The differences in results, between Western blot and FCM, could be because the major responder band in the two populations was different and the antibodies used have different affinities towards these bands. Secondly, depending on the localization of the 21 kd and 25 kd rac1 proteins in the cell, in FCM, the antibody could have had altered accessibility.

fMLP stimulated transport of rac1 to the cell membrane is negligible in CML PMNL

In unstimulated normal PMNL, rac1 expression was less in cytoplasm and more on the membrane. On stimulation, the fluorescence intensity increased, but the distribution pattern of rac1 remained same. In the majority of unstimulated CML samples rac1 was distributed everywhere. On fMLP treatment, rac1 distribution remained unaltered. In both, changes in rac1 levels seen by LCM matched with that observed by FCM, and were independent of morphological changes (Additional file 3: Figure S3 & Additional file 4: Figure S4).

The major difference in the distribution of rac1 between normal and CML PMNL was that normal PMNL showed a higher concentration of rac1 on the membrane, suggesting that CML PMNL could be defective in translocation of rac1 to the cell membrane. Alternatively, in view of the high binding of LCM antibody to the 25 kd band, the major portion of peripheral rac1 could be 25 kd suggesting higher expression of post-translationally modified rac. If this were true, then access to 21 kd would be lowered. This could lead to weak fluorescence in stimulated CML PMNL. However, unaltered rac1 distribution on stimulation indicated an absence of significant changes in rac1 localization.

fMLP stimulated degradation of rhoA is slower in CML PMNL

In the Western blots about 50% normal and 60% CML samples showed a drop in rhoA levels at early time points of fMLP stimulation, resulting in a 20% drop in average rhoA levels. In normal, the drop gradually increased to a significant level (Figure ​(Figure2a)2a) (Table ​(Table1a).1a). But in CML the decrease was statistically significant only at 45 min of stimulation (Table ​(Table11a).

Higher rhoA expression in unstimulated CML PMNL, as compared to that in normal PMNL, was not statistically significant. But on stimulation, differences between the rhoA levels of the two populations increased gradually, resulting in significantally lower (2-3 folds) levels in normal at later time points.

In FCM analysis of normal PMNL, at early time points of stimulation mixed response was seen. Later, a majority of the samples showed decrease in rhoA (Figure ​(Figure2e,2e, Table ​Table1b).1b). In CML PMNL, about 50% samples showed a drop in rhoA at early time points, but eventually showed an increase. As a result, at the later time points, rhoA levels in stimulated CML PMNL remained at par with the basal level (Table ​(Table11b).

A comparison between normal and CML PMNL showed that unstimulated normal PMNL as well as those during early stimulation have higher rhoA expression. But, at 45 min of stimulation, the picture reversed. Though the trend seen for rhoA expression was similar by Western blotting and FCM, the latter did not yield significant differences.

Intracellular distribution of rhoA is comparable in normal and CML PMNL

In the majority of samples, unstimulated normal and CML PMNL showed cytoplasmic rhoA. In 20% normal and 40%CML samples, unstimulated PMNL showed rhoA in the peripheral region below the F-actin layer. In both, rhoA distribution remained unaltered on fMLP treatment. Changes in rhoA levels were similar to those seen using FCM, and were not associated with morphological changes (Figure ​(Figure3a3a).

Co-localization of F-actin with rhoA

In unstimulated and stimulated normal and CML PMNL, peripherally concentrated F-actin did not co-localize with rhoA, while most of the diffused cytoplasmic actin co-localized with rhoA (Figure ​(Figure3a).3a). This was reflected in the lack of statistically significant differences in the co-localization coefficient of unstimulated (0.63 ± 0.04) and fMLP-stimulated PMNL (0.63 ± 0.02). CML PMNL showed lower co-localization coefficients as compared to the normal (0.57 ± 0.02). Moreover, co-localization coefficients were more scattered in stimulated CML PMNL (range 0.13 - 0.86) than that in normal PMNL (range 0.29 - 0.86). Less than one values of average co-localization coefficients in normal and CML PMNL further supported the observation of lack of colocalization of major part of F-actin with rhoA. In contrast to this, in normal and CML PMNL, all rhoA was co-localized with F-actin.

Some variation was seen within the unstimulated normal population with respect to co-localization of F-actin with rhoA. To group the majority of normal samples as a tight population and to segregate samples that behaved differently from the rest, a cut off percentage (20 percentile) was applied. All the samples above the cut off were considered as "normal" and all the samples below the cut off were categorized as "non-normal". The percentage of samples behaving as non-normal was similar under unstimulated and stimulated conditions (Table ​(Table22).

To segregate CML samples from the normal samples, the same cut off was applied to the CML PMNL. In CML, under unstimulated conditions, 32% of the samples behaved as non-normal. On stimulation, the percentage of non-normal samples increased to 45% and to 55% at 0.5 min and 30 min of fMLP stimulation, respectively (Table ​(Table2).2). Thus, 0.5 and 30 min of fMLP stimulation appeared to be critical to differentiate between normal and CML PMNL.

Clinical samples and cell lines

After taking written consent, peripheral blood was collected from healthy volunteers (control) and CML patients in chronic phase; before commencement of therapy and processed simultaneously. PMNL were isolated on a ficoll-hypaque gradient [54] and immediately used for the experiments.

Bcr-abl expressing cell lines K562 and BaF3/bcr-abl/T315I were used along with bcr-abl negative cell line HL-60, as a control. The cell lines were maintained in RPMI 1640 containing 1X AB-AM and 10% (v/v) fetal bovine serum, at 37°C in a humidified atmosphere containing 5% CO₂.

PMNL stimulation

PMNL were stimulated at 37°C with 10^-8M fMLP for various time durations. For Western blotting, cells were quickly pelleted at 4°C and lysed in Laemmli's sample buffer containing CompleteTM protease inhibitor [Roche, Mannheim, Germany]. For flow cytometric and microscopic analysis, cells were fixed with paraformaldehyde. Unstimulated PMNL were used as control.

Western blotting

Lysates form one million cells were loaded on 16% SDS-PAGE. After confirming equal protein loading by staining the blots, actin and GTPases were detected. Protein band density was quantitated as mean area density, using the software-Labwork.

Immunofluorescent staining

Fixed PMNL were permeabilized with Triton-X100 and stained using FITC-anti-ras or TRITC-anti-rac or anti-rho followed by Alexa 488-GAM. For F-actin staining, TRITC-phalloidin was used [4]. The cells were analysed by FCM using FACScalibur (Becton-Dickinson) and LCM MRC1024 (Biorad) using Lasersharp software. A minimum of 10,000 cells were analysed by FCM. The median channel number was taken as a measure of fluorescence intensity. To precisely study localization of the molecules, Z-series images of 1 μm thickness were analysed by LCM.

Growth inhibition assay

Cell lines were treated with inhibitor of rhoA GTPase activation - C3 exoenzyme (10 μg/ml), ROCK I inhibitor - Y-27632 (10 μM) and ASODN targeted against rhoA (1 μM). Cell proliferation was monitored at 24 and 48 hrs., by using CCK-8 assay [55]. Effect of treatment on cell growth was expressed as percent growth inhibition.