Counting Cytokinesis Proteins Globally and Locally in Fission Yeast

See allHide authors and affiliations

Science  14 Oct 2005:
Vol. 310, Issue 5746, pp. 310-314
DOI: 10.1126/science.1113230


We used fluorescence microscopy to measure global and local concentrations of 28 cytoskeletal and signaling proteins fused to yellow fluorescent protein (YFP) in the fission yeast Schizosaccharomyces pombe. Native promoters controlled the expression of these functional YFP fusion proteins. Fluorescence measured by microscopy or flow cytometry was directly proportional to protein concentration measured by quantitative immunoblotting. Global cytoplasmic concentrations ranged from 0.04 (formin Cdc12p) to 63 micromolar (actin). Proteins concentrated up to 100 times in contractile rings and 7500 times in spindle pole bodies at certain times in the cell cycle. This approach can be used to measure the global and local concentrations of any fusion protein.

To understand any complicated cellular process, one must know the global and local concentrations of the participating proteins. Concentrations reveal stoichiometries and are prerequisites for mathematical modeling of biological systems. Unfortunately, concentration measurements are rarely available. In the case of cytokinesis, only one estimate of the concentration of myosin-II in the cleavage furrow is available (1). Antibodies can be used to estimate the cellular abundance of a few native proteins (2) or thousands of proteins carrying molecular tags (3) but not their local concentrations. Light microscopy of cells expressing proteins with a fluorescent tag is an attractive approach, but previous calibration methods depended on assumptions that are difficult to verify, such as the assumption that the quantum yield and detection efficiency are the same inside and outside cells (46). Even with careful microscopy, absolute concentration may be unknown, because a fluorescent fusion protein was expressed from a plasmid along with endogenous native protein. Tests to verify the functionality of the fusion protein are often lacking.

Here we show that the fluorescence of functional fluorescent fusion proteins expressed from native promoters is directly proportional to the number of molecules in live cells. This calibration allowed us to measure the global and local concentrations of 28 different cytoskeletal and signaling proteins as the fission yeast Schizosaccharomyces pombe goes through the cell cycle. This approach is applicable to a medium-scale analysis of proteins participating in any cellular process, even without genomic integration of the fusion protein.

We integrated the coding sequence for monomeric yellow fluorescent protein (mYFP) or YFP into the genome of S. pombe at the N or C termini of 27 different proteins, so the fusion proteins were expressed from their endogenous promoters (7, 8). We varied the position of mYFP and the size of linker to find functional constructs. We ruled out partial loss of function by observing normal colony formation and cellular morphology under both standard and stressful conditions and by observing the lack of synthetic phenotypes when crossed to strains with other mutations (9). The abilities of these tagged constructs to function normally suggest that their expression levels are not substantially different from those of the native proteins. We confirmed this by immunoblotting Arp3p and by microscopy of Myo2p, IQGAP Rng2p, and pombe Cdc15 homology (PCH) protein Cdc15p (fig. S1) (9).

We combined two measurements to determine the fluorescence per fusion-protein molecule. First, we collected stacks of confocal sections through whole cells to measure the total and local fluorescence from each strain (Fig. 1A and fig. S2). For most tagged proteins, the mean fluorescence intensity was constant from cell to cell, with standard deviations from the mean intensity of only 8 to 37% (Table 1). Second, we measured the average number of mYFP fusion molecules per cell by quantitative immunoblotting (Fig. 1B). After background subtraction, the mean fluorescence per cell measured microscopically was directly proportional (R = 0.99) to the average number of molecules per cell for seven fusion proteins (Fig. 1C). Flow cytometry confirmed the linear relationship between fluorescence and concentration for all 27 proteins (Fig. 1C and fig. S3). The variance was smaller by microscopy, especially for global concentrations <0.5 μM. The total fluorescence of Fim1p–mYFP was the same whether concentrated in patches or dispersed by treating cells with Latrunculin A (fig. S4), so quenching was not an issue even when proteins were crowded together in the cell. Similarly, the fluorescence of formin Cdc12p and the Unc45/Crolp/She4p–related (UCS) protein Rng3p with triple YFP tags was three times that of single YFP tags. Because of photobleaching and low signal-to-noise ratios, the measured concentrations of low abundance proteins are less accurate than those of abundant proteins (fig. S3).

Fig. 1.

Measurement of protein concentrations by fluorescence microscopy, flow cytometry, and quantitative immunoblotting. (A and D) Differential-interference contrast and fluorescence micrographs of living wild-type (wt) cells and cells expressing integrated polo kinase Plo1p-mYFP (SPBs marked by arrows), myosin regulatory light chain Rlc1p-mYFP, and fimbrin Fim1p-mYFP at endogenous levels, or YFP-actin from a plasmid (strain JW1206). Fluorescence micrographs of 12 z-sections spaced at 0.6-μm intervals were projected into a two-dimensional image using (A) a sum intensity projection or [(D) and (A, inset)] maximum intensity projection. The inset in (A) shows the broad band marked by the arrowhead. Numbered cells in (D) expressed YFP-actin at low levels but were included in measurements. (B) Quantitation of septin Spn1p-mYFP by immunoblotting with antibody to YFP. In lanes 1 to 8, a standard curve was generated with 0 to 1.2 ng of purified 6His-mYFP mixed with 5 μl of wt cell extract. Lanes 9 to 12 have duplicate samples of 5.0 and 2.5 μl of cell extract, giving Spn1p-mYFP signal in the linear range of standard curve. A nonspecific band from cell extract provided a convenient loading control. (C) The correlation of average fluorescent molecules per cell from immunoblots (upper x axis) and cytoplasmic concentration (lower x axis) with cell-size–corrected integrated mYFP fluorescence intensity per cell from microscopy (mean ± 1 SD; solid circles and darker error bars; y = 0.0676x, R = 0.99) and fluorescence intensity from flow cytometry (mean ± 1 SD; open squares and lighter error bars; y = 0.0676x, R = 0.99) for strains expressing integrated mYFP fusion proteins. (E) Measurements of actin concentration by immunoblotting using antibodies to YFP (upper blot) and antibodies to actin (lower blot). Dilutions of cell extract of a strain with YFP-actin plasmid (JW1206) or wt JW729 grown in minimal medium for 36 hours and then in rich medium for 4 hours, and standards of purified 6His-mYFP (upper blot) and S. pombe actin (lower blot) with or without cell extract from wt cells were separated by SDS-polyacrylamide gel elecrophoresis. Scale bars, 5 μm.

Table 1.

Global cytoplasmic concentrations, mean molecules per cell, and local accumulation in actin patches, spindle pole bodies, or the division site for 28 proteins measured by fluorescence microscopy and immunoblotting.

Protein (number of cells analyzed for global concentration; local concentration; number of patches) Exposure time/slice (ms) Global cytoplasmic concentration (μM) Mean polypeptides per average cell with a volume of 92 μm3 Local accumulation %View inline [mean (maximum observed)] Local concentration [mean polypeptides (mean concentration, μM)]
Actin patch proteins
YFP-actin Act1pView inline (302; 2; 89) 69 0.78 ± 0.71 17,600 ± 16,000 13 (16) 145 ± 89 (29)
Actin Act1pView inline (118 × 106) Imm/blot 63.2 ± 10.5View inline (1.43 ± 0.24) × 106View inline >13 2,700 ± 1,700 (530)
Arp2 (Arp2p) (104; 6; 168) 69 2.88 ± 0.35 46,600 ± 5,700 10 (11) 212 ± 94 (42)
Arp3 (Arp3p) (86; 6; 158) 69 4.12 ± 0.45 66,700 ± 7,300 7 (8) 210 ± 87 (42)
ARPC1 (Arc1p/Sop2p) (85; 6; 199) 69 2.49 ± 0.29 40,300 ± 4,700 15 (17) 208 ± 79 (41)
ARPC3 (Arc3p/Arc21p) (85; 6; 165) 69 2.39 ± 0.22 38,700 ± 3,600 12 (14) 185 ± 73 (37)
ARPC5 (Arc5p/Arc16p) (94; 6; 165) 69 1.88 ± 0.14 30,500 ± 2,300 12 (13) 193 ± 76 (38)
Capping protein Acp2p (42; 2; 69) 99 1.19 ± 0.16 19,200 ± 2,600 17 (19) 90 ± 48 (18)
Fimbrin Fim1p (121; 4; 121) 69 5.34 ± 0.56 86,500 ± 9,100 15 (21) 507 ± 290 (100)
Spindle pole body proteins
SPB protein Sad1pView inline (58; 58) 198 0.15 ± 0.05 3,300 ± 1,100 31 (52) 450-1,030 (900-1,120)
Polo kinase Plo1pView inline (65; 38) 399 0.29 ± 0.06 6,600 ± 1,400 1 (6) 30-220 (33-440)
SIN kinase Cdc7p (103; 22) 399 0.24 ± 0.08 4,000 ± 1,300 5 (13) 0-440 (0-480)
Cytokinesis proteins
Mature contractile ring
Anillin-like Mid1pView inline (94; 23) 300 0.09 ± 0.02 2,100 ± 500 40 (68) 700 ± 200 (4)
Myosin-II Myo2p kanrView inline (53; 13) 300 0.45 ± 0.08 7,300 ± 1,400 27 (50) 2,900 ± 400 (20)
Myosin-II ELC Cdc4p (54; 15) 78 4.75 ± 0.67 77,000 ± 10,800 22 (31) 24,900 (165)
Myosin-II RLC Rlc1p (45; 10) 399 0.60 ± 0.09 9,600 ± 1,500 18 (28) 3,200 ± 600 (28)
IQGAP Rng2p kanrView inline (112; 17) 300 0.17 ± 0.04 2,700 ± 600 35 (62) 1,300 ± 100 (10)
mYFP-Cdc15p kansView inline (102; 16) 198 2.13 ± 0.33 35,600 ± 5,400 21 (34) 16,100 ± 2,300 (94)
Formin Cdc12pView inline (98; 9) 600 0.04 ± 0.01 600 ± 200 11 (26) 300 ± 50 (3)
Actin Act1pView inline (118 × 106) Imm/blot 63.2 ± 10.5View inline (1.43 ± 0.24) × 106View inline 4 ∼76,000 (460)
UCS protein Rng3pView inline (72; 12) 600 0.12 ± 0.03 1,900 ± 400 3 (8) 60 ± 20 (0.5)
Rng3p in myo2-E1View inline (42; 13) 198 0.32 ± 0.11 6,800 ± 2,400 30 (50) 4,200 ± 1,600 (28)
Alpha-actinin Ain1p (101; 10) 300 0.22 ± 0.03 3,600 ± 500 8 (12) 500 ± 100 (4)
Myosin-II Myp2p (89; 14) 399 0.38 ± 0.07 6,100 ± 1,100 21 (28) 2,000 (15)
Septin Spn1p (159; 24) 198 0.63 ± 0.10 10,300 ± 1,600 35 (50) 7,000 ± 800 (21)
Septin Spn4p (131; 28) 198 0.50 ± 0.07 8,100 ± 1,200 34 (50) 6,100 ± 1,200 (18)
Anillin-like Mid2p (116) 198 0.11 ± 0.19 1,800 ± 3,100 NA NA
Protein kinase C Pck2p (102; 19) 399 0.27 ± 0.04 4,300 ± 600 13 (24) 800 ± 100 (6)
Rho GEF Rgf1p (89; 9) 300 0.27 ± 0.05 4,300 ± 700 5 (8) 200 ± 30 (1)
Rho GEF Rgf3p (44; 6) 999 0.20 ± 0.08 3,200 ± 1,300 4 (13) 200 ± 40 (1)
Chitin synthase Chs2p (97; 9) 600 0.13 ± 0.07 2,100 ± 1,100 3 (8) 100 ± 30 (0.5)
  • View inline* Percent of total molecules localized to actin patches, SPB(s), or the cell-division site (excluding medial patches).

  • View inline Actin, Sad1p, Plo1p, and Mid1p were not excluded from the nucleus (4044). We assumed equal concentrations of these proteins in the cytoplasm and nucleus.

  • View inline The average of the two methods using S. pombe actin as standard as shown in fig. S5.

  • View inline§ Strain analyzed with (kanr) or without (kans), the kanMX6 selectable marker.

  • View inline Triple YFP tag gave three times the signal of single YFP and less variance in the measurements.

  • Knowing the fluorescence per molecule in live cells, we measured the global concentrations of 20 other proteins tagged with YFP or mYFP from the integrated fluorescence intensity of asynchronous cells (Table 1). Point counting stereology of electron micrographs established that cytoplasm is 29% of the total cellular volume. We used this number to calculate cytoplasmic concentrations from the total fluorescence.

    None of the several different YFP-actin constructs could replace the native actin gene, so we expressed N-terminally tagged YFP-actin at low levels from a plasmid, under the control of attenuated 81nmt1 promoter. Cells tolerated YFP-actin and incorporated it into actin patches with capping protein (10) but not into the contractile ring, likely because formin Cdc12p cannot handle YFP-actin (Fig. 1D). The fluorescence intensity of YFP-actin expressed from the plasmid varied widely from cell to cell, with the standard deviation (0.71 μM) equal to the mean 0.78 μM (Figs. 1D and 2A). Quantitative immunoblots gave cytoplasmic actin concentrations of 31.3 ± 1.3 μM in minimal medium (Fig. 1E) and 63.2 μM in rich medium (fig. S5). Cell size differed by less than 5% under these conditions. Immunoblots with YFP antibody showed that an average of 1.8% of the actin was tagged with YFP in cells grown on minimal medium (Fig. 1E). To estimate the number of actin molecules in individual cells, we determined the ratio of that cell's volume to the average volume of all the cells in the population.

    Fig. 2.

    Measurement of protein concentrations by fluorescence microscopy. (A) Cytoplasmic concentration of fimbrin (solid circles) and YFP-actin (solid triangles) as a function of cell volume. Mean concentrations and cell volumes ± 1 SD are shown. (B) Local accumulation of proteins in actin patches. Top: Fluorescence micrographs (maximum intensity projection) of interphase and septating cells expressing Fim1p-mYFP. Bottom: Box plots of molecules of YFP or mYFP fusion proteins per patch in 70 to 200 patches in 2 to 6 cells for each protein. Box plots display mean (marked with ×), median (line inside the box), and outliers (cicles). The box represents the middle 50% data and the bars represent the top and bottom 25% data. (C) Local accumulation of proteins in SPBs during cytokinesis. SPB separation defines time zero. The cell-cycle time was calculated from the separation of the SPBs and septum diameter. Top: Micrographs (sum intensity projection) of cells expressing mYFP (YFP for Cdc7p) fusion proteins. Bottom: Time course of the mean accumulation ± 1 SD of three fusion proteins in SPBs. Scale bars, 2.5 μm.

    Global cytoplasmic concentrations of cytokinesis proteins (Table 1) range from 0.04 μM formin Cdc12p (600 molecules per cell) to 63.2 μM actin (1.43 million molecules per cell). The concentration of myosin-II light chain Cdc4p (4.8 μM) is 10-fold higher than that of Myo2p (0.45 μM), which is in keeping with evidence for functions beyond its partnership with myosins (1114). Most of the proteins studied are more abundant in S. pombe than in haploid strains of S. cerevisiae (table S2), in part because of the 1.7-fold higher protein content per cell of S. pombe (15, 16).

    Most protein concentrations were constant with cell volume, an estimate of position in the cell cycle (Fig. 2A and fig. S6). This confirms with better resolution previous immunoblotting measurements on bulk samples of synchronized cells for formin Cdc12p (17), both types of myosin-II heavy chain (18), and polo kinase Plo1p (19). One exception is the anillin-like protein, Mid2p, which is only detected during septation and cell separation (fig. S6E), as reported (20). The concentrations of anillin-like protein Mid1p and PCH protein Cdc15p are constant across the cell cycle, in spite of fluctuations in mRNA levels (19, 2123), so the ratios of these mRNAs to their proteins vary over time.

    The biochemical reactions depend on local concentrations, so we used fluorescence and our standard curve (Fig. 1C) to measure local concentrations. Actin patches concentrate at cell poles and at the cell-division site where the cell wall grows (Fig. 2B). The molecular composition of each patch varies over its 20-s lifetime (24) as it forms, matures, and dissociates (24, 25), so measurements on patches with random ages vary considerably (Fig. 2B and Table 1; standard deviations 38 to 62% of means). Assuming that the ratio of YFP-actin to native actin is at least as high in patches as the cell as a whole and assuming that patches are spheres 250 nm in diameter (26), an average patch has at least 2700 actin molecules (minimal local concentration 530 μM). Patches also contain 202 Arp2/3 complexes (40 μM; with 1:1:1:1:1 stoichiometry of 5 subunits), 90 capping proteins Acp2p (18 μM), and a remarkable number, 507, of the actin filament crosslinker fimbrin Fim1p (100 μM). The fractions of each protein in patches range from 6 to 21% (Table 1). Assuming that each Arp2/3 complex nucleates an actin filament as a branch on an older filament, the filaments in patches consist of at least 13 subunits (≥30-nm-long) and half are capped. This minimum length is similar to that of filaments of 20 subunits in patches isolated from S. cerevisiae (27, 28).

    Spindle pole bodies (SPBs) have both stable and transient components (Table 1 and Fig. 2C) that organize microtubules for mitosis and regulate cytokinesis (29, 30). These plaque-like oblate ellipsoids are 0.18 μm in diameter and 0.09 μm thick (0.05 μm early in mitosis) (31). Throughout the cell cycle, about 30% of Sad1p (an SPB marker protein) concentrates up to 7500-fold in SPBs relative to the global cytoplasmic concentration. When mature SPBs separate early in mitosis, the ∼1000 molecules of Sad1p divide equally between the two daughter SPBs, which then recruit ∼5 molecules of Sad1p each minute as the cell cycle progresses (Fig. 2C). Early in mitosis, 6% of polo kinase Plo1p (220 molecules per SPB) concentrates transiently on both SPBs and, to a lesser extent, in the mitotic spindle to trigger the septation initiation network (SIN) (30). About 10 min later, the SIN pathway kinase Cdc7p starts to accumulate at ∼10 molecules per minute in the new SPB (32), peaks at ∼440 molecules late in anaphase, and then drops to undetectable levels (Fig. 2C). Thus, the signal from Plo1p is only amplified by a factor of two in terms of accumulation of Cdc7p, the first kinase downstream from the guanosine triphosphatase Spg1p in the SIN pathway, and most of both Plo1p and Cdc7p are free to function throughout the cytoplasm. Fluorescence recovery after photobleaching (FRAP) experiments also show dynamic association of other SIN pathway components, Spg1p and Sid2p kinase, with SPBs (33).

    Fission yeast begin cytokinesis when anillin-like Mid1p exits from the nucleus and initiates assembly of at least seven proteins in a broad band of 50 to 120 dots at the cleavage site (Fig. 3A) (34). Given ∼75 dots, each would have about 21 molecules of Mid1p, 43 myosin-II heavy chains Myo2p, 35 myosin regulatory light chains Rlc1p, a vast excess of 289 myosin light chains Cdc4p, 23 IQGAP Rng2p, 22 Cdc15p proteins, and only 2 Cdc12p formin dimers. Thus the stoichiometry of Mid1p to dimeric myosin-II molecules to IQGAP to Cdc15 is close to 1:1:1:1. Formin homodimers nucleate and remain attached to the barbed end of elongating actin filaments (35). Thus each dot could grow about 2 actin filaments anchored by Cdc12p, sufficient for myosin-II to pull the dots together into a continuous contractile ring early in mitosis. The numbers of both Cdc4p and Cdc15p in dots exceed their known binding partners included in this study [myosin-II heavy chain Myo2p and IQGAP Rng2p (13, 14), and Cdc12p (36)] by a factor of 10, consistent with evidence for other binding partners (myosins-V and -I) and functions.

    Fig. 3.

    Local accumulation and concentrations of mYFP fusion proteins at the cell-division site measured by fluorescence microscopy. SPB separation defines time zero. The cell-cycle time and stage (8) ofeach cell were estimated from the morphology and diameters of the contractile ring and septum. (A and C) Fluorescence micrographs (maximum intensity projection) of cells at various stages expressing mYFP fusions of myosin-II Myo2p (JW1110), PCH protein Cdc15p (JW1063), α-actinin Ain1p, and septin Spn1p. The brightness and contrast of images were linearly adjusted to show the contracted ring. Graphs plot the local accumulation [(A) and (C)] (mean ± 1 SD) and local concentrations (B and D) of molecules in contractile rings and entire broad band including spaces between the dots. The data for Cdc15p after 60 min show the protein in the septum region after ring contraction. In (A) to (C), the green symbols and lines use scale on the right. All others use scale on the left. Scale bar, 2.5 μm.

    Our measurements provide the first quantitative inventory of contractile ring proteins. Given a circumference of 10.3 μm and about 20 actin filaments in a cross section estimated from published electron micrographs (37), an uncontracted actin ring has ∼206 μm of filaments consisting of ∼76,000 subunits (given 370 actin subunits per micrometer of filament), which implies a local actin concentration of ∼460 μM. If each of the 150 Cdc12p dimers in a mature contractile ring nucleates an actin filament, an average filament is ∼1.4 μm long. A mature contractile ring includes ∼2900 molecules of Myo2p (Fig. 3, A and B), enough for 14 myosin heavy chains (∼one myosin-II minifilament) per micrometer of actin filament. When activated by the UCS protein Rng3p, myosin-II moves actin filaments at 0.5 μm/s (14). If the actin filaments are about 1 μm long, a minimum of 3 or 4 sarcomere-like contractile units would encircle the equator and constrict the circumference of an unloaded contractile ring at ∼3 to 4 μm/s, which is ∼103 times faster than observed. More likely, the number of contractile units in series is >4, increasing the maximum rate of contraction. If all the myosin-II heads in such a contractile unit were active, they would produce ∼1500 pN of force. Thus, either the load is considerable or few myosin heads are active. The ratio of one Rng3p per 50 myosin-II heads in the contractile ring (Table 1 and fig. S7) supports the latter hypothesis. In the temperature-sensitive myo2-E1 mutant, the concentration of Rng3p is higher by threefold globally and ∼60-fold in the contractile ring, perhaps to compensate for the minimal motor activity of this mutant (14). Large pools of most contractile-ring proteins remain in the cytoplasm during cytokinesis (Table 1) available for exchange with assembled rings, as observed in FRAP experiments (38, 39).

    The number of conventional myosin-II (Myo2p, Cdc4p, and Rlc1p) molecules in contractile rings is roughly constant from the time that a ring condenses through constriction (Fig. 3), so the local myosin concentration and force-producing capacity increase dramatically as the diameter of the ring declines. Cdc15p not only participates in ring assembly, but its number of molecules increases ∼10-fold as rings mature during anaphase in preparations for septum formation. As rings constrict, the local concentrations of actin-binding proteins IQGAP Rng2p, Cdc15p, formin Cdc12p, and alpha-actinin Ain1p are nearly constant, so all are lost in proportion to the decline in the volume of ring. Septins Spn1p and Spn4p arrive late at the division site, forming rings with 15 to 22 μM septin that do not constrict as the septum forms (Fig. 3, C and D).

    If used with caution, microscopy with fluorescent fusion proteins provides a precision measuring tool for quantitative biology. Our calibration method can be used to determine the concentration of any protein in yeast or other organisms with homologous recombination. Episomal expression of a YFP-fusion protein with measurement of the ratio of tagged and untagged proteins can be used where homologous gene replacement is impossible.

    Supporting Online Material

    Materials and Methods

    Figs. S1 to S7

    Tables S1 and S2

    References and Notes

    References and Notes

    View Abstract

    Stay Connected to Science

    Navigate This Article