Reliable Detection of Dead Microbial Cells by Using Fluorescent Hydrazides

We have developed a new method for accurate quantification of dead microbial cells. This technique employs the simultaneous use of fluorescent hydrazides and nucleic acid dyes. Fluorescent hydrazides allow detection of cells that cannot be detected with currently used high-affinity nucleic acid dyes. This is particularly important for nongrowing bacterial populations and for multicellular communities containing physiologically heterogeneous cell populations, such as colonies and biofilms.Many different approaches are used to assess the viability of bacterial cells. Currently, the most-used method is determination of the number of CFU, which reflects the ability of cells to reproduce. However, this method is not reliable because a fraction of live bacterial cells cannot divide under standard growth conditions (2) and because some bacteria are killed due to oxidative stress that occurs upon plating (3). Another approach detects dead cells through staining with fluorescent dyes that have high affinities for nucleic acids (RNA and DNA). Examples of this class of dye include Sytox Green (SG) (7). Cells incorporating these stains are then identified using flow cytometry or fluorescence microscopy. Discrimination between live and dead cells is based on the selective entry of dyes such as SG into dead cells, whose membrane integrity is compromised. However, the accessibility of DNA to SG staining also depends on the bacterial cell cycle stage. Cells that die during stationary phase are poorly stained by SG, probably because the DNA topology is altered (7). In addition, it is not possible to detect dead cells whose nucleic acids are degraded.In order to detect dead cells that escape detection by the methods described above, we used an Alexa Fluor hydrazide (AFH) dye. AFH dyes are low-molecular-weight, bright, photostable fluorescent molecules that are generally used as cell tracers by microinjection into eukaryotic cells. Because hydrazine components interact with carbonyl groups (aldehydes and ketones), AFH can be used for the detection of carbonylated proteins (1, 10). The quantity of carbonylated proteins, which are irreversibly damaged, increases after various lethal stresses such as oxidative stress, heat shock, and acidic stress (4). In stationary phase, bacteria also accumulate carbonylated proteins (5). As AFH cannot pass freely across the functional membranes of living cells, we hypothesized that it can be used for the detection of dead cells by tagging carbonylated proteins even when cells are devoid of nucleic acids. For the development of this method, we used Escherichia coli as a model organism.We first compared the results of staining E. coli cells killed by heat treatment with SG and Alexa Fluor 633 hydrazide (AF633H; both purchased from Invitrogen, Carlsbad, CA). Cells from 1- and 15-day-old liquid cultures were killed by incubation at 95°C for 10 min, which reduced the CFU counts in the cultures from approximately 2 × 10⁸/ml to undetectable levels (below 10 CFU/ml). Aliquots of cells taken before and after heat treatment were stained with the two dyes and analyzed with a FACSAria cell sorter and flow cytometer (Becton Dickinson Biosciences, San Jose, CA) (Fig. ​(Fig.1).1). Among cells from the 1-day-old cultures, less than 1% collected before treatment and 99.5% collected after treatment were stained with the two dyes. The 15-day cultures contained large proportions of dead cells even before treatment, as indicated by a decrease in the CFU count from approximately 2 × 10⁹/ml on day 1 to 1 × 10⁸/ml on day 15 (a reduction of 95%). In 15-day cultures, 26 and 89% of cells before heat treatment were stained with SG and AF633H dyes, respectively. After heat treatment, CFU levels were again undetectable, and 99.9% of cells from 15-day cultures were stained with AF633H, while only 36% were stained with SG. This difference in staining of dead cells by SG and AF633H indicates that the ability of dead cells to be stained by SG decreases dramatically with cell age but that the staining of dead cells by AF633H increases with cell age. This may be due at least partly to the fact that AF633H can stain cells devoid of nucleic acids while SG cannot. Reconstruction experiments in which heat-treated and untreated cells from 1- and 15-day-old cultures were mixed in fixed ratios and stained by the two dyes gave expected values (Table ​(Table1),1), showing that AF633H can be used for precise quantification of dead cells. In addition, we observed that AF633H staining remained stable after cell fixation with paraformaldehyde (see Fig. S1 in the supplemental material).Open in a separate windowFIG. 1.Levels of fluorescence of heat-killed microbial cells stained by AF633H and SG are indicated by fluorescence intensity histograms for untreated microbial cells (white histograms) and heat-treated cells (brown histograms). Cells were stained with AF633H (left panels) and SG (right panels). E. coli (MG1665) cells were from 1- and 15-day-old cultures. D. radiodurans, B. subtilis, and S. cerevisiae cells were from 1-day-old liquid cultures. The presence of two peaks of fluorescence for cells stained with SG probably indicates the existence of cells with different numbers of copies of the chromosome.

TABLE 1.

Detection of dead E. coli cells in liquid cultures by AF633H and SG staining^a

% of heat-treated cells	% of SG-stained cells in:		% of AF633H-stained cells in:		% of expected dead cells in:
	1-day culture	15-day culture	1-day culture	15-day culture	1-day culture	15-day culture
0	0.6	26.3	0.4	84.7	0.7	88.5
25	27.45	31.6	26.6	87.1	25	91.4
50	48.3	29.6	48.1	92	50	94.2
100	98.75	36.1	99.7	99.9	100	100
Chi-square test result	0.95	3.4 × 10−36	0.96	0.94

Open in a separate window^aHeat-treated cells were mixed with untreated cells at different ratios, and the cell mixtures were simultaneously stained with AF633H and SG in order to estimate which of the two dyes is more reliable for the detection of dead cells. The number of stained cells was compared to the number of dead cells expected in each different mixed population. The probability that the number of stained cells corresponded to the known number of dead cells was estimated using a chi-square test. The expected number of dead cells in each population of untreated bacteria was defined as the total number of stained cells, i.e., the sum of the cells stained by SG, the cells stained by AF633H, and the cells stained by both dyes. In a 1-day culture, these groups were 0.3, 0.1, and 0.3% of the population, respectively; in a 15-day culture, they were 4, 62, and 22.5% of the population, respectively.We then examined mid-exponential-phase cultures treated with hydrogen peroxide (H₂O₂) (see the supplemental material). Cells were exposed to 20 mM H₂O₂ for 15 min in Luria-Bertani (LB) medium. The CFU count showed that 75% of cells were not capable of growing after treatment. Using flow cytometry, we observed that 45 min after the end of treatment, 77% of cells were stained by AF633H (Fig. 2C and D), which correlates very well with the CFU count. The same experiment was performed using an E. coli strain with the deletion of the gene coding for the RpoS sigma factor, which regulates a set of genes endowing the cells with resistance to various stresses, including oxidative stress (9). When rpoS-deficient cells were treated with H₂O₂, 100% were killed and stained by AF633H (data not shown).Open in a separate windowFIG. 2.Detection of oxidative-stress-killed E. coli cells by CLSM and flow cytometry. To visualize AF633H staining at the single-cell level and to evaluate the timing of carbonylation processes in cells after oxidative stress, aliquots of cells either exposed to 20 mM H₂O₂ or untreated were inoculated onto microscope cavity slides containing a solid matrix of M9 minimal medium-agarose with AF633H. The staining of cells with AF633H was monitored by CLSM using a Zeiss LSM 510 microscope for 1 h. (A) Untreated cells; (B) H₂O₂-treated cells; (C and D) flow cytometry analyses of AF633H-stained untreated cells and H₂O₂-treated cells, respectively. t, time.In order to trace the kinetics of protein carbonylation, the timing of AF633H staining of individual cells treated or not treated with H₂O₂ was monitored by confocal laser scanning microscopy (CLSM) (Fig. 2A and B; see also Movie S1 in the supplemental material). In the population of untreated bacteria, 1.6% of cells were stained after 1 h of incubation with AF633H. This indicates that AF633H very poorly stains healthy, undamaged cells. Between 1 and 45 min after treatment with H₂O₂ (with a peak after 30 min), individual cells became fluorescent. Forty-five minutes after treatment, 83% of cells were stained, and this percentage did not increase with further incubation. This result was in quantitative agreement with the CFU count, which showed that H₂O₂ treatment rendered 82% of cells incapable of growth.After determining that AFH accurately detects cells killed by different stress treatments, we wanted to detect cells that die without exogenous toxic compounds under standard laboratory growth conditions. Using the cell sorter, we isolated two populations from an overnight liquid culture of E. coli, those that were stained by AF633H and those that were not. We sorted single cells into wells of a 96-well plate that contained the growth medium and checked the capacities of the cells to grow. Less than 5% (standard deviation SD], 3%) of AF633H-stained cells were capable of growing, while 96% (SD, 2.6%) of cells not stained by AF633H were capable of growing (Fig. ​(Fig.3A3A).Open in a separate windowFIG. 3.Analyses of dead cells from liquid cultures and aging colonies. (A) The live/dead status of cells stained with AF633H was investigated by cell sorting. Cells from an overnight culture were sorted into a 96-well plate filled with LB medium. AF633H-stained cells were sorted into the 48 wells on the right side of the microplate, while cells that were not stained with AF633H were sorted into the 48 wells on the left side of the microplate. The abilities of the sorted cells to reproduce were assessed by incubating the plate at 37°C and adding IPTG (isopropyl-β-d-thiogalactopyranoside) and X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) to reveal β-galactosidase activity and facilitate visualization of the cells. (B) Percentages of cells in E. coli colonies stained by AF633H alone, by SG alone, and by both AF633H and SG. (C) Results from dual staining of the cells from a 4-day colony with the membrane potential dye DiOC₂(3) and AF633H. (D to F) Three-dimensional views (obtained by CLSM) of a 7-day colony which was stained with SG (D) and stained with AF633H (E) and a merged image in which cells stained with AF633H and SG appear pink (F).In nature, bacteria are found mostly in structured communities, such as colonies and biofilms (6). Structured bacterial communities cause many problems for industry and for public health because they are highly resistant to various types of physical, chemical, and biotic stresses, including the host immune system. Dead cells in structured communities play an important role in the development and spreading of the communities (11, 13). Therefore, we tested the validity of our method for the quantification of dead cells in aging E. coli colonies. Cells from 1- and 11-day-old colonies were stained with SG and AF633H and analyzed by flow cytometry (Fig. ​(Fig.3B;3B; see also Fig. S2 in the supplemental material). Three populations of stained cells were observed: (i) cells stained only by AF633H, (ii) cells stained by both dyes, and (iii) cells stained only by SG. Over time, the percentages of cells in these three categories changed significantly. In 2-day-old colonies, 3% of all cells were stained, of which 10, 16, and 90% belonged to populations i, ii, and iii, respectively. In 11-day-old colonies, 75% of cells were stained. Among them, 55, 32, and 11% belonged to populations i, ii, and iii, respectively. These variously stained populations of cells may represent cells that die in different metabolic states and/or different phases of cell deterioration after death. Therefore, we conclude that without the use of AF633H, the majority of dead cells in old colonies would pass undetected.We also determined the numbers of cells within colonies which had active metabolism. We used the cyanine dye DiOC₂(3) (3,3′-diethyloxacarbocyanine iodide) (12), which allows the estimation of membrane potential. In 1-day-old colonies, 93% of cells were metabolically active, while 40% of cells in 4-day-old colonies and 7% of cells in 7-day-old colonies were metabolically active (see Fig. S3 in the supplemental material). As these populations represented only fractions of the cells capable of growing according to the CFU count, we concluded that metabolically active cells coexist with the dormant cells, i.e., those that are alive but metabolically inactive. Simultaneous staining with DIOC₂(3) and AF633H showed that cells stained by AF633H did not accumulate DIOC₂(3) (Fig. ​(Fig.3C).3C). Therefore, we confirm that without the use of AF633H, the majority of dead cells in old colonies would pass undetected.We showed that SG and AF633H freely diffuse inside a bacterial colony (see the supplemental material). This allowed us to monitor death patterns by using CLSM without disruptions of the aging colony. For example, Fig. 3D to F show the spatial distribution of the dead cells in a 7-day-old colony.These results demonstrate that this method can be used for Gram-negative bacteria. To determine if the method is generally relevant to microorganisms, we assayed 1-day-old cultures of the Gram-positive bacteria Bacillus subtilis and Deinococcus radiodurans and the yeast Saccharomyces cerevisiae in which the cells were either heat inactivated or untreated. In all cases, AF633H staining clearly distinguished between live and dead populations (Fig. 1E to J), indicating that this technique can be broadly useful in microbiology. Considerable fluorescence from live D. radiodurans and B. subtilis cells stained with SG was observed. This background fluorescence is due probably to the binding of this dye to the cell surface, as reported previously by Lebaron et al. (8). The level of background fluorescence was much lower with AF633H. The shift in fluorescence intensity between live and dead cells was 2 log with AF633H, whereas it was only 1 log with SG.In conclusion, we developed a method consisting of the simultaneous use of fluorescent hydrazide and a high-affinity nucleic acid dye that allows precise quantification of dead microbial cells. Fluorescent hydrazides allow detection of cells that cannot be detected with currently used nucleic acid dyes, which may represent a large part of the biomass.