Fuhrman and Azam, 1982; Bell, 1993), subject to ... Newell, 1986; Bååth, 1990) protists (Pedrós-Alió et ..... easily altered during the manipulation of the sam-.
SCI. MAR., 61(2): 111-122
SCIENTIA MARINA
1997
Limitations of 3H-thymidine incorporation in measuring bacterial production in marine systems* MARIAN UNANUE and JUAN IRIBERRI Departamento de Inmunología, Microbiología y Parasitología, Facultad de Ciencias, Universidad del País Vasco, Apdo. 644 E-48080 Bilbao, Spain.
SUMMARY: Bacterial production estimates are essential in order to assess the role of bacterioplankton in the flux of carbon in the ocean. Several methods have been developed to quantify bacterial production in marine systems, but the estimation of DNA synthesis rate from measurements of the rate of tritiated thymidine incorporation has been considered the most promising since its publication in 1980. Due to its high sensitivity and apparent simplicity it has been extensively used and modified. However, our current knowledge of thymidine bacterial metabolism is still insufficient and moreover, there are some methodological issues that need to be reviewed and clarified. The accurate application of the method in natural waters requires several precautions like the extraction and purification of the DNA, the measurement of isotope dilution by extracellular and/or intracellular non-labelled thymidine, and the estimation of a conversion factor to transform thymidine incorporation into bacterial production. All these methodological requirements make the thymidine method not appropriate for routine measurements. Its application to field studies without considering these prerequisites leads to erroneous estimations of bacterial production. Implications in our understanding of the microbial ecology of marine systems are discussed. Key words: Bacteria, production, thymidine, sea. RESUMEN: LIMITACIONES DEL MÉTODO DE LA INCORPORACIÓN DE 3H-TIMIDINA EN LA MEDIDA DE PRODUCCIÓN BACTERIANA EN SISTEMAS MARINOS. – La estimación de la producción bacteriana es esencial para determinar la función del bacterioplankton en el flujo de carbono en el océano. Se han desarrollado varios métodos para cuantificar la producción bacteriana en los sistemas marinos, pero la estimación de la tasa de síntesis de ADN a partir de la medida de la tasa de incorporación de timidina tritiada ha sido considerado el más prometedor desde su publicacion en 1980. Debido a su alta sensibilidad y aparente simplicidad ha sido ampliamente utilizado y modificado. Sin embargo, nuestro conocimieno actual sobre el metabolismo bacteriano de la timidina es todavía insuficiente y además hay ciertos aspectos metodológicos que necesitan ser revisados y aclarados. La correcta aplicación del método requiere algunas precauciones como la extracción y purificación del ADN, la estimación de la dilución del isótopo por timidina no marcada de origen extracelular y/o intracelular y la determinación de un factor de conversión para transformar la incorporación de timidina en producción bacteriana. Todos estos requisitos metodológicos hacen que el método de la incorporación de timidina no sea apropiado para medidas rutinarias. Su aplicación a estudios de campo sin considerar estos requisitos previos conduce a estimaciones erróneas de la producción bacteriana, cuyas implicaciones en nuestra comprensión de la ecología microbiana de los sistemas marinos serán discutidas. Palabras clave: Bacteria, producción timidina, mar.
*Received March 12, 1996. Accepted October 15, 1996. BACTERIAL PRODUCTION FROM THYMIDINE INCORPORATION 111
INTRODUCTION The significance of bacterioplankton in the flux of carbon and energy in pelagic marine ecosystems has been emphasized during the last two decades due to the emergence of new methodologies. A great advance was the development of sensitive methods to estimate bacterial production and growth rates. The methods most widely used rely on the uptake of radiolabelled precursors of nucleic acids and proteins: the incorporation of [3H]adenine (Karl, 1982; Karl, 1993) that does not appear to be specific to bacteria and therefore seems to give an estimate of total microbial production; the incorporation of [3H]leucine into proteins (Kirchman et al., 1985; Kirchman et al., 1986; Kirchman, 1993) that appears to be sensitive and reliable and has been used frequently in recent years; and the incorporation of [3H]thymidine into bacterial DNA (Fuhrman and Azam, 1980; Fuhrman and Azam, 1982; Bell, 1993), subject to criticism and controversy since 1980 (Moriarty, 1986; Bell, 1990; Robarts and Zohary, 1993). All these methods have advantages and disadvantages that have been deeply discussed in several reviews (Moriarty, 1986; Riemann and Bell, 1990; Ducklow and Carlson, 1992). However, most data about bacterial production in pelagic marine ecosystems are based on the use of the [3H]thymidine method (Ducklow and Carlson, 1992), in spite of the fact that this methodology relies on several assumptions still considered controversial. Bacterial growth leads to cell division and requires the synthesis of new cellular material, carbohydrates, lipids, proteins, RNA and DNA. The replication of the bacterial chromosome must occur before cell division, and consequently the rate of synthesis of DNA can be used as an index of bacterial growth rate. Thymidine is one of the four deoxyribonucleosides participating in the DNA synthesis, and hence the incorporation of [methyl3 H] thymidine into DNA can potentially give a measurement of bacterial production. These are the theoretical principles, but the application of the method to natural bacterial communities has captured the attention of microbial ecologists during the last fifteen years. Apparently the method to measure bacterial production from thymidine incorporation is extremely simple: samples have to be incubated with [methyl-3H] thymidine at nM concentrations during a short period of time (usually less than an hour), followed by the extraction of DNA in ice112 M. UNANUE and J. IRIBERRI
cold 5% TCA (trichloroacetic acid). The extracted material is collected on membrane filters and radioassayed by liquid scintillation counting. Finally, the amount of thymidine incorporated is converted to cell numbers and biomass by using conversion factors. Its apparent simplicity has seduced microbial ecologists. During the last fifteen years, the method has been modified and used in a broad range of habitats by applying different protocols, but usually, many of the assumptions inherent to the method have not been properly addressed. On the other hand, attempts to establish a universal protocol have not been successful, although we have to point out the valuable contributions of Bell (1993) as well as Kirchman and Ducklow (1993) for the Handbook of Methods in Microbial Ecology. At present, a deeper knowledge of the method and its pitfalls, as well as the study of thymidine metabolism by bacteria have led researchers in microbial ecology to question its validity. As a consequence it is recommended that several complementary methods to measure bacterial production are used; i.e. thymidine incorporation and leucine incorporation (Chin-Leo and Kirchman, 1988). During balanced growth of bacterioplankton both methods should give similar estimates of bacterial production. However, in nature, balanced growth is never fulfilled and under unbalanced growth conditions the comparison of both methods will provide additional information about the physiological state of bacterioplankton. Considering its advantages and disadvantages, the thymidine method is probably one of the best at our disposal, although we have to be aware that there are some uncertainties in its theoretical and practical aspects that have important implications for the understanding of the marine microbial ecology. We will focus this work on the advances and perspectives in measuring bacterial production from thymidine incorporation, and on their implications when trying to estimate carbon flux through the microbial loop. Universality and specificity of thymidine incorporation for growing bacteria The first assumption to be discussed is the universality and specificity of thymidine for growing heterotrophic bacteria. This is a necessary prerequisite for using the thymidine method. There are three questions related to this premise.
Bacteria are the only microorganisms which incorporate thymidine at low concentrations and short incubation times. Microorganisms other than bacteria have been investigated to determine if they incorporate thymidine. Cyanobacteria (Fuhrman and Azam, 1982; Robarts and Wicks, 1989), fungi (Fallon and Newell, 1986; Bååth, 1990) protists (Pedrós-Alió et al., 1993) and algae (Bern, 1985; Rivkin, 1986) have been reported not to incorporate thymidine at least at the concentrations and incubation times normally used in bacterioplankton studies. Thymidine is taken up by all growing heterotrophic bacterial cells. This assumption has been investigated by a number of authors and the results are contradictory. Fuhrman and Azam (1982) reported that virtually all the active heterotrophic bacteria were able to incorporate thymidine. From microautoradiography, other workers (Novitsky, 1983; Douglas et al., 1987; Pedrós-Alió and Newell, 1989) found that bacterial production rates from thymidine incorporation may be underestimated since not all growing bacteria are able to use exogenous thymidine (Table 1). In samples from the Baltic Sea, the North Sea and the northeastern Mediterranean Sea, Zweifel and Hagström (1995) showed that only a minor fraction (2-32%) of total bacterial counts can be scored as bacteria with nucleoids, which represents the maximum estimate of the number of bacteria able to use exogenous thymidine.
On the other hand, several authors showed that isolates of heterotrophic bacteria (Vibrio sp., Flavobacterium sp., Pseudomonas sp.) were unable to incorporate thymidine, probably due to the absence of thymidine transport systems and/or thymidine kinase activity, the enzyme needed to incorporate exogenous thymidine to the salvage pathway for the synthesis of DNA (Pollard and Moriarty, 1984; Davis, 1989; Jeffrey and Paul, 1990). It is striking that thymidine kinase activity has not been found in any species of the genus Pseudomonas (Robarts and Zohary, 1993). Data on anaerobic bacteria show that some but not all organisms are able to incorporate thymidine. Therefore, bacterial growth rates in anoxic environment estimated from rates of thymidine incorporation may be underestimated (McDonough et al., 1986; Winding, 1992; García-Cantizano et al., 1994). Wellsbury et al. (1993) investigated the ability of different isolates of anaerobic physiological groups to incorporate exogenous thymidine: sulphate-reducing bacteria, fermentative heterotrophs, purple sulphur bacteria, acetogens and methanogens. The only obligate anaerobes in which thymidine incorporation into DNA was demonstrated were those of the genus Clostridium. Anaerobically growing Bacillus sp. also incorporated thymidine. Consequently the use of the thymidine method in anaerobic environments is not recommended as a reliable method to measure bacterial production. With regard to chemolithotrophic bacteria, the obligate aerobe Thiobacillus ferrooxidans also cannot incorporate thymidine (Wellsbury et al., 1993). Three strains of nitrogen-dependent chemolithoautotrophic bacteria and two strains of methylotrophic bacteria
TABLE 1. – Percentage of cells incorporating thymidine from microautoradiographic studies in marine systems.
% related to total AODC
Samples
Reference
11.2 - 16.8 5.9 - 16.5 8.8 34 - 52 2.6 9.2 - 19.7 1 - 69 20 - 80 67 - 75 7 - 82
Nova Scotia, Canada, coastal Nova Scotia, Canada, slope Nova Scotia, Canada, eddy Scripps Pier Coastal lagoon, Spain Halifax Harbour, Canada Sapelo Island Danish Fjord Chesapeake Bay Chesapeake Bay
Douglas et al. (1987) Douglas et al. (1987) Douglas et al. (1987) Fuhrman and Azam (1982) García-Cantizano et al. (1994) Novitsky (1983) Pedrós-Alió and Newell (1989) Riemann et al. (1984) Tabor and Neihof (1982) Tabor and Neihof (1984)
BACTERIAL PRODUCTION FROM THYMIDINE INCORPORATION 113
were also incapable of thymidine uptake (Johnstone and Jones, 1989). However, Kraffzik and Conrad (1991) found thymidine uptake in pure cultures of chemolithoautotrophic (CO, H2) and methanotrophic bacteria. From microautoradiographic studies GarcíaCantizano et al (1994) also detected labelled autotrophic cells after short incubations in the dark. Non-growing cells do not incorporate thymidine Mården et al. (1988) and Davis (1989) found that non-growing bacteria are able to incorporate thymidine even into DNA. These studies demonstrated that some bacteria under starvation can transport thymidine into the cell, and the nucleoside can be incorporated into DNA in the absence of cell division. In consequence, the use of the thymidine method in oligotrophic systems may lead to overestimates of bacterial production. In conclusion, some of the inherent assumptions about universality and specificity of thymidine incorporation to growing bacteria have to be questioned if not rejected. It seems to be confirmed that thymidine is mainly incorporated by bacteria at nanomolar concentrations and short incubation times, but there is also evidence that some growing bacteria are not capable of incorporating thymidine, which implies that bacterial production will be underestimated. Concerning this issue, realistic estimates of bacterial production might be achieved if the conversion factors to transform thymidine incorporation into bacterial production were specifically characterized for each system.
situation to accurately estimate bacterial growth rates occurs when the de novo pathway is not functioning. Then the only supply of dTMP is from the salvage pathway and therefore dTMP is mainly synthesized from exogenous thymidine. It is assumed that for water samples the isotope dilution by extracellular and intracellular sources of thymidine can be prevented by using saturating concentrations of tritiated thymidine. It is necessary to determine the saturating concentration in previous experiments and repeat the concentration tests frequently, or at least when presuming that environmental conditions are changing. Heinänen (1993) and García-Cantizano et al. (1994) showed that saturating concentrations can change both in time and space. Bell (1993) recommended the use of 20 nM, although lower concentrations can be enough in oligotrophic systems (Robarts and Zohary, 1993). The use of saturating concentrations minimizes the problem, but when analysing samples with high levels of particulate material, like river samples, sediments or marine snow, it is important to determine the extent of the isotope dilution and thereafter to correct thymidine incorporation rates. Several approaches to estimate the extent of isotope dilution have been applied to a wide range of habitats (Pollard and Moriarty, 1984; Bell, 1986; Jeffrey and Paul, 1988; Moriarty and Pollard, 1990; Tibbles et al., 1992; Heinänen, 1993), but this issue remains controversial, since it has been reported that intracellular isotope dilution is not accounted for by any isotope dilution analysis, when the limiting step for the incorporation of thymidine is the reaction catalysed by the enzyme thymidine kinase and not thymidine uptake.
Isotope dilution Nonspecific Macromolecular Labelling Once thymidine is transported into the cell, it must be converted to dTMP (thymidine monophosphate) by the enzyme thymidine kinase, and subsequently to dTDP (thymidine diphosphate) and dTTP (thymidine triphosphate) before incorporation to DNA. This is the salvage pathway of thymidine metabolism in bacterial cells. A potential problem is the dilution of the [3H]thymidine by extracellular or intracellular sources. The extracellular dilution comes from the presence of high concentrations of non labelled thymidine in the natural environment or any nucleic acid component which may compete with thymidine for the same enzymes in pyrimidine salvage pathways (Jeffrey and Paul, 1988). Intracellular dilution is due to the synthesis of dTMP from other nucleosides by the de novo pathway or when labelled thymidine is degraded within cells (Moriarty, 1986). The optimal 114 M. UNANUE and J. IRIBERRI
The thymidine used for the estimate of bacterial production in microbial ecology is usually labelled in the methyl group and it has been widely reported that label can also be incorporated into macromolecules other than DNA such as proteins, RNA and lipids. For that reason, incubation times must be as short as possible in order to reduce nonspecific labelling. Our knowledge of bacterial thymidine metabolism is still insufficient to explain why nonspecific macromolecular labelling occurs and the distribution of the label is variable. Cho and Azam (1988) suggested that marine bacteria can use thymidine for protein synthesis during nitrogen starvation. It has also been hypothesized that thymidine may be used as a carbon source rather than as a DNA precursor (Servais et al., 1987; Robarts and Wicks, 1989; Brittain and Karl, 1990).
TABLE 2. – Percentages of tritium in the DNA, RNA and protein fractions related to the label in the total macromolecules in marine samples.
% DNA
% RNA
% Protein
Samples
Reference
86 - 89 34 - 86 80 8 - 32 3 - 28 13 - 64 58 69 - 120 80 - 95 60 - 85 90 - 98 0 - 80 0 - 35 6 - 64 12 - 56 82 25 6 - 95 21 53 68 nr 71 - 99 55 21 - 83 43 - 65 66 - 82 29 - 80 20 15 - 51 33 - 110
nr nr nr 14 - 74 47 - 92 nr nr nr nr nr nr nr 34 - 67 29 - 49 47 - 82 nr 50 nr 79 47 32 nr nr 20 nr nr nr nr nr 0.1 - 61 nr
nr nr nr 2 - 76 3 - 34 nr nr nr nr nr nr nr 20 - 59 7 - 45 6 - 29 nr 25 nr nr nr nr 30 - 100 nr 25 nr nr nr nr nr 8 - 24 nr
Marine snow Coastal waters, North Spain Baltic Sea Seawater Sediments Southern California Salt-marsh, South Carolina North Atlantic Coastal waters, California Southern California Bight Antarctica, surface Antarctica, deep San Francisco Bay Monterrey Bay San Francisco Bay Scripps Pier, Saanich Inlet Bayboro Harbor, Florida Waters of Southwest Florida Hawaii, offshore surface Hawaii, offshore deep Hawaii, coastal waters Delaware Estuary Sapelo Island Coastal waters, Belgium Mediterranean Sea, surface Mediterranean Sea, deep North Sea Scripps Pier Biétri Bay, Ivory Coast Coastal water, Gulf of Guinea Several aquatic systems
Alldredge and Gotchalk (1990) Barcina et al. (1992) Bell (1986) Brittain and Karl (1990) Brittain and Karl (1990) Cho and Azam (1988) Chrzanowski and Zingmark (1989) Ducklow et al. (1992) Fuhrman and Azam (1980) Fuhrman and Azam (1982) Hanson and Lowery (1983) Hanson and Lowery (1983) Hollibaugh (1988) Hollibaugh (1988) Hollibaugh (1994) Hollibaugh et al. (1980) Jeffrey and Paul (1988) Jeffrey et al. (1990) Karl (1982) Karl (1982) Karl (1982) Kirchman and Hoch (1988) Riemann et al. (1987) Servais et al. (1987) Servais et al. (1987) Servais et al. (1987) Servais et al. (1987) Simon and Azam (1989) Torréton and Bouvy (1991) Torréton and Bouvy (1991) Wicks and Robarts (1987)
nr: not reported
In reviewing the literature the percentage of label in different fractions was found to be very variable (Table 2) and it is accepted that most variability can be attributed to the fractionation procedure used. Currently, there is not a simple and quick standardized protocol for the extraction and purification of DNA and other macromolecules. At present the methods used include acid-base hydrolysis (Fuhrman and Azam, 1982; Riemann and Søndergaard, 1984; Hollibaugh, 1988), enzymatic fractionation (Servais et al., 1987) and phenol-chloroform extraction (Wicks and Robarts, 1987). All these methods have been widely modified from their appearance and show advantages and disadvantages extensively discussed by Robarts and Zohary (1993) in their excellent review. Another way to face up to the problem of the nonspecific labelling and the absence of an appropriate method to extract the labelled DNA is to assume that the bulk of the label in cold TCA precipitates is DNA. Data on bacterial production rates in marine
systems have been obtained from the measurement of the label in cold TCA precipitates. This method is faster than applying an extraction procedure, and the results are not more uncertain. These data would be realistic if the label were mainly incorporated into DNA, as was reported by Fuhrman and Azam (1980) in the introduction of the method. On the contrary, these data would be overestimates of the real rates. However, the conversion factor to transform thymidine incorporation into bacterial production can be used as a correction factor if it is estimated following the empiric method by measuring the label into cold TCA precipitates (see below). Conversion factors Finally, the use of accurate conversion factors to convert thymidine incorporation into bacterial production has been considered one of the most critical issues. The first conversion factor to transform thymidine incorporation rates into numbers of cells was
BACTERIAL PRODUCTION FROM THYMIDINE INCORPORATION 115
calculated by Fuhrman and Azam (1980) based on several assumptions: labelled DNA comprises 80% of the total labelled macromolecules in the cold TCA precipitate, total bacterial DNA contains 25% mol thymidylic acid residues and the amount of DNA per cell ranges from 7.47x10-16 to 4.82x10-15 g. They found a conversion factor of 2.0 x1017 to 1.3 x1018 cells produced per mol of thymidine incorporated. The conversion factors estimated following this approach are usually referred to as theoretical conversion factors. Subsequently, the authors modified this value by introducing a series of corrections based on empirical measurements for the systems under study, such as DNA content of bacterioplankton and the intracellular isotope dilution, and advised that an specific value should be estimated for each system (Fuhrman and Azam, 1982). The conversion factors corrected for these empirical measurements were more realistic, 1.7x1018 cells mol-1 for the nearshore and 2.4x1018 cells mol-1 for the offshore (>10 km). Kirchman et al. (1982) proposed an empirical approach to derive the conversion factor from bacterioplankton cultures. The aim of this approach was to compare the increase in thymidine incorporation to the increase in bacterial abundance or biomass during the incubation of bacterial assemblages; water samples are filtered through 0.6 µm - 2 µm pore size filters to remove grazers, and diluted with 0.2 µm filtered water (usually 1:10) to allow bacterial growth. Thymidine incorporation rates as well as bacterial abundance and bacterial biovolume are measured over time (for a more detailed description of the procedure see Kirchman and Ducklow, 1993). The advantage of this method compared to the theoretical approach is that it allows estimates of specific conversion factors for each system. Its main disadvantage is that bacterial physiology can be easily altered during the manipulation of the samples (filtration, dilution, incubation etc.) which can affect the conversion factors. The main assumption for the calculation of the conversion factor from the empirical approach is that thymidine incorporation and production of new cells are coupled processes. However, it is commonly found (Kirchman et al., 1982; Ducklow and Hill, 1985; Iriberri et al., 1990a; Bjørnsen and Kuparinen, 1991) that thymidine incorporation increases faster than cell number or biomass, indicating unbalanced growth (Figure 1). Kirchman et al. (1982) hypothesized that this uncoupling might be due to the presence of non growing cells and proposed a correction, which would provide an addi116 M. UNANUE and J. IRIBERRI
FIG. 1. – Changes in thymidine incorporation rates and bacterial abundance in diluted bacterioplankton cultures (Salvaje, Beach, Spain).
tional estimate of the percentage of active cells (Iriberri et al., 1990a). There might be some other explanations for the differential rates of increase of abundance and incorporation like continuous changes in growth rate during the incubation, and selection of stronger thymidine incorporators (Ducklow et al., 1992) as it is indicated by the increase in thymidine incorporation per cell with time of incubation (Figure 2).
FIG. 2. – Chage in thymidine incorporation per cell during the incubation time in diluted bacterioplankton cultures (Salvaje Beach, Spain).
Three methods have been developed to estimate conversion factor from the empirical approach: 1) the derivative method (initially proposed by Kirchman et al., 1982 and later modified by Ducklow and Hill, 1985). Conversion factors are estimated from the y-intercept of Ln (cells) versus time, multiplied by the growth rate determined from the increase in cells numbers over time, and divided by the y-intercept of Ln (thymidine incorporation rates) versus time. 2) the integrative method (Fuhrman and Azam, 1980; Riemann et al., 1987). Conversion factors are calculated by dividing the number of bacterial cells produced in a time period by the total amount of radioactive label incorporated during that period. 3) the cumulative method (Bjørnsen and Kuparinen, 1991). This is related to the integrative method because it calculates the slope of cell numbers at a time point versus the incorporation rates integrated to that time point. It is a way of averaging the factors calculated by the integrative method over time. Advantages and disadvantages of these methods have been discussed by Ducklow et al. (1992) and Kirchman and Ducklow (1993), but currently there is no agreement about the best method for calculating conversion factors. When bacterial growth is balanced all methods provide similar values. Problems arise when thymidine incorporation and cell production are not coupled. Table 3 shows the conversion factors estimated by the empirical approach in coastal waters of the North of Spain (unpublished data). We used the modified derivative method, the integrative method and the cumulative method. Conversion factors varied depending on the method used. The derivative method provided the highest values, while we found very similar values using the integrative and cumulative methods. TABLE 3. – Conversion factors (1018 cells mol-1) calculated from the empirical approach by using different calculation methods. Data from a marine coastal system, Salvaje Beach in the North of Spain (unpublished data). Samples
Derivative
Integrative
Cumulative
11 January 19 January 3 February 17 March 11 May 25 May 1 June
1.37 7.95 4.33 5.52 3.28 7.42 1.58
2.50 1.32 0.68 3.76 1.22 1.64 0.34
2.22 2.09 0.73 4.12 1.73 1.44 0.49
Mean ± SE
4.49 ± 0.99
1.64 ± 0.44
1.83 ± 0.45
Conversion factors reported in the literature show a great variability (10x1018 cell mol-1) (Table 4). However, during the last few years, most studies have estimated values