Capabilities and limitations of DGGE for the analysis of hydrocarbonoclastic prokaryotic communities directly in environmental samples

Abstract Prokaryotic communities in pristine and oil‐contaminated desert soil, seawater, and hypersaline coastal soil were analyzed using culture‐dependent and culture‐independent approaches. The former technique was the dilution‐plating method. For the latter, total genomic DNA was extracted and the 16S rRNA genes were amplified using a universal bacterial primer pair and primer pairs specific for Actinobacteria, Gammaproteobacteria, and Archaea. The amplicons were resolved using denaturing gradient gel electrophoresis (DGGE) and sequenced, and the sequences were compared to those in GenBank. The plating method offered the advantages of capturing the targeted hydrocarbonoclastic microorganisms, counting them and providing cultures for further study. However, this technique could not capture more than a total of 15 different prokaryotic taxa. Those taxa belonged predominantly to the genera Alcanivorax, Pseudoxanthomonas, Bosea, Halomonas, and Marinobacter. The individual isolates in culture consumed between 19 and 50% of the available crude oil in 10 days. Although the culture‐independent approach revealed much more microbial diversity, it was not problem‐free. The subdivision primers exhibited satisfactory specificity, but they failed to capture all the available taxa. The universal bacterial primer pair ignored Actinobacteria altogether, although the primer pair specific for Actinobacteria captured many of them, for example, the genera Geodermatophilus, Streptomyces, Mycobacterium, Pontimonas, Rhodococcus, Blastococcus, Kocuria, and many others. Because most researchers worldwide use universal primers for PCR, this finding should be considered critically to avoid misleading interpretations.

One of the problems is related to the specificity of the primers used for amplification of the 16S rRNA genes. The use of "universal" PCR primers is known to result in ignoring minor constituents of the microbial community because the predominant constituents will prevail in the amplification product. To address this issue, group-specific PCR primers, for example, for actinobacterial (Stach, Maldonado, Ward, Goodfellow, & Bull, 2003) and gammaproteobacterial (Mühling, Woolven-Allen, Murrel, & Joint, 2008) 16S rRNA gene fragments, have been developed. However, there are no extensive studies on the specificity strictness of such primers. The major objective of work in this paper, which emphasizes hydrocarbonoclastic microbial communities, is to contribute to filling this information gap. The results are expected to provide researchers in the field of environmental microbiology with useful guidelines for the molecular analysis of microbial communities. It may be argued that the molecular approaches adopted in this study are "old," while deep sequencing is now the method of choice. While this is certainly true, many laboratories worldwide still use the "old" techniques (e.g., Corsellis, Krasovec, Sylvi, Cuny, & Militon, 2016;Pasumarthi & Mutnuri, 2016;Wu, Dick, Li, & Chen, 2016) and this situation will probably continue in the future. Furthermore, extensive studies are found in the literature, whose interpretation would benefit from the results of this study. One motivation behind this work was to prove that whatever advances in techniques for microbiology analyses would be reached, such techniques will still always have limitations.

| Environmental samples
The environmental samples used in this study represented three ecosystems: a desert soil sample from the Kadma area, north of Kuwait City; a seawater sample from the Arabian Gulf coast at Salmiya, in the middle of Kuwait City; and a hypersaline soil sample from the "sabkha" at the Khiran coast, south of Kuwait. Pristine samples were collected in sterile containers, transported to the laboratory, and subjected to processing on the same day.
Relevant environmental parameters at the sampling sites were measured in triplicate samples taken 20 cm apart using the quality checker WQC-24, Japan, and APOGEE, USA. These parameters were the pH values; dissolved oxygen content of the seawater; temperature; concentrations of sodium chloride, nitrate, ammonium and phosphate; and total organic carbon content. The oxygen content in soil was measured using an ICTO2 Soil Oxygen Sensor (Armidale, Australia) (Cary & Holder, 1982).

| Experimental design
Each of the pristine samples was used for microbiological analysis directly (designated "pristine") or after having been artificially polluted with crude oil (designated "contaminated"). Aliquots of 50 g of pristine desert soil were suspended in 50 ml aliquots of sterile tap water in screw-capped 250 ml conical flasks. Portions of seawater (100 ml) were dispensed into 250 ml flasks. Hypersaline coastal soil aliquots, 50 g in conical flasks, were suspended in 50 ml aliquots of hypersaline water (2 mol/L NaCl) from the same site.
Contaminated samples were prepared by adding 0.3% w/v light Kuwaiti crude oil to the conical flasks. Flasks in triplicate containing pristine and contaminated samples were incubated on an electrical shaker at 110 rpm and 30°C for 1 month before their contents were analyzed.

| Culture-dependent analysis of hydrocarbonoclastic microorganisms
A selective mineral medium with oil vapor as a sole source of carbon and energy (Sorkhoh, Ghannoum, Ibrahim, Stretton, & Radwan, 1990) was used. Medium aliquots for the analysis of seawater and hypersaline soil were provided with 3% and 12% NaCl, respectively.

| Determination of oil consumption by individual isolates
Cell suspensions of 1 ml from a common stock (loopful in 10 ml water) of the tested microorganism were inoculated into 250 ml screw-capped flasks containing 50 ml aliquots of the mineral medium (Sorkhoh et al., 1990) and 0.3% light Kuwaiti crude oil. Control flasks were prepared similarly and autoclaved. Three replicates were prepared throughout. The flasks were inoculated, tightly screwcapped, and incubated on an electrical shaker at 110 rpm and 30°C for 10 days. Residual hydrocarbons were recovered from each flask with three 20 ml aliquots of pentane. The combined extract was brought to 60 ml with pentane, and 1 μl was analyzed via gas liquid chromatography (GLC) using a Varian 3900 (USA) as described earlier (Al-Mailem, Eliyas, Khanafer, & Radwan, 2015). The percentage of oil consumption was calculated as the percentage reduction in total hydrocarbon peak area in the GLC profiles based on the total areas of peaks of hydrocarbons recovered from the autoclaved controls.

| Culture-independent analysis of samples using specific primers for amplification
Total genomic DNA was extracted from the sample, purified, and PCR amplified using the four different primer sets as follows: (1) universal primer pair for bacteria (Santegoeds, Ferdelman, Muyzer, & Beer, 1998): GM5F with the sequence 5′-CCTACGGGAGGCAGCAG-3′ and 907R with the sequence 5′-CCGTCAATTCMTTTGAGTTT-3′; (2)  For denaturing gradient gel electrophoresis (DGGE) analysis, forward primers with GC clamps were used. The acrylamide gel concentration used was 50%-70% for the actinobacterial primer pair and 40%-60% for the remaining three pairs. The DGGE was performed as described previously (Al-Awadhi et al., 2013). The amplicon bands on the gels were cut out, amplified, and sequenced using the specific primers as described above. The sequences were compared with those of the closest relatives in the GenBank database.

| Environmental parameters
The three ecosystems studied, desert soil, seawater, and hypersaline coastal water, shared the characteristic of being poor in inorganic (nitrate, ammonia, phosphate) and organic carbon nutrients (Table 1).
The standard deviation values for the triplicates were <5% of the mean values. However, these three ecosystems varied dramatically in their NaCl contents. In the hypersaline coastal soil, the salinity was significantly (p < .05) higher than in the seawater ecosystem, in which F I G U R E 1 Typical DGGE profile of 16S rDNA amplicons (using universal bacterial primers) in total DNA extracted from desert soil, seawater, and hypersaline soil samples (for band sequencing see Table 4). PD, pristine desert soil; CD, contaminated desert soil; PSW, pristine seawater; CSW, contaminated seawater; PS, pristine hypersaline soil; CS, contaminated hypersaline soil the salt concentration was significantly (p < .05) higher than in the desert soil. Consequently, the dissolved oxygen contents were lowest in the hypersaline ecosystem. Oxygen solubility in water decreases with increasing salinity. In this context, the initial step of microbial attack on the hydrocarbon substrate requires molecular oxygen (Ratledge, 1978).
T A B L E 4 Sequencing of the 16S rDNA bands in the DGGE gel, Figure 1 Band

| Hydrocarbonoclastic microbial communities captured using the culture-dependent method
The results in Table 2 show that the dilution-plating method using a mineral medium with oil as a sole source of carbon and energy revealed hydrocarbonoclastic microbial numbers in the magnitude of 10 3 to 10 7 colony forming units (CFU) g −1 . The lowest numbers were counted in the pristine hypersaline coastal soil. However, it was in this extreme ecosystem that oil addition resulted in the highest increases in microbial numbers, from 10 3 to 10 7 CFU g −1 , that is, approximately 10,000-fold. Increases in numbers of hydrocarbonoclastic microorganisms in response to oil contamination also occurred in the desert and seawater samples but by approximately only two-to ninefold.   (Klug & Markovetz, 1971;Leahy & Colwell, 1990). The pristine and contaminated hypersaline coastal samples each contained one gammaproteobacterium, Halomonas sp. and Marinobacter sp., respectively, but surprisingly F I G U R E 2 Typical DGGE profile of 16S rDNA amplicons (using Actinobacterial-specific primers) in total DNA extracted from desert soil, seawater, and hypersaline soil samples (for band sequencing see Table 5). PD, pristine desert soil; CD, contaminated desert soil; PSW, pristine seawater; CSW, contaminated seawater; PS, pristine hypersaline soil; CS, contaminated hypersaline soil     ( limited capturing capacity, revealed two Actinobacteria and one taxon belonging to Bacilli in the studied samples (Tables 2 and 3). It will be shown below that all the studied samples were rather rich in taxa belonging to this subdivision when the actinobacterial primer pair was used for amplification.

| Microbial communities captured by the cultureindependent method using the universal bacterial primer pair
An interesting result is that the oil-contaminated samples always contained fewer bands than the pristine samples. It is possible that oil enriches hydrocarbonoclastic members at the expense of the nonhydrocarbonoclastic ones. Among the few bands enriched by oil, was band 34 of Phenylobacterium, which is hydrocarbonoclastic   Table 4 with those in Table 2 clearly shows that the culture-independent method was much more powerful in capturing diverse bacterial taxa than the culture-dependent method. As the culture-independent approach failed to provide information of the hydrocarbonoclastic potential of the taxa, we cited for the microbial names in Table 4 the pertinent references in the available literature that confirm the probable hydrocarbonoclastic nature of the respective organisms. Needless to say, several of the other yet-unstudied taxa may also contain hydrocarbonoclastic activity. Figure 2 shows a typical DGGE plate resolving 16S rRNA genes resulting from the amplification of total genomic DNA extracts from the studied environmental samples using the specific actinobacterial primer pair. The identities of amplified bands from this gel are provided in Table 5.

| Microbial communities captured by the culture-independent method using the actinobacterial primer pair
A total of 24 amplicon bands were successfully sequenced. In addition, several other bands failed to be sequenced. The primer pair exclusively amplified Actinobacteria. While this could be considered a point of strength of the culture-independent approach, it is also a point of weakness. The richness of actinobacterial species in Table 5 compared with their absence in Table 4 shows that depending on a universal bacterial primer pair alone, as frequently reported in the literature, leads to ignoring many actinobacterial species actually present in the  Table 5, many of the taxa listed above were reported earlier as hydrocarbonoclastic. It is also clear that fewer bands were present in oily than in pristine samples. The former were apparently those of bacteria with high hydrocarbonoclastic potential. The two species belonging to the two genera, Streptomyces and Mycobacterium, captured by the culture-dependent method (Table 2) were different from the species captured by the culture-independent method.
F I G U R E 3 Typical DGGE profile of 16S rDNA amplicons (using Gammaproteobacteria-specific primers) in total DNA extracted from desert soil, seawater, and hypersaline soil samples (for band sequencing see Table 6). PD, pristine desert soil; CD, contaminated desert soil; PSW, pristine seawater; CSW, contaminated seawater; PS, pristine hypersaline soil; CS, contaminated hypersaline soil 3.5 | Microbial communities captured by the culture-independent method using gammaproteobacterial primers Figure 3 illustrates a typical DGGE plate resolving 16S rRNA genes resulting from amplification of total genomic DNA extracted from the studied experimental samples using the specific gammaproteobacterial primer pair. The identities of amplified bands on this gel are provided in Table 6.
As many as 25 bands were successfully sequenced and identified.
Several other bands failed to be sequenced. This relatively large number of amplicon bands reflects the diversity of the gammaproteobacterial taxa in the microbial communities. As should be expected, many more gammaproteobacterial bands were found in the seawater and hypersaline coastal soil than in the desert soil.
The primer pair exclusively amplified Gammaproteobacteria (  ( 3.6 | Microbial communities captured by the cultureindependent method using the archaeal primer pair Figure 4 shows the resolution of 16S rRNA genes resulting from the amplification of total genomic DNA extracted from the samples using the archaeal primer pair for amplification. The identities of amplified bands on this gel are available in Table 7.
Only eight bands were successfully sequenced and identified.

| CONCLUSIONS
Technical bias problems related to the molecular analysis of microbial communities in environmental samples are well documented. Such problems may be due to template annealing in the amplification of 16S rRNA genes (Suzuki & Giovannoni, 1996), template-to-product ratios in multi-template PCR (Polz & Cavanaugh, 1998), limitations inherent in 16S rRNA gene interspecies heterogeneity (Dahllöf et al., 2000), single DGGE bands not always representing single bacterial strains (Sekiguchi et al., 2001), primer mismatch, annealing temperature, and PCR cycle numbers affecting the 16S rRNA gene-targeting (Sipos et al., 2007), intraspecific polymorphism of 16S rRNA genes (Cui, Zhou, Oren, & Liu, 2009) and differential 16S rRNA gene amplification by primers (Al-Awadhi et al., 2013). The novelty of this study is that it sheds light on additional bias problems other than those reviewed above. Using new environmental samples, we confirmed F I G U R E 4 Typical DGGE profile of 16S rDNA amplicons (using archaeal primers) in total DNA extracted from desert soil, seawater, and hypersaline soil samples (for band sequencing see Table 7). PD, pristine desert soil; CD, contaminated desert soil; PSW, pristine seawater; CSW, contaminated seawater; PS, pristine hypersaline soil; CS, contaminated hypersaline soil T A B L E 7 Sequencing of the 16S rDNA bands in the DGGE gel, Figure 4 Band here our recent finding (Al-Awadhi et al., 2013) that the frequently used bacterial universal primers ignore the major subdivision of Actinobacteria, although representatives of this subdivision may actually be quite frequent in the studied samples. Neglecting this fact would lead to misinterpretation of old and new findings on microbial community composition in environmental samples.
It was not the major objective of this study to discuss the frequency of hydrocarbonoclastic prokaryotes in relation to the prevailing environmental parameters. Although the major objective was to evaluate common techniques used in the microbial analysis of environmental samples, the former subject could not be completely neglected. One conclusion of this paper is that for a comprehensive view of the microbial communities of environmental samples, a culture-dependent method should be adopted along with the culture-independent approach using primers specific for various subdivisions. Each individual analysis gives its own unique list of microorganisms. The collective result would provide a microbial community structure closest to reality, as suggested previously (Shade et al., 2012).
It may be argued that hydrocarbonoclastic microorganisms in environmental samples could have been analyzed using alkane-specific primers. However, experience in our laboratory revealed repeatedly that the alkB genes frequently failed to be demonstrated even in clas-