We then mad dilutions of this culture that corresponded to OD600 values of 0.1, 0.2, 0.4, 0.6, 0.8, 1.0, and 2.0. Each of these dilutions was then used to make dilutions of 1 in 105 , 1 in 106 , and 1 in 107 , or 1/10,000, 1/1000,000, and so on. Dilutions were then plated on LB/Tetracycline agar with 50µL, 100µL, 150µL, or 200µL plating volumes. To determine whether cultures contained enough biomass for sequencing, we tested the DNeasy Blood and Tissue Kit and the PowerSoil DNA Isolation Kit for efficacy of DNA extraction. Briefly, the liquid above the sedimented cells in the 22-hour cultures from each of three wells was aspirated and kept. This liquid, the sedimentary culture, and the two different types of negative controls from 3 wells were each divided into two volumetrically equal aliquots and pelleted by centrifugation at 6,000 x g. The supernatant was discarded, and the pellets were resuspended in 200µL 1X PBS, pH 8.0. For each set of the two pellets from the same well, we extracted one pellet following the DNeasy kit short protocol and the other following the manufacturer’s instructions of the PowerSoil kit. Each individual well was therefore extracted by two different kits, raspberries in pots and each type of sample was extracted in triplicate by both kits. Extracted DNA concentrations were measured by Qubit 2.0 with the high-sensitivity double-stranded DNA assay kit.
The four-day cultures were similarly processed and extracted, with the key difference that each well was divided into 6 aliquots instead of 2, and 3 of which were extracted with the PowerSoil kit and the other 3 with the DNeasy kit. After extraction and quantification, we compared the means and standard deviations of the DNA concentrations from the two kits.DNA extraction and sequencing were performed at the UC Davis Host Microbe Systems Biology Core. The extraction protocol was a slightly modified version of the protocol for the Earth Microbiome Project . Each flash-frozen pellet was resuspended in 400µL of 1X PBS while the 1.0mL frozen liquid samples were simply defrosted.Half of all samples was extracted – 200µL of the resuspended pellets and 500µL of the defrosted liquid – and the other half reserved as samples for quality control or potential resequencing. After extraction, 1µL of each sample was used for PCR amplification of the 16S rRNA V4 region with the 515F/806R primer pair , the products of which were quantified and normalized to 100ng. The PCR library was then pooled and cleaned on a Pippin Prep gel and quality-checked on a bioanalyzer, after which the molar concentrations were determined by quantitative PCR . Sequencing of properly diluted libraries was performed on the Illumina MiSeq platform using the paired-end method with a length of 253bp. Quality control of the raw sequences was performed first in QIIME by HMSB , and then formally reperformed in R with the mothur software. The QIIME-processed reads were not used in analysis, so those methods will not be detailed here. For the quality control process in mothur, 3,332,845 contigs were constructed from raw reads, size-selected to be between 240 to 275 basepairs, and trimmed to eliminate ambiguous reads.
Denoising and chimera identification with UCHIME were performed for additional quality control, and chimeric sequences were removed with VSearch. Non-bacterial sequences were removed based on version 132 of the full-length SILVA database. The resulting 1,932,473 contigs were clustered into operational taxonomic units based on the same SILVA reference database with 97% sequence identity, approximating species-level taxa. These OTUs were used to construct a dense biological observation matrix table for analysis.Bio-informatics analysis was first performed in QIIME by HMSB, and then formally reperformed in R with version 1.38.0 of the phyloseq software package. In this analysis, we examined the sequencing depth, the number of OTUs, and the inverse Simpson’s index of the samples. We considered the prevalence of organisms at the phylum level to gain an understanding of the general compositional structure of the bacterial community. Then, we investigated the correlation between sequencing depth and diversity by examining rarefaction curves and plots of read counts vs. diversity indices. To minimize the presence of potentially spurious OTUs, we rarefied samples to an even depth of 20,000 reads and reexamined the number of OTUs and inverse Simpson’s index values. We also used the absolute counts and relative abundances of the controls and cultures to examine the biomass in controls and cultures relative to the E. coli spike-ins, thereby verifying that there was minimal contamination throughout the cultivation, DNA extraction, and sequencing processes.
After analyzing the process to verify minimal contamination, we removed the most prominent spike-in OTU from the read counts of the rarefied samples, converted read counts to relative abundances, and performed Principal Coordinate Analysis on the liquids and sedimented cultures using Bray-Curtis dissimilarity measures. We observed sample clustering in the PCoA plots to extract potential correlations between sample clusters and biological/compositional characteristics.Visual inspection of the cultures from the surface modification experiment revealed that while differences in modifications did not seem to lead to differences in the control wells, , they might have resulted in some differences in the cultures. Plates with mixed charges seemed to allow for more even attachment and distribution of cells than the plates with negatively surface alone . However, viability of cells and extent of attachment were difficult to assess quantitatively without extensive cell-counting experiments. In the absence of quantitative differences, we decided to use the well plates with mixed charges, based on the perceived evenness of attachment of cells , as well as previous research that has shown that bacteria tend to have net negative cell surface charges at neutral pH. The SHI medium used in the preliminary experiments has a pH between 7.0 and 8.0, in which bacterial cells would retain their negative cell surface charges. These negative charges would enable cells to attach more strongly on surfaces with mixed charges than on surfaces with negative charge alone. Since attachment and biofilm formation are integral parts of the formation process of dental plaque bacterial communities, we chose to incubate cultures in the mixed-charge plates going forward.We extracted each type of sample – negative control without pellicle , negative control with pellicle , liquid above sedimented cells in cultures, and sedimented cells – by two commercial kits, in triplicate for each kit. Table 1 contains the DNA concentration values of these extractions. Negative controls, incubated with media that underwent decontamination by centrifugation, yielded average concentrations of 0.0479 ad 0.0753 ng/µL when extracted with the DNeasy kit. These values were much lower than the liquids and sedimented cells extracted by the same kit . The PowerSoil extraction, on the other hand, also led to negative controls having lower DNA concentrations than the liquids and sedimented cells, but not by as great of differences as concentrations from the DNeasy kit; the two controls have similar concentration values , and the sedimented cells contained 25-30 times as much DNA as the controls, blueberries in containers growing but the difference between the controls and the liquid from the PowerSoil extraction was not as great as that between the controls and liquid from the DNeasy extraction. Compared to the DNeasy kit, the PowerSoil kit returned significantly lower DNA concentrations for Control 1 and the liquid – determined by a two-tailed student’s t-test with an value of 0.05 – but not significantly lower concentrations for Control 2 or sedimented cells . Both kits yielded DNA concentrations from the liquid samples that were much lower than those from the sedimented cell samples, validating that the contents of the kits and the processes were at least effective and unbiased in terms of the amount of DNA extracted. Extractions with the DNeasy kit generally led to higher DNA concentrations than those with the PowerSoil kit, but not all differences were statistically significant.We also extracted four-day cultures with both kits. Figure 6 shows the averages and standard errors of the DNA concentrations from the extractions.
With the exception of the one of the Control 1 wells, controls contained less than 0.5 ng/µL of DNA. The PowerSoil kit, much like in the extractions of the 22-hour cultures, yielded significantly lower concentrations of DNA than DNeasy in all except for two of the liquid samples . The unexpectedly high concentration of DNA in a single well was not an issue, as will be shown in later parts of preliminary experiments as well as later phases, as it was extremely likely that this control contained bacteria from OTUs specific to the dental plaque community. In terms of comparing the two kits, despite the lower DNA yield, the PowerSoil kit yielded much more consistent concentrations of DNA for any given type of sample from both the four-day cultures and the 22-hour cultures. It is unlikely that the differences in DNA concentrations exhibited by the DNeasy kitr eflected biological differences, in other words, differences in the number of cells across the wells, and this is an especially salient fact in the context of consistently low bacteria presence in the controls. Furthermore, the PowerSoil kit uses both chemical and physical processed to lyse cells, targeting both Gram-positive and Gram-negative bacteria more equitably than kits that exclusively employ one or the other lysis mechanism. It has been shown to yield more representative genomic profiles, as it does not lead to a bias toward Gram-negative organisms. Probably for this reason as well as the ease of use and ease of scaling, this kit has been an integral part of the EMP protocols for many years and was therefore a crucial step in the procedure used by HMSB. This research, the adoption of the kit by a professional genomic sequencing center, as well as our own results, led to our decision to choose the PowerSoil kit for DNA extraction. We kept this kit in reserve for the potentiality that we would later extract the DNA in the lab instead of outsourcing to a sequencing center.Analysis of the read counts, categorized by sample type and spike volume, showed that all samples except for the unspiked controls yielded more than 10,000 reads . The unspiked controls yielded substantially lower than 10,000 reads,providing evidence that centrifuging the medium helped decontaminate the controls. Read counts in other samples were comparable to the sequence counts in the E. coli samples , indicating that the biomass in the cultures as well as the depths of sequencing were sufficient for our goals. The number of OTUs varied across samples, albeit with no consistent trends . To begin with, the number of OTUs in both types of controls increased with spike-in volumes, but not along a unified, singular direction; the number of OTUs seemed to first increase with the spike-in volume and then decrease . On the other hand, the number of OTUs in the liquid consistently decreased with increasing spike-in volumes; the highest number of OTUs was observed in unspiked liquid samples and the lowest in liquids spiked with 500µL of E. coli. In cultures, spike-ins contributed to the changes in the number of OTUs similarly to the way they did in the controls. To examine how differences in sample size affect the number of OTUs discovered, we studied the correlation between diversity and sequencing depth using three distinct indices that emphasize different aspects of sample diversity. The first one, Shannon index, mathematically represents the entropy associated with predicting the type of an individual in a community. Simply put, this index associates higher number of individual types in a community with higher uncertainties of predicting the type of an individual randomly drawn from the community. The Simpson’s index represents the probability that two randomly drawn individuals would belong to the same type, or in the context of the preliminary experiments, the probability that two randomly drawn reads in a given sample belong to the same OTU. This index ranges between 0 and 1. Values close to 0 describe communities of high diversity in terms of both richness – the number of types – and evenness – how evenly distributed the individuals are among types; values close to 1 describe communities that are low diversity in both richness and evenness.