DNA extraction of bacterial cells using a semi-automated filtration system

The COVID-19 pandemic lockdown created problems with importing of commercial kits resulting in extended turnaround times for consumable deliveries. One way to circumvent this was to use an inexpensive optimized in-house method for DNA extraction from water. • The DNA extraction methods were optimized on a 96-well plate using a semi-automated filtration system to increase the number of samples from 24 to 96 at a time in 2 h. The DNA extraction method optimizations included: (a) Guanidium thiocyanate method plus dilution series of celite to determine DNA binding capacity; (b) QIamp 96 Qiacube HT kit (Qiagen®); (c) Guanidium thiocyanate with the celite replaced with a binding buffer. • The in-house DNA extraction methods and adapted in-house DNA extraction method were compared to QIamp 96 Qiacube HT kit (Qiagen®), which is used on a 96-well semi-automated filtration system. The results showed maximum capacity of the 96-well filter plates was 400 μℓ broth (OD600 = 0.45 = 3.6 × 108 cells/mℓ) before the 96-well filters blocked. • When the methods were compared, there was no significant difference between the in-house DNA extraction method with 1:420 celite dilution (P-value = 0.126) and the adapted in-house method with binding buffer (P-value = 0.298) DNA yield or amplification of PCR products.


Specifications table
Biochemistry, Genetics and Molecular Biology More specific subject area; Semi-automated DNA extraction Method name; Semi-automated DNA extraction method Name and reference of original method; (1) Boom

Growth and maintenance of bacterial strains
A commensal non-pathogenic E. coli (ComEC) strain (ATCC 25922) was used in this study for the loading capacity experiment. The strain was cultured from frozen glycerol stock on Plate Count Agar (PCA) (Oxoid, UK) and incubated under aerobic conditions at 37 °C for 16 h. Single colonies were enriched in 100 m sterile distilled water using the Colilert® Quanti-Tray® (IDEXX) and incubated under aerobic conditions at 37 °C for 16 h. After incubation a total of 20 m of the broth was removed from 10 wells and aliquoted into a 50 m Falcon tube. The absorbance reading at 600 nm (OD 600 ) of this broth was 0.45 correlating to 3.6 × 10 8 cells/m .
One environmental water sample (river) was tested where 100 m of a river water sample was enriched in the Colilert® Quanti-Tray® (IDEXX) and incubated under aerobic conditions at 37 °C for 16 h. After incubation a total of 20 m of the broth was removed from 10 wells and aliquoted into a 50 m Falcon tube. The OD 600 of the sample was 0.49, i.e. equivalent to 3.92 × 10 8 cells/m .

Buffer composition and preparation
The preparation of the celite, lysis buffer and washing buffer used for the in-house DNA extraction and adapted in-house DNA extraction method was as follows:

Lysis buffer
Lysis buffer was prepared to a final concentration of 5 M Guanidium thiocyanate, by dissolving 120 g guanidinium thiocyanate (GuSCN) (Sigma Aldrich) in 100 m of 0.1 M hydroxymethyl amino methane-hydrochloric acid (Tris-HCl) (pH 6.4) (Sigma Aldrich) in a 500 m Schott bottle. The bottle was heated to 60 °C to dissolve the GuSCN. After heating, 22 m of a 0.2 M ethylenediamine tetra-acetate (EDTA) (pH 8.0) with 2.6 m triton X-100 (Sigma Aldrich) solution was added to the suspension. The suspension was mixed and dispensed into 50 m Eppendorf tubes and 0.5 m of celite suspension was added to remove any contaminating DNA from the buffer. The final solution was left to stand for at least 1 h at room temperature with sporadic mixing. The celite was pelleted by centrifugation at 30 0 0 rpm for 10 min (NeoFuge-15R, Heal Force, Vacutec®) and the supernatant was transferred into sterile 50 m Eppendorf tubes wrapped in aluminium foil (stable 3 weeks at room temperature).

Binding buffer
Binding buffer was prepared to a final concentration of 5 M Guanidine hydrochloride and 40% isopropanol. The bottle was heated to 60 °C to dissolve the Guanidine hydrochloride. The solution was filter sterilised using 25 m syringe (Laboratory and Scientific Equipment) and 0.2 μm syringe filter (Laboratory and Scientific Equipment), stable at room temperature.

Wash buffer
Wash buffer was prepared to a final concentration of 5 M Guanidium thiocyanate, up by dissolving 120 g GuSCN and 100 m of 0.1 M Tris-HCl (pH 6.4) in a 500 m Schott bottle, heated to 60 °C to dissolve the GuSCN and dispensed into 50 m Eppendorf tubes. Thereafter, 0.5 m celite suspension was added to each tube to remove contaminating DNA from the buffer as described above.

Wash ethanol
A 70% (v/v) ethanol solution was prepared with sterile distilled water.

Multiplex polymerase chain reaction (PCR)
The extracted DNA was amplified with a single multiplex PCR (m-PCR) targeting 11 E. coli genes from commensal and pathogenic E. coli to visually compare the DNA amplification efficiency of the DNA extracted with the various methods when viewed on the agarose gels.
All virulence genes were amplified in a BIO-RAD® T100 TM Thermal Mycycler under the following conditions: enzyme activation at 95 °C for 15 min, 35 cycles of DNA denaturation at 94 °C for 45 s, annealing at 55 °C for 45 s, elongation at 68 °C for 2 min with a final elongation step at 72 °C for 5 min. For the negative control reaction mixture, the template DNA was replaced with sterile PCR grade water and the positive control reaction contained the combined extracted plasmids from the transformed E. coli strains [2] .
Amplified products were separated according to their molecular size on a horizontal agarose gel slab [2.5% (w/v)] containing ethidium bromide (0.5 μg/m ) using TAE buffer. Electrophoresis was conducted at 80-100 Volts for 30-50 min and viewed under UV light (Gene Genius Bio Imaging System, Vacutec®). The relevant sizes of the DNA fragments were estimated by comparing their electrophoretic mobility to that of a standard 100 bp marker (Quick-Load®, New England BioLabs®).

In-house DNA extraction method
The DNA extraction method used as starting point for the experiments was based on the method reported by Omar et al. [3] who used a modification of the Boom et al. [1] protocols. The in-house DNA extraction method used guanidium thiocyanate to lyse the bacterial cells and silicon dioxidebased material (such as celite) to bind genomic and plasmid DNA [1] . In the 96-well semi-automated system, the celite added a second DNA binding step to the 96-well filter plate to increase the DNA yield.
Briefly the methodology is as follows ( Fig. 1 ): Colilert® Quanti-Tray® broth (200 μ ) was aliquoted into each of the 96-well preparation plate (2 m V-bottom, Greiner bio-one (Cat log no. 780285-914)). To this 700 μ of lysis buffer was added and incubated at 70 °C for 10 min. Thereafter, 250 μ of 100% ethanol was added and incubated at 56 °C for 10 min. Then, 50 μ of celite was added to the mixture and incubated for 10 min at room temperature, a series of celite dilutions was prepared ( Fig. 1 ). The suspension was transferred into the 96-well filter plate (Whatman, Merck Life Science) and vacuum filtered to remove supernatant. The celite pellet was subsequently washed twice with 400 μ of wash buffer followed by vacuum filtration to remove supernatant. Two more wash steps with 400 μ of 70% (v/v) ethanol followed this. The pellet was vacuum dried before an elution plate (Eppendorf, Merck Life Science) was placed below the filter plate. Thereafter, 150 μ of preheated at 56 °C PCR water was added, the pellet was re-suspended and waited for 2 min before vacuum filtration to elute the DNA.

Adapted in-house DNA extraction method
The adaptation of the in-house DNA extraction method was to replace celite solution with a binding buffer. The protocol was thus using 200 μ Colilert® Quanti-Tray® broth to which 700 μ of lysis buffer was added and incubated at 70 °C for 10 min. Thereafter, 200 μ of binding buffer was added and incubated for 10 min at room temperature. The suspension was transferred into the semi-automated 96-well filter plate (Whatman, Merck Life Science) and vacuum filtered to remove supernatant. The remaining protocol was followed as described above.

QIAamp 96 DNA QIAcube HT kit (Qiagen®) commercial kit
Colilert® Quanti-Tray® broth 200 μ was aliquoted into each of the 96-well preparation plate. To this 100 μ of VXL as instructed by the manufacturer was added and incubated for 5 min at room temperature. Thereafter, added 20 μ proteinase K and incubated at room temperature for 15 min, then 350 μ of ACB buffer was added. The suspension was transferred into the 96-well filter plate and vacuum filtered to remove supernatant. Two wash steps with 600 μ of AW1 and AW2 buffer were added separately followed by vacuum filtration. Followed by 600 μ of 96% ethanol and vacuum filtration to remove supernatant. The pellet was vacuum dried before a 96-well elution plate was placed below the 96-well filter plate. Thereafter, 150 μ of PCR water preheated at 56 °C was added, the pellet re-suspended and waited for 2 min before vacuum filtration to elute the DNA.

Experimental design
Originally, the in-house DNA extraction method was optimised to be used with manual spin columns adapted from Borodina et al. [5] . However, we wanted to adapt the in-house DNA extraction method [3] to be used on the semi-automated 96-well vacuum filtration system. To test the method on 96-well filter plates, factors such as filter blocking from celite, binding capacity of the filter (with and without the celite) and potential influence of proteinase K (used in the commercial kit) were tested. This was achieved by testing the following using one bacterial solution with each variable tested with six repeats: (1) Original in-house DNA extraction method with undiluted celite.
(2) In-house DNA extraction method with a series of celite dilutions (1:10; 1:60; 1:420). From the results obtained, a question arose regarding the loading capacity on the filters. A dilution series (1 ml, 800 μ , 600 μ , 400 μ and 200 μ ) of the bacterial suspension removed from the Colilert® Quanti-tray® system to determine the concentration of cells allowed to be filtered before blocking the vacuum filtration system. This experiment was only performed in triplicate using the in-house DNA extraction method with 1:420 celite dilution and the adapted in-house DNA extraction method, with no celite and ethanol.

Statistical analysis
The extracted DNA was quantified using Qubit TM fluorometer (Invitrogen USA) with the Quanti-It kit and reported as μg/m as specified by the manufacturer. Statistical analysis was performed using Graphpad Prism® version 9 and IBM SPSS statistics version 28. The underlying normality (Shapiro-Wilk test) and homogeneity of variances (Levine's test) were tested to allow for further analysis for one-way ANOVA and Dunnett T3 Post-hoc parametric tests. The non-parametric test was used to check for any contradictions to the parametric tests, because there were less than 30 observations per sample. The Kruskal-Wallis tests were used to see if there were significant differences in the mean scores on the dependent variables across the groups. This test is an alternative to one-way ANOVA. The Mann-Witney U test was used to find out where these differences lie. This test is an alternative to post-hoc test.

Results and discussion
Owing to lockdown regulations and difficulties experienced in importing commercial DNA extraction kits for the 96-well semi-automated system, we wanted to optimize the manual in-house DNA extraction method [3] and adapt it for the semi-automated 96-well vacuum filtration system for high throughput of analysis, additionally determine how comparable the in-house DNA extraction method is to commercially available 96-well DNA extraction kits [4] . Currently, with the manual inhouse DNA extraction method, 24 samples can only be analyzed at a time, while with the semiautomated system, 93 samples plus 3 controls can be analyzed.
The results for comparison of the three methods: (A) In-house DNA extraction method [3] ; (B) Adapted in-house DNA extraction method, i.e. removal of celite and ethanol and an inclusion of the binding buffer; (C) Commercial DNA extraction method (Qiagen®) and the series of experiments conducted using the three methods from the enriched river water sample are shown in Fig. 2 . The agarose gel pictures for the PCR from the enriched river water sample are shown in Fig. 3 .
The river water sample was an environmental water sample and could contain the presence of commensal and pathogenic E. coli genes in a complex mixture. The 11-gene E. coli m-PCR confirmed that a variety of these genes could be successfully amplified from an environmental sample showing the presence of atypical Enteropathogenic E. coli (aEPEC), Enterotoxigenic E. coli (ETEC), Enteroaggregative E. coli (EAEC) and the E. coli toxin in the river water sample. The mdh is a malate dehydrogenase gene used as a control to confirm microbiology results in case no pathogenic genes were tested. No false positives or inhibition were indicated in the m-PCR as the external control ( gapdh ) was detected in the samples ( Fig. 3 ).
The nine methods ( Fig. 2 ) were statistically analyzed for the DNA yield as described in Section 1.8. The underlying normality (Shapiro-Wilk test) and homogeneity of variance (Levine's test) were firstly tested because of the small data set. The homogeneity of variance assumptions was not violated, which allowed for further one-way ANOVA analysis ( Table 1 ). In normality the assumptions were not violated; however, non-parametric tests were used to check for any contradictions on the results provided by the parametric tests. From the analysis, non-parametric tests did not contradict the results ( Table 1 ).
When the in-house DNA extraction method and the adapted in-house DNA extraction method were compared to the commercial testing kit (Qiagen®) ( Table 1 ), using one-way ANOVA and Dunnett T3 post-hoc test to determine in which pairs of methods significant differences lie. Highly significant interactions between the in-house DNA extraction method and QIAamp 96 DNA QIAcube HT DNA extraction kit with proteinase K (Qiagen®), except for celite dilution 1:420. This is expected since the celite with no dilution, dilution of 1:10 and 1:60 blocked the wells. No significant interactions between QIAamp 96 DNA QIAcube HT DNA extraction kit with and without proteinase K (Qiagen®); Fig. 2. Box-Whisker plot comparing mean DNA yield obtained with the nine DNA extraction methods. Methods were compared to the to QIAamp 96 DNA QIAcube HT DNA extraction kit with proteinase K (Qiagen®) and statistically significant differences ( p ≤ 0.05) are indicated by "+ " with non-significant difference indicated with "++ ".
in-house celite dilution 1:420 and the three adapted in-house DNA extraction methods ( Table 1 ). Furthermore, the robust test of equality of means was done, using Brown-Forsythe, thus further confirming the F distribution between the groups (20.683) is the same as the one-way ANOVA therefore, no contradictions. Adams [6] stated that data should be presented in a way which would allow the reader to observe the amount of variation inherent to the experiment, for example, mean,  standard deviation (SD) and confidence intervals. Biological systems are subjected to variation and experimental imprecision and some statistics can reveal differences that are not otherwise discernible [7] . The coefficient of variation (CV) was used to measure intra-assay reproducibility from well to well and inter-assay variation from assay to assay, the smaller the CV the better the assay. Therefore, the CV was included for these analyses. The SD and CV from Table 1 indicate that the in-house DNA extraction method with celite dilution 1:420 and the three adapted in-house DNA extraction methods were lower than the QIAamp 96 DNA QIAcube HT DNA extraction kit with proteinase K (Qiagen®).
To determine the loading capacity, a dilution series (1 m , 80 0 μ , 60 0 μ , 40 0 μ and 200 μ ) was removed from the Colilert® Quanti-tray® system to determine the concentration of cells allowed to be filtered before blocking the vacuum filtration system. Fig. 4 indicates the binding capacity for the DNA extraction methods.
The result illustrates that 200 μ ( ∼7.9 × 10 7 cells) and 400 μ broth ( ∼1.4 × 10 8 cells) filters through with no blockages and consistent yields are recovered. In conclusion, the in-house DNA extraction method with 1:420 celite dilution and the three adapted in-house DNA extraction with binding buffer method provided comparable DNA yields to the Qiagen® commercial 96-well DNA extraction kit.
The aim of this method was to develop an inexpensive DNA extraction method using 96-well plates, which is comparable to the available DNA extraction 96-well commercial kit. The method was developed for use with E. coli grown in the selective media of the Colilert® Quanti-Trays for PCR testing as shown in Fig. 3 . In line with this, the DNA concentrations obtained ( ∼30-40 ng) are typical growth observed in the Colilert® Quanti-Trays and are sufficient for the ≤1 μg DNA required in a PCR reaction [8,9] . Additional steps (such as homogenization and DNA precipitation) can be tested if other bacterial enrichments or samples will be used with this method to increase the DNA yield. Several studies have demonstrated variability in their DNA recovery efficiency. Studies have shown the DNA recovery efficiency never exceeded 50% and some of the commercial kits tested only recovered 2% on the input DNA [10,11] .

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.