Bacterial isolates and mucus sample collection
Glycerol stocks of P. multocida, M. haemolytica, and H. somni isolates were obtained from Purdue University’s Indiana Animal Disease Diagnostic Laboratory (ADDL), as previously described [16]. These isolates were originally cultured by ADDL as a part of routine diagnostic testing from lung/nasopharyngeal sample submissions and identified using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) analysis. Mucus samples were obtained from steers (n = 5) approximately 12 months of age and 600 lbs. in weight that had not been given antibiotics for at least 100 days at a feedlot in Indiana (Purdue Animal Care and Use Committee Approval #1906001911) using rayon-tipped sterile double swabs designed for general specimen laboratory use (BD 220135, Becton, Dickinson, and Company, Franklin Lakes, NJ, USA). Each animal was restrained in a livestock handling chute and the animal head was restrained to minimize movement. The nostrils were wiped with paper towels to remove excess mucus. One double swab was inserted into both nostrils sequentially at a depth of approximately 5 cm. The swabs were swirled in tubes with 200 µL of DNA-free water and aliquoted for use in LAMP reactions for both on-farm and in-lab settings.
Bacterial DNA isolation and quantification
P. multocida, M. haemolytica, and H. somni were isolated and final DNA concentrations were quantified according to Mohan et al. [16]. Specifically, all pathogens were streaked on tryptic soy agar plates supplemented with defibrinated sheep blood (blood agar). P. multocida and M. haemolytica were incubated aerobically at 37 °C for 16–18 h while H. somni was incubated in a 5% CO2 atmosphere at 37 °C for 2–3 days by using BD GasPak™ EZ container systems (BD 260672) with BD BBL™ CO2 gas generators (BD 260679). Individual colonies of each bacterial species were picked from the blood agar plates. P. multocida and M. haemolytica were inoculated into brain–heart infusion (BHI) broth and H. somni was inoculated into tryptic soy broth (TSB). All were incubated in the same conditions as the plates.
DNA extraction of each pathogen was carried out by taking 2 mL of saturated liquid culture and following the PureLink™ Genomic DNA Mini Kit (Catalog #K182002, Invitrogen, Waltham, MA, USA) procedure. Final DNA concentrations (ng/µL) of eluted extracts were measured using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen P11496).
Colorimetric quantitative LAMP assay (qLAMP)
The colorimetric assay was conducted by modifying the previously published procedure [16]. Specifically, in the colorimetric assay, the New England Biolabs’ Warmstart Colorimetric LAMP 2 × Master Mix was used. The mix was coupled with Antarctic Thermolabile uracil DNA glycosylase (UDG) and deoxyuridine triphosphate (dUTP) to minimize carryover contamination throughout the experiment. In-house validation experiments have confirmed that UDG/dUTP does not affect reaction performance at the concentration used. The LAMP solution comprised 12.5 µL of this mix (40 mM Tris–HCl, 20 mM (NH4)2SO4, 100 mM KCl, 16 mM MgSO4, 2.8 mM dNTPs,0.28 µM dUTP, 0.64 U/µL Warmstart® Bst 2.0 DNA polymerase, 0.6 U/µL Warmstart® Reverse Transcriptase [RTx], 4 × 10–4 U/µL Antarctic Thermolabile UDG, 200 mM Phenol red, 0.2% Tween 20, pH 8.8@25 °C) (Catalog # M1800L, New England Biolabs, Ipswich, MA, USA), 2.5 µL of a 10 × LAMP primer mixture (10× concentration: 2 µM F3, 2 µM B3, 4 µM LF, 4 µM LB, 16 µM FIP, 16 µM BIP), 5 µL of DNA-free water, and 5 µL of DNA or mucus containing solution. Reactions were pipetted into wells of clear 96-well FrameStar® skirted flat optical bottom PCR plates (Catalog #1149V67, Thomas Scientific, Swedesboro, NJ, USA). Wells were sealed with adhesive PCR plate seals (Catalog #AB-0558, Thermo Fisher Scientific Waltham, MA, USA) and inserted into a CLARIOstar Plus (BMG Labtech Cary, NC, USA), a multi-mode plate reader with temperature control, for real-time colorimetric measurement. Spectra were collected from 350 to 750 nm with a step size of 5 nm for 60 cycles lasting approximately 60 s each. Reaction plates were incubated at 65 °C using the CLARIOstar Plus.
Each step in LAMP preparation (primer addition, template DNA loading, and reaction incubation/measurement) was conducted in separate lab spaces in order to minimize false positives due to amplicon aerosol contamination. To further reduce contamination, RNase AWAY™ Surface Decontaminant (Thermo Fisher Scientific 14-754-34) was thoroughly applied to all working surfaces and lab gloves before and after use and wiped completely with Kimwipes to prevent residue formation.
Data analysis
Absorbance measurements for each minute at 430, 560, and 620 nm wavelengths were extracted, and the data were normalized using the formula (Equation 1):
$${\text{Colorimetric}}\,{\text{absorbance}}\,{\text{ratio}} = \frac{{{\text{Absorbance at }}\,430{\text{ nm}} - {\text{Absorbance at }}\,620{\text{ nm}}}}{{{\text{Absorbance at }}\,520{\text{ nm}} - {\text{Absorbance at }}\,620{\text{ nm}}}}$$
(1)
The absorbance at 620 nm was used as a baseline, and the 430 nm and 520 nm wavelengths were used to mark the change in color of phenol red from red to yellow. The resulting ratios were plotted against time in Microsoft Excel.
Colorimetric threshold
A one-to-one mixture of pH 7.2 phosphate-buffered saline (PBS, Thermo Fisher Scientific 20012050) and pH 8.5 Tris–HCl (Catalog #SD8141, Bio Basic Amherst, NY, USA) was prepared. Using 0.1 M NaOH and 0.1 M HCl, the solution was adjusted to range between pH 6.0 and 8.0 with increments of approximately 0.2 pH units. 5 µL of the buffer solution were added to 12.5 µL of Warmstart® Colorimetric LAMP 2 × Master Mix and 7.5 µL of DNA-free water. Each condition was added to a 96-well FrameStar® skirted flat optical bottom PCR plate in triplicate, sealed with a PCR film, and inserted into the CLARIOstar Plus to obtain measurements for three minutes. Data were analyzed as explained above and the values at each minute were averaged. From plotting the data, a colorimetric absorbance ratio was selected as the threshold according to the color changes observed so that colorimetric absorbance ratios above the threshold were considered positive and colorimetric absorbance ratios below the threshold were considered negative.
An image of the plate was taken using an Epson Perfection V800 Photo scanner (Catalog #B11B223201, Amazon, Seattle, WA, USA) with settings set to professional mode, 48-bit color image type, and 720 dpi resolution. The image of the sample closest to the threshold value was processed using ImageJ to find the RGB values. Those RGB values were in turn used to calculate Hue Saturation Values (HSV) values. The Hue threshold was used to determine positive (above Hue threshold) versus negative (below Hue threshold) results in other assays.
Primer screening and limit of detection (LOD)
Colorimetric absorbance ratios were obtained from qLAMP experiments using a 2 × DNA dilution factor (10 000–78.125 copies of DNA per reaction). Note that we are including the fraction of copies simply to indicate the dilution factor, the actual number of copies in a reaction would be rounded to the nearest whole number. All primer sets in Additional file 1 were tested and each concentration of the template included three replicates for each primer set. DNA-free water was used as a control (no-template control, NTC).
Primer sets were scored by annotating the number of sufficient amplification reactions—defined as any replicate whose colorimetric absorbance ratio at 60 min was greater than 3.0—for each template concentration [including (NTC)] for each primer set. Any replicate that was deemed as sufficient amplification in the NTC was designated as a false positive. Any missing data for an entire template concentration was set at a constant value equal to the maximum colorimetric absorbance ratio observed across all primer sets at all concentrations. In contrast, any missing data for any given time point was filled with the value of the previous time point.
Primer sets were scored by first calculating the maximum colorimetric absorbance ratio and reaction time for each replicate at each template concentration (excluding NTC) for a given primer set. The average and standard deviation of these values were then calculated for each template concentration for a given primer set. Reaction time was defined as the first time point at which the absorbance ratio was greater than 3.0. For each primer set, the average of each one of these four metrics (average and standard deviation of maximum intensity and reaction time) was calculated across all template concentrations to assign a primer set metric (e.g., primer set average maximum colorimetric absorbance ratio). The LOD for each concentration was then calculated as the minimum template concentration where all replicates sufficiently amplify and all replicates of template concentrations above this minimum template concentration also sufficiently amplify. For cases where all replicates for all tested template concentrations amplified, the LOD was set at the lowest non-zero template concentration if there were less than three false positives. If all NTC reactions amplified (i.e., three false positives) or no replicates amplified at any template concentration, the LOD was set at −1.
For the overall scoring of primer sets, ineligible primer sets (as designated by an LOD of −1) were automatically assigned an overall score of 0 and withdrawn from further scoring. All eligible primer sets were then assigned a weighted overall score, Sk, for a primer set, k, using the following expression:
$$S_{k} = w_{{\overline{I}}} \cdot \left( {1 - \frac{{\max \left( {\overline{I}} \right) - I_{k} }}{{Range\left( {\overline{I}} \right)}}} \right) + \mathop \sum \limits_{n} w_{n} \cdot \left( {1 - \frac{{\min \left( n \right) - n_{k} }}{Range\left( n \right)}} \right)$$
$$n \in \left( {\sigma \left( I \right), \,\overline{{t_{rxn} }} ,\,\sigma \left( {t_{rxn} } \right), \,LOD,\,FP } \right)$$
$$w_{n} = \left[ {\begin{array}{*{20}c} {35} \\ 5 \\ {30} \\ 5 \\ 5 \\ {20} \\ \end{array} } \right]for \begin{array}{*{20}c} {\overline{I}} \\ {\sigma \left( I \right)} \\ {\overline{{t_{rxn} }} } \\ {\sigma \left( {t_{rxn} } \right)} \\ {LOD} \\ {FP} \\ \end{array}$$
where \(\overline{I},\sigma \left( I \right), \overline{{t_{rxn} }} ,\sigma \left( {t_{rxn} } \right), LOD,FP\) is the set average maximum colorimetric absorbance ratio, set standard deviation of the maximum colorimetric absorbance ratio, set average reaction time, set standard deviation of the reaction time, set LOD, and the number of false positives for a given primer set, respectively. The range defined above is the maximum value minus the minimum value for a given set metric across all eligible primer sets. If the range for a given set metric was 0 (i.e., all primer sets had the same value), that set was given the full weight allotted for that set metric.
Primers with the highest scores were selected as the best primer sets to detect the bacteria of interest using a python script (Additional file 2). The worst LOD for the three selected primers was set as the LOD to be used in other experiments. Images of the plates were taken at 0 and 60 min using the Epson Perfection V800 Photo scanner.
Combinatorial experiment with one, two, or three species spiked into water
Colorimetric qLAMP assays were performed for 60 min using 1250 copies per reaction of gDNA of one, two, and/or three bacteria P. multocida, M. haemolytica, and H. somni in the same reaction. Each condition was repeated nine times in nine separate wells of 96-well plates. Images of the plates were taken at 0 and 60 min using the Epson Perfection V800 Photo scanner. Colorimetric absorbance ratios were calculated as explained above, and the resulting data were plotted against time. The finalized data were also analyzed in receiver operating characteristic (ROC) curves by using the colorimetric threshold previously determined and assessing positive versus negative reactions for each primer set. The highest number obtained from subtracting the false positive rate (Equation 2) from the true positive rate (Equation 3) was selected as the time threshold for that specific primer.
$${\text{False}}\,{\text{positive}}\,{\text{rate}}\,({\text{FPR}}) = \frac{{{\text{no}}{.}\,{\text{of}}\,{\text{false}}\,{\text{positives}}}}{{{\text{no}}{.}\,{\text{of}}\,{\text{false positives}} + {\text{no}}{.}\,{\text{of}}\,{\text{true}}\,{\text{negatives}}}}$$
(2)
$${\text{True}}\,{\text{positive}}\,{\text{rate}}\,({\text{TPR}}) = \frac{{{\text{no}}.{\text{ of true positives}}}}{{{\text{no}}.{\text{ of true positives }} + {\text{no}}.{\text{ of false negatives}}}}$$
(3)
Precision cooker experiments (on-farm and in-lab)
LAMP reactions were prepared in individual domed PCR tubes (Thermo Fisher Scientific AB0337) using 12.5 µL New England Biolabs’ Warmstart® Colorimetric LAMP 2 × Master Mix, 2.5 µL of primer mix, 5 µL of DNA free water, and 5 µL of mucus sample. An Anova Culinary AN500-US00 Sous Vide Precision Cooker (Amazon B08CF6Y4WF) was filled with water and set to 149 °F (65 °C). The temperature of the water was verified in the lab using an Hti HT-04 Thermal Imaging Camera (Additional file 3). The tubes were submerged in the water on the right side (the region with a relatively homogenous temperature of 65 °C) either by taping them to the inside of the precision cooker with heat-resistant 3/4-inch autoclave tape (Thermo Fisher Scientific 15904) or by using PCR tube holders designed and 3D-printed in-lab with a Formlabs Form 3B 3D printer (Formlabs, Somerville, MA, USA) using high-temperature resin v2 and 0.1 mm layer thickness (Additional files 4 and 5). The tubes were removed from the water after 60 min.
The experiment was performed in-lab using the usual procedures to avoid contamination (RNase AWAY™ spray, separation of lab spaces, etc.) and on-farm. For the on-farm experiment, the reagents were prepared in the lab, and the addition of mucus was done on-farm using a 0.5–10 µL single-channel pipette with no additional measures to avoid contamination (Additional file 6). The mucus addition on-farm happened no more than 30 min after extraction from the steers, while the mucus addition in the lab was done 4 days after collecting the samples (the samples being stored at −80 °C in the meantime). The samples were stored in water so that the test matrix would be similar in the lab and on the farm.
Images of the tubes were taken at 0 and 60 min. Images of the tubes in-lab were taken using the Epson Perfection V800 Photo scanner and images of tubes in-farm were taken using a Samsung Galaxy A50. All images obtained were adjusted by using the white balance tool on Adobe Lightroom to obtain a relatively uniform background. The RGB values of each solution were extracted at 60 min using ImageJ and Hue values were calculated to differentiate positive and negative results. Shadows and glows on the images were avoided during this process to increase the accuracy of the results. The Hue scale indicated on a color wheel from 0° to 360°. Red/pink color is around 0–15° and 345–360°, orange/yellow is around 30–60°. Since we set a Hue value of 35 as cut-off (higher than 35 is a positive reaction), the red/pink color on the high end (close to 360°) was simply set to 0 to avoid confusion.
When comparing the LAMP farm results with PCR, having 2 out of 3 LAMP reactions show the same result as PCR was considered agreement.