Enhanced Reproducibility and Precision of High-Throughput Quantification of Bacterial Growth Data Using a Microplate Reader

Leili Abkar; Florentin M. Wilfart; Marta Piercey; Graham A. Gagnon

doi:10.3791/63849

JoVE Journal > Environment

Environnement

Enhanced Reproducibility and Precision of High-Throughput Quantification of Bacterial Growth Data Using a Microplate Reader

Published: July 27, 2022

doi:

10.3791/63849

Leili Abkar¹, Florentin M. Wilfart², Marta Piercey³, Graham A. Gagnon⁴

¹Chemical and Biological Engineering Department,University of British Columbia, ²Process Engineering and Applied Science,Dalhousie University, ³Food Science and Technology,Dalhousie University, ⁴Center for Water Resources Studies, Civil and Resource Engineering,Dalhousie University

Summary

Here, a high-throughput protocol is presented to measure growth data, including growth curves, growth rate, and maximum growth rate. The protocol was verified and validated using two biofilm-producing bacteria. The results and approach applied in this study can be expanded to other high-throughput protocols using microplate readers.

Abstract

This study aimed to develop a repeatable, reliable, high-throughput protocol to monitor bacterial growth in 96-well plates and analyze the maximum growth rate. The growth curves and maximum growth rates of two bacterial species were determined. Issues including (i) lid condensation, (ii) pathlength correction, (iii) inoculation size, (iv) sampling time interval, and (v) spatial bias were investigated. The repeatability of the protocol was assessed with three independent technical replications, with a standard deviation of 0.03 between the runs. The maximum growth rates of Bacillus mycoides and Paenibacillus tundrae were determined to be (mean ± SD) 0.99 h⁻¹ ± 0.03 h⁻¹ and 0.85 h⁻¹ ± 0.025 h⁻¹, respectively. These bacteria are more challenging to monitor optically due to their affinity to clump together. This study demonstrates the critical importance of inoculation size, path length correction, lid warming, sampling time intervals, and well-plate spatial bias to obtain reliable, accurate, and reproducible data on microplate readers. The developed protocol and its verification steps can be expanded to other methods using microplate readers and high-throughput protocols, reducing the researchers' innate errors and material costs.

Introduction

Developing interest in multi-omics manipulation, including mechanism and metabolic studies of bacteria, emphasizes the importance of high-throughput and automated methods such as recording growth data¹^,². Growth data comprising kinetic parameters, such as maximum growth rates, can help characterize bacterial responses to different physical, chemical, and antibacterial conditions. Growth rate data are a standard response variable utilized to uncover potential genotype-phenotypes linkages¹ or indicate the microbial safety and shelf life of food produce³^,⁴. Techniques such as adaptive laboratory evolution⁵^,⁶^,⁷, genome-wide screening, certain chemical assays⁸, and various forward genetic screens⁹ rely on growth rates to evaluate the results.

Optical density (OD) measurements of bacterial cultures are a standard microbiological method to monitor bacterial growth. OD measurements are often recorded at a wavelength of 600 nm, relying on light scattering and the cell density¹⁰^,¹¹. The Beer-Lambert law explains the OD values' dependency on the concentration (i.e., cell density, cell number), path length, and absorptivity coefficient. The geometry and optical system of a spectrophotometer influence the OD readings¹¹. Classical methods of OD measurements can be very time- and labor-intensive, and the data can carry a variety of human errors. In this protocol, a microplate reader is used to decrease the analyst time¹²^,¹³ and the chance of biological contamination. High-throughput analysis using microplate readers is broadly applied in different microbiology areas, such as screening biofilm-producing bacteria¹⁴^,¹⁵, bacterial growth inhibition¹⁶, yeast cell growth¹⁷, the determination of antifungal susceptibility¹⁸, and toxicity screening of nanomaterials¹⁹.

A few researchers have published bacterial growth rate protocols using a microplate reader¹²^,²⁰^,²¹. However, a thorough protocol that examines the reliability of collected data has not been fully established. It is reported that factors such as the type of species²²^,²³^,²⁴ and sealing tapes impact the repeatability due to the oxygen transfer inadequacy in a 96-well plate²⁵^,²⁶. Delaney et al. reported large clusters of Methylorubrum extorquens (wild-type strain) in the growth medium when using a microplate reader, which caused extremely noisy growth data²⁴. The issue was resolved by removing the genes associated with biofilm production²⁴. Due to the secretion of extracellular polymeric substances, biofilm-producing bacteria have a greater affinity to coalesce together and create cell clusters. Therefore, it is more challenging to monitor their growth using light scattering techniques (e.g., spectrophotometers and microplate readers).

This protocol aims to establish steps to obtain reproducible data in a high-throughput method using a microplate reader. Bacillus mycoides and Paenibacillus tundrae were used due to their fast growth and biofilm-producing ability, which are traditionally challenging in manual and automated approaches. Factors such as (i) pathlength correction, (ii) condensation on the lid, (iii) inoculum size, (iv) sampling time interval, and (v) spatial bias were investigated to assess the reliability and reproducibility of the data. This protocol presents steps for accurately monitoring bacterial growth and measuring specific growth rates using a microplate reader.

Protocol

NOTE: All steps in this protocol must be followed in sterile conditions (i.e., between two flames or a biosafety cabinet). All materials and tools are autoclaved for 20 min. See the Table of Materials for details about all materials, equipment, and software used in this protocol. Gloved hands are disinfected, kept wet with hand disinfectant or 70% alcohol solution for at least 1 min, and not removed from the safety cabinet afterward. Otherwise, the disinfecting procedure must be repeated before introducing hands back into the safety cabinet. CAUTION: Ensure the disinfectant is completely evaporated before using an open flame.

NOTE: Two bacteria were isolated from drinking water biofiltration as explained previously²⁷ for their ability to produce biofilm. They were identified by the full-16S rRNA sequencing and submitted to NCBI as Bacillus mycoides (SAMN10518261) and Paenibacillus tundrae (SAMN10452279).

1. Preparing Bacterial Stock in Glycerol

Weigh out 3 g of Tryptic soy broth (TSB) powder and dissolve it in 100 mL of distilled water. Autoclave the broth for 20 min and let it cool down to room temperature. Add 50 mL of TSB to a 50 mL flask.
Transfer the bacteria to agar plates²⁷^,²⁸. Using a loop, pick a colony of bacteria from the agar plate. Add the loop's content into the flask by gently swirling the loop. Incubate the flask overnight at 30 °C ± 1 °C with orbital shaking at 150 rpm.
NOTE: Temperature and shaking need to be adjusted according to the strain type.
Prepare another 50 mL flask with 47.5 mL of TSB. Transfer 2.5 mL of the overnight culture to the flask. Incubate it for 5-6 h to reach OD₆₀₀ = 0.6 0.05 (to get to the mid-exponential phase).
Transfer the culture to a 50 mL sterilized centrifuge tube. Centrifuge the culture at 2,200 × g for 10 min. Decant the broth gently to avoid losing any cells collected at the bottom of the tube.
Add 5 mL of phosphate-buffered saline (PBS) to the same tube and centrifuge at 2,200 × g for 10 min. Repeat this step 3x.
Adjust the OD₆₀₀ of the cells to 0.6 0.05 by using a pipette to titrate an adequate amount of PBS solution. Accurately record the added amount of PBS for future use.
Titrate glycerol to the suspension to reach 20% v/v. Cap the tube and vortex for 15 s.
Transfer 1.5 mL of the suspension to 2 mL cryovials. Put the cryovials in an appropriate box and store them in a freezer at −80 °C.

2. Prepare Overnight Culture.

Let the cryovials with bacterial stock thaw at room temperature for 30 min or until the content is liquid.
NOTE: It is recommended not to accelerate the thawing process.
Prepare a sterilized tube with 10 mL of TSB. Transfer 50 µL of the stock suspension to the tube. Incubate the tube overnight (16-18 h) at 30 °C ± 1 °C with orbital shaking at 150 rpm.

3. Prepare the Inoculum.

Prepare a tube with 10 mL of sterilized TSB. Transfer 500 µL of the overnight culture to the tube. Incubate it for 6-8 h to reach OD₆₀₀ = 0.6 0.05 (or get to the mid-exponential phase).
Transfer the culture to a 15 mL centrifuge tube. Centrifuge the tube at 2,200 × g for 10 min. Decant the broth gently.
Add 5 mL of PBS and centrifuge at 2,200 × g for 10 min. Repeat this step 3x.
Adjust the OD₆₀₀ of the cells to 0.6 0.05 by titrating PBS solution using a pipette. Accurately record the added amount of PBS for future use.

4. Transferring Growth Medium to the Microplate Reader

Choose a transparent, sterile, flat-bottom well plate with a lid (i.e., 96-well plate, 6-well plate, 12-well plate).
Measure the pathlength correction for the microplate using DI water before running the protocol, as explained in step 1.10.
Draw a reference table based on the selected well plate to specify the samples' positions before loading. For example, draw an 8 x 12 table for a 96-well plate.
NOTE: The reference table avoids confusion during the incubation and result collection steps.
Place the following items in a biosafety cabinet: a microplate with a lid, pipettor 1,000 µL and 200 µL, 1,000 µL and 200 µL pipette tips, several cryovials, a 25 mL beaker, the growth medium, and the inoculum.
NOTE: The mentioned items should be transferred to the biosafety cabinet aseptically.
Add 10 mL of growth medium to the beaker. Transfer the calculated amount of inoculum to the beaker with the growth medium.
NOTE: The inoculum ratios in this protocol are 1% and 5%. For example, to achieve a 5% and 1% inoculum ratio, add 500 µL and 100 µL of the inoculum to 9.5 mL and 9.9 mL of the growth medium, respectively.
After adding the inoculum, gently shake the beaker to get a uniform distribution. Dispense 200 µL of the inoculated medium into the designated wells using the reference table.Choose at least three wells to act as controls by adding 200 µL of the growth medium (with no inoculum).
NOTE: Ensure that the control wells are clearly marked in the reference table. If any growth (contamination) is observed in these wells, the experiment should be repeated.
Dispense 200 µL of DI water into the edge wells where the evaporation rate is higher.Place the lid gently before removing the 96-well plate from the safety cabinet. Place the 96-well plate in the plate reader carefully, ensuring no sudden movements.

5. Microplate Reader Settings

Turn on the microplate reader and then use the embedded software to create the customized protocol. In Task Manager, open protocol and choose Create New | Standard Protocol. Open Procédure and adjust the settings as follows:
1. Set the temperature to 30 °C (or the targeted values) using Set Temperature.
2. Click on Incubator On and then set the Gradient à 1 °C.
  NOTE: To avoid unreliable data due to condensation on the lid, ensure that Gradient is set to 1 °C.
3. Choose the Preheat option, ensuring uniform heat distribution along the well plate.
4. Open Start Kinetic and choose the desired incubation duration using the Run time (e.g., 12:00:00 or 24:00:00, etc.) and Interval between readings for 00:30:00 (represents 30 min) or 01:00:00 (represents 1 h).
  NOTE: Reading each 96-well plate takes approximately 1 min.
5. Set the Shaking settings as follows: open Shake; set Shake Mode à Linear; click on Continuous Shake; and adjust the Frequency to 567 cpm (or any desired value).
6. Set the OD reading settings as follows: click on Read | Absorbance; select Endpoint/Kinetic and Monochromators; set Wavelength à 600 nm; and click Validate | Save to save it as a new protocol with the appropriate name and date.

6. Recording Growth Data

Open Task Manager | Read Now.
Choose the saved protocol, then click on OK to begin the protocol.
Save the readings for data analysis.

7. Data Analysis

Transfer the OD₆₀₀ results (growth data) into spreadsheet format. In the results section, click on the image option for each well to observe the final growth curves.
Open the spreadsheet file and rearrange the recorded data based on the reference table. Subtract the mean value of the blank reads from the other reads.
NOTE: Blanks are the OD readings at time zero. In this protocol, OD₆₀₀ values are considered for calculation. Cell density can be used if the calibration curve for OD₆₀₀ vs. cell count is available.

8. Growth Rate Determination

Use the transferred data in the spreadsheet created in step 1.6. Use the method described below and equation (1) to evaluate the growth rate.
(1)
where µ is the growth rate, OD_i is the OD₆₀₀ in each time point, OD_i_-1 is the initial OD₆₀₀ value in the last time point, t_i and t_i_-1 are the time differences between the two mentioned points (e.g., 0.5 for 30 min time intervals, 1 for 1 h time intervals).
Calculate the mean and standard deviation of five successive growth rates using the desired data analysis software. Note that the largest mean with the lowest standard deviation is the maximum growth rate (μ_m) ¹².

9. Determining Spatial Bias

NOTE: Microplate readers and plates introduce bias in results. It is crucial to assess the spatial biases of the plates used in a specific microplate reader to ensure the reliability and reproducibility of the results. To achieve that, use the following steps:

Repeat steps 1.2-1.6.
Record the data and analyze according to step 1.7 and step 1.8 to calculate the growth rate for each well.
Create a heatmap to visualize the growth rate changes in the 96-well plate.
Perform the statistical analysis (see the Supplemental Tables), ensuring no significant differences among the wells.
NOTE: It is essential to use the same type of bacteria and the same inoculum size in all wells.

10. OD reading validation and pathlength correction factor

NOTE: Determining the pathlength correction factor is crucial to validate the data and ensure reliability and validity across different devices.

Use step 1.2 and step 1.3 to prepare the overnight inoculum of B. mycoides.
Prepare three sterilized 50 mL flasks and add 49.5 mL of fresh TSB to each flask. Add 500 µL of the inoculum to each flask in a safety cabinet aseptically. Add 50 mL of non-inoculated TSB to a sterilized flask, which will serve as the control.
Incubate the flasks at 30 °C and 150 rpm. Every 30 min, take samples from each flask aseptically and read and record OD₆₀₀ in the spectrophotometer.
Add 9.9 mL of TSB to a 25 mL beaker. Add 100 µL of inoculum to the beaker and shake gently.
Add 200 µL of the mixture in the beaker to the six wells in a select column in a 96-well plate.Add 200 µL of non-inoculated TSB to three wells as a control. Fill the edge wells with 200 µL of DI water.
Place the lid and gently transfer the 96-well plate to the microplate reader. Follow step 1.5 and step 1.6 to record the growth data.
Plot the OD₆₀₀ from the microplate reader against the OD₆₀₀ from the spectrophotometer. Note that the slope is the offset between the two sets of readings, called the pathlength correction factor.
NOTE: The incubator and microplate reader incubation should be simultaneously done on the same day.

Representative Results

OD reading validation and pathlength correction factor
Split samples of B. mycoides culture were taken at different time points and measured using the microplate reader and the spectrophotometer (Figure 1A). This step was taken to validate the results across different devices. The OD₆₀₀ data correlated but did not match (Figure 1B). The correlation was linear with a slope of 0.55 (95% confidence interval [CI]: 0.53-0.58, R² = 0.99), indicating a constant offset (slope) between the two instruments. The goodness of fit test found the root mean square error (RMSE) and standard error of estimate (Sy.x) to be 0.01, confirming high correlation accuracy.

Figure 1: OD reading validation and pathlength correction factor. (A) OD₆₀₀ data from the microplate reader (squares) are compared with the readings from the spectrophotometer (circles) for the split samples, and (B) the correlation between data from the two instruments is displayed. The dashed line shows the 95% confidence interval for the fitted line. The measured slope is 0.55, which matches the pathlength correction factor of 0.54 ± 0.01 (mean ± SD). Each point is an average of three measurements, and the error bars are standard deviations. Abbreviations: CI = confidence interval; SD = standard deviation; OD₆₀₀ = optical density at 600 nm. Please click here to view a larger version of this figure.

Further investigation of this mismatch revealed the pathlength correction factor as the root cause. The pathlength correction factor was recalculated according to the manufacturer's instructions and was determined to be 0.54 0.01 (mean SD) for the specific sample volume (200 µL), well plate, and lid type. This factor normalizes the OD readings from a microplate reader to correspond to the light absorbance in a standard cuvette (i.e., 1 cm path length). This step shows the absolute necessity to use the pathlength correction factor to obtain accurate OD readings. It is noteworthy that the factor can be applied retrospectively to correct results.

Spectrophotometers are based on horizontal photometry, which depends on the physical geometry of the cuvette. In contrast, microplate readers are based on vertical photometry, which varies by the sample volume and, thus, the solution height. The microplate reader requires a pathlength correction to produce comparable results consistent with the literature¹¹. Uncorrected data from the microplate reader would underestimate the maximum growth rate and OD₆₀₀ readings.

Lid condensation
Another issue encountered was condensation buildup on the 96-well plate lid. This influenced each well's pathlength, which caused irregular jumps in OD readings and, thus, noisy data. Lid warming of 1 °C above the sample temperature (30 °C) resulted in substantial noise reduction (see Figure 2A). Significant condensation buildup was visible after 24 h without lid warming, as depicted in Figure 2B compared to Figure 2C using the lid-warming step included in the protocol.

Figure 2: Lid condensation. (A) Data represent the OD readings with lid condensation buildup (squares) and after setting the lid-warming step (triangles). The standard deviation is shown as the band on the square symbols and as error bars on the triangular symbols.(B) Condensation buildup on the lid during 24 h incubation. (C) No condensation was observed when using a temperature gradient of 1 °C over 24 h of incubation on the lid. Abbreviation: OD = optical density. Please click here to view a larger version of this figure.

Hart et al. also observed heavy condensation using a 96-well plate to study time-lapse fluorescence microscopy, despite incubating it in a temperature-controlled environment²⁹. They resolved this by warming the lid 0.68 °C ± 0.44 °C above the sample temperature with an additional heating plate²⁹. In the current study, a temperature graduation of 1 °C in the lid-warming step inhibited the accumulation of condensation during growth data collection.

Inoculation size
The 1% and 5% inoculum sizes directly impacted the standard deviation and error between OD₆₀₀ readings in the microplate reader. The fluctuation is higher at the end of both species' log and stationary phases. A larger inoculation size (i.e., 5%) (Figure 3A,B) and the affinity of the bacteria to produce biofilm could result in dense biofilm clusters, falsely reflecting on the sample OD₆₀₀ readings. Specifically, the OD₆₀₀ readings may become unreliable for sensitive species as time progresses. Cells can agglomerate and clog the light path, resulting in highly noisy data. It is expected to observe a similar impact with a higher inoculation size for other bacteria. This study subsequently utilized a smaller inoculation size (i.e., 1%) to resolve this issue (Figure 3C,D).

Figure 3: Growth curves for two species. Growth curves for two species are shown, with 5% inoculation and a 1 h time interval: (A) Paenibacillus tundrae and (B) Bacillus mycoides, and with 1% inoculation and a 30 min sampling time interval: (C) P. tundrae and (D) B. mycoides. The three runs, including run 1 (circle), run 2 (square), and run 3 (triangle), represent separate technical replicates. The x-axis shows growth time in hours, and the y-axis represents OD values at 600 nm. Each point averages six measurements. Error bars represent the standard deviation. The growth curves with 1% inoculation have a lower standard deviation, and each point is consistently within the standard deviation. Abbreviation: OD = optical density. Please click here to view a larger version of this figure.

Sample reading interval
A high enough temporal resolution is required to ensure accurate calculation of the maximum growth rate using the analytical method in step 1.7. Reading intervals must be set so that the log phase readings include more than five time points. As shown in Figure 3A,B, 1 h reading intervals barely contain five-time points in the log phases for both species. In comparison, 30-min time intervals (Figure 3C,D) provide sufficient data to identify the highest slope in the log phase. A 1 h reading interval underestimates the specific growth rate as it cannot capture the highest slope. For detailed results, see Supplemental Table S1.

Growth data
Applying pathlength correction, lid heating, an inoculation size of 1%, and 30 min reading intervals, the growth curves for B. mycoides and P. tundrae were recorded. The results are depicted in Figure 3C,D. The X- and Y-axes show time (h) and OD₆₀₀ values, respectively, collected automatically by the microplate reader. Each of the three runs represents a technical replication on a different date.

The growth rates of the two species in each run were calculated and presented in Figure 4A,B. To assess the repeatability between runs, the growth rates in all runs were compared and were not significantly different (non-parametric ANOVA, Kruskal-Wallis, P > 0.05). A post-hoc Dunn's test also showed no significant differences between any two runs. This demonstrates the repeatability and accuracy of the protocol to obtain reliable results for high biofilm-producing bacteria. Additionally, a standard deviation of 0.03 for each run proved the repeatability of this protocol. See Supplemental Table S2, Supplemental Table S3, and Supplemental Table S4 for detailed results.

Figure 4: Maximum growth rate (h⁻¹) in three separate runs. The maximum growth rate (h⁻¹) is shown in three different runs for (A) Bacillus mycoides and (B) Paenibacillus tundrae. The maximum growth rate for each run is represented as mean and 95% CI. Comparing the maximum growth rates using the Kruskal-Wallis test showed no significant differences, B. mycoides with P = 0.29 and P. tundrae with P = 0.69, demonstrating repeatability between the runs. (C) Comparing the growth rates (h⁻¹) of P. tundrae (circles) and B. mycoides (squares) shows a significant difference (Mann-Whitney, P < 0.0001). Abbreviation: CI = confidence interval. Please click here to view a larger version of this figure.

The maximum growth rates for the two species were significantly different (Mann-Whitney, P < 0.0001) (Figure 4C). This illustrates the protocol's ability to differentiate bacterial growth rates even for biofilm-producing species with high sensitivity. Averaging the growth rate data from technical and biological replications resulted in a maximum growth rate (h⁻¹) of 0.99 h⁻¹ ± 0.03 h⁻¹ for B. mycoides and 0.85 h⁻¹ ± 0.025 h⁻¹ for P. tundrae. These results align with a maximum growth rate of 1.08 h⁻¹ for B. mycoides (30 °C) in the literature³⁰.

Figure 3A,B shows that the obtained growth curves in three separate runs, using 1% inoculum size, were relatively similar and within their collective standard deviation for both bacterial species. Researchers have investigated the impact of the inoculum preparation on the growth rate variability¹². Hall et al. suggested that variability of ±2.5% for 6-12 replicates/experimental run is reasonable, which aligns with the variability in this study¹². Mira et al. suggested a V-factor of <0.053 as acceptable variability in a trial³¹. They defined the V-factor as the standard error of the mean divided by the mean. A lower V-factor reflects a low variability within a run³¹.

Spatial bias
Another consideration is the spatial bias of wells distributed in a 96-well plate. A culture of P. tundrae was used to measure the bias between wells. Figure 5 shows the heatmap of the growth rate distribution of the mentioned culture in a 96-well plate. The edge wells were filled with distilled water since the evaporation rate is higher there than in the non-edge wells²⁰. Changing color intensity represents varying growth rate values.

Figure 5: Spatial bias of the 96-well plate. The spatial bias of the 96-well plate is shown. A culture of Paenibacillus tundrae is distributed in all wells except the edge wells. The growth rate is calculated and displayed in a heatmap. The color graduation and intensity are on the scale adjacent to the heatmap. Please click here to view a larger version of this figure.

At first glance, the location of the well seems to have a significant influence on the obtained growth rate results. However, when comparing the mean of each column, there was no significant difference (non-parametric ANOVA, P = 0.089, alpha = 0.05). The detailed results for each column and statistical results of the post comparison test are shown in Supplemental Table S5.

Literature discussing the spatial bias of a 96-well plate is sparse. However, Kurokawa and Ying examined the spatial bias of a 96-well plate while developing a growth rate procedure²⁰. They observed variation and suggested using multiple wells at various locations to consider the spatial bias. This study confirmed the spatial variation in a 96-well plate. However, the columns' mean growth rates were not significantly different. Thus, it is recommended to use the same column to ensure repeatability.

Figure 6: Fast, reliable, and high-throughput quantification of bacterial growth. Please click here to view a larger version of this figure.

Supplemental Table S1: Growth rates for two isolated species with two different time intervals. Please click here to download this Table.

Supplemental Table S2: Post-hoc multiple comparisons for two species in three separate runs with 1% inoculum and a sampling time interval of 30 min. Please click here to download this Table.

Supplemental Table S3: Post-hoc multiple comparisons for two species in three separate runs with 1% inoculum and a sampling time interval of 30 min. Please click here to download this Table.

Supplemental Table S4: Growth rate (h⁻¹) measured for each species in three separate runs in two forms of the mean (95% confidence interval and ± standard deviation) using an inoculum size of 1% and a 30 min OD reading time interval. Please click here to download this Table.

Supplemental Table S5: Post multiple comparison test between columns to investigate the spatial bias for P. tundrea. Please click here to download this Table.

Discussion

Microplate readers allow for obtaining consistent and repeatable growth rates. This technology minimizes human error and enables high-throughput sampling. The small amount of culture required per sample makes this approach an attractive, low-cost alternative to cell counts using flasks or test tubes. Microplate readers allow a large sample size, increasing the statistical power and subsequently facilitating reliable growth rate calculations while keeping costs and labor low.

This article presents a detailed protocol for measuring and analyzing growth data. The protocol was verified with B. mycoides and P. tundrea. These bacteria were isolated from a full-scale drinking water biofilter due to their ability to produce biofilm, making the growth data assessment more challenging.

Several significant factors for reliable and accurate measurements were identified: pathlength correction, lid condensation, inoculum size, reading interval, and spatial bias (Figure 6). Pathlength correction is of essential importance for accurate results and can be determined and applied automatically or retrospectively. Data verified by cross-calibrating with a spectrophotometer showed the importance of pathlength correction for a specific sample volume, lid, and well plate type.

Lid condensation during data collection led to erroneous and noisy results. It was resolved by raising the lid 1 °C above the incubation temperature. With this protocol, it is crucial to consider the inoculum size. A too large inoculum size led to erratic readings when density increased during the incubation. This is likely due to the light path sensitivity to the sample’s clustered areas.

Reading intervals impacted the growth rate calculation. The reading intervals need to be frequent enough to capture more than five points in the log phase when using the numerical method described. Otherwise, the resultant growth rate will be underestimated and inaccurate. Thus, accurate and reliable growth rate calculation requires adjusting the protocol to the appropriate reading intervals. It is highly recommended to try higher time interval frequencies (e.g., 20 min, 10 min) when using numerical growth rate models to capture changes with higher resolution and eliminate the risk of under-sampling. The spatial bias showed that the columns’ growth rates were not significantly different. This finding necessitates evaluating the spatial bias for the applied well microplate, lid, and microplate.

Although the repeatability of data in different runs was investigated and showed no significant differences in growth rates, it is worthwhile to rerun the experiments to account for between-run variations. While this protocol has considered several factors for undefined bacteria, other parameters such as batch variation of undefined growth media or the age of the prepared media should be considered and evaluated.

Divulgations

The authors have nothing to disclose.

Acknowledgements

This work was funded by the Natural Sciences and Engineering Research Council (NSERC) / Halifax Water Industrial Research Chair in Water Quality and Treatment (Grant No. IRCPJ 349838-16). The team of authors also would like to acknowledge the help of Anita Taylor in reviewing this article.

Materials

Centrifuge	Eppendorf	5810 R
Centrifuge tubes – 15 mL	ThermoFisher- Scientific	339650	Sterile
Centriguge tubes – 50 mL	ThermoFisher- Scientific	339652	Sterile
Disposable inoculating loop , 10 µL	Cole-Parmer	UZ-06231-08	Sterile
Erlenmeyer flasks – 250 mL	Cole-Parmer	UZ-34502-59	Glass
Isopropanol	ThermoFisher- Scientific	396982500	≥99.0
Phosphate Buffer Saline	Sigma-Aldrich	P4417
Pipett tips 1,000 µL	ThermoFisher- Scientific	UZ-25001-76
Pipett tips 10 mL	ThermoFisher- Scientific	UZ-25001-83
Pipett tips 200 µL	ThermoFisher- Scientific	UZ-25001-85
Pipett tips 5 mL	ThermoFisher- Scientific	UZ-25001-80
Pipettor 1,000 µL	Cole-Parmer	UZ-07909-11
Pipettor 10 mL	Cole-Parmer	UZ-07909-15
Pipettor 200 µL	Cole-Parmer	UZ-07909-09
Pipettor 5 mL	Cole-Parmer	UZ-07859-30
Tryptic Soy Broth	Millipore	22091	Suitable for microbiology

References

Reuß, D. R., et al. Large-scale reduction of the Bacillus subtilis genome: Consequences for the transcriptional network, resource allocation, and metabolism. Genome Research. 27 (2), 289-299 (2017).
Sparkes, A., et al. Towards Robot Scientists for autonomous scientific discovery. Automated Experimentation. 2 (1), (2010).
Zwietering, M. H., Jongenburger, I., Rombouts, F. M., van t Riet, K. Modeling of the bacterial growth curve. Applied and Environmental Microbiology. 56 (6), 1875-1881 (1990).
Pla, M., Oltra, S., Esteban, M., Andreu, S., Palop, A. Comparison of primary models to predict microbial growth by the plate count and absorbance methods. BioMed Research International. 2015 (6), 1-14 (2015).
Choe, D., et al. Adaptive laboratory evolution of a genome-reduced Escherichia coli. Nature Communications. 10 (1), 935 (2019).
Dykhuizen, D. E., Dean, A. M. Enzyme activity and fitness: Evolution in solution. Trends in Ecology and Evolution. 5 (8), 257-262 (1990).
McDonald, M. J. Microbial experimental evolution – A proving ground for evolutionary theory and a tool for discovery. EMBO Reports. 20 (8), 1-14 (2019).
Lum, P. Y., et al. Discovering modes of action for therapeutic compounds using a genome-wide screen of yeast heterozygotes. Cell. 116 (1), 121-137 (2004).
Cagnon, C., et al. Development of a forward genetic screen to isolate oil mutants in the green microalga Chlamydomonas reinhardtii. Biotechnology for Biofuels. 6 (1), 178 (2013).
Stevenson, K., McVey, A. F., Clark, I. B. N., Swain, P. S., Pilizota, T. General calibration of microbial growth in microplate readers. Scientific Reports. 6 (1), 38828 (2016).
Matlock, B. C., Beringer, R. W., Ash, D. L., Page, A. F., Allen, M. W. Differences in Bacterial Optical Density Measurements between Spectrophotometers. Technical Note. ThermoScientific. , (2011).
Hall, B. G., Acar, H., Nandipati, A., Barlow, M. Growth rates made easy. Molecular Biology and Evolution. 31 (1), 232-238 (2014).
Rolfe, M. D., et al. Lag phase is a distinct growth phase that prepares bacteria for exponential growth and involves transient metal accumulation. Journal of Bacteriology. 194 (3), 686-701 (2012).
Djordjevic, D., Wiedmann, M., Mclandsborough, L. A. Microtiter plate assay for assessment of Listeria monocytogenes biofilm formation. Applied and Environmental Microbiology. 68 (6), 2950-2958 (2002).
O’Toole, G. A. Microtiter dish biofilm formation assay. Journal of Visualized Experiments. (47), e2437 (2011).
Campbell, J. High-throughput assessment of bacterial growth inhibition by optical density measurements. Current Protocols in Chemical Biology. 3 (3), 1-20 (2012).
Toussaint, M., Conconi, A. High-throughput and sensitive assay to measure yeast cell growth: A bench protocol for testing genotoxic agents. Nature Protocols. 1 (4), 1922-1928 (2006).
Goughenour, K. D., Balada-Llasat, J. -. M., Rappleye, C. A. Quantitative microplate-based growth assay for determination of antifungal susceptibility of Histoplasma capsulatum yeasts. Journal of Clinical Microbiology. 53 (10), 3286-3295 (2015).
Qiu, T. A., et al. Growth-based bacterial viability assay for interference-free and high-throughput toxicity screening of nanomaterials. Analytical Chemistry. 89 (3), 2057-2064 (2017).
Kurokawa, M., Precise Ying, B. -. W. Precise, high-throughput analysis of bacterial growth. Journal of Visualized Experiments. (127), e56197 (2017).
Bredit, F., Romick, T. L., Fleming, H. P. A rapid method for determination of bacterial growth kinetics. Journal of Rapid Methods and Automation in Microbiology. 3, 59-68 (1994).
Delaney, N. F., et al. Development of an optimized medium, strain and high-throughput culturing methods for Methylobacterium extorquens. PLOS ONE. 8 (4), 62957 (2013).
Haire, T. C., et al. Robust microplate-based methods for culturing and in vivo phenotypic screening of Chlamydomonas reinhardtii. Frontiers in Plant Science. 9, 235 (2018).
McBirney, S. E., Trinh, K., Wong-Beringer, A., Armani, A. M. Wavelength-normalized spectroscopic analysis of Staphylococcus aureus and Pseudomonas aeruginosa growth rates. Biomedical Optics Express. 7 (10), 4034 (2016).
Sieben, M., Giese, H., Grosch, J. H., Kauffmann, K., Büchs, J. Permeability of currently available microtiter plate sealing tapes fail to fulfil the requirements for aerobic microbial cultivation. Biotechnology Journal. 11 (12), 1525-1538 (2016).
Zimmermann, H. F., John, G. T., Trauthwein, H., Dingerdissen, U., Huthmacher, K. Rapid evaluation of oxygen and water permeation through microplate sealing tapes. Biotechnology Progress. 19 (3), 1061-1063 (2003).
Abkar, L., Gagnon, G. A. Biological responses to P-limitation in indigenous bacteria isolated from drinking water. AWWA Water Science. 3 (5), 1248 (2021).
Sanders, E. R. Aseptic laboratory techniques: Plating methods. Journal of Visualized Experiments. (63), e3064 (2012).
Hart, S. F. M., Skelding, D., Waite, A. J., Burton, J. C., Shou, W. High-throughput quantification of microbial birth and death dynamics using fluorescence microscopy. Quantitative Biology. 7 (1), 69-81 (2019).
Periago, P. M., Abee, T., Wouters, J. A. Analysis of the heat-adaptive response of psychrotrophic Bacillus weihenstephanensis. International Journal of Food Microbiology. 79 (1-2), 17-26 (2002).
Mira, P., Barlow, M., Meza, J. C., Hall, B. G. Statistical package for growth rates made easy. Molecular Biology and Evolution. 34 (12), 3303-3309 (2017).