通过喷墨打印机辅助的自动化棋盘阵列法与手工时间-杀菌法进行抗菌协同作用测试

摘要

抗菌药物协同作用测试用于评估两种或多种抗生素联合使用的效果，通常采用以下两种方法之一进行：棋盘格稀释法或时间-杀菌试验。本文介绍了一种基于自动喷墨打印机辅助的棋盘格稀释法协同作用技术，以及一种经典的时间-杀菌协同作用研究方法。

摘要

随着多重耐药（MDR）病原体的发病率持续上升，且超过了新型抗菌药物的研发速度，针对多重耐药细菌的治疗亟需新的策略。其中一种策略是联合用药，即联合使用两种或多种抗生素来治疗单药可能无效或对其中一种或两种药物均无效的感染。当两种药物联用产生的效应大于各自效应之和时，即被视为具有协同作用。体外协同活性的研究是评估药物组合潜在疗效的重要第一步。目前已有两种主要的体外协同作用检测方法：棋盘格法和时间-杀菌试验。本文介绍了一种利用喷墨打印技术实现自动化的棋盘格法，可提高该技术的效率与准确性，同时介绍了一种标准的手动时间-杀菌协同试验方法。自动化棋盘格法可用于高通量筛选，而手动时间-杀菌试验则可提供关于协同活性及杀菌效果的额外补充数据。

棋盘法阵列是标准最低抑菌浓度（MIC）检测的一种改进方法，该方法将细菌与不同浓度组合的抗生素共同孵育，并在过夜孵育后评估其生长抑制情况。手动进行棋盘法阵列实验需要一系列繁琐且易出错的计算和稀释步骤。本文介绍的自动化方法利用喷墨打印技术，实现了所需抗生素储备液体积的自动计算与分配。在时间-杀菌协同试验中，将细菌与目标抗生素单独或联合孵育，并在24小时过程中定时取样进行定量培养。该实验结果可判断抗生素组合是否具有协同作用，是否具有杀菌效应，并提供随时间变化的细菌抑制和杀灭数据。

引言

多重耐药（MDR）细菌病原体的传播，特别是碳青霉烯类耐药肠杆菌科（CRE）等多重耐药革兰阴性菌，使临床医生在成功实施抗感染治疗方面面临越来越有限的选择¹，而新型抗菌药物研发进展缓慢进一步加剧了这一问题^2,3。抗菌协同作用是指两种药物联用时产生的效应大于各自单独作用之和，这种机制为挽救现有抗生素用于治疗多重耐药菌感染提供了可能，即使这些细菌对其中一种或两种抗生素单独使用时具有耐药性。本文所述技术提供了两种互补的体外协同作用检测方法：联合使用时，可使研究人员高效地筛选感兴趣的抗菌药物组合是否具有协同活性（自动化棋盘格阵列法），并进一步评估在筛选阶段发现的有前景组合所表现出的抑制和杀菌动力学特征（手工时间-杀菌法）。

体外协同作用检测最常用的方法之一是棋盘格阵列法，该方法是最低抑菌浓度（MIC）检测的一种改进形式，通过一系列浓度组合来检测两种不同抗生素对某种细菌分离株的抑制活性^4,5。当两种药物联合使用时若其活性超过相加效应，则认为该组合具有协同作用⁶。然而，手动设置棋盘格阵列涉及一系列计算以及稀释和移液步骤，操作繁琐且容易引入人为误差。这些限制因素导致协同作用检测主要局限于对少量抗生素组合和细菌分离株进行回顾性评估，且不同研究之间的结果并不总是一致^7,8,9,10,11。此外，协同作用检测的复杂性也导致其在临床微生物学实验室中难以开展，并使得临床联合治疗研究中几乎缺乏体外协同作用检测的数据^12,13。

为了提高棋盘格阵列法的效率和通量，我们采用了一种本实验室先前开发的自动化MIC检测技术，该技术利用喷墨打印技术将抗生素储备液精确且一致地分配至微孔板的孔中¹⁴。该平台消除了复杂计算和多次移液步骤的需要。只要用户输入所需抗生素的浓度范围及其储备液浓度，配套软件即可自动计算并分配相应的抗生素体积，从而生成二维棋盘格阵列。我们最初在一组碳青霉烯类耐药肠杆菌（CRE）分离株上测试了该方法¹⁵，随后重点研究了含多粘菌素（colistin）的药物组合对多粘菌素耐药分离株的抗菌活性¹⁶。多粘菌素通常作为最后手段的药物，主要用于治疗多重耐药（MDR）革兰阴性病原菌感染^17,18，而多粘菌素耐药会使原本已为多重耐药的细菌几乎成为泛耐药菌¹⁹，因此这类细菌成为开发新型治疗策略的理想对象，即通过联合使用细菌对其单独不敏感的药物来实现协同作用。我们发现，多粘菌素与蛋白合成抑制剂类抗生素米诺环素（minocycline）联用时表现出极高的协同作用率，即使对这两种药物各自耐药的菌株亦是如此，这可能是由于多粘菌素对即使耐多粘菌素的细菌仍具有亚抑制浓度下的通透化效应。本文选择该组合作为示例进行阐述。值得注意的是，协同作用测试也可用于评估两种单独使用时均有效的药物联用后是否具有增强的疗效。

自动化棋盘格阵列法可实现快速、高通量的协同作用检测。然而，棋盘格阵列法也存在局限性。作为一种改良的最小抑菌浓度（MIC）检测方法，该法仅能提供关于细菌生长抑制的数据，而无法反映杀菌效果，且不能提供抗生素随时间变化的作用信息。相比之下，手工操作的时间-杀菌协同试验虽然更为耗时费力，但可在24小时的时间过程中提供关于抑制和杀菌的双重信息^20,21。我们对少量分离株进行了时间-杀菌分析，以验证棋盘格阵列法的结果，并确定所发现的协同组合是否同样具有杀菌作用。

棋盘式阵列法和时间-杀菌协同作用方法均可提供有关药物组合活性的重要信息，尤其适用于评估针对高度耐药细菌病原体的潜在新型治疗方案。这些方法也存在固有的局限性。标准的微量肉汤稀释法最小抑菌浓度（MIC）检测已知存在一个两倍稀释度的预期误差范围²²，当两种药物在棋盘式阵列中联合检测时，该误差范围会进一步增大。协同作用的标准定义仅将药物联合使用时在各自MIC的四分之一浓度下仍具有活性的情况视为协同作用⁶，这一定义已考虑了预期的变异性；但这种变异性（被认为源于生物和实验技术波动的共同作用²³）不可避免地导致对协同作用结果可靠性的不确定性。目前协同作用检测尚缺乏建立完善的质量控制标准，这也是当前的一大局限。或许所有协同作用检测方法中最显著的局限在于，体外检测结果与临床治疗效果之间尚缺乏明确的相关性，即当药物组合用于患者治疗时，其体外结果与实际临床结局之间的关联尚未确立²⁴。更简单且更快速的协同作用检测方法，例如本文所述的自动化棋盘式阵列法，可能有助于将体外协同作用检测整合到临床试验或其他患者结局评估中，从而在未来更好地阐明体外与体内效应之间的关系。

我们在此介绍的自动化棋盘阵列法为多种组合的高通量筛选提供了可行方案，并可快速评估非常规、 "高风险-高回报" 无需投入大量时间和资源即可实现组合。随后我们将演示的时间-杀灭法可为组合的协同活性提供额外支持性信息，并有助于表征其杀菌活性及抗菌动力学特征。

方案

CAUTION: Use appropriate safety procedures when working with bacteria. Wear gloves and a lab coat at all times. Perform work in a biosafety cabinet if aerosols will be generated or working with high risk pathogens.

NOTE: Twenty to 24 h before starting experiments, streak out the bacterial isolate(s) to be tested (from a colony-purified, minimally passaged stock frozen at -80 °C in tryptic soy broth with 50% glycerol stock) onto a blood agar plate. Incubate the plate at 35 °C in ambient air.

1. Inkjet Printer-assisted Automated Checkerboard Array Synergy

1. Make antimicrobial stock solutions (colistin and minocycline).
   1. Determine antibiotic stock solution concentrations based on solubility of antibiotics and desired final concentrations in checkerboard array. Make 10 mg/mL stocks of colistin and minocycline for this example. Use the CLSI M100 document to determine appropriate solvents for each antibiotic²⁵. Both colistin and minocycline are water-soluble; because the D300 inkjet printer requires the addition of surfactant for proper aqueous fluid handling, dissolve the antibiotics in ultrapure deionized water with 0.3% polysorbate 20.
   2. Weigh out antibiotic powder using an analytical balance and calculate volume of solvent needed to obtain goal stock concentration.
      1. If the antibiotic is supplied as a salt (e.g. colistin sulfate, minocycline hydrochloride) or in hydrated form (e.g. meropenem trihydrate), or if it is reported by the manufacturer to have less than 100% purity, perform a potency calculation²⁶ to determine the quantity of solvent required.
      2. Follow this example of minocycline hydrochloride with a stated purity of 900 μg/mg:
         Assay purity: 900 μg/mg
         Water content: None
         Active fraction: 0.926 (obtained by dividing the molecular weight of minocycline (457.48 Da) by the molecular weight of minocycline hydrochloride (493.94 Da)).
         Potency = (Assay purity) * (1 – water content) * (active fraction)
                            = (900 μg/mg) * (1) * (0.926) = 833.4 μg/mg or 83.34%
         Then determine the volume of solvent required as follows:
         Volume (mL) = [Weight (mg) * Potency (μg/mg)] ÷ [Concentration (μg/mL)]
         So, for example, if 34.7 mg of minocycline hydrochloride powder is weighed out, use the following calculation to determine the volume of solvent required to make a 10 mg/mL solution:
         Volume = (34.7 mg) * (833.4 μg/mg) = 2.89 mL
                                10,000 μg/mL
   3. Pour antibiotic powder into a 15 mL conical tube and add the appropriate volume of water plus 0.3% polysorbate 20. Vortex until dissolved.
   4. Aliquot antibiotic stock solution into 0.5 mL microcentrifuge tubes and store at -80 °C until ready for use.
2. Perform quality control (QC) of antimicrobial stocks for use in checkerboard array experiments at least one day prior to synergy testing so that QC results can be evaluated before using the stock for synergy testing.
   NOTE: The QC technique described here is identical to the technique that would be used for minimum inhibitory concentration (MIC) testing of individual drugs and can be used as such with any strains of interest to the investigator.
   1. Prepare bacterial suspension.
      1. Take an aliquot of each antibiotic stock out of the -80 °C freezer to start thawing while preparing bacterial suspension. Vortex once thawed to ensure that the antibiotic is in solution.
      2. Select an appropriate QC strain and determine the acceptable MIC range for drugs being tested based on Table 5A-1 in CLSI M100²⁵. For the drugs here, use E. coli ATCC 25922; the MIC ranges for this strain are 0.25-1 μg/mL for minocycline and 0.25-2 μg/mL for colistin.
      3. Select a range of antibiotic concentrations to test that will include the entire QC range. Use the range of 0.0156 μg/mL to 8 μg/mL for minocycline and colistin for ATCC 25922.
      4. Add 1 mL of 0.9% sodium chloride to a 12 mm x 75 mm round bottom glass culture tube. Select one or two colonies from an overnight plate of ATCC 25922 and vortex gently to suspend.
      5. Check the concentration of bacteria using a McFarland reader. Adjust as needed by adding more 0.9% sodium chloride or more bacteria to achieve a 0.5 McFarland turbidity reading.
      6. Make a 1:300 dilution of the 0.5 McFarland suspension by adding 100 μL of the suspension to 30 mL of cation-adjusted Mueller-Hinton broth (CAMHB) in a 50 mL conical tube to reach a final cell density of 5x10⁵ CFU/mL, as recommended by CLSI²⁷.
      7. Using a sterile inoculating loop, isolation streak a drop of the starting inoculum onto a blood agar plate to confirm inoculum purity, and incubate at 35 °C in ambient air.
   2. Add antimicrobials to a flat-bottom, square-well, clear, untreated 384-well plate using the D300. Perform this step immediately after preparing bacterial suspension so that suspension can be added to the plates within 15 min of preparation²⁶.
      1. Turn on the D300 inkjet printer and the associated computer. Open the software program.
      2. Start a new file. Above the image of the plate grid, right-click on Plate 1 and choose Edit plate. Select the appropriate plate type (384 well) and additional volume (50 μL).
      3. Add fluids (i.e., antibiotic stocks) to the protocol by clicking the plus sign next to Fluids on the left-hand panel. Add two fluids (colistin and minocycline).
         1. Hover over the panel that appeared for Fluid 1 and click the pencil to edit. Name the fluid “Colistin”, change Class to “Aqueous + Tween 20”, change Concentration to 10,000, and change concentration units to μg/mL (note that stock concentration is 10 μg/mL, i.e., 10,000 mg/mL). Leave Dispense by at Concentration and leave the remainder of the fields at their default settings. Click OK.
         2. Repeat the procedure above for Fluid 2 (minocycline).
         3. Click the Current Protocol tab at the top of the screen and change Concentration (mass) to μg/mL to determine the units used for final well antibiotic concentrations.
      4. Select 10 wells in the grid by clicking and dragging, then click Titration at the top of the screen. For Specify titration using select Highest concentration, for Fluid choose Colistin, for Highest Concentration enter 8 (make sure units are μg/mL), and for Distribution select 1:2 (50%). Leave default values in place for the rest of the window and click OK.
      5. Repeat the procedure above for minocycline to generate the minocycline titration.
      6. Save the protocol, and then click the Run button at the top left.
      7. Click the Start button. Load a 384-well plate (with lid removed) into the plate holder and press Loaded under the “Load Plate 1 – Synergy” prompt.
         NOTE: This prompt is chosen by the software and does not indicate that synergy testing is being performed. Place a T8+ cassette into the cassette slot and press Loaded under the Load a T8+ cassette prompt.
      8. When prompted, add antibiotic stock solution to the indicated reservoirs on the cassette. Follow instructions on the screen for proper loading and dispense carefully to avoid getting any bubbles in the solution. After each solution is added, press the Filled button.
      9. Once the inkjet printer has added antibiotic stock in appropriate volumes to each well and the Run completed box appears, click Close, remove the plate, and turn off the D300.
   3. Add bacterial suspensions to 384-well plate and incubate plate.
      1. Pour the previously prepared bacterial suspension into a sterile reagent reservoir.
      2. Use a multichannel pipette to add 50 μL of bacterial suspension to all antibiotic-containing wells. Add 50 μL of CAMHB without bacteria to an empty well; this will be the negative control well to confirm sterility of the media.
      3. Place plate in a 35 °C ambient air incubator and incubate for 16-20 h²⁶.
         NOTE: A different duration of incubation may be required if organisms other than Enterobacteriaceae are being tested; consult CLSI M100²⁵ for organism-specific recommendations.
   4. Read plate on a microplate reader at an optical density of 600 (OD₆₀₀) and analyze results.
      1. Using a spreadsheet program, shade cells with an OD₆₀₀ value of ≥0.07 green, indicating growth, and cells with a value of <0.07 red, indicating no growth.
         NOTE: These values were determined based on visual inspection of growth vs. no growth and correlation with OD readings for these experiments; OD₆₀₀ readings from wells containing media alone were consistently below 0.07. Appropriate cutoffs may differ with different plate readers and bacteria.
      2. Determine the MIC for each drug. The MIC is the lowest concentration of drug at which bacterial growth is inhibited. If the MIC is within the expected QC range according to the CLSI M100 document²⁵, the stock solution is appropriate for use.
3. Prepare bacterial suspension for checkerboard array.
   1. Take an aliquot of each antibiotic stock out of the -80 °C freezer to start thawing while preparing bacterial suspension. Vortex once thawed to ensure antibiotic is in solution.
   2. Add ~1 mL of 0.9% sodium chloride to a 12 mm x 75 mm round bottom glass culture tube. Select one or two colonies from an overnight plate of bacteria (in this case, E. coli strain FDA-CDC 0494) and vortex gently to suspend these in the 0.9% sodium chloride.
   3. Check the concentration of bacteria using a McFarland reader. Adjust as needed by adding more 0.9% sodium chloride or more bacteria to achieve a 0.5 McFarland turbidity reading.
   4. Make a 1:300 dilution of the 0.5 McFarland suspension by adding 100 μL of the suspension to 30 mL of CAMHB in a 50 mL conical tube to reach a final cell density of 5 x 10⁵ CFU/mL²⁷.
4. Add antimicrobials to a flat-bottom, square-well, clear, untreated 384-well plate using the D300.
   NOTE: Perform this step immediately after preparing bacterial suspension so that suspension can be added to the plates within 15 minutes of preparation²⁶.
   1. Turn on the inkjet printer, start a new file, and add fluids to the protocol as in steps 1.2.2.1-1.2.2.3.
   2. Generate the synergy grid.
      1. Click the Synergy icon at the top of the screen and proceed through the steps. For Type select Two or more fluids factored together. For Plate, it is not necessary to exclude any wells; click Next on this step without making changes.
      2. On the Titration tab, enter the antibiotic concentrations and placement.
         1. In order to add minocycline in decreasing doubling dilutions from 32 to 0.031 μg/mL, in addition to a negative well with no antibiotic, down the y axis, enter 12 for Titration levels on the left panel. For Specify titration using, select Highest concentration; for Fluid, select Minocycline; for Highest concentration, enter 32 (make sure the unit is set at μg/mL; if it is not, close the Synergy dialog box and change under the Current Protocol tab). Make sure that the Include 0 value box is checked. Change Distribution to 1:2 (50%).
         2. Repeat these steps for colistin on the right-hand panel, using 12 titration levels and a highest concentration of 16. Click Next.
      3. On the Layout tab, choose Titration levels of first 2 fluids determine the number of rows and columns in a layout grid. Click Next. If the grid appears as expected, click Finish.
   3. Save the protocol, then click the Run button at the top left.
   4. Click the Start button. Load a 384-well plate (with the lid removed) into the plate holder and press Loaded under the Load Plate 1 – Synergy prompt. Place a T8+ cassette into the cassette slot and press Loaded under the Load a T8+ cassette prompt.
   5. When prompted, add antibiotic stock solution to the indicated reservoirs on the cassette. Follow instructions on the screen for proper loading and dispense carefully to avoid getting any bubbles in the solution. After each solution is added, press the Filled button.
   6. Once the D300 dispenser has added antibiotic stock in appropriate volumes to each well and the Run completed box appears, click Exit, remove the plate, and turn off the D300.
5. Add bacterial suspensions to 384-well plate and incubate plate.
   1. Pour the previously prepared bacterial suspension into a sterile reagent reservoir.
   2. Use a multichannel pipette to add 50 μL of the suspension to all wells in the checkerboard array. Add 50 μL of CAMHB without bacteria to an empty well; this will be the negative control well to confirm sterility of the media. Incubate in a 35 °C ambient air incubator for 16-20 h²⁶.
6. Read plate on a microplate reader at OD₆₀₀ and analyze checkerboard array results.
   1. First, check the purity plate and ensure that the isolated colonies are of a single morphology that is consistent with the expected morphology of the organism being tested.
   2. Using a spreadsheet program, shade cells to indicate growth and no growth as in step 1.2.4.1.
   3. Determine the MIC for each drug. For a drug that does not inhibit bacterial growth at the highest concentration tested, the MIC is considered to be off-scale.
   4. For each well in which growth is inhibited, determine the fractional inhibitory concentration (FIC) for each antibiotic based on that antibiotic’s MIC (see Figure 1B and Figure 2B).
      NOTE: The FIC is the ratio of the concentration of antibiotic in a well in which growth is inhibited to its MIC; so for a drug with an MIC of 8 μg/mL, a well containing 8 μg/mL of that drug has an FIC of 1, while a well containing 4 μg/mL has an FIC of 0.5.
   5. Calculate the fractional inhibitory concentration index (FIC_I) value for each well in which growth is inhibited as the sum of the FICs of each of the drugs in that well.
   6. Determine the lowest FIC_I at which growth is inhibited (minimum FIC_I). If the minimum FIC_I is £0.5, consider the combination synergistic; if 0.5-4, consider the combination indifferent; and if >4, consider the combination antagonistic. If the combination is synergistic at some concentration combinations but antagonistic at others, note this result but consider the combination overall antagonistic.

2. Time-kill Synergy Testing

1. Make antimicrobial stock solutions. If this step is performed ahead of the experiment, freeze stocks at -80 °C until ready for use.
   1. Determine antibiotic stock solution concentrations based on solubility of antibiotics and desired final concentrations in time-kill studies. In this example, make colistin and minocycline stocks at a concentration of 1 mg/mL. Use the CLSI M100 document to determine appropriate solvents for each antibiotic²⁵. Use water, as recommended, for both colistin and minocycline.
   2. Weigh out antibiotic powder using analytical balance and calculate volume of solvent needed to obtain goal stock concentration. If needed, perform a potency calculation prior to determining the quantity of solvent required, as described in step 1.1.2.1 above.
   3. Pour antibiotic powder into 15 mL conical tubes and add the appropriate volume of water. Vortex until dissolved.
2. Using the manual broth microdilution method described in CLSI M07²⁶, perform QC of antimicrobial stocks for use in time kill synergy experiments. Perform this step at least one day prior to synergy testing so that QC results can be reviewed before using the stock.
   1. Select an appropriate QC strain and determine the acceptable MIC range for drugs being tested based on Table 5A-1 in CLSI M100²⁵. For the drugs here, use E. coli ATCC 25922; the MIC ranges for this strain are 0.25-1 μg/mL for minocycline and 0.25-2 μg/mL for colistin.
   2. Prepare antibiotic-containing broth microdilution plates.
      1. Select the highest concentration of antibiotic to be tested so the entire QC range can be included. Use a range of 0.016 to 8 μg/mL for minocycline and colistin for ATCC 25922.
      2. Dilute the antibiotic stocks to a working solution in CAMHB at two times the highest concentration needed (because it will be diluted 1:1 with the suspension of bacteria). In this example, dilute both stocks from 10 mg/mL to 16 μg/mL.
      3. Using a multi-channel pipette, add 100 μL of each of the 2x antibiotic suspensions to a well in the first column of a clear, round-bottom, untreated 96-well plate and add 50 μL of plain broth (i.e., without antibiotic) to each well of the subsequent columns.
      4. Remove 50 μL of antibiotic-containing broth from each well in the first column and add to the wells in the second column. Pipette up and down several times to mix the contents, generating an antibiotic concentration half that of the concentration in the first column.
      5. Repeat step 2.2.2.4 with each column, so that a series of serial two-fold dilutions, each with a volume of 50 μL, is prepared. Change pipette tips between each dilution step if desired to eliminate the possibility of antibiotic carryover. Note that the resultant concentrations are all still 2x the final concentrations, as they will subsequently be diluted 1:1 with bacterial suspension.
      6. Do not add any antibiotic to the final two columns, as these will be the negative and growth control columns.
   3. Prepare bacterial suspension.
      1. Prepare a 0.5 McFarland suspension from an overnight plate of E. coli ATCC 25922 in 0.9% sodium chloride as described in steps 1.2.1.5-1.2.1.6.
      2. Make a 1:150 dilution of the 0.5 McFarland suspension by adding 50 μL of the suspension to 7.5 mL of CAMHB.
         NOTE: The final bacterial suspension will be diluted 1:300 once it is mixed 1:1 with the antibiotic solution, reaching the CLSI-recommended cell density of 5 x 10⁵ CFU/mL²⁷).
   4. Add bacteria to the microplate and incubate.
      1. Add 50 μL of the bacterial suspension to each well, except in the 11^th column. Add 50 μL of CAMHB to the 11^th column (negative control column).
      2. Incubate plate at 35 °C for 16-20 h.
         NOTE: A different duration of incubation may be required if organisms other than Enterobacteriaceae are being tested; consult CLSI M100²⁵ for organism-specific recommendations.
      3. Read plate for growth visually using transmitted light and, for each antibiotic, determine the lowest concentration in which there is no growth; this is the MIC. Consult the CLSI M07 document for additional details on visual interpretation of MIC²⁶. If the MIC is within the expected QC range, the stock solution is appropriate for use.
3. Start initial culture.
   1. Make a 0.5 McFarland suspension of test organism in sterile 0.9% NaCl as described above.
   2. Add 100 μL of the 0.5 McFarland suspension to 5 mL of CAMHB in a 25 mm x 150 mm glass round bottom culture tube with stainless steel closure and vortex gently to mix.
   3. Using a sterile inoculating loop, isolation streak a drop of the diluted suspension onto a blood agar plate to confirm inoculum purity and incubate at 35 °C in ambient air.
   4. Replace closure on tube and incubate in a test tube rack on a shaker at 35 °C in ambient air for at least 3 h, until logarithmic-phase growth is reached (see step 2.6.1). Proceed to step 2.4 while the culture is in the incubator.
4. Prepare antimicrobial solutions in 25 mm x 150 mm glass culture tubes.
   1. Take out antimicrobial stock aliquots from -80 °C freezer to thaw. Vortex once thawed to ensure antibiotic is in solution.
   2. While initial culture is incubating, add 10 mL of CAMHB to five autoclaved 25 mm x 150 mm glass culture tubes and add antimicrobial stock solutions as follows.
      NOTE: For a synergy study, at least one drug should be at a concentration that does not affect the growth curve individually²⁸; this can be determined by evaluating the effects of individual drug concentrations prior to the synergy study.
      1. Tube 1: Add appropriate quantity of antibiotic #1 to obtain target final antibiotic concentration. In this case, add 10 μL of 1 mg/mL colistin stock to obtain a final colistin concentration of 1 μg/mL, as this is a concentration that is ineffective against the strain used in this example.
      2. Tube 2: Add appropriate quantity of antibiotic #2 to obtain final antibiotic concentration to be tested. In this case, add 10 μL of 1 mg/mL minocycline stock to obtain a final concentration of 1 μg/mL, a concentration that is ineffective against the strain being used in this example.
      3. Tube 3: Add the same quantity of antibiotic #1 and antibiotic #2 as used in tubes 1 and 2. In this case, add 10 μL of 1 mg/mL minocycline stock and 10 μL of 1 mg/mL colistin stock.
      4. Tube 4: Add no antibiotics; this will be the growth control tube.
      5. Tube 5: Add no antibiotics; this will be the negative control tube.
5. Prepare 96 deep well polypropylene plates with 2 mL wells for serial dilutions by adding 900 µL of sterile 0.9% sodium chloride to rows B-H of columns 1-5 with a multichannel pipette.
6. Prepare starting inoculum and add to tubes.
   1. Once the initial culture has reached logarithmic growth phase (~3 h for Klebsiella pneumoniae, the organism used in this example), remove the culture tube from the shaker, vortex gently, transfer ~1 mL of suspension to a 12 mm x 75 mm glass culture tube, and check density with a McFarland reader.
      1. If it is less than 1.0 McFarland, return tube to the shaker and incubate longer. If it is greater than 1.0 McFarland, add CAMHB to the tube, vortex gently, and re-sample, repeating the process until the suspension is at 1.0 McFarland.
   2. Add 100 μL of the 1.0 McFarland suspension to tubes 1-4 and vortex gently.
7. Sample aliquots from each culture and perform serial ten-fold dilutions.
   1. At time 0 (immediately after adding bacteria to the tubes) and at 1, 2, 4, 6, and 24 h, remove a 150 μL aliquot from each culture tube by tilting the tube so that only the sterile pipette tip enters the tube and not the unsterile pipettor shaft during aliquot withdrawal. Add aliquots, respectively, to consecutive wells in the first row of the previously prepared 96 deep well plate. Return tubes to a test tube rack on a shaker in a 35 °C ambient air incubator immediately after removing aliquots at each time point.
   2. Using a multichannel pipette, remove 100 μL from row A, add to row B (which contains 900 μL of 0.9% sodium chloride), and pipette up and down 4-5 times to mix, creating a 1:10 dilution. Discard tips following each dilution step to prevent carryover of bacteria, which can lead to falsely elevated colony counts.
   3. Repeat step 2.7.2 for rows B-H with new pipette tips for each row.
8. Plate diluted samples for colony counts using the drop plate method^29,30.
   1. Label Mueller-Hinton agar plates with the antibiotic conditions and dilution to be plated.
   2. Using a multichannel pipette and extra-long tips (to ensure that tips reach into suspension), remove 10 μL from each well in column one and dispense carefully in a row onto the appropriately labeled plate. If small (100 mm diameter) plates are used, dispense 3 rows (each consisting of drops from rows A-H of a single column) per plate; on large (150 mm diameter) plates, dispense 8 rows per plate.
   3. Allow drops to dry completely (~15 min).
   4. At 24 h, place a 10 μL drop taken directly from the negative control tube in an indicated area of one of the plates to test for sterility. Invert plates and incubate overnight at 35 °C in ambient air.
9. Count colonies and calculate cell density. Mark colonies with a fine-tip permanent marker on the reverse of the plate to avoid double-counting or missing colonies.
   1. First, check the purity plate and ensure that the isolated colonies are of a single morphology that is consistent with the expected morphology of the organism being tested.
   2. For each dilution series, identify drops with 3-30 colonies (typically one drop per dilution series). Count the colonies in these drops and record the count along with the dilution factor.
      1. If there are no drops in a dilution series with 3-30 colonies, count the colonies in the last drop with >30 colonies and the first drop with <3 colonies (these should be adjacent drops).
   3. For each dilution series, calculate number of colony forming units per milliliter (CFU/mL) in the sample based on the number of colonies in the drop using the following formula: CFU/mL = n(1/d)(100) where n is the number of colonies, d is the dilution factor (1 for undiluted sample (row A), 0.1 or 10^-1 for the first 1:10 dilution (row B), 0.01 or 10^-2 for the second 1:10 dilution (row C), and so on, and the constant 100 accounts for the fact that the total volume of the drop is 10 μL, while the final value is expressed in CFU/mL, i.e., CFU/1,000 μL. Use a spreadsheet containing formulas that calculate CFU/mL from colony count to simplify this process.
      1. For dilution series where more than one drop was countable (or where two drops had to be counted because no drop fell in range), average the final CFU/mL counts for all counted drops.
      2. Because the lower limit of detection is 300 CFU/mL (3 colonies in the undiluted drop), record and plot colony count as £300 CFU/mL for dilution series in which there are <3 colonies in the undiluted drop.
   4. Inspect the sterility control drop from time 24; if any growth is observed in this drop, the results of the experiment should not be used.
10. Graph and analyze results.
    1. Plot growth curves from the three antibiotic-containing cultures and the growth control on the same graph. Plot time on the x axis and CFU/mL, using a logarithmic scale, on the y axis.
    2. Calculate the difference in CFU/mL between the combination tube at time 24 and the most active single agent at time 24. If the difference is ≥2 log₁₀, consider the combination synergistic. Then calculate the difference in CFU/mL between the combination tube at time 24 and at time 0. If the difference is ≥3 log₁₀, consider the combination bactericidal.

结果

图1A 展示了一项棋盘格阵列协同实验的网格结果，其中米诺环素（浓度范围为0–32 μg/mL）与多粘菌素（浓度范围为0–16 μg/mL）联用，针对大肠杆菌（E. coli）菌株FDA-CDC 0494进行测试。数值表示在600 nm光密度（OD₆₀₀）下的分光光度读数。OD₆₀₀ 值低于0.07的孔（肉眼观察无细菌生长）以红色阴影标示，而OD₆₀₀ 值为0.07或以上的孔（肉眼观察有细菌生长）以绿色阴影标示。对于每种药物，最低抑菌浓度（MIC；加粗显示）是指能够抑制细菌生长的最低药物浓度。米诺环素的最低抑菌浓度为32 μg/mL，多粘菌素的最低抑菌浓度为8 μg/mL。图1B 保留了相同的着色方式，但在细菌生长被抑制的孔中，数值被替换为部分抑菌浓度指数（FIC_I）值。其计算方法如下：在每个孔中，每种药物的部分抑菌浓度（FIC）通过将该孔中的抗生素浓度除以该药物的MIC得到，FIC_I 则为两种药物FIC值的总和。FIC_I 值为0.5的孔（通常作为协同作用的临界值）以虚线边框标示，FIC_I 值最低的孔（0.094）以加粗显示。由于最小FIC_I 值处于协同作用范围内，因此该药物组合被认为具有协同效应。

图2A 和 图2B 显示的网格与 图1A 和 图1B 中的类似，但在本例中，该组合对测试菌株（K. pneumoniae 分离株 BIDMC 4）未表现出协同作用，因为抑制生长的最低FIC_I为1，该值>0.5。

图3展示了棋盘式协同试验中出现多个跳过孔的光密度读数结果（Enterobacter cloacae 复合群分离株 BIDMC 27）。跳过孔是指尽管相邻孔中抗生素浓度更高但仍出现细菌生长抑制的孔。这一现象在标准 MIC 测试中亦有报道，可能源于不同孔之间细菌生长特性的生物学差异，以及某些抗生素对细菌接种量微小变化的敏感性^23,31,32。若在棋盘式阵列中出现一个以上的跳过孔，则舍弃该次实验结果并重新进行检测。

图4展示了三种药物组合针对肺炎克雷伯菌（K. pneumoniae）分离株BIDMC 32的时程杀菌协同作用结果示例。y轴以对数刻度表示菌落计数，x轴表示时间（小时）。红色条形和数字表示含药物组合试管中初始接种量与24小时后细菌浓度之间的差值，而蓝色条形和数字表示含药物组合的试管与仅含最有效单药的试管在24小时后细菌浓度之间的差值。图4A显示了多粘菌素与米诺环素组合的结果；该组合具有协同作用（24小时时，接受组合处理与仅接受最有效单药处理的细菌浓度差值 ≥2 log₁₀ CFU/mL），且具有杀菌效应（从初始接种量到24小时浓度的下降值 ≥3 log₁₀ CFU/mL）。图4B显示了多粘菌素与克林霉素组合的结果；该组合具有协同作用，但不具杀菌效应。该组合抑制了细菌生长（单独任一药物均无此作用），但未杀死细菌。图4C显示了多粘菌素与红霉素组合的结果；该组合既无杀菌效应，也无协同作用。

图1：棋盘法阵列结果展示协同作用（米诺环素 + 多粘菌素测试针对 E. coli 菌株 FDA-CDC 0494 (A) 棋盘式阵列的分光光度读数与生长结果解读。单元格中的数值为 600 nm 处的光密度读数（OD₆₀₀）。OD值的细胞₆₀₀ OD值低于0.07（目视观察无生长）以红色阴影标注，而OD值₆₀₀ 值为 0.07（对应于通过肉眼观察的生长）的部分用绿色标出。B) 部分抑菌浓度指数（FIC_我）计算。表示生长或无生长的阴影部分已保留。多粘菌素和米诺环素的数值沿 x- 和 y-轴分别表示抑制分数浓度（FIC），即该列或该行中药物浓度与其单独使用的最小抑制浓度（MIC）的比值。每个单元格中的数值即为FIC_我，或该孔中两种药物的FIC之和。由粗虚线边框标出的大区域包含具有FIC的孔_我 0.5。粗边框的孔表示FIC值最低的孔_我 生长受到抑制，或最低FIC_我由于最低FIC_I 若为0.5，则该组合被视为具有协同作用。 请点击此处以查看此图的放大版本。

图2：无协同作用的组合棋盘法阵列结果（米诺环素 + 多粘菌素测试针对 K. pneumoniae 分离 BIDMC 4） (A600 nm 处的光密度值及棋盘格阵列结果的生长解释，如所述 图1A. (B) 部分抑菌浓度指数（FIC_我）计算，方法如前所述 图1A由于最低FIC_我 是 >0.5时，该组合不被视为具有协同作用。 请点击此处查看此图的放大版本。

图3：由于跳过孔位导致无法解读的棋盘式阵列结果（米诺环素 + 多粘菌素用于检测对抗 阴沟肠杆菌 复合分离株 BIDMC 27 600 nm 处的光密度值及棋盘式阵列结果的生长情况解读，如所述 图1A出现多个跳孔现象，即尽管相邻孔中抗生素浓度更高且有细菌生长，但某些孔中细菌生长仍受到抑制。结果无法解读，需重复实验。 请点击此处以查看此图的放大版本。

图4：三种组合针对肺炎克雷伯菌（K. pneumoniae）分离株BIDMC 32的时杀协同作用结果。 菌落计数以对数刻度表示在y轴上，时间（小时）表示在x轴上。组合组在24小时时的细菌浓度与起始接种量之间的差异由红色条形和数字表示。如果从起始接种量到24小时浓度的下降≥3 log₁₀ CFU/mL，则认为该组合具有杀菌作用。含有组合的试管与仅含最有效单药的试管在24小时时细菌浓度之间的差异由蓝色条形和数字表示；如果浓度降低≥2 log₁₀ CFU/mL，则认为该组合具有协同作用。（A）黏菌素（CST）+ 米诺环素（MIN），一种既具有协同作用又具有杀菌作用的组合。（B）黏菌素 + 克林霉素（CLI），一种具有协同作用但无杀菌作用的组合。（C）黏菌素 + 红霉素（ERY），一种既无协同作用也无杀菌作用的组合。这些结果最初作为一项研究的一部分发表，该研究探讨了含黏菌素组合对黏菌素耐药肠杆菌科（Enterobacteriaceae）的协同活性，我们在此证明黏菌素可与多种仅对革兰阳性菌个别有效（例如克林霉素）或主要有效（例如红霉素）的抗生素产生协同作用¹⁶。（注意：红霉素在此菌株的棋盘格试验中显示协同作用，但在时杀试验中未显示，因此在此选作非协同组合的示例。）我们假设，已知通过破坏革兰阴性菌外膜通透性发挥作用的黏菌素，可在耐黏菌素的革兰阴性菌中产生亚抑制浓度下的通透化效应，从而允许如克林霉素等通常无法进入革兰阴性菌细胞的药物进入。本图A部分经 Brennan-Krohn、Pironti 和 Kirby 2018¹⁶ 修改，版权 © 美国微生物学会，《抗菌剂与化疗》，62(10)，2018，pii: e00873-18，doi: 10.1128/AAC.00873-18。 请点击此处查看此图的放大版本。

讨论

本文所述的两种方法均可提供有关联合使用抗菌药物时其活性相对于各自单独使用时活性的信息。自动化喷墨打印机辅助数字分配法是对《临床微生物学操作手册》中所述方法的一种改进³³，而时间-杀菌法则更严格遵循同一参考文献中的相应方案³⁴。

在棋盘格阵列法中，确定每孔所需添加的抗菌剂储备液体积以及这些体积的分配均由自动化完成，从而消除了手动操作棋盘格阵列时可能遇到的一些主要误差来源。然而，研究者仍必须确保原始储备液的配制浓度符合预期，并且目标终浓度已正确输入D300软件中。将抗菌剂悬液加入384孔板的孔中起初可能具有挑战性，需格外小心，以确保移液器吸头准确进入相应孔位，并防止液体溅至孔壁边缘。可使用自动化液体处理仪替代手持多通道移液器，以提高细菌悬液加入孔板的速度和准确性。如实验方案所述，D300系统需要添加表面活性剂聚山梨酯20（P-20），以实现正常的液体处理。已有文献指出，在标准肉汤微量稀释法中，使用浓度为0.002%的另一种表面活性剂聚山梨酯80可降低对多粘菌素MIC <2 µg/mL的微生物的多粘菌素MIC值。^35,36 我们实验室先前已证实，与参考BMD方法相比，浓度高达0.0015%的P-20对D300辅助测定的MIC结果无影响¹⁴。在本实验示例中，P-20的最高浓度为0.0014%。

我们在进行某些棋盘式阵列实验时遇到的一个问题是大量孔位被跳过。这种情况在某些抗生素中出现的比例明显偏高。具体而言，在针对一组碳青霉烯类耐药肠杆菌科细菌（Enterobacteriaceae）的药物联用筛选中，我们发现521次试验中有49次（9.4%）因多个孔位被跳过而无法使用，其中12种测试抗生素中的2种（磷霉素和头孢吡肟）占到了46次（94%）。此类较高发生率可能更常见于对接种量效应特别敏感的药物^31,32,37。值得注意的是，美国临床与实验室标准协会（CLSI）不推荐使用肉汤稀释法检测磷霉素的敏感性²⁵，原因是该方法所得结果的可靠性存疑，这可能解释了本研究中该药物结果不稳定的现象。根据研究者的偏好，可对自动化棋盘法进行一些修改。例如，若实验室工作流程需要，可将抗菌药物直接加入已含有细菌悬液的孔板中，而非加入空孔中。尽管本研究使用了384孔板，但该方法也可在96孔板中进行，只需相应调整孔内液体体积。采用96孔板格式可能有助于减少对接种量微小变化特别敏感的抗生素出现跳孔现象。在计算FIC_I时，可能会遇到最小抑菌浓度（MIC）超出检测范围（即高于所测试浓度）的情况，包括被测药物对目标微生物单独作用时无活性的情形。在此类情况下，可假设MIC为比最高测试浓度高一个稀释度的值来计算FIC。这是一种最保守的策略，因为它假设在协同作用测试中观察到抑制效应的任何稀释度下，FIC均取可能的最大值。例如，若实际MIC比最高测试浓度高出两个倍比稀释度，则相应的FIC值将比保守估计值低两倍，依此类推。

为了在时间-杀菌实验中准确评估药物的杀菌活性，必须确保细菌处于对数生长期，尤其是在测试细胞壁活性抗生素时尤为关键²⁸。对于本示例中使用的快速生长细菌（K. pneumoniae），在振荡条件下培养3小时即可达到该生长阶段，但不同微生物可能需要不同的培养时间。通常情况下，菌液应呈现明显但不过度的浑浊。可通过在连续时间点（例如每30分钟一次，持续4–6小时）取样进行菌落计数，绘制生长曲线，以确定合适的培养时间³⁸。时间-杀菌实验中初始接种量的设定也至关重要。目标初始接种浓度约为5 × 10⁵ 至1 × 10⁶ CFU/mL。本文所述的稀释方法（将100 μL 1.0麦氏浊度标准的菌悬液加入10 mL培养基中）适用于肺炎克雷伯菌（Klebsiella pneumoniae）及其他我们已测试过的肠杆菌科（Enterobacteriaceae）菌种。若在使用其他微生物的实验中，初始接种菌液密度显著高于或低于此范围，则可能需要调整稀释比例。（可通过测定0.5或1.0麦氏浊度菌悬液的平板菌落计数，确定该浊度对应的活菌数量，再计算初始菌悬液需稀释的倍数，以获得目标终浓度。）在分析协同作用实验的平板计数结果时，若发现含抗生素试管的初始接种量显著低于生长对照组的初始接种量，可能提示存在抗生素残留效应，或在将细菌加入含抗生素试管至取样涂板的短暂时间内发生了极快速的细菌杀灭。若在系列稀释中，未稀释样本点的菌落数量低于后续稀释度的菌落数量，则提示可能存在抗生素残留效应。已有多种方法被提出用于避免该效应，包括将单次取样液均匀涂布于整块平板³⁸，或在涂板前将样本离心、弃去上清液，并用无菌生理盐水重悬³⁹。在时间-杀菌实验的每个时间点，研究者均需高效且准确地从每支培养管中取出等份样本，并进行系列稀释。操作过程中的延迟，尤其是在连续且早期的时间点，可能导致培养物在非培养、非振荡条件下停留过久；而粗略的加样或稀释操作则可能导致平板计数结果不准确。与涂布平板法（将每种稀释液100 μL均匀涂布于整块琼脂平板）相比，本研究所描述的滴液平板法速度更快，所需琼脂平板数量显著减少，且计数更迅速，因为每个液滴的最大可计数菌落数为30，而涂布平板通常可计数多达300个菌落。然而，若研究者更熟悉涂布平板法，也可选择该方法。若使用多通道移液器加样后液滴发生相互融合，可改用单通道移液器分别点加间距更大的液滴。根据我们的经验，在加样前将平板冷却至4 °C，似乎可减少液滴过度扩散现象。

本文所述技术的一个局限性在于，两种协同作用检测方法（棋盘格阵列法和时间-杀菌法）的结果并不总是一致。由于大多数已发表的协同作用研究仅使用其中一种方法，而非同时结合两种方法，因此可能难以整合这两类检测的数据。由于我们开发的自动化棋盘格阵列法操作简便且通量高，我们实际上将其作为一种筛选手段，用于在更多分离株中测试药物组合，并确定哪些浓度组合具有协同作用。随后，我们针对在棋盘格阵列中表现有效的组合和浓度，开展了少量的时间-杀菌实验。值得注意的是，由于棋盘格检测通常在微量肉汤稀释体系中进行，而时间-杀菌检测使用较大的反应体积（类似于宏量肉汤稀释），我们发现两种方法所得的FIC值有时存在差异，通常在时间-杀菌实验中需要更高的浓度才能观察到活性。此前已有报道指出，当比较革兰阴性杆菌的宏量与微量肉汤稀释MIC检测结果时²⁶，以及当时间-杀菌实验中使用的较大接种量与微量肉汤稀释和棋盘格阵列检测中标准接种量进行比较时³²，也会出现类似现象。棋盘格阵列检测的一个具体局限性在于微量肉汤稀释MIC测试本身固有的变异性²²。尽管协同作用的FIC_I 判断阈值已在数学上对这种变异性进行了校正⁶，但这种变异性仍不可避免地引发人们对棋盘格阵列结果可靠性和一致性的担忧。

由于所有体外协同作用检测方法本身存在局限性（包括在人工培养基中培养细菌、抗生素浓度恒定以及作用时间有限），通过这些方法获得的结果必须通过补充技术进行验证和进一步评估。此类方法包括体外药代动力学/药效学（PK/PD）研究（例如中空纤维感染模型⁴⁰）、动物模型，以及最终的人体PK/PD和疗效研究。本文所述的自动化棋盘格阵列方法提供了一种快速筛选潜在协同作用组合的手段，从而更有针对性地利用上述技术。进一步实现这些方法的自动化，以及更系统地研究体外参数与临床结果之间的关系，对于扩大协同作用检测的应用规模并提高其临床适用性具有重要意义。

材料

本文使用的材料清单

姓名	公司	目录编号	评论
Escherichia coli 菌株 ATCC 25922	ATCC	25922	质控菌株
0.5 mL 微量离心管	USA Scientific	1605-0000
1 L 0.22 µm 瓶顶过滤器	Thermo Scientific Nalgene	597-4520
12 mm × 75 mm 硼硅酸盐玻璃圆底培养管	Fisherbrand	14-961-26
15 mL 圆锥形离心管	Phenix	SS-PH15
15 × 100 mm 或 15 × 150 mm 水解酪蛋白琼脂平板	Thermo Scientific	R01620 或 R04050
25 mm 不锈钢密封盖（用于 25 × 150 mm 玻璃培养管）	Bellco	2005-02512
25 × 150 mm 硼硅酸盐玻璃圆底培养管	Bellco	2011-25150
348 孔无菌透明平底未处理微孔板（带盖）	Greiner Bio-One	781186
50 mL 圆锥形离心管	Phenix	SS-PH50
50 mL 无菌试剂储液槽	Corning	4870
96 深孔聚丙烯微孔板（每孔 2 mL）	Fisherbrand	12-566-612
96 孔无菌透明圆底未处理微孔板（带盖）	Evergreen	222-8032-01R
阳离子调节水解酪蛋白肉汤	BD Diagnostics	212322
硫酸多粘菌素	Alfa Aesar	J60915
D300e 控制软件	HP/Tecan
DensiCHEK Plus McFarland 比浊仪	bioMérieux	21250
Excel 电子表格软件	Microsoft
超长 SHARP 10 µL 精密屏障吸头	Denville Scientific	P1096-FR
HP D300 数字分配仪	HP/Tecan
HP D300 T8+ 加样卡盒	HP/Tecan	30097370
盐酸米诺环素	Chem-Impex	14302
Picus 12 通道 10–300 µL 移液器	Sartorius	735461
聚山梨酯 20	Fisher Bioreagents	BP-337	商品名：Tween 20
氯化钠	Fisher Chemical	S271
分光光度计	Tecan	Infinite M1000 PRO
Xplorer 12 通道 50–1200 µL 移液器	Eppendorf	2231000328