原子力显微镜结合红外光谱作为探测单个细菌化学特性的工具

摘要

原子力显微镜-红外光谱法（AFM-IR）为细菌研究提供了强大的平台，能够实现纳米级分辨率。该技术可在单细胞水平上对细菌进行亚细胞结构变化的成像（例如细胞分裂过程中）以及化学成分的比较研究（例如由药物耐药性引起的差异）。

摘要

原子力显微镜-红外光谱法（AFM-IR）是一种新型的联用技术，能够以纳米级分辨率同时表征样品的物理特性和化学组成。通过将原子力显微镜（AFM）与红外光谱（IR）相结合，克服了传统红外光谱空间分辨率的限制，实现了20–100 nm的分辨率。这为红外光谱在微米以下样品探测中的广泛应用开辟了道路，而此类应用此前无法通过传统红外显微镜实现。AFM-IR非常适用于细菌研究，可在单细胞及细胞内水平提供光谱和空间信息。由于全球健康问题日益严峻，且细菌感染尤其是抗菌耐药性的快速发展的前景不容乐观，迫切需要一种能够在单细胞和亚细胞水平进行表型探测的研究工具。AFM-IR有望满足这一需求，通过实现对单个细菌化学组成的精细表征。本文提供了针对细菌研究应用的AFM-IR单光谱采集和映射模式的完整样品制备与数据采集方案。

引言

Bacteria are single cell prokaryotic organisms, occurring in various shapes and sizes, typically in the range of several hundred nanometers to micrometers. They exist in a variety of habitats and are essential to the existence of life. Within the human body, the majority of bacteria present in the gut are harmless and many are in fact beneficial¹. However, several bacterial species are pathogenic and cause a range of infectious diseases. Bacterial infections can lead to the development of sepsis and septic shock: a life-threatening condition, resulting from the body’s response to an infection². Sepsis is a global major health threat, with high prevalence worldwide and severe mortality rates. In 2017 alone, an estimated 50 million cases of sepsis were recorded worldwide, with 11 million of those resulting in death (approximately 20%)². Furthermore, a decrease in patient’s survival chances, due to delayed therapy, was shown to occur in an hourly manner^3,4.

Bacterial infections are treated with antibiotics. The severity of potential consequences of bacterial bloodstream infections (BSIs), together with a clear significance of quick initiation of antimicrobial therapy, prompt the need for immediate antibiotics administration. However, as the current diagnostic approaches used in clinical practice (e.g., blood culturing) require a relatively long time, antibiotics administration often occurs prior to positive BSI diagnosis⁵. This factor leads to extensive overuse of antibiotics, which—together with excessive antibiotic use in other sectors such as agriculture—creates a severe evolutionary pressure towards the development of antimicrobial resistance (AMR)^6,7. AMR is currently one of the most pressing global health issues^7,8 and, by 2050, is predicted to become the leading cause of death⁹. The development of resistance, together with the spread of AMR strains is occurring at an alarming pace^7,8,9 and exceeds, by far, the rate of discovery of new antibiotics¹⁰. New resistant phenotypes are continuously emerging worldwide, while
research dedicated towards understanding the AMR-related changes is often slow and limited by available approaches¹¹. In addition, the commonly used methods, such as polymerase chain reaction (PCR) and whole gene sequencing (WGS), focus only on genotypic changes. These are not sufficient to reveal the mechanisms of resistance¹¹, prompting an urgent need for a research tool enabling to understand the chemical composition of bacteria.

Infrared spectroscopy (IR) provides a molecular characterization of the sample and thus is a promising candidate for phenotypic bacterial probing. Since its early applications¹², a great magnitude of examples of its use was demonstrated in the literature^13,14. These include phenotypic-based identification of bacteria on genus¹⁵, species¹⁶, and strain^17,18 level. However, the spatial resolution of conventional IR is restricted to several microns due to the wavelength diffraction spatial resolution limit¹⁹. Since the size of majority of bacteria lies below that limit (e.g., Staphylococcus aureus ≈ 400 nm in diameter), conventional IR is not applicable for probing at the single-cell or intracellular level.

The spatial resolution limitation was recently overcome by combining IR spectroscopy with Atomic Force Microscopy (AFM-IR). In this instance, the IR absorption is detected indirectly, through thermal expansion of the material^19,20,21,22. In brief, the absorption of IR radiation results in a local temperature increase. This can be measured either directly²³ or through the measurement of oscillation of the AFM cantilever probe, resulting from force impulse created by IR absorption^20,21. The combinatory AFM-IR technique enables to achieve spatial resolution approaching 20 nm, providing simultaneous information about local physical properties of a sample (AFM) and its chemical composition (AFM-IR). Collection of both, single spectra from selected spots and mapping of the intensity of selected wavenumber values within a chosen area are possible.

Considering the achievable spatial resolution of AFM-IR, it is evident that the technique opens the possibility of chemical/phenotypic probing of single bacterium cell and their intracellular composition²⁴. Hitherto, several examples of the application of AFM-IR for single bacteria were demonstrated in the literature^{19,20,21,22,25,26,27,28}. These involve single spectral analysis^19,21,22 and mapping at the subcellular level^{19,22,25,26,27,28}. For example, the ability to detect intracellular lipid vesicles²⁷ and viruses²⁸ within single bacterium has been described. These results demonstrate the usefulness of AFM-IR for nanoscale studies of single bacteria and clinically relevant pathogens¹⁹.

Hence, we present a sample preparation and collection method for AFM-IR data of multilayer, monolayer and single cell bacterial samples. The protocol described herein was applied to study different species of bacteria²² and the changes in their chemical composition. In particular, the in vivo development of vancomycin resistance and daptomycin non-susceptibility was investigated in clinical pairs of S. aureus¹⁹. Both, vancomycin intermittent resistance and daptomycin non-susceptibility in S. aureus (VISA and DpR) emerged relatively recently, following the increased use and introduction of these antibiotics to clinics, constituting a significant medical problem. Furthermore, in particular, the mechanism of daptomycin non-susceptibility still remains elusive, impeding alternative drug development^19,29. The presented protocol focuses on provision of reliable AFM-IR spectra of single bacteria, which can further be analyzed using a variety of chemometric approaches, according to the experimental aims. It additionally includes the mapping approach, which is applicable for intra-cellular studies.

方案

All work conducted with pathogenic bacteria should be undertaken with appropriate safety measures in place. These include working in a laboratory with adequate biosafety level and in a biosafety cabin (PC2) as well as careful decontamination of work area with an appropriate disinfectant, e.g., 80% ethanol solution. Appropriate PPE must be worn all the time.

1. Preparation of solvents and materials

1. Solvents: Use ultrapure water as a solvent. Use purified water, autoclaved prior to the experiment to avoid any potential cross-contamination.
2. Substrate: Use any of these substrates for AFM-IR, e.g., ZnSe, CaF₂, BaF₂, etc. Since AFM-IR is, in principle, a non-destructive technique, one can apply a variety of other research tools to the same sample post AFM-IR analysis. For instance, correlation of the results with Raman spectroscopy can be performed if Raman grade CaF₂ or BaF₂ slides are used.
3. Use glass vials instead of plastic tubes as plastic can contaminate the sample.

2. Sample preparation for AFM-IR

1. Growth/incubation of sample
   1. Grow bacteria in liquid media or on solid plates. Select the type of medium, growth conditions (e.g., temperature, availability of oxygen) and growth time according to the specific requirements of the species of bacteria under investigation. For example, for S. aureus Heart Infusion (HI) agar plates can be used, with growth for 16 h in 37 ˚C in aerobic conditions.
      NOTE: To achieve the best results, the growth/incubation should yield enough bacteria that would allow the collection of a micro-pellet of sample. The specific number of colony forming units or bacterial cells depend on the type and size of the bacterium.
2. Sample deposition
   1. Using a sterile loop, carefully collect bacteria from the colonies on the agar plate and transfer them to a glass tube. Collect bacteria only from the top of the colonies. If collecting samples from a liquid culture, using a pipette, transfer approximately 1 mL of the bacterial suspension to a glass tube. The volume can be modified depending on the bacterial load.
      NOTE: It is important to attempt to not collect (or minimalize as much as possible the collection of) any medium from underneath the colony. The subsequent steps of sample preparation aim to remove any potential residual of media. Minimization of the potential medium residual from the beginning enables the spectral acquisition of data from purified bacterial cells. Steps 2.2.2 and 2.2.3 apply to samples prepared from agar plates. For samples prepared from liquid media, move to step 2.2.4.
   2. Add 1 mL of ultrapure water to the tube. Vortex until the collected bacterial pellet are no longer visible at the bottom of the tube (typically 1–2 min).
   3. Estimate the rough turbidity of the solution using, e.g., McFarland standards by visual comparison³⁰ between the prepared solution and McFarland standards. If the turbidity of bacterial suspension appears to be very low, add more bacteria from the plate using a sterile loop and vortex again. Repeat until the rough turbidity of the solution is comparable to McFarland standards 0.5 and 1. This will generally yield a good amount of bacterial pellet.
   4. Centrifuge the bacterial suspension at 3, 000 x g for 5 min to obtain a pellet.
      NOTE: Centrifugation parameters can be modified to obtain bacterial pellet. Caution should be taken if increasing the g-force, to not induce breakage of bacteria (especially in case of Gram-negative bacteria).
   5. Using a pipette, gently remove the supernatant from above the pellet. Add 1 mL of ultrapure water to the tube and vortex to re-suspend the pellet. Subsequently, centrifuge the sample as was done in step 2.2.4.
   6. Repeat the washing procedure (steps 2.2.2 and 2.2.4) at least three times. In case of collection of the initial sample from liquid media, repeat the procedure at least four times (media removal followed by three washes).
   7. After the final wash, remove the supernatant, add ultrapure water and vortex for at least 2 min. Subsequently, deposit 5 µL of the sample on the substrate (e.g., Raman grade CaF₂).
   8. If the desired thickness of the sample is a multilayer of bacteria, leave the sample to air-dry.
   9. If the desired thickness is monolayer or individual bacteria, immediately after depositing the sample (step 2.2.7) add between 20–100 µL of ultrapure water and mix gently with a pipette tip. Leave to air-dry.
      NOTE: The exact volume of water can vary between experiments as it is dependent on many factors (e.g., size of the organism, density of the pellet, etc.) and is, therefore, best determined empirically. Preparation of a series of samples with varying volumes of ultrapure water added enables one to select a sample with the desired thickness/density of bacteria. The thickness/density of bacteria can be easily visualized via AFM in the subsequent stages. Examples of AFM images from monolayer and single cell samples are shown in Figure 1A–H.
   10. Mount the substrate on an AFM metal specimen disk using double-sided adhesive tape.

3. Instrument preparation

NOTE: The instrumental procedures described here are for the instrument listed in the Table of Materials. The detail instrumental procedure may differ slightly from the one described here if using a newer model of the AFM-IR instrument.

1. Switch on and initialize the instrument by pressing the Initialize button. Ensure that the laser shutter is in the Open position for the laser test.
2. If a purging system is set up, purge the instrument with N₂ by turning on the flow of N₂. Adjust the nitrogen purge to achieve a stable humidity level (for example, 20%). Ensure that the humidity does not fluctuate during measurements and between background and sample data collection. Allowing approximately 20 min for the humidity levels to stabilize is recommended.
3. Load the sample into the sample chamber by pressing the Load button. Sample loading is conducted through the software wizard. While operating the software wizard, first focus on the tip, using arrows to move the microscope stage in the Z-direction and click on Next. Secondly, adjust the data collection spot, using arrows guiding the in-plane movement and align the AFM laser and AFM detector using the knobs on the top of the AFM head. Subsequently, focus on the sample surface by moving the microscope stage in Z-direction.
   NOTE: Detailed illustrations of each step of sample loading are provided in the software manual³¹. Focusing on the sample should be conducted with care. When approaching the sample surface in the Z-direction, use a slow motor speed.
4. Approach the sample without engaging by clicking on the Approach button.

4. Data collection

1. Background
   1. Prior to data acquisition, collect the background. For background collection, ensure that the laser shutter is in the Open position. Select the spectral range and resolution (depending on the aim of the analysis) and the number of scans and number of co-averages of background. These are generally recommended to be high (e.g., 1024 scans and 3 co-averages).
      NOTE: In general, spectral resolution of 4 cm^-1 or 8 cm^-1 and spectral ranges of 3,200 cm^-1–2,800 cm^-1 and 1,800 cm^-1–900 cm^-1 are recommended.
   2. After acquisition of the background, save the background file. The file is not stored automatically. Change the laser shutter position to Close.
2. Sample – single spectra
   1. Press the Engage button to engage to the sample. The system will begin to approach the sample surface, until direct contact is detected.
      NOTE: Set point used in this work ranged between 0.15–2 V and the feedback gains (I Gain and P Gain) would typically be set to 3 and 10. NIR2 contact probes are commonly used with nanoIR2 system (model: PR-EX-nIR2-10, resonance frequency (kHz): 13 +/−4 kHz, spring constant (N/m): 0.07−0.4 Nm^-1).
   2. Collect an AFM image to visualize the surface. In the first instance, scan a larger area (e.g., 50 x 50 µm) with lower spatial resolution (e.g., 200 x 200 points) (Figure 1I).
      NOTE: AFM-IR data is always collected in contact mode, however, the AFM data can be collected in contact or tapping mode.
   3. From the AFM height/deflection image, select a specific area of interest and re-image it with higher spatial resolution (Figure 1J–K). Ensure that the speed of data collection is appropriate, with slow tip movement (e.g., Scan Rate 0.2–0.4 Hz).
   4. Select the measurement spot (e.g., single bacterium) and move the tip to the spot.
   5. Align the IR laser. For this purpose, use a wavenumber at which the sample will absorb. For biological materials this can be, e.g., amide I (1655 cm^-1). Make sure that the Band Pass Filter is off and click on Start IR. The right graph in the nanoIR meter (FFT of the deflection displayed as amplitude vs. frequency) should show at least one clear peak and the left graph (deflection vs. time) should have a periodic waveform. If this is not the case, proceed to optimize the IR spots.
      NOTE: Even if the Fast Fourier Transform (FFT) and deflection show expected profile, it is recommended to conduct the optimization of IR spots for at least several wavenumbers, where bands are expected.
   6. Optimize the hot spots for the IR data collection using the selected wavenumber values. It can be helpful to use a conventional IR spectrum of the bacteria (e.g., ATR spectrum of bacterial pellet) to identify the positions of the bands and use them to optimize the hot spots. Select various wavenumber values (e.g., 8–10) from various spectral regions.
      NOTE: If conventional IR spectrum of bacteria of interest could not be collected prior to AFM-IR data collection, bacterial spectra available in the literature can be used as a rough guidance. The outcome of the optimization of an IR spot is an image, that presents a map of the FFT magnitude signal at each x and y location. The location with largest signal is selected automatically. Examples of such images are given in software manual³¹.
   7. After optimizing the IR spots for selected wavenumber values, define the parameters of spectral data collection: spectral region, spectral resolution, number of scans, and applied power and input these into the appropriate windows in the software. The spectral resolution should match the background resolution and the spectral region should be within the spectral region for which background was collected.
      NOTE: A generic initial set of parameters could be: spectral range: 3200 cm^-1–2800 cm^-1 and 1800 cm^-1–900 cm^-1, spectral resolution: 4 cm^-1 or 8 cm^-1, number of scans: 512–2048.
   8. If needed, adjust the laser power depending on the signal. In general, values between 8%–10% of laser power should be sufficient for good quality of signal. Higher values can be used with caution, as they may result in sample damage.
      NOTE: The percent laser power can vary depending on the type of IR laser. The percent values given here are for OPO laser.
   9. Click on Acquire to collect AFM-IR spectrum.
   10. Re-collect the AFM data from the same area after collection of the AFM-IR spectrum. This is highly recommended as it will reveal any potential drift and/or destructive influence on the sample.
   11. If the AFM-IR spectrum is satisfactory and no destructive influence on the sample is observed, proceed with data collection. If needed, define a series of points for data collection using the Array option and collected AFM height or deflection image. This option allows to collect spectra consecutively from each point with the same spectral parameters as defined for a single spectrum.
   12. If the AFM image collected after collection of AFM-IR spectrum reveals destructive influence on the sample (typically a burned spot), reduce the power; select a different spot and repeat steps 4.3.8–4.3.11.
   13. If the signal in AFM-IR spectrum is not satisfactory, check the correctness of optimization of IR spots (step 4.3.6). If it is correct, increase the laser power slightly and repeat steps 4.3.7–4.3.11. This can be repeated until satisfactory signal is achieved.
3. Sample – imaging approach
   NOTE: It is highly recommended to record a single AFM-IR spectrum of the bacterium prior to collecting an intensity distribution image for a selected wavenumber value.
   1. Record an AFM image of the chosen sample area. To do this, first collect AFM image of a larger area with lower spatial resolution (e.g., 50 x 50 μm, 200 x 200 points), then select a region of interest and collect an AFM image with increased spatial resolution (as illustrated in Figure 1I–K).
   2. Select the wavenumber values for AFM-IR imaging.
   3. Ensure that the IR spot of the laser is optimized for the selected wavenumber values (step 4.3.6). If the IR spot is not optimized for some wavenumbers (no clear maximum), optimize it for them.
   4. Define the parameters of the imaged area: the width and height, number of data points in X and Y direction.
      NOTE: If the consecutive selection of spots from the previous AFM images is applied (as demonstrated in Figure 1I–K), the width and height fields will be automatically filled up, upon marking of the area.
   5. Define the parameters of spectral signal acquisition: the wavelength, number of scans, and laser power.
      NOTE: The number of scans needs to be kept within reason. 64 or 32 scans will typically allow a sufficient amount of signal.
   6. Define the parameters of AFM tip movement by clicking on the Scan Rate. The higher the number of scans in the previous step and the number of data points in X direction, the slower will the tip movements need to be. Lack of adjustment between these parameters will result in too fast movement of the tip, preventing the actual acquisition of defined number of scans from each point.
      NOTE: For example, for an appropriate collection of IR signal with 64 co-additions and 200 points, set the Scan Rate as 0.07 kHz.
   7. Make sure that the Enable IR imaging box is ticked.
   8. Begin imaging. AFM-IR of the intensity of signal at the selected wavenumber will be collected simultaneously with the AFM data from that area.
      NOTE: When the OPO laser is used, it is possible to additionally collect simultaneously contact resonance peak frequency image. This can be used to obtain information about the relative stiffness of the sample at different locations.
   9. Use the Capture Sequence window to set a consecutive collection of AFM-IR data from the same area with the same parameters, but for different wavenumber values. To do that, open the Capture Sequence window, type in each wavenumber, and define the applied laser power (for each wavenumber).
   10. Export the collected data (AFM and AFM-IR, single spectra and imaging) into various formats and analyze it using methods adequate for specific research aims.

结果

所述方案能够根据样品的初始浓度和所加水量，在基底上获得多种类型的细菌细胞分布。图1展示了利用该方案制备的革兰氏阳性菌（S. aureus）和革兰氏阴性菌（Escherichia coli）单层膜及单细胞样品的原子力显微镜图像（高度和偏转模式）示例。

此处描述的方案可用于单个细菌的细胞内和细胞外结构的原子力显微镜-红外光谱（AFM-IR）成像。该应用的一个示例如下所示 图2，展示了在细胞分裂过程中监测空间定位化学变化的结果 S. aureus 细胞。尽管空气干燥通常被视为细菌制备的一种固定方法，但细菌本身对外界因素（如温度）具有很强的抵抗力，且已有报道称其可在脱水条件下存活³². 此处展示的结果来自风干样品。通过原子力显微镜成像观察并监测了细胞分裂前发生的隔膜形成过程图2A–D通过连续采集同一区域的12张图像（单张图像采集约需20分钟）完成。 图2A–D 显示4张选定的原子力显微镜（AFM）图像，每张图像采集间隔时间约为40分钟。所形成的结构（隔膜）高度为45 nm。在AFM高度图和偏转图中均可清晰观察到所形成的隔膜（图2E–F）。从细胞及隔膜区域记录的AFM-IR光谱（图2G，起点位置已标记 图2F) 在比较前均以酰胺 I 谱带为基准进行归一化处理，以最小化数据采集点之间样品厚度差异带来的影响。隔膜的 AFM-IR 光谱特征表现为在 1240 和 1090 cm⁻¹ 处的谱带相对强度更高^-1 与从细胞区域采集的AFM-IR谱相比，这些归属于细胞壁组分中的碳水化合物和磷酸二酯基团（包括例如肽聚糖和壁磷壁酸）²².

所述方案也可用于比较多个不同样本间的单个光谱。该应用示例及其结果如下所示 图3 和 图4本研究旨在确定万古霉素间歇性耐药在体内发育过程中所发生的化学变化 S. aureus （VISA）。为此，采集了患者的临床配对样本，其中亲本菌株在患者入院时、抗生素治疗（万古霉素敏感）前分离获得 S. aureus，VSSA）以及同一患者在使用抗生素后因治疗失败而分离出的子代菌株。样本进一步在琼脂培养基上培养，并根据方案进行制备（图3A–B）。AFM-IR 谱图采集自 VSSA 和 VISA 的多个单个细菌（及多个样品），随后采用多种化学计量学方法进行分析。图3C).

在VSSA和VISA细胞之间未观察到形态学差异（图4A–C）。然而，原子力显微镜-红外光谱（AFM-IR）图谱（图4D,F）及其二阶导数图谱（图4E,G）显示，耐药菌株与敏感菌株在化学组成上存在明显差异。与细胞壁组分相关的碳水化合物和磷酸二酯基团的特征峰（特别是1088 cm^-1处的峰）在耐药菌株中的相对强度明显高于敏感菌株。值得注意的是，所有记录的光谱（VISA：81条，VSSA：88条）均表现出较小的标准偏差，表明来自同一菌株的不同样品所记录的光谱数据具有良好的可重复性，且无法区分同一菌株不同样品间的光谱差异。观察到的这些差异提示，与敏感菌株相比，耐药菌株的细胞壁增厚，该结果与其他文献报道一致^33,34。

图1：用于AFM-IR测量的不同细菌样品的代表性AFM图像。根据在基底上的稀释程度，本方案可获得细菌的多层膜、单层膜以及单细胞样品。代表性AFM图像包括：(A–D) 单层膜样品和 (E–H) 单细胞样品，其中(A,B,E,F)为革兰氏阳性菌（S. aureus），(C,D,G,H)为革兰氏阴性菌（E. coli）。(A,C,E,G)为高度图像，(B,D,F,H)为对应的悬臂偏转图像。成像区域尺寸：(A-D,G,H) 20 × 20 µm，(E,F) 50 × 50 µm。(I–K) 为AFM-IR映射区域的连续选择过程，以单个S. aureus细胞为例，通过逐步提高空间分辨率的AFM成像实现。每幅图像均采集200 × 200个数据点，随着成像区域尺寸减小，空间分辨率逐步提高。成像区域尺寸：(I) 40 × 40 µm，(J) 20 × 20 µm，(K) 2.24 × 2.24 µm。(I)中的黑色方框标示了(J)中成像的区域，(J)中的黑色方框标示了(K)中成像的区域。请点击此处查看该图的放大版本。

图 2：通过 AFM-IR 监测 S. aureus 细胞分裂。（A–D）S. aureus 细胞的原子力显微镜（AFM）图像，显示细胞分裂前隔膜的形成过程。成像区域大小：2 × 2 µm。这些图像选自一个更大的图像序列（每 20 分钟记录 12 幅图像），代表每 40 分钟记录的数据。（E–F）在细胞隔膜形成结束时记录的 AFM 高度和偏转图像，图中标记了 AFM-IR 光谱采集的位置。成像区域大小为 1.17 × 1.15 µm。新形成结构的高度为 45 nm。（G）从细胞区域（黑色）和隔膜区域（红色）（在 F 中标记）记录的 AFM-IR 光谱，波数范围为 1400–900 cm^-1。两条光谱均以酰胺 I 带归一化，显示隔膜区域细胞壁组分的相对强度增加。本图经 K. Kochan 等人²² 修改。 请点击此处查看此图的放大版本。

图3：原子力显微镜-红外光谱（AFM-IR）研究抗菌耐药性的实验设计概览。（A）样品来源与初始制备：敏感亲本菌株取自患者在抗生素治疗前的样本，耐药子代菌株则取自同一患者在抗生素治疗失败后的样本（体内耐药性发展）。细菌在心浸液（Heart Infusion, HI）琼脂培养基上于37 °C培养16小时。（B）AFM-IR后续样品制备，包括样品收集、细菌沉淀物洗涤（3次）及样品沉积。（C）AFM-IR数据采集与分析：AFM高度图像与AFM-IR光谱（1800–900 cm^-1）。AFM成像区域大小为1.7 × 1.4 µm。AFM-IR光谱采集自细胞中部。随后采用化学计量学方法对数据进行分析，包括层次聚类分析。本图经K. Kochan等¹⁹修改。 请点击此处查看此图的放大版本。

图4：万古霉素中间体化学变化研究的原子力显微镜与原子力显微红外光谱结果 S. aureus （VISA）与万古霉素敏感株相比 S. aureus 临床配对中的VSSA。 AFM图像的（A–BVISA 和（C) VSSA 单细胞样本。成像区域的大小：(A，C) 40 × 40 µm，(B2.56 × 2.45 µm。D–E平均AFM-IR光谱及其（F–G）二阶导数，用于：（D，FVISA 和（E, GVSSA 细胞在 1800–900 cm 的光谱范围内^-1谱图显示的是81个（VISA）和88个（VSSA）独立谱图的平均值，并同时呈现标准差（SD）。所有独立谱图在归一化后进行平均。主要谱带已在图中标出。F–G此图已根据 K. Kochan 等人的研究修改。¹⁹. 请点击此处以查看此图的放大版本。

图5：金黄色葡萄球菌（S. aureus）细胞在选定波数下连续采集AFM-IR图谱时的图像漂移。（上排）：AFM图像与相应的（下排）AFM-IR图谱同步记录，后者基于选定波数处红外信号的强度。波数值（966、1055、1079、1106、1234、1398、1454、1540、1656、1740 cm^-1）标注在下排图像上方。每组图像（AFM图像与AFM-IR图谱）均在前一组图像采集完成后立即记录（每组约耗时40分钟）。成像/映射区域的尺寸为：1.54 × 1.57 µm。图像之间可明显观察到漂移现象。请点击此处查看该图的放大版本。

讨论

红外光谱技术在表征多种生物样品化学组成方面的有效性已得到广泛认可。在过去十年中，红外光谱已成为细菌研究中一种极具前景的工具^{12,13,14,15,16,17}。由于该技术是少数能够通过化学组成实现表型表征的方法之一，因此在微生物学领域持续受到广泛关注。然而，传统傅里叶变换红外显微技术的主要缺点在于其空间分辨率有限，难以对细菌进行单细胞及亚细胞水平的研究。事实上，细菌体积微小不仅对红外技术构成挑战，也限制了绝大多数分析技术的应用。因此，可用于细菌单细胞和亚细胞研究的实验手段极为有限。原子力显微镜与红外技术的结合能够突破红外光谱的空间分辨率限制，为细菌研究提供一种新型工具，实现对化学组成的纳米尺度探测。

该技术不仅限于单细胞研究，还可用于检测多种不同厚度的样品。毫无疑问，干净且细致的样品制备对于获得高质量图像至关重要。本文提供的方案可用于制备多种细菌的多层、单层和/或单细胞样品（图1）。所获得的样品状态取决于多个因素，包括初始细菌载量、洗涤后的稀释倍数以及在基底上的进一步稀释。将洗涤后的菌体沉淀稀释后、沉积到基底前，通常可获得足量样品，用于制备多个样本。因此，为了在基底上获得理想的样品分布，通常建议制备一系列不同稀释梯度的样品。对于以采集原子力显微-红外光谱（AFM-IR）而非亚细胞成像为主要目标的研究，调整样品量（例如，从单层变为多层）可能有助于增强信号强度。

样本制备中的另一个关键环节是适当去除培养基残留物。根据所选用的样本培养方法，样本可从液体培养基或琼脂平板中收集。在这两种情况下，样本中均可能存在培养基残留，尽管从琼脂平板收集的样本中残留量显著较少。由于细菌培养基含有大量各种生物成分，因此确保充分去除培养基残留至关重要。我们建议对琼脂平板样本进行三次超纯水洗涤，对从液体培养基中收集的样本至少进行四次洗涤。如有需要，可增加洗涤次数；但为了不同样本之间的可比性，各样本间的洗涤次数应保持一致。本方案采用水而非磷酸盐缓冲液（PBS）或生理盐水等溶剂进行洗涤。PBS和生理盐水在空气干燥后均会形成晶体，可能对细菌造成损伤。此外，两者在红外光谱中均会产生强烈的吸收峰，尤其是PBS在指纹区含有多个强吸收峰。目前无法使用生理盐水或PBS是该技术的一个重要局限性。通常情况下，用水洗涤不会对细菌产生破坏性影响；但操作时仍需谨慎，且应尽可能缩短细菌与水接触的时间。若样本制备流程需在洗涤阶段暂停，建议在去除水分后将样本保持为沉淀状态。这一点对于细胞壁较薄、更易破裂的革兰氏阴性菌尤为重要。

为确保获得准确且高质量的原子力显微红外（AFM-IR）数据，数据采集过程中的若干方面至关重要。首先，正确采集背景信号是数据获取的基础。尤其需要在整个背景采集过程以及背景与样品采集之间维持稳定的湿度水平。为实现这一点，我们建议使用氮气对仪器进行吹扫，并将湿度水平控制在不超过25%。在高湿度地区，若不进行吹扫将可能带来显著限制。其次，必须强调对红外光斑进行适当优化的重要性。为获得最佳结果，预先了解吸收峰最大值的位置将非常有益。例如，可利用细菌沉淀物的传统红外光谱来确定样品中预期出现的吸收峰位置。若无法获取此类数据，替代方案是参考文献中已有的红外光谱，或从细菌中合理预期的吸收峰位置开始优化（如酰胺I带和酰胺II带）。第三，在数据采集过程中，需特别强调谨慎选择激光功率的重要性（以实现良好的信噪比），因为过高功率可能对样品造成破坏。推荐的功率取决于样品厚度，仪器手册中提供了大致的指导建议³¹。我们建议通过采集测量后的原子力显微镜（AFM）图像，经验性地检验样品状态，以识别任何潜在的破坏性影响。此外，在采集AFM-IR光谱前后，对同一区域采集AFM图像，可有效验证未发生样品漂移，从而确认光谱确实来源于细胞中选定的位置。在采用成像模式时，即连续在选定波数下成像红外强度时，漂移的可能性尤为值得关注。图5展示了一个实例。实验开始时定义了成像区域，并期望该区域在所有波数下保持一致。然而，在每幅AFM高度图（及相应的红外波数强度图）之间均可见明显漂移，每幅图谱的采集时间约为40分钟。因此，对于采集成像数据的用户，我们建议始终选择略大于目标样品的成像区域，以确保即使发生漂移，目标样品仍保留在成像范围内。

该方案的潜在局限性包括无法在生理溶液（如生理盐水或PBS）中以水合状态收集数据，如上所述。此外，特别是在高湿度环境中，通常需要使用氮气吹扫。此外，该方案仅能检测小至100 nm的生物体，因此无法用于更小结构的分析。尽管可通过使用不同激光器（例如量子级联激光器，可实现20 nm的空间分辨率）来克服此限制，但这也会带来光谱范围受限以及难以获得良好的信噪比等问题。最后，对柔软表面进行检测时可能存在挑战，探针可能无法准确感知表面而继续下压超过接触点，直至发生断裂。虽然细菌样品通常不会出现此类问题，但在检测较软样品时可能发生。在此类情况下，建议尝试在样品附近的基底洁净表面上进行探针接触。

该方案可用于多种类型的细菌学研究，包括不同样本间的比较研究以及亚细胞水平的检测。根据研究目的，可采用化学计量学方法对单个光谱和成像模式的数据进行分析³⁵。此外，通过增加固定步骤，该方案还可修改后应用于其他生物材料（如真菌、酵母、细胞等）。

材料

本文使用的材料清单

姓名	公司	目录编号	评论
AFM金属样品圆盘	PST ProSciTech Pty Ltd	GA530-15	推荐使用15 mm
Anasys AFM-IR nanoIR2	Anasys Instruments		型号：nanoIR2
nanoIR2接触模式探针	Bruker / Anasys Instruments	-	型号：PR-EX-NIR2
Heraeus Pico 17 微型离心机	Thermo Scientific	-	-
Matlab	Mathworks Inc	-	多元数据处理软件
1.5 mL 微量离心管	Heathrow Scientific	HEA4323	可用其他品牌微量离心管替代
nanoIR2仪器	Bruker / Anasys Instruments	-	-
PLS工具箱	Mathworks Inc	-	Matlab图形用户界面
选用的细菌培养基（例如HBA哥伦比亚平板）	Thermo Fisher	PP2001	所列培养基仅为示例，可根据实验类型更换其他类型
选用的细菌菌株	-	-	来源取决于研究目的（如患者分离株、ATCC菌株等）
基底材料（例如拉曼级CaF2）	Crystran	CAFP13-2R	推荐尺寸：13 mm Ø × 2.0 mm
1000 µl 移液器吸头	Axygen	T-1000-B	-
200 µl 移液器吸头	Axygen	T-200-C	-
0.5–10 µl 移液器吸头	Axygen	T-300-R	-
超纯水	-	-	-