An optimized method for the detection and spatial distribution of aminoglycoside and vancomycin antibiotics in tissue sections by mass spectrometry imaging

Abstract

Suboptimal antibiotic dosing has been identified as one of the key drivers in the development of multidrug-resistant (MDR) bacteria that have become a global health concern. Aminoglycosides and vancomycin are broad-spectrum antibiotics used to treat critically ill patients infected by a variety of MDR bacterial species. Resistance to these antibiotics is becoming more prevalent. In order to design proper antibiotic regimens that maximize efficacy and minimize the development of resistance, it is pivotal to obtain the in situ pharmacokinetic–pharmacodynamic profiles at the sites of infection. Mass spectrometry imaging (MSI) is the ideal technique to achieve this. Aminoglycosides, due to their structure, suffer from poor ionization efficiency. Additionally, ion suppression effects by endogenous molecules greatly inhibit the detection of aminoglycosides and vancomycin at therapeutic levels. In the current study, an optimized method was developed that enabled the detection of these antibiotics by MSI. Tissue spotting experiments demonstrated a 5-, 15-, 35-, and 54-fold increase in detection sensitivity in the washed samples for kanamycin, amikacin, streptomycin, and vancomycin, respectively. Tissue mimetic models were utilized to optimize the washing time and matrix additive concentration. These studies determined the improved limit of detection was 40 to 5 μg/g of tissue for vancomycin and streptomycin, and 40 to 10 μg/g of tissue for kanamycin and amikacin. The optimized protocol was applied to lung sections from mice dosed with therapeutic levels of kanamycin and vancomycin. The washing protocol enabled the first drug distribution investigations of aminoglycosides and vancomycin by MSI, paving the way for site-of-disease antibiotic penetration studies.

1 |. INTRODUCTION

The emergence of antimicrobial resistance and the increasing prevalence of multidrug-resistant (MDR) and extremely drug-resistant (XDR) bacterial pathogens are a global health crisis.^1–3 In addition to causing treatment failure that results in increased morbidity and mortality, suboptimal dosing has been identified as one of the major contributors to the emergence of antimicrobial resistance.^4,5 Several factors contribute to suboptimal dosing, but perhaps one of the most impactful and overlooked, and an area where there is limited clinical data, is the levels or concentration of a given antibiotic at the site of infection where it is needed for efficacy.⁵ Investigating the penetration of antibiotics into the site of infection, at concentrations and dosing times above the minimum inhibitory concentration (MIC) required for sterilization, is key to developing optimal dosing regimens. A wide variety of pharmacokinetic and pharmacodynamic (PK-PD) models have been established to aid in predicting the probability of clinical efficacy and thus dose optimization.^6–9 None of these models, however, consider the in situ spatial distribution of antibiotics within the sites of infection. Additionally, infection sites are often complex microenvironments that have varying levels of inflammation, necrosis, and vasculature, which significantly influences antibiotic penetration. Antibiotic penetration into these sites cannot be elucidated using conventional PK-PD techniques.

The introduction of mass spectrometry imaging (MSI) to traditional PK-PD enabled the label-free detection of drugs, their metabolites, and biomarkers of toxicity while maintaining the spatial information necessary to evaluate drug distribution in heterogeneous environments.^10–16 By directly overlaying the MSI drug distribution data with its corresponding histologically stained section, MSI offers the advantage of identifying cell- and region-specific drug penetration and accumulation. While many studies have evaluated antibiotic PK in normal, non-infected tissue, few MSI studies have focused on antibiotic PK-PD at sites of infection.^17–21 One of the most fundamental studies published in this field demonstrated the power of MSI when utilized for antibiotic PK-PD during the evaluation of anti-tuberculosis (TB) drugs.²¹ This group demonstrated that antibiotic penetration into TB lesions correlated to sterilizing activity and treatment shortening. These findings explained why certain antibiotics such as rifampicin had sterilizing capabilities and why another drug, clofazimine, had limited mouse efficacy and clinical utility.^22–24 This relationship would not have been established without MSI. This approach has become fundamental in identifying optimal TB dosing regimens and aiding in the development of new anti-infective TB therapies.²⁵ These studies demonstrate the significance of adequate measurements of the lesion-specific pharmacokinetic profiles of antibiotics and the critical role that MSI plays in understanding the relationship between antibiotic penetration and therapeutic efficacy. Such studies are vital to avoid sub-optimal dosing that will result in poor clinical outcome and the development of MDR and XDR pathogens.

The aminoglycosides, amikacin, kanamycin, and streptomycin, are broad-spectrum antibiotics that are used to treat a variety of severe and MDR/XDR bacterial infections, including those caused by methicillin-resistant Staphylococcus aureus (MRSA), and MDR Pseudomonas aeruginosa, Klebsiella pneumoniae, and Acinetobacter baumannii.^26,27 Amikacin and kanamycin are also used as part of a second-line therapeutic combination to treat MDR tuberculosis infections.³ Vancomycin is a tricyclic glycopeptide that is used to treat severe, life-threatening gram-positive bacterial infections, including those caused by MRSA.²⁸ Both classes of antibiotics are associated with quite severe dose-dependent nephrotoxicity,^29,30 and aminoglycosides are known to cause irreversible ototoxicity,^31,32 but they remain vital therapeutic options for treating critical and life-threatening infections. Resistance to these classes of antibiotics, however, is becoming more and more prevalent^33,34 and highlights the essential requirement for infection site PK-PD to protect the utility of these important therapeutics.^35,36 Several studies have focused on methods to determine optimal dosing for aminoglycosides and vancomycin as there remains controversy over the concentration, route of administration, frequency of dosing, and treatment duration.^37–43 This is further complicated by the recognition that different infection site locations and the resulting infections caused require different dosing regimens for clinical efficacy. All studies are aimed at improving efficacy, minimizing their toxicity, and preventing the increasing levels of resistance. None of these studies, however, concentrate on comprehensive PK-PD at the sites of infection. To date, the only data for both aminoglycosides and vancomycin have focused on clinical outcome or quantification using homogenized tissue and bone samples, thereby losing all spatial information.

The penetration of these drugs at the site of infection using MSI for lesion PK has never been ascertained before. This is primarily due to the poor ionization efficiency of aminoglycosides and ion suppression effects caused by more abundant and readily ionizable endogenous lipids.^44,45 Aminoglycosides are particularly problematic and are commonly derivatized to improve their ionization efficiency prior to quantification by mass spectrometry.⁴⁶ Multiple approaches have been developed to enhance MSI detection sensitivity of low abundant and poorly ionizable drugs and endogenous molecules. These have included solvent washing,^47–53 metal-based nanoparticle implantation,⁵⁴ on-tissue derivatization,^55–60 and post-ionization (matrix-assisted laser desorption ionization [MALDI]-2).^61–64 Tissue washing protocols were established as a rapid and effective way to enhance MSI sensitivity by significantly removing endogenous ion suppressants.⁴⁷ Washing protocols have been tailored to their specific application based on the physiochemical properties of the analytes of interest and the ion-suppressing molecules.⁴⁹ This is to ensure maximum efficacy while avoiding potential delocalization of the analytes of interest, as delocalization would result in misleading data interpretation. To date these protocols have been successfully used to enhance the MSI detection sensitivity of proteins^47,48, lipids⁵², metabolites⁵⁰ and drug compounds.^49,65

In this study, we evaluated the effects of solvent washing times and matrix salt additive concentration on the MSI detection of aminoglycosides and vancomycin directly from tissue sections. We first applied a tissue spotting method to evaluate efficacy and potential delocalization. This was followed by a tissue mimetic model to identify optimal washing times and quantify salt additive effects on detection limits. The mimetic model also enabled the determination of the improved sensitivity in μg/g of tissue. Finally, the optimized protocol was applied to mouse tissues as a proof-of-concept study. Collectively, our results validate a method that enables the MSI detection of these key antibiotics in tissue sections, for the first time.

2 |. EXPERIMENTAL

2.1 |. Materials

Sodium chloride was purchased from Sigma-Aldrich (St. Louis, MO, USA); 2,5-dihydroxybenzoic acid (DHB), hematoxylin and eosin (H&E) stains and solvents, all high-performance liquid chromatography (HPLC)-grade solvents, and the Superfrost^™ Plus slides were purchased from the Fisher Scientific Company (Waltham, MA, USA).

2.2 |. Animals

All animal studies were performed with prior approval by the Institutional Animal Care and Use Committee (IACUC) of the Center for Discovery and Innovation, Hackensack Meridian Health, New Jersey. Female CD-1 mice (Charles River Labs), 4–6 weeks old, received single intravascular dose of kanamycin at 25 mg/kg (the aminoglycoside efficacious dose in mice against nontuberculous mycobacterial disease⁶⁶) or a subcutaneous dose of vancomycin at 100 mg/kg. Three mice per cohort were euthanized 2 h post-dose, the lungs were collected, snap frozen in dry ice, and stored at −80°C until use. Lung tissue was collected from female New Zealand white (NZW) rabbits (Millbrook Farm, Concord, MA), for the preparation of the tissue spotting and tissue mimetic models. Rabbits were maintained under specific pathogen-free conditions with access to food and water ad libitum, as previously described.⁶⁷

2.3 |. Standard stock solutions for tissue spotting

Stock solutions of kanamycin, amikacin, streptomycin, and vancomycin were prepared at 2.5 mg/ml in 50% MeOH. These were then used to prepare working concentrations ranging 0.13–6.73 pmol/0.2 μl.

2.4 |. Preparation of the tissue mimetic

A cylinder-shaped tissue mimetic model composed of six layers was prepared based on a protocol published by Barry et al.⁶⁸ and was modified accordingly. In brief, approximately 1.5 g (~0.25 g/layer × six layers) of rabbit lung tissue was homogenized, without additional solvent, unsing the gentleMACS^™ Octo Dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany). This enables homogenates that are single cell suspensions. Six layers of mimetic were made containing 0, 1, 5, 10, 20, and 40 μg/g of kanamycin, amikacin, streptomycin, and vancomycin. Kanamycin and amikacin were prepared in separate mimetic models to avoid inaccuracies that may occur due to their structural similarities. The mimetic models were snap frozen and stored at −80°C until use.

2.5 |. Experimental workflow

Lung tissue and the lung mimetic models were sectioned at 10 μm using a Leica CM 1860 cryostat (Leica Biosystems, Buffalo Grove, IL, USA), thaw mounted onto Superfrost^™ Plus slides and stored at −80°C until use. All slides were desiccated for 15 min upon removal from the −80°C freezer. For the tissue spotting experiments, 0.2 μl of the working concentrations of each drug were spotted across control lung sections. The sections were then air-dried for 20 min prior to the tissue washing procedure and matrix deposition.

For tissue washing, the tissue sections with the spotted drug concentration ranges and the tissue mimetic model underwent tissue washes in a solution of chloroform that had been stored at −20°C.Each slide had two sections on, one at the top and one at the bottom; the bottom section of each slide underwent tissue washing. The tissue samples were washed by dipping and agitating the slides and in a glass coplin jar filled with chloroform. A fresh solution was used for every wash. A schematic of this process is shown in Figure S2. Several washing times were carried out to identify the most optimal; these included 15, 30, 60, and 90 s using both a single solution for each washing time and fresh solutions for 2–3 × 30 s washes each. Mouse PK tissue sections were washed for 15 and 90 s. The slides were air-dried for 10 min before matrix deposition.

Matrix deposition was carried out using the HTX M5 sprayer (HTX Technologies LLC, Chapel Hill, NC, USA). DHB at a concentration of 20 mg/ml (50% MeOH, 0.1% TFA) with and without NaCl, utilized as a matrix additive, was deposited onto the tissue sections using the following settings; 60 μl/min flow rate; 60°C nozzle temperature; 55 kPa. nitrogen; 3 mm track spacing; 900 mm/min velocity; crisscross pattern. A total of 26 and 16 passes were applied to the tissue sections and tissue mimetics, respectively.

2.6 |. MALDI-MSI data acquisition

MALDI imaging experiments were performed using a Bruker SolariX 7 T FT-ICR mass spectrometer (Bruker Daltonics, Billerica, MA, USA) equipped with a smartbeam II Nd:YAG laser, 355 nm, and a dual ESI/MALDI ion source. Data were acquired in the positive ion mode over the mass range m/z 150–3000. The transient time was 0.7340 s, resulting in an estimated resolving power of 99,000 at m/z 400. The small laser setting was used, accumulating 300 laser shots in smart walk mode, operated at a frequency of 2 kHz and 15–17% laser power. The pixel size was set to 75 × 75 μm for the tissue spotting and mimetic samples and 50 × 50 μm for the mouse lung tissue samples. For aminoglycosides, continuous accumulation of selected ions (CASI) was used. The quadrupole (Q1) mass was set to m/z of the sodium adduct for each drug compound with a CASI mass window of m/z 200. The time of flight was set to 0.5. For vancomycin, CASI was not used. The Q1 mass was set to 1250, and the time of flight was set to 1.2. Following data acquisition, the slides were washed with 70% ethanol, stained with H&E, and scanned using the Pannoramic DESK II DW scanner (3DHISTECH, Budapest, Hungary).

2.7 |. MALDI-MSI data analysis

Data were processed using the SCiLS Lab MVS, version 2020a Pro (Bruker Daltonics, Billerica, MA, USA). The ion maps were color-coded using the “Fire” scale. The H&E images were overlaid with the ion maps. For tissue spotting data, regions of interest (ROIs) were drawn around each spot to obtain average signal intensities that were used to calculate intensity fold changes and plot calibration curves for each antibiotic.

3 |. RESULTS AND DISCUSSION

3.1 |. Tissue washing efficacy and evaluation of delocalization

The investigation of delocalization effects is important for all new washing protocols as delocalization of target analytes from their actual site of distribution would lead to false data interpretation. In order to limit delocalization, the washing solvent needs to be selected based on the solubility of the analytes of interest.⁴⁹ Aminoglycosides and vancomycin are insoluble in chloroform and ethyl acetate. Chloroform was selected for method development based on the ability of this solvent to reduce ion-suppressing lipid species from tissue, while maintaining cell membrane structure and integrity. To evaluate the efficacy of the chloroform washing protocol on the detection of these antibiotics and to investigate any potential delocalization effects, drug concentrations ranging from 0.13–6.73 pmol/0.2 μl were spotted directly onto control sections of lung tissue. This tissue spotting procedure is commonly used for drug quantitation by MALDI-MSI as this method enables evaluation of detection sensitivity and the ion suppression effects caused by endogenous molecules within the local tissue environment.^69–71 This method has also been utilized to evaluate the efficacy and any potential delocalization effects of a controlled pH tissue washing protocol on the detection of cimetidine and imipramine.⁴⁹

Due to their structures, only the sodium adducts of kanamycin, amikacin, and vancomycin were detected following MSI analysis of the on-tissue spotting experiments. This ion formation is commonly reported during the MALDI MS analysis of carbohydrate/glycan type compounds such as these.^72,73 Different tissue washing times were investigated, ranging from 15 to 90 s. The results presented in Figure 1 for kanamycin, amikacin, streptomycin, and vancomycin were taken from the 3 × 30 s wash and demonstrate a significant increase in detection sensitivity in the washed samples compared to the unwashed samples. The signal for each antibiotic remained within the area where it was spotted, demonstrating minimal-to-no delocalization with this method, this is demonstrated more clearly in supplementary Figure S2a. This is important as the penetration of antibiotics into sites of infection within tissue sections inform on the ability of the drug compound to reach the target area needed for bacterial sterilization. Delocalization would thus result in inaccurate lesion PK data.

FIGURE 1.

MALDI MS images of the tissue spotting analysis of amikacin, kanamycin, streptomycin, and vancomycin. Spots of five different antibiotic concentrations on washed (W) and unwashed (U) tissue sections are shown. The fold change of the average intensity for the highest concentration of each antibiotic deposited on the lung section is shown in the bottom right of each panel under their respective spot

The fold change increase in detection sensitivity for washed compared to the unwashed lung sections was calculated from average intensities taken from ROIs, drawn around the spots that had the highest concentration deposited, as previously reported.⁶⁵ The fold change increase varied depending on the antibiotic analyzed. In ascending order, kanamycin, streptomycin, and amikacin demonstrated a 5-, 15-, and 40-fold improvement, respectively. Vancomycin, the most readily ionizable of the compounds, demonstrated a 54-fold increase in signal intensity (Figure 1). Technical replicates were investigated to ensure reproducibility of the method and the results from these studies are presented in Figure S1.

3.2 |. Method optimization—Lung tissue mimetic

The tissue spotting method mentioned above, even though widely used to simulate an in vivo biological environment and to establish quantitative calibration curves, does not fully recapitulate the actual situation of extracting endogenous drugs from tissue. The tissue mimetic model, based on tissue homogenates that better mimic both ion suppression and extraction efficiency of drug compounds, provides quantitative MALDI imaging data with improved accuracy as it is based on μg/g of drug concentrations in tissue.⁷⁴ For these reasons, tissue mimetic models of kanamycin and streptomycin were used to further optimize washing times and matrix additive concentration. Initial experiments extrapolated on the tissue washing times carried out during the spotting analysis. Washing times included 30, 60, and 90 s in a single solution of chloroform, in addition to 2 × and 3 × 30 s washes, in two and three different solutions of chloroform, respectively. This was to evaluate the relationship between washing times, number of washes and the removal of ion-suppressing endogenous lipid species compared to the limit of detection (LOD) for each antibiotic under investigation. In the unwashed samples, both antibiotics were barely detectable at 40 μg/g of tissue, as demonstrated in the top panel of Figure S3 for each compound. Kanamycin demonstrated an improved LOD from 40 μg/g of tissue to 10 μg/g following tissue washing. Variations in the washing time from 30–60 s demonstrated an increase in signal intensity, but washing times in excess of 60 s provided diminishing return, as shown in Figure S3. Streptomycin, the most readily detectable of the aminoglycosides, demonstrated improved LOD from 40 to 5 μg/g of tissue in the washed versus unwashed samples, respectively. Again, improvements were observed with the 30 and 60 s washes, with slight increases in the signal intensity for this aminoglycoside when using the 3 × 30 s washing protocol. The improvement in LOD and detection sensitivity between the unwashed and washed samples directly correlates with the reduced detection of ion-suppressing lipid species, as is shown in Figure S4. Phosphatidylcholines (PC) predominate in positive ion mode as they readily ionize and are known to cause suppression of pharmaceutical compounds, metabolites, and lipid species that do not ionize as readily.^50,65 The most abundant PC in lung tissue, PC(16:0/16:0), is detected at high intensity in the unwashed sample, followed by medium intensity in the 30 s washed sample, and low intensity in the samples washed for 60 s and more. These results demonstrate a direct correlation between the removal of ion-suppressing lipids and the improved LOD of the aminoglycosides.

Carbohydrate-related molecules such as the aminoglycosides and the tricyclic glycopeptide, vancomycin, are predominantly detected as alkali metals (sodium and potassium) during MALDI MS analysis.^75,76 Introduction of alkali salts as cationization agents increases the signal intensity of such molecules thereby improving their detection sensitivity during MALDI analysis.^77,78 With this in mind, the next stage of optimization focused on the use of sodium additives to the MALDI matrix solvent system, as this ion formation was detected with the highest abundance in these studies. Minimal improvement was observed in the LODs or signal intensity of kanamycin and streptomycin using the addition of 5 or 10 mM NaCl, as shown in Figure S5. As improvements in matrix deposition/crystallization were observed with 5 mM NaCl, this was included in the optimized solvent system going forward.

The optimized washing (3 × 30 s) and matrix additive (5 mM NaCl) protocol was then utilized along with the MS CASI function for the analysis of all three aminoglycosides and vancomycin tissue mimetic models. The results presented in Figure 2 demonstrated an improved LOD from 40 μg/g in the unwashed sections to 5 μg/g in the washed sections for vancomycin and streptomycin. Kanamycin and amikacin demonstrated an improved LOD from 40 μg/g in the unwashed sections to 10 μg/g in the washed sections, indicating a significant enhancement in their detection sensitivity. It is believed that further enhancement of matrix deposition cycles would increase the achievable LOD further as only 16 spraying cycles were used for the mimetic models, compared to 26 cycles for the tissue spotting experiments. The latter cycle number is more commonly used for both lipid^79,80 and drug^17,21 analysis using DHB at ~20 mg/ml. Using the MS CASI function also improved the LOD and detection sensitivity of all antibiotics in the unwashed samples as can be observed by comparing Figure 2 with Figure S3.

FIGURE 2.

MALDI MS images of amikacin, kanamycin, streptomycin, and vancomycin from lung tissue mimetic models comparing washed versus unwashed sections. Antibiotic concentration in each mimetic section is shown at the top of the figure in μg/g of lung tissue. Representative mimetic H&E stained optical images are shown below each MSI image

3.3 |. Antibiotic imaging in mouse tissue sections

To evaluate and validate the in vitro protocol in an in vivo setting, mice were dosed with therapeutic levels of kanamycin, the most difficult of the aminoglycosides to detect, and vancomycin. Experiments initially utilized the 3 × 30 s washing protocol as optimized in the tissue mimetic section. Successful detection of these antibiotics was achieved in both washed samples; little-to-no signal was observed in the unwashed samples. The longer washing times in lung samples taken from dosed mice did result in slight delocalization, which was particularly evident for kanamycin, as shown in Figure S7. A lower abundant signal for kanamycin was detected in the region of bronchi, a large empty airway that is void of tissue, thus indicating the longer washing times are causing slight delocalization. Based on these results the washing times were reduced to 15 s and this limited the delocalization effects observed with kanamycin as minimal signal was detected in non-tissue regions or regions surrounding the outside of the lung sections. This washing time in lung tissue resulted in the sensitive detection of both kanamycin and vancomycin compared to the unwashed sections, with minimal-to-no signs of delocalization, as shown in Figures 3 and 4. The technical replicates and further demonstration of the minimal-to-no delocalization with the 15 s washing protocol are presented in supplementary Figures S8 and S9. The results demonstrate a significant improvement in detection sensitivity that correlates with a decrease in signal for endogenous lipids that are known to cause ion suppression. The PC headgroup ion that is commonly detected from the labile fragmentation of PCs during the MALDI MS process was also detected with reduced intensity in the washed versus unwashed section, as shown in the left bottom corner of Figure 3. This ion was chosen to display the effects of tissue washing on PC lipids in the kanamycin sample as the CASI window used for MSI acquisition did not include the commonly observed PCs detected in the m/z 650–1000 mass range. PC(16:0/16:0) demonstrated a significant reduction in signal intensity in the washed section compared to the unwashed section (Figure 4). The intensity of the head group presented in Figure 3 is higher than the intact lipid presented in Figure 4 due to the PC headgroup ion signal accumulating from the labile fragmentation of sphingomyelins (SM), PCs, and lysoPCs. The right bottom corner of both Figures 3 and 4 demonstrates no signal was detected for the dehydrated cholesterol ion in the kanamycin and vancomycin washed samples, respectively. Technical replicates were investigated to ensure reproducibility of the method and the results from these studies are presented in Figures S8 and S9. Collectively, these results demonstrate that reduction of ion-suppressing endogenous lipids enabled the detection of aminoglycosides and vancomycin that would not be possible otherwise.

FIGURE 3.

MALDI MS images of washed and unwashed mouse lung sections, therapeutic level of kanamycin at 25 mg/kg. Images of several major lipid species or lipid fragment ions in the washed and unwashed samples are shown below each antibiotic image. The corresponding H&E stained sections utilized for MSI are shown besides the antibiotic images. Scale bar represents 3 mm

FIGURE 4.

MALDI MS images of washed and unwashed mouse lung sections, therapeutic level of vancomycin at 100 mg/kg. Images of several major lipid species or lipid fragment ions in the washed and unwashed samples are shown below each antibiotic image. The corresponding H&E stained sections utilized for MSI are shown besides the antibiotic images. Scale bar represents 3 mm

Comparison of the H&E stained sections in the washed and unwashed samples confirms that the solvent washing had negligible influences on tissue integrity (Figure S6). This is of great importance when histological features of infection sites (e.g., immune cell populations, hyperplasia, and vessels) need to be analyzed in correlation with the imaging data.

4 |. CONCLUSION

An optimized sample preparation protocol was designed to specifically target the key frontline antibiotics, kanamycin, amikacin, streptomycin, and vancomycin, to enable their MSI detection for the first time. Method development, optimization, and verification were achieved in a multi-step process utilizing tissue spotting, tissue mimetic, and mouse tissue sections. Combined, these results demonstrated that the washing protocol was reproducible and significantly improved the detection sensitivity of all antibiotics tested. The addition of NaCl to the matrix solution, while having minimal impact on the detection sensitivity and LOD, did improve matrix deposition and could thus positively influence the achievable spatial resolution. When applied to mouse PK studies, the developed protocol enabled the detection and spatial distribution of kanamycin and vancomycin that was not achievable in the unwashed samples. By comparing the detection sensitivity and data quality of the antibiotics to commonly detected lipid species, we have shown that the improved sensitivity is due to reduced complexity in the gas phase caused by the reduction in readily ionizable lipid species. Although chloroform as a tissue washing solvent was only tested for these aminoglycosides and glycopeptide, it holds potential for the plethora of other antibiotics within the same classes or with similar physiochemical properties. We have provided the first protocol that will enable MSI of aminoglycosides and vancomycin at the site of infection. Future studies will apply this method to lesion PK of amikacin and kanamycin in TB-infected lung granulomas and vancomycin in infected bone tissue.