Here, we describe a protocol using laser capture microdissection coupled with LC/MS analysis to spatially-quantify drug distributions within pulmonary tuberculosis granulomas. The approach has broad applicability to quantifying drug concentrations within tissues at high spatial detail.
Tuberculosis is still a leading cause of morbidity and mortality worldwide. Improvements to existing drug regimens and the development of novel therapeutics are urgently required. The ability of dosed TB drugs to reach and sterilize bacteria within poorly-vascularized necrotic regions (caseum) of pulmonary granulomas is crucial for successful therapeutic intervention. Effective therapeutic regimens must therefore contain drugs with favorable caseum penetration properties. Current LC/MS methods for quantifying drug levels in biological tissues have limited spatial resolution capabilities, making it difficult to accurately determine absolute drug concentrations within small tissue compartments such as those found within necrotic granulomas. Here we present a protocol combining laser capture microdissection (LCM) of pathologically-distinct tissue regions with LC/MS quantification. This technique provides absolute quantification of drugs within granuloma caseum, surrounding cellular lesion and uninvolved lung tissue and, therefore, accurately determines whether bactericidal concentrations are being achieved. In addition to tuberculosis research, the technique has many potential applications for spatially-resolved quantification of drugs in diseased tissues.
The ability to spatially resolve and quantify drug levels is a crucial requirement for determining whether anti-tuberculosis drugs reach bacterial subpopulations within pulmonary lesions at sterilizing concentrations1. Of particular importance is determining drug penetration into the necrotic core of the lesion (called caseum), which typically contains the highest number of bacilli and may be poorly accessible to drugs due to the absence of vascularization.
Traditional methods to assess lesion penetration, which involve homogenization of excised pulmonary lesions followed by solvent extraction and liquid chromatography mass spectrometry (LC/MS) analysis, are highly sensitive and selective for the drugs of interest. However, these methods offer poor spatial information, limited to the size of the original homogenized tissue. Mass spectrometry-based imaging approaches, such as matrix-assisted laser desorption ionization (MALDI)2,3, desorption electrospray ionization (DESI)4 or liquid-enhanced surface extraction5,6 offer highly spatially-resolved imaging capabilities, but direct quantification can be extremely challenging or impossible due to heterogeneous ion suppression effects and differing extraction efficiencies of analyte from the various cell or tissue types7. Additionally, most direct tissue MS imaging approaches are inherently less sensitive than LC/MS due to the lack of chromatographic separation of endogenous species competing for ionization and the lower solvent extraction efficiency of the drug from tissue.
Laser capture microdissection (LCM) combined with LC/MS analysis has been routinely applied to isolate and characterize distinct tissue regions for proteomic studies8,9 and recently utilized for drug quantification in dosed animal tissue10. Here we present an optimized protocol applying LCM combined with LC/MS (LCM-LC/MS) analysis to quantify anti-TB drugs within distinct granuloma compartments. In the laser capture microdissection process, a UV laser is focused through the microscope objective onto the tissue section, which cuts and isolates the desired tissue area by following a path defined by the user. For gravity-assisted LCM (the technique used for this research), the tissue section is mounted onto a thin polymer membrane slide (PET or PEN) and the tissue is captured in a collection tube cap sited below the slide. The drugs are extracted from the excised tissue and quantified using standard LC/MS approaches. The amount of tissue required to be collected is ultimately determined from the expected concentration of the drug present in the tissue and the sensitivity of the LC/MS method. For most analyses of drugs dosed at therapeutic levels and analyzed using a routine triple quadrupole mass spectrometer, 3 million µm2 (3 mm2) of tissue surface area is sufficient.
This protocol describes the powerful combination of spatial profiling and full quantification offered by LCM-LC/MS, providing absolute drug concentrations within all compartments of TB granulomas. The technique may also be applied to determining drug concentrations in many different diseased tissues providing vital drug discovery and development information.
All animal studies were carried out in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health with approval from the Institutional Animal Care and Use Committee of the NIAID (NIH), Bethesda, MD.
1. Animal Experiments and Tissue Collection
This section of the protocol describes animal procedures and sample collection under Biosafety Level 3 (BSL3) conditions. Detailed protocols of the Mycobacterium tuberculosis aerosol infection procedure and drug administration protocols in rabbits have been described previously11,12.
2. Tissue Sectioning
3. Microdissection
4. Extraction and LCMS Analysis
5. Method Validation
An overview of the LCM-LC/MS approach is shown in Figure 1. After sterilizing the tissue by gamma-irradiation, all subsequent steps (from tissue sectioning onwards) take place outside of BSL3 conditions. Figure 2 shows the lesion biopsy sections before and after tissue isolation by LCM. Necrotic and cellular areas of TB lesions can be easily identified and isolated by visual inspection of optical images alone (without the requirement to refer to histologically-stained adjacent tissue sections). The dissection process produces a clean cut with minimal disturbance to the surrounding tissue, and dissecting 3 million µm2 (3 mm2) of each region of interest from the lesions takes approximately 1 hour in total.
The extraction efficiency and stability of the LCM-LC/MS method was assessed using Ethambutol (EMB)-spiked lung homogenate (Table 1). Complete extraction of the drug was observed, and no drug stability issues were detected through the dissection and extraction process. The LCM-LC/MS extraction and quantification protocol was further validated by comparing to established LC/MS quantification methods applied to the same tissues. Due to the inherent heterogeneity of necrotic pulmonary TB lesions and the poor spatial specificity offered by standard tissue homogenization, we validated the LCM-LC/MS method by directly comparing drug concentrations in uninvolved lung analyzed by both analytical techniques (due to its relatively high tissue homogeneity). Table 2 shows the drug concentrations in uninvolved lung from biopsies taken from three Ethambutol-dosed rabbits as evaluated by: 1) homogenizing 25 µm thick tissue sections and analyzing by standard LC/MS, and 2) LCM-LC/MS analysis of uninvolved lung areas taken from an adjacent 25 µm thick tissue section. The data shows that two approaches produce consistent quantification data, demonstrating the suitability for routine spatial quantification.
We have applied LCM-LC/MS to spatially-quantify many existing and novel anti-TB drugs within pulmonary lesions. Figure 3A shows example data from MTB-infected rabbits dosed steady-state with Ethambutol. LCM-LC/MS enabled full quantification of the drug resolved within caseum, cellular lesion, and uninvolved lung tissue areas. EMB was observed to penetrate well into the lesion and reach sterilizing concentrations within the necrotic caseum. The corresponding MALDI-MS image of EMB distribution acquired from an adjacent tissue section is shown in Figure 3B. The qualitative MALDI-MS images correlates well with the quantitative LCM-LC/MS data with lower drug concentrations detected in the necrotic caseum in comparison to the cellular rim. LCM-LC/MS data was validated by direct comparison to tissue homogenates analyzed by standard LC/MS.
Figure 1: Schematic of the LCM-LC/MS process. Rabbit lung biopsies containing necrotic lesion along with surrounding uninvolved lung are collected and frozen. Cryosections are cut onto thin PET membranes and areas of uninvolved lung, cellular, and caseous lesion are dissected and isolated for quantification by LC/MS. Adapted from Zimmerman et al.12 Please click here to view a larger version of this figure.
Figure 2: Lesion (A,B) and uninvolved lung (C,D) as they appear when viewed using the microscope before (A,C) and after (B,D) microdissection. Caseous lesion areas appear darker and 'cracked' in the optical image scan (A, green outline). Cellular lesion is lighter in color and more solid in structure (A, red outline). Uninvolved lung should be sampled at least 5 mm away from the lesion border and appears red/pink in color (C, cyan outline). Scale bars (black, blue, and purple) = 400 µm. Note that only solid areas of uninvolved lung should be selected to avoid including alveolar spaces and bronchioles in the total surface area of tissue collected (as shown in in D). Please click here to view a larger version of this figure.
Figure 3: Example LCM-LC/MS dataset from lesion biopsies taken from two rabbits dosed with 100 mg/kg EMB for 7 days. (A) Favorable penetration of the drug into all lung and lesion compartments was observed. Drug concentrations quantified by LCM-LC/MS (hollow bars)/MS were in strong agreement with those quantified from homogenized dissected lesions by standard LC/MS (solid bars, mean ± standard deviation, n = 3 – 8). The minimum concentrations required to kill 99% of extracellular replicating bacilli (MBC99) and 99% of intracellular bacilli in macrophages (iMBC99) are indicated. (B) Top panel: MALDI-MS image showing the distribution of EMB [M+H]+ ion (m/z 205.193) within an adjacent tissue section. Note that the MALDI-MS image suggests poor penetration of EMB into the caseum due to the present drug concentrations being below the lower limit of detection (LOD) of the technique. However, the spatial specificity and superior LOD of LCM-LC/MS show the drug is reaching sterilizing concentrations within all lesion compartments including the caseum. Lower panel: Hematoxylin & Eosin stained tissue section directly adjacent to the section used for MALDI MSI. Cellular (C) and necrotic granulomas (NG) were present in the tissue. The caseum core is outlined in white. Adapted from Zimmerman et al.12 Please click here to view a larger version of this figure.
Dissected area (µm2) | Measured EMB conc. (ng/g) (n = 3) | EMB recovery (%) (n = 3) |
3 million | 10633 (±404) | 106 (±4) |
5 million | 10057 (±1132) | 101 (±11) |
10 million | 10563 (±1128) | 105 (±11) |
Table 1: Extraction efficiency of the LC-LC/MS method for quantifying EMB in lung. Complete recovery of EMB was observed in the EMB spiked lung homogenate dissected tissues for all evaluated tissue volumes.
Rabbit ID | LC/MS EMB (ng/g) | LCM-LC/MS EMB (ng/g) | Difference (%) |
899 | 2910 | 3320 | 14 |
904 | 2010 | 1870 | -7 |
911 | 2150 | 2370 | 9 |
Table 2: Comparison of EMB quantification in uninvolved lung biopsies from 3 dosed rabbits by LC/MS and LCM-LC/MS. Equivalent drug concentrations were detected by both methods and no loss of signal due to degradation or extraction during the LCM-LC/MS process was observed.
Spatially-resolved quantification of drugs within pulmonary TB lesions is required to determine whether drug exposure reaches sterilizing concentrations to bacterial populations residing within the different lesion compartments. The LCM-LC/MS method described here enables absolute quantification of anti-TB drugs within all lesion compartments, including the bacteria-rich caseum, using only 1 – 3 tissue sections in total. Traditional tissue homogenization and LC/MS approaches for drug quantification in tissue often lack the spatial specificity to resolve specific lesion compartments and even when it is possible, there is significant potential to cross contaminate cellular and necrotic lesion areas during the manual extraction process.
The LCM-LC/MS approach has several key advantages to mass spectrometry-based imaging technologies. Primary among these are the fully quantitative capabilities of the method and the additional sensitivity of the LC/MS analysis due to resolving drug from interfering/suppressing endogenous species via chromatographic separation. However, MALDI-MSI provides more detailed spatial information regarding drug distribution. As both MALDI-MSI and LCM-LC/MS/MS require the same sample preparation steps to generate frozen tissue sections, the two techniques can be performed in tandem on the same tissue biopsy providing full quantification alongside highly-detailed imaging of drug distribution. An example of the requirement for higher analytical sensitivity is shown in Figure 3. Only very low levels of EMB signal were detected in the central necrotic granuloma region of the MALDI MS image (shown in Figure 3B), suggesting caseum penetration of the drug is poor. However, due to the superior limit of detection of LCM-LC/MS the drug was clearly demonstrated to reach sterilizing concentrations within that compartment and the drug concentration present was predominantly undetectable by MALDI (Figure 3A).
Tissue staining protocols are typically used prior to LCM for proteomic applications to enable easy identification of tissue areas and the cell populations located within. Routine histochemical stains such as H&E are not compatible with drug quantification, as the many solvent washing steps involved would delocalize the drug from the tissue section. Adjacent cut sections can be stained with H&E and used as a guide for microdissection. However, this is not usually necessary, as the lesion compartments can be optically resolved under the standard LCM brightfield microscope (Figure 2).
Optimization of the sectioning and microdissection steps is crucial for successful lesion quantification. During development of the method, we evaluated two different membrane materials, polyethylene terephthalate (PET) and polyethylene naphthalate (PEN), as substrate for mounting the tissue sections for microdissection. PEN membranes suffered from increased static charging in comparison to PET membranes and approximately 30% of all dissected tissue areas were lost due to static attraction between the dissected tissue region and the membrane. For this reason, PET membranes were chosen for future dissections due to the significantly reduced static attraction observed (only 5% of regions dissected from PET membranes were lost during the LCM process).
The stability of drugs within tissue sections is an important consideration when designing an LCM-LC/MS study. During the sectioning and dissection process, the tissue sections are exposed to lab temperature and light for a period of at least 1 hour. Other issues to consider include the extraction efficiency of the drug from the tissue sections, as there is no homogenization step involved (extraction is only performed by vortex mixing and sonication). In our experience, the extraction does not suffer from the lack of tissue homogenization due to the high extraction solvent to tissue ratio used and the thinness of the tissue sections dissected. Our observations are in agreement with a previously-described online LCM-LC/MS method developed to quantify propranolol in microdissected tissue sections of brain and liver, where complete extraction of drug from tissue was observed from 20 µm and 40 µm thick sections10. We validated the efficiency of drug extraction and the drug stability during the LCM process by directly comparing lung tissue concentrations (from the same rabbit and lung lobe) analyzed by LCM-LC/MS with concentrations determined by standard tissue homogenization and LC/MS (an example is shown in Table 1). This comparison allows us to determine whether any change of signal is occurring between the two methodologies.
The density of dissected tissue regions is also an important consideration when quantifying drug levels based upon the tissue surface area or volume. This is of particular concern for lung and lesions biopsies, in which the uninvolved lung tissue areas have an overall lower density than cellular and caseum (less tissue covering the same relative surface area) due to the presence of multiple open bronchioles and alveolar spaces. The effects of this difference in density can be mitigated by carefully drawing around the open spaces to avoid including them within the cumulative surface area of collected tissue (as shown in Figure 2).
An additional limitation is that the presented LCM-LC/MS method only quantifies total drug concentration within the isolated tissue (in common with all tissue homogenization, solvent extraction, and LC/MS quantification approaches) and does not resolve protein-bound drug concentrations from unbound fractions. Microdialysis is an alternative approach for quantifying unbound drug within tissue and enables accurate quantification of free drug concentrations reaching extracellular populations of bacteria13. However, the technique would be best applied as a complementary approach, as it lacks the spatial specificity of LCM-LC/MS and only quantifies soluble drug levels within tissue extracellular fluid, not the intracellular content.
We have combined, for the first time, LCM and LC/MS to spatially quantify antibiotics at the site of tuberculosis infection. The methodology is tremendously powerful since it can be applied to any small molecule drug used in any disease state. Indeed, we have recently quantified an antifungal drug candidate in a mouse model of abdominal candidiasis14. Differential drug partitioning into heterogeneous tumor compartments (including necrotic cores) is a primary concern in the treatment of cancer, and a critical area of research in cancer drug discovery15. LCM-LC/MS is ideally suited to approach these questions. Furthermore, LCM-LC/MS can be used for biomarker discovery to quantify metabolic, lipidomic and proteomic changes occurring in tissue regions and cell populations during disease pathogenesis.
The authors have nothing to disclose.
We thank Paul O'Brien, Marizel Mina and Isabella Freedman for animal experiments, Jacquie Gonzalez and Danielle Weiner from NIH/NIAID for help with gamma irradiation of rabbit tissues prior to laser capture microdissection and Jansy Sarathy for manuscript thoughts and advice. This work was supported by funding from the Bill and Melinda Gates Foundation (OPP1174780) and NIH shared instrumentation grant 1S10OD018072. We thank Eliseo A. Eugenin for providing access to the Leica LMD 6500 microscope and sharing expertise and advice. The purchase, and ongoing support of, the LMD 6500 was funded by The National Institute of Mental Health grant, MH096625, the National Institute of Neurological Disorders and Stroke, NS105584, PHRI funding (to E.A.E) and GSK contributions (to E.A.E).
New Zealand White rabbits | Covance | N/A | |
HN878 Mycobacterium tuberculosis | BEI Resources | NR-13647 | |
Ketathesia (Ketamine) 100 mg/mL C3N | Henry Schein Animal Health | 56344 | |
Anased (Xylazine) 100 mg/mL | Henry Schein Animal Health | 33198 | |
Euthasol (pentobarbital sodium and phenytoin sodium) Solution | Virbac | 710101 | |
Acetonitrile (LC-MS grade) | Fisher | A955-212 | |
Methanol (LC-MS grade) | Fisher | A456-212 | |
Formic Acid (LC-MS grade) | Fisher | A117-50 | |
Water (LC-MS grade) | Fisher | W6212 | |
0.2 mL flat-cap PCR tubes | Corning | 07-200-392 | |
Steel frames, PET-membrane | Leica | 11505151 | |
Premium Frosted Microscope Slides | Fisher | 12-544-2 | |
96 Deep well plate 2.0ML PP RB | Fisher | NC0363259 | |
Zorbax SB-C8 column (4.6 by 50 mm; particle size, 3.5 μm) | Agilent | 820631-001D | |
"Zipper” Seal Sample Bags | Fisher | 01-816-1B | |
Name | Company | Catalog Number | Comments |
Equipment | |||
CM1850 cryostat | Leica | Discontinued | Leica CM1860 is the current model |
Laser Microdissection System 6500 | Leica | Discontinued | Leica LMD 6 is the current model |
Agilent 1260 Infinity II HPLC | Agilent | ||
API 4000 QTRAP Mass Spectrometer | Sciex |