Original Article |
|
1 Department of Veterinary Pathology, College of Veterinary Medicine, Iowa State University, Ames, USA;
2 National Animal Disease Center, Agricultural Research Services, United States Department of Agriculture, Ames, Iowa, USA
Corresponding Author: Jack M. Gallup, Department of Veterinary Pathology, College of Veterinary Medicine, Iowa State University, Ames, IA 50011, USA. Tel: 515-294-5844; Fax: 515-294-5423; E-mail: eag@iastate.edu.
Note: Fatoumata B. Sow and Jack M. Gallup contributed equally to this work
Running title: LCM-DERIVED RNA sample qPCR assay set-up using PREXCEL-Q
![]() |
ABSTRACT |
![]() |
INTRODUCTION |
![]() |
EXPERIMENTAL DESIGN |
![]() |
ANTICIPATED RESULTS |
![]() |
ACKNOWLEDGEMENTS |
![]() |
CONFLICT OF INTEREST |
![]() |
REFERENCES |
![]() |
ABSTRACT
![]() |
---|
The ability to reliably analyze cellular and molecular profiles of normal or diseased tissues is frequently complicated by the inherent heterogeneous nature of tissues. Laser Capture Microdissection (LCM) is an innovative technique that allows the isolation and enrichment of pure subpopulations of cells from tissues under direct microscopic examination. Material obtained by LCM can be used for downstream assays including gene microarrays, western blotting, cDNA library generation and DNA genotyping. We describe a series of LCM protocols for cell collection, RNA extraction and qPCR gene expression analysis. Using reagents we helped develop commercially, we focus on two LCM approaches: laser cutting and laser capture. Reagent calculations have been pre-determined for 10 samples using the new PREXCEL-Q assay development and project management software. One can expect the entire procedure for laser cutting coupled to qPCR to take approximately 12.5-15 h, and laser capture coupled to qPCR to take approximately 13.5-17.5 h.
KEY WORDS: LCM; laser capture; microdissection; microsection; laser cutting; laser catapulting; PREXCEL-Q; PCR; qPCR; RT; gene expression; real-time PCR; quantitative PCR; qPCR software
![]() |
INTRODUCTION![]() |
---|
Relatively recent advances in science including PCR and LCM have allowed the rapid analysis of normal and diseased tissues at the DNA, RNA and protein levels. The etiology and pathology of many diseases can be linked to alterations in genes, gene products, and the signaling pathways stimulated by these gene products. The molecular alterations observed may be specific to a cell type, and attempts at correlating such defects to the cells in question can be hampered by the cellular heterogeneity of tissues. This can limit the meaning of biological data obtained since the many different cell types which comprise a certain tissue may not be represented individually by analysis of and information from whole-tissue. Further, the use of whole-tissues renders it impossible to determine which cellular constituents contribute to the expressed gene signal of interest, especially since a two-fold signal difference is considered significant in gene expression studies. In addition, proteomic and genomic techniques rely heavily on the procurement of homogeneous cell populations. While cell cultures and cell lines are frequently used to study pure cellular populations in a controlled milieu, the genetic information obtained in vitro may not necessarily represent the molecular events occurring in the actual tissue environment. Detailed molecular and biochemical analyses of in vivo interactions require the ability to analyze specific cell populations within their heterogeneous tissue environment.
LCM (1-4) is a recently developed technology that provides the means to isolate or enrich single cell types or unique cellular structures from heterogeneous tissues while preserving the original tissue’s morphology and without introducing contamination from surrounding cells. As its name implies, the LCM technique is based on the use of a near infra-red laser with pinpoint precision fitted to an inverted microscope. The principle steps of LCM have an elegant simplicity: a tissue sample is mounted on a slide, and cells of interest are visualized (morphologically, or based on the use of a marker specific to the cell type). A transparent 100 μm-thick ethylene-vinyl acetate film coated on a cap is then placed over the tissue section by pulling the cap holder (loaded with a cap) over the tissue, then lowering the cap film-side down onto, and in direct contact with the tissue. The diameter of the laser beam can be adjusted from 7.5 to 30 μm, depending on the size of the cell or group of cells one wishes to select. The low-energy laser, administered in pulses, causes the thermoplastic film to melt, bind to, and lift the targeted cells out of the tissue. No damage occurs to the biological macromolecules collected as the energy coming from the laser is absorbed completely by the film, and the pulsing of the laser is only performed for milliseconds. All unwanted cells are left behind in the original tissue, which could be further dissected if needed, provided the tissue is well preserved. The samples captured by LCM can be immediately harvested for molecular analyses. This technique is very well suited for the isolation of single cells or small groups of cells. The Arcturus PixCell II Laser Microdissection apparatus is an excellent system for isolating cells of interest by laser capture.
Recently, a new generation of microdissection apparatuses has been unveiled: laser cutting (MMI, Leica), laser catapulting (PALM), and scanning laser microdissection (XMD) (5) systems. The PALM microlaser/microbeam systems are based on the ability of the laser to microdissect tissues and to pressure-catapult the collected cells into a collection or resuspension-lysis buffer. This is often referred to as a precise “non-contact” laser pulse system. Here, an ultra-violet (UV-A) laser with a beam spot of less than 1 μm in diameter is used to cut selected cells. After microdissection, the cells are catapulted directly (against gravity) into the lid of a Zeiss PALM 0.5 ml microfuge tube, which minimizes contamination from neighboring tissue and eliminates the possibility of contamination by way of direct mechanical contact with the source sample from which selected regions or cells are being taken.
Diversity of applications of LCM
The LCM technology has been used widely in cancer research, therapeutic efficacy studies, forensics, drug interactions, and toxicity assessments. In studies involving host and pathogen interactions, it is possible to identify the first cells targeted by invading pathogens, differentiate infected cells from non-infected cells, and examine the pattern of viral or bacterial distribution. In studies involving drug interactions and therapeutic efficacy, it is possible to determine where the drug goes, how it affects safety and efficacy in tissues, how cells respond to treatment by comparing whole tissue to a specific structure of the tissue, and even identify critical safety biomarkers. Protein studies on LCM-derived cellular material can be performed as well. Although this protocol summarizes an LCM-based approach to study gene expression by qPCR in ovine macrophages, it can be adapted to study any animal cell type. LCM for plant material is not addressed in this protocol, but, with appropriate use of other fixatives preceding IHC, and proper adjustments to laser power strength and duration, any plant cell of interest can be similarly accessed.
Limitations of the LCM and LCM-qPCR techniques
There are several drawbacks associated with LCM. Some of these relate to sampling issues, such as the stability of the isolated material (e.g. RNA degradation) and to the quantity of material. Frequently, it may be necessary to pool material from multiple slides/tissues to get enough samples for downstream analyses. Probably the most obvious limitations of LCM are those associated with it being a microscope-based technology. Since LCM scopes are inverted by design, they must focus first through the glass slide itself before they obtain the tissue or cell image of interest. In addition, it is impractical in LCM to coverslip sections, thus recognizing tissue morphology is often more difficult. In some cases, LCM can be coupled to immunofluorescence microscopy (as in this study), but such assays rely on the availability of antibodies specific to desired cellular markers. Other microscopy-based concerns relate to the inability to use certain difficult-to-section tissues (e.g. bone or mineralized tissue, tooth pulp, some plant material) as well as to the strength or weakness of tissue adherence to the slides in the first place (e.g. too strong, and collection can be impossible, too weak, and the subject tissue may fall off of the slides entirely, before LCM can be performed). In addition, LCM-collected cells can be contaminated by material adjacent to the subjects of interest if the tissue is not suitably adherent to the chemically-treated glass substrate used.
There is also a current need to improve and/or come up with more consistent global profiling methods using LCM-coupled qPCR and qPCR in general. Recent concerns have been voiced as to the rare use of LCM and lack of awareness about qPCR inhibition (Bustin, S. How Reliable is Your qPCR Data? Drug Discovery & Development, March 01, 2006, http://www.dddmag.com/reliability-of-qPCR-data.aspx). qPCR also requires extensive calculations, standard curves, and optimal reaction efficiencies. Attention to these details takes time and can limit investigators to assessment of one or just a few genes. If inhibition is not eliminated, or if calculations and optimizations are not performed correctly, qPCR will yet generate results; however, they can be very inaccurate. This is especially true for LCM-qPCR because the sample size is very tiny and therefore, mistakes become magnified. Unfortunately, investigators can be completely unaware that their data is faulty. To address many of the concerns above, we have invented PREXCEL-Q (P-Q) (6, 7), a program that allows swift calculations of reagent and sample needs for every aspect of qPCR. In addition, the program has a built-in function specifically tailored for LCM-qPCR on which we have published previously (8, 9) and have further optimized for lung macrophage studies.
P-Q
The P-Q program (7) has been a very helpful implement in addressing the common time-consuming perfunctory concerns with qPCR setups by automatically calculating amounts of all needed reagents (primers, probes, master mix), gauging total sample material needed, assisting directly with appropriate standard curve designs, identifying the dynamic dilution range of sample material within which qPCR inhibition is absent and target amplification efficiencies are highest, automatically generating drawings and printouts summarizing reaction formulations, and estimating total cost of the assay. P-Q has also been designed to address parameters related specifically to LCM, such as pg of nucleic acid per cell-type, number of cells collected per sample, volume of each final collected sample and the most conservative use of the collected material in creating standard curves are all calculated by the program.
MIQE and the RDML Consortium
Finally, it is important to remind all current users of qPCR technology, and all publishers of qPCR data, to follow the responsible guidelines set forth by MIQE (or MIqPCR): Minimum Information for Publication of Quantitative Real-Time PCR Experiments, which is part of the Real-time PCR Data Markup Language (RDML) Consortium created by Stephen Bustin and presently tended to by Vladimir Benes, Jeremy Garson, Jan Hellemans, Jim Huggett, Mikael Kubista, Reinhold Mueller, Tania Nolan, Michael Pfaffl, Gregory Shipley, Jo Vandesompele, Carl Wittwer, Steve Lefever, Andreas Untergasser and Filip Pattyn. The aim of MIQE, which is coordinated under the umbrella of MIBBI (Minimum Information for Biological and Biomedical Investigations) is to provide authors, reviewers and editors specifications for the minimum information that must be reported for a qPCR experiment in order to ensure its relevance, accuracy, correct interpretation and repeatability. A checklist, which should be submitted along with the paper, is available for authors in preparing a manuscript employing qPCR. This organization and its philosophy were created in effort to standardize the way people perform and report their qPCR studies (http://medgen.ugent.be/rdml/guidelines.php). Following these guidelines will encourage better experimental practice, allowing more reliable and unequivocal interpretation of quantitative PCR results (http://www.sabustin.org/). The RDML Consortium’s aim is to encourage and foster the use of a universal qPCR data exchange format. The RDML format will facilitate the exchange of data between instruments and data analysis software, between different users and even allow submitting qPCR data to central repositories or as supplemental data to a paper. MIQE (MIqPCR) guidelines have been developed by Stephen Bustin in collaboration with MIBBI to assure that data files contain the minimal information allowing unambiguous interpretation of the data. The central focus of this stalwart endeavor is to create a universal qPCR data format that can be used by anyone regardless of qPCR instrument and analysis software and to supplement this format with guidelines and tools to achieve maximum benefits with minimal burden for users. Founders of this organization define its members as “anyone who helps developing, makes suggestions and comments or just declares support for the RDML initiative.” Further information on the RDML Consortium can be found in the following websites: info@rdml.org, http://www.rdml.org, http://sourceforge.net/projects/rdml/.
Alternative methods
Flow cytometry can be used to isolate cells of interest from a suspension, but it relies on the use of a specific cellular marker for selection, and commonly requires enzymatic digestion or other treatments to isolate of cells from solid tissue samples. Histopaque gradient methods for isolating different leukocyte populations can also be met with difficulties when contaminating platelets are carried over into the final samples, and when centrifuge speeds, temperature of material and tube-type usage have not been optimized. Similar problems are found using Percoll gradients to isolate tissue leukocytes (e.g. gut lymphocytes (IELs, LPLs)). Additional concerns with histopaque and Percoll isolations include limited quantities of blood or tissue samples, low target cell population(s), and cost. Thus, LCM remains the best procedure for the isolation and enrichment of specific cells from samples immobilized on a solid matrix.
Recent responses obtained from a MIQE questionnaire revealed that only 12% of scientists performing qPCR on a regular basis use LCM technology (Bustin, S. How Reliable is Your qPCR Data? Drug Discovery & Development, March 01, 2006, http://www.dddmag.com/reliability-of-qPCR-data.aspx). In addition, qPCR coupled with LCM might be considered too daunting to undertake, as both techniques are often rife with complexities not often surmounted by experienced investigators and novices alike. We offer here a fail-safe approach to LCM-coupled qPCR in the hope to increase its user-ship – in keeping with the wishes of leaders in the qPCR field (10). In revisiting this important technique, we have found several ways to improve its user-friendliness and reliability: 1) use of the P-Q program to facilitate qPCR setups; 2) use of a defined kit that incorporates reagents for both LCM RNA extraction and subsequent qPCR (which we recently co-developed with Invitrogen) and a correlate master mix kit used for genomic DNA contamination analysis in all sample isolates; 3) use of the two particular master mixes mentioned in this study is convenient in that they can both use the same thermocycling program, and can therefore be used simultaneously on the same plate; 4) demonstration of using LCM on a difficult tissue, such as lung, from which epithelial cells (8, 9) and macrophages can be collected. For simplicity’s sake, in this manuscript, we show the reagents required for performing LCM-qPCR on 10 samples throughout.
Conclusions
Since its discovery in 1997, the LCM technique has undergone many refinements and helpful modifications (2-4). The analysis of purified cells populations from selected regions of a tissue can now be automatically performed by the use of a computer-controlled stage combined with the LCM microscope. Improvements in DNA and RNA extraction procedures have allowed the increased yield and purity of isolated material for downstream molecular analysis. It is also possible to collect enough sample material for protein isolation, profiling, and discovery. With our current approach, we can repeatedly and reliably analyze basal and induced gene expression of cells obtained from diseased animals (respiratory syncytial virus, RSV, for example (11)) for comparison to control animals, thus adding a powerful dimension to the study of host factors regulating disease progression. The present protocol represents a tightly-defined approach to LCM-coupled qPCR, one which should be routinely implemented and advocated as a standard operating procedure.
![]() |
EXPERIMENTAL DESIGN![]() |
---|
MATERIALS
Reagents:
|
|
REAGENT SETUP
Primer and probe design. We designed all primers and probes mentioned in this work using Primer Express version 2.0 (Applied Biosystems). Primers and probes designed in this fashion require an adjustment of the final qPCR reaction Mg+2 concentration to 5.5 mM. Be aware that different primer-probe designing programs generate primers and probes with different ionic strength requirements of the final master mix in order to attain the intended (program-calculated) melting temperature (Tm) value of each oligo. Be certain to use the recommended Mg+2 concentration calculated or suggested by the particular software and/or particular master mix you decide to use. Detailed accounts of these and other important considerations that accompany good primer and probe design can be found elsewhere (8).
Master mix considerations. Two avenues are available when choosing a master mix to use for qPCR. Although we speak only of using a “one-step” qPCR master mix (which contains both RT and Taq enzymes) in this study, a “two-step” format (which uses a qPCR master mix that does not contain RT enzyme, only a variety of Taq) is also available. By definition, a “two-step” qPCR procedure, implies that RNA has been converted to cDNA in a separate step preceding qPCR. LCM-derived RNA samples can be converted to cDNA (with or without linear amplification) and stored at -80°C for several years. Next, the cDNA can be subjected to qPCR using an appropriate master mix (without RT enzyme) to complete the qPCR – the same kind of master mix can also be used in the event that your LCM-derived samples are genomic or organismal DNA as the template is already “DNA.”
Testing for genomic DNA contamination in LCM-derived RNA samples. Even though proper use of the CellsDirect™ One-Step qRT-PCR Kit with ROX includes a DNase-treatment step of LCM-RNA isolates, carryover DNA can still be present. In order to be certain that signals generated by qPCR are truly resultant of transcriptomic (mRNA) as opposed to genomic nucleic acid (gDNA) amplifications, it is imperative to include sample wells that contain master mix with no RT enzyme. Such reactions are called “no RT control” reactions (NRC), wherein only contaminating gDNA (if present) will be amplified and thus serve as an indicator as to what degree the RNA samples are gDNA-free. In our studies, we have found it most advantageous to use a highly expressed gene such as a reference gene (formerly known as a “housekeeping gene”) for this assessment. We used ovine ribosomal protein S15 (ovRPS15) as the reference gene here since a nearly homogeneous cell population was being studied. Since reference genes are normally abundant in samples and/or tissues, they should provide the best chance of finding any indication of gDNA presence. gDNA signal does not contribute significantly to genuine sample signals if a difference in Ct values of greater than 5 (between the NRC master mix and the sample master mix) is observed. gDNA contribution to genuine transcriptomic signal can be mathematically obtained as follows: 1/EAMP(NRC Ct – Sample Ct), where “EAMP” = Exponential amplification = 10 -1/m, m = slope of target standard curve, “Ct” = threshold cycle, and “NRC” = no RT control (we use a reference gene as qPCR target for this). Contaminating gDNA qPCR signals 5 or more cycles away from genuine transcript-generated qPCR signals have a nearly negligible impact on final qPCR results, e.g. 1/(EAMP of 2)5 = ∼3.13% signal contribution when E = 100% Fig. 1).
In our experience with non-LCM-based qPCR, using Turbo DNase (Ambion/ABI), contaminating gDNA signals have always been greater than 12 cycles (but typically 13 cycles or more) away from our genuine transcriptomic target qPCR signals, indicating to us that gDNA-related contributions to each of our genuine/intended one-step qPCR target amplifications have been clearly minimized (<1/212 = ∼0.024% signal contribution when E = 100%, where E = EAMP - 1). Realize, however, that these contributions compound upon themselves when both target and reference gene signals are affected by contaminating signal contributions, so, knowing the extent of this contribution is absolutely necessary in any experiment using a reference gene to quantify gene expression. NRC reactions are always necessary to prove that your LCM-qPCR Ct signals are not merely the result of amplified gDNA in cases where the chosen primers or probes do not span a genomic intron6. In the present study, we used Amplification Grade DNase I (as provided in the Invitrogen kit used for one-step qPCR) for the DNase treatment of our LCM samples.
Reference gene considerations. Although the use of reference (housekeeping) genes has come under fire over recent years due to their variable expression in many tissue and cell samples, qPCR performed on RNA from homogeneous cell populations collected by LCM is perhaps the best candidate for their use in normalizing qPCR gene expression data. Assessment of reference genes in total RNA from whole tissue samples is inherently subject to the differential contributions of reference gene transcripts from a myriad of cell types - thereby rendering useless the advantage initially sought by using a reference gene in the first place. LCM-collected homogeneous cell samples offer an intuitively better paradigm wherein the use of a reference gene for normalization purposes is well-suited. However, any normal cell, when compared to its neoplastic incarnation, runs the same risk of variable reference gene expression and thereby, caution is again advised when using a reference gene to normalize gene expression data gathered from two similar but metabolically different cell types (12). In this manuscript, we have chosen to use the reference gene, ovRPS15 as it has demonstrated great stability in many of our particular qPCR studies for several years now, with the exception of studies involving qPCR analyses of total RNA extracted from tissues of animals at different stages of ontogeny. Although we use only one reference gene here, other investigators choose to test up to 10 reference genes until they find one, two or three that are stable in their particular system. SABiosciences offers an excellent platform in which to test multiple reference genes in mouse, rat and human qPCR metabolic pathway studies. Be certain to test whatever reference gene (or genes) you choose for stability in your particular system. The ideal reference gene should generate the same Ct value in all samples when all samples are loaded into the qPCR reactions equally.
EQUIPMENT
LCM-qPCR Procedure
1) Procurement of tissue ( ♦TROUBLESHOOTING see Table 9)
(A) Tissue collection and storage = TIMING 2 h
Freshly-necropsied lung tissues (the example in this paper is sheep lung) are placed into plastic disposable cryomolds containing OCT. More OCT is applied over the tissues, and these are then placed onto blocks of dry ice until the tissues embedded in OCT are frozen to a solid white (Fig. 2). All samples are then transferred immediately to -80°C for storage. ⌈PAUSE POINT Samples are stable for years in this state.
(B) Tissue sectioning (per 5 sections, assessed in duplicates for a total of 10 samples) ♦ TIMING 30 min-1 h
(C) Slide preparation (per 5 slides, assessed in duplicates for a total of 10 samples) ♦ TIMING 2 min
2) Immunofluorescence (IF) staining of leukocyte populations in frozen tissue sections (? TROUBLESHOOTING see Table 3)
(A) Fixation and blocking
All the following steps are performed at 4°C on a slide rack or slide-mailers (Fig. 5A and Fig. 5B), unless otherwise specified: IF ♦ TIMING 42 min.
(B) Staining procedure
The following steps can be performed using Option A or Option B depending on which secondary conjugates are available or desired:
First option ♦ TIMING 45 min (hands-on) – but includes a 16-24 h incubation
Second option ♦ TIMING 2 h 21 min
3) Hematoxylin staining = TIMING 7 min< (? TROUBLESHOOTING see Table 3)
(A) Hematoxylin staining and dehydration of the tissue
The procedure can be carried out on slide racks as follows:
(B) Processing of slides for LCM
Slides are allowed to dry (of xylene) 1 to 2 min before each laser cutting or laser capture event inside a dessicator. Slides are then ready for collection of cells (Fig. 6).
4) Laser Cutting/Capture Microdissection (for 10 caps) (? TROUBLESHOOTING see Table 3)
(A) Laser cutting
The PALM apparatus from Zeiss (Fig. 7) was used: ♦ TIMING 2 h 30 min
(B) Laser capture
The PixCell IIe Laser Capture Microscope (Arcturus; now Molecular Devices; Fig. 8) was used: ♦ TIMING 3 h 30 min
5) RNA extraction procedure (? TROUBLESHOOTING see Table 3)
(A) Laser Cutting RNA isolation (for 10 caps) ♦ TIMING 1 h 10 min
Note: most reagents used in this procedure can be found in the CellsDirectTM kit. See Box 4 for procedure summary:
(B) Laser Capture RNA isolation (for 10 caps)♦ TIMING 1 h 10 min
Note: most reagents used in this procedure can be found in the CellsDirectTM kit. See Figure 14 for procedure summary:
6) qPCR procedure (per plate per target) (? TROUBLESHOOTING see Table 3)
The PREXCEL-Q-based LCM approach was used (6, 7).
(A) 1 day prior to qPCR: Labeling of tubes, master mix preparation, machine programming ♦ TIMING 1 h
(B) Day of qPCR ♦ TIMING 3 h
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
![]() |
ANTICIPATED RESULTS![]() |
---|
Even though LCM collects miniscule amounts of material, it is possible for users to see collected cells on the collection cap (which look like scratch marks to the naked eye, or like actual cells under the microscope), or inside the resuspension-lysis buffer inside the lid (for laser cutting) by microscopic inspection. For 500 macrophages collected, using 10 pg of total RNA per macrophage, we obtained an approximate yield of 5 ng of total RNA. Each RNA sample was brought up into a final volume of 250 μl which provided enough material, from 10 such samples, to assess 8 targets of interest by qPCR (replete with 3-point standard curves, individual sample assessments – all in duplicate, and NRC wells for genomic DNA analysis, see Box 7 and Box 8). Abundantly expressed target transcripts are typically able to generate better standard curves than rarely expressed targets. For example, of the 8 targets assessed, MCP-1 (15) was among 7 targets which were able to generate reasonable standard curves, with very acceptable amplification efficiencies, whereas TLR8 did not. We reasoned that, since the other 7 targets behaved nicely, RNA integrity was not the problem. This suggests that TLR8 is simply a rare target in our samples. When DNase treating the RNA samples with reagents not included in the CellsDirect™ One-Step qRT-PCR Kit with ROX, one can expect varying degrees of success. For example, we tested the kit’s Amplification Grade DNase I in comparison to Ambion’s TURBO DNase I, and found that the kit’s Amplification Grade DNase I worked better in this application. When using IF to label specific cells to be collected by LCM, non-specific background fluorescence can make it difficult to identify the cells sought for study. When comparing expression of a cell-specific (mRNA) transcript from LCM-derived material to whole tissue-derived material, one can expect lower Ct values (thus higher expression) from the enriched LCM samples with or without normalization to an appropriate, stable reference gene (e.g. ovRPS15 in this study). Additional considerations for this section can be found elsewhere (3, 4).
![]() |
ACKNOWLEDGEMENTS![]() |
---|
The authors would like to thank Margie Carter and the Image Analysis and Confocal Microscopy Facility of Iowa State University Office of Biotechnology, Dr. Pat Schnable, Dr. Wei Wu and Marianne Smith of the Plant Sciences Institute/Carver Co-Laboratory for the use of their LCM (PALM) equipment; Toni Christofferson, Diane Gerjets, Jennifer Groeltz-Thrush for their expert help with histology. Funding by NIH NIAID RO1 AI062787.
![]() |
CONFLICT OF INTEREST![]() |
---|
The authors declare that no conflicting interests exist.
![]() |
REFERENCES![]() |
---|