• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Laser-controlled microdissection of tissues opens a window of new opportunities

Posted on: Thursday, 2 October 2003, 06:00 CDT

Summary

Gene expression analysis using total RNA of bulk tissue usually cannot assign specific messages to particular cell types. Cell- specific RNA expression profiling, though, may be crucial for a better understanding of the role of each distinct cell type within a physiological or pathophysiological setting. RNA profiling based on laser-controlled microdissection (LCM) of defined cells of a tissue now provides a useful tool for studying molecular crosstalk between different cell types within a tissue. The LCM technique allows for efficient isolation of single cells with no or very low contamination of surrounding tissue components, simultaneously leaving the intracellular structure and molecules intact. In this review, different issues of the LCM technique and the RNA amplification procedure for microarray analysis are discussed. An exemplary summary of results obtained from gene profiling of epithelial and stromal cells from human prostate tumors is presented, demonstrating the power of LCM-based molecular analysis. Finally, we discuss the potential use of the LCM technique i) to study the transcriptome of distinct cells from formalin-fixed and paraffin-embedded tissues in subcellular RNA profiling and ii) high resolution proteomic and metabolistic studies.

Key words: Laser-controlled microdissection (LCM) Prostate tumor - RNA amplification - Tumor-stromal interactions - Tissue fixation - LCM applications

Introduction

The inception and implementation of high-throughput methods such as transcriptome profiling (analysis of the identities and quantities of all messenger RNA sequences expressed in a tissue or a cell), proteomic approaches (qualitative and quantitative analysis of specific proteins and peptides), and standard methods of clinical chemistry for determination of metabolites, hormones, drugs etc. would, at least conceptually, pave the way for analysis of disease markers and response to therapy of a particular patient. This approach has already been supported by expression profiling of non- Hodgkin lymphomas and breast cancer [17, 20]. However, such a development would have been impossible, if there had not been a preceding number of studies on differences of the transcriptome in tumors vs. normal tissue [4] in relation to clinical outcome.

Early transciptome research using bulk tumor or diseased tissues

The first studies of this kind have used total RNA from bulk tumor tissues or disease-associated tissues, because the amounts of mRNA or total RNA needed for reliable profiling were between 500 ng to 1 [mu]g mRNA or 50 to 100 [mu]g of total undegraded RNA. This approach resulted in defining a number of messages that were more than two-fold over- or underexpressed in tumor tissue vs. normal tissue [4]. By necessity, such an approach could not consider the heterogeneity of tumor cells and tumor-stromal interactions within the same tumor, nor could it assign specific messages to particular cell types. It may also have 'lost' specific and therefore potentially very useful markers owing to averaging out, as expression levels are relatively low (e.g. the expression of stromelysin-2 in prostate cancer epithelial cells) [9]. Expression profiling of bulk tissue alone would cause extreme difficulties in detecting putative marker mRNAs at very early stages of tumor development, in residual disease, or in the beginning of disease dissemination, during which a few cells only express mRNAs relevant to diagnosis and prognosis.

Expression profiling of microdissected cells from diseased tissues

The advent of microdissection methods opened a new window of opportunities regarding the assignment of specific messages, proteins/peptides, and even metabolites to particular cell populations. Laser-controlled microdissection (LCM), in particular, offers the possibility of obtaining morphologically identified target cells with absent or low cross contamination by other cell types. Initially, two major methods of laser-controlled microdissection opened the field, both requiring assistance by specific membranes:

(i) For the PALM, method, which we used for analysis of markers for epithelial and stromal cells from human prostate tumors [9], cryostat cuts are deposited on glass slides covered with a membrane (for details, see www.palm-microlaser.com). The positions of the desired cell populations are electronically circumscribed under the microscope, coherent cell fields are cut out automatically by cold laser ablation, and catapulted into a microtube by damage-free laser- induced propulsion (see Fig. 1). In our and others' experience, initially frozen slides may be used up to 30 min at room temperature without mRNA degradation. We isolated RNA with the Stratagene Absolutely RNA Nano kit, and monitored RNA quality by capillary electrophoresis on Agilent Bioanalyzer 2100 nanoRNA 6000 or picoRNA 6000 chips.

(ii) For the Arcturus procedure (see www.arctur.com), cryocuts are covered by a proprietary membrane that is 'glued' to cells by laser warming. By recovering the membrane, attached cells or organelles are separated from the section and then harvested from the membrane. Both instruments are not walk-away instruments: in a time-consuming procedure, a pathologist or a person appropriately trained in discrimination of cells has to select the areas to be microdissected, even if these deviate from the 'ideal' morphology; cross-contamination or exogenous RNAse contamination should be avoided. Whether automation of the time-consuming process might be possible in the more evident cases, e.g. by coupling to image analysis procedures, remains to be demonstrated. Membranes from both companies are said to be inert to downstream methods, in particular, no or low retention of RNAs is claimed.

Fig. 1. Laser-controlled microdissection (LCM) of a human prostate gland. The position of a prostate gland from a hematoxylin- stained human tissue section is electronically circumscribed under the microscope. The coherent tissue area is then cut automatically by cold laser ablation and finally catapulted into the cap of a microtube by laser-induced propulsion.

Fig. 2. Flow scheme of our optimized RNA amplification method (see [1, 9] and [Kenzelmann et al., submitted]). Nanogram amounts of total RNA are reverse transcribed and double-stranded (ds) cDNA is synthesized. The subsequent first round of T7 RNA polymerase-based in-vitro transcription results in a 5-20 fold mass conversion of complementary (c)RNA. The second round of ds cDNA synthesis uses a random priming step which may lead to 5'end-truncated cDNA molecules. The subsequent second round of T7-based in-vitro transcription results in a 200-400 fold cRNA amplification. This step also includes labeling of the cRNA with biotin for downstream Affymetrix GeneChip(TM) profiling. The whole procedure allows for a > 10^sup 5^-fold faithful amplification (correlation coefficient r > or = 0.9) of the original mRNA population. (V = A, C, G; B = T, G, C)

The nanograms of total RNA obtained from such microdissections are appropriate for reverse transcription into cDNAs by sensitive Reverse Transcriptases, but would be relatively useless for expression profiling if not coupled to an efficient amplification/ labeling step. While optimized PCR amplification of modified cDNA appears the simplest, least expensive and time-consuming approach [11], the method may distort quantitative relationships between messages, i.e., doubts remain as to whether low-abundance messages would be adequately represented by this method. Like other working groups [1, 6, 11, 21, 22], we have developed protocols [9, Kenzelmann et al., submitted] that can be used without major distortion of quantitative message interrelationships. One of our two successful protocols is a modification of the method of Baugh et al. [1], the other is based on Clontech's SMART protocol [22, Kenzelmann et al., submitted]. The success of the robust first protocol [9] (see Fig. 2 for flow scheme) is evident from the fact that Affymetrix mouse chip data from 8 [mu]g and 20 ng (less than 1,000 intact cells) of the same mouse total RNA are correlated by r = 0.88, including sequences from the 5' end (which are by necessity underepresented in amplification protocols owing to a random priming step), 3' end sequences, and ESTs. We found that messages possessing specific sequence-intrinsic properties, such as a high G-C or A-T content and internal poly A sequences, reproducibly behaved as false positives with respect to differential expression (for list, see Kenzelmann et al., submitted). If constant outlier sequences, 5' and 3' ends and EST sequences were removed, the correlation coefficient rose to > = 0.9, in other words, 400-fold different starting RNA amounts in our hands reproducibly gave similar absolute levels of individual messages, a fact that compares advantageously with other protocols. As expected, the major part of the lowest-abundance messages with a 'Present' call using 8 [mu]g of RNA were underrepresented using 20 ng of the same RNA, because sampling errors come into play during the RT reaction. This effect increases with decreasing amounts of total RNA; therefore, at present, in our view, the effective limit for representation of low-abundance messages may be around 20 ng of total RNA, which is, in theory, obtained from about 1,000 intact cell\s. On the other hand, a few low-abundance messages are detected only by the amplification process that would otherwise have escaped detection. Owing to the amplification-inherent errors mentioned and biological noise [e.g. 8, 16], some messages may represent expression 'flashes' and have to be corroborated by additional tests (e.g. real-time PCRs) with regard to their general validity. As expected, microdissection and expression profiling of prostate tumor cells has shown advantages over profiling of bulk tissue: i) a resolution was obtained that is similar to that of immunohistochemistry, ii) putative tumor markers and therapeutic targets could be clearly discriminated (increased fold changes compared to bulk tissue analysis), iii) some putative markers reported by other studies before were assigned to stromal cells rather than to epithelial tumor cells, iv) the gap between primary tumor cells and established prostate tumor cell lines widened, and v) the relationships between tumor epithelial and stromal cells could be studied [9, and work in progress]. The latter is a very important aspect, as there is increasing evidence that tumor stroma participates in shaping the growth rate and invasiveness of epithelial tumor cells [for reviews see 2, 23]. We found that i) cross contamination of epithelial and stromal cells was absent or low using absolute levels of keratin and follistatin messages, ii) a number of messages that were more highly expressed in epithelial tumor cells than in surrounding 'normal' cells increased to similar or higher levels in tumor fibroblasts, iii) few fibroblast mRNAs increased or decreased specifically in tumor fibroblasts over non-cancer fibroblasts. Is this a manifestation of an epithelial-mesenchymal transition [18] that affects both epithelial and fibroblastic cells? There is probably induction crosstalk, via surface contact or secreted products, between epithelial and stromal cells to be studied intensely. At any rate, those 'converging' messages might provide interesting markers and targets for gene-specific therapeutic approaches.

Our studies also revealed that it would be difficult, if not in exceptional situations, to harvest single immune, neuroendocrine, and endothelial cells from prostate tumors by LCM owing to the topography of these cells. Specific (immuno)cytochemical staining protocols for such target cells could be immensely helpful provided the staining leaves messenger RNAs intact; such protocols have been described [10], but are not yet generally available and confirmed.

Obviously, LCM-dissected tissues may also be used for analysis of mutations in specific genes, such as p53 (which is rarely mutated in prostate cancer) and KLF6 (Kruppel-like factor 6) [Muhlbauer et al., Br. J. Cancer, in press]. Further applications based on DNA microarrays might study for instance methylation patterns of specific genes, such as the glutathione S-transferase (GSTP1) in tumors, or genomic changes, such as gene amplification/gene loss and gene-gene fusion [Hergenhahn et al., Current Genomics, in press].

Pathologists are confronted with the question of whether formaldehyde-fixed, paraffin-embedded sections might be submitted to LCM/profiling. Since aqueous solvents and paraffin were used for preparation of the blocks, extraction (=loss) or relocalization of messages and hydrophobic components of the tissue can be expected. In our experience, intact messenger RNA can be obtained from such blocks or sections only occasionally and might actually result from insufficient fixation. As we and our associates have demonstrated, quantitative RT-PCR of such samples with retained quantitative relationships between messages is feasible, but results in 50 to 100- fold lower concentrations compared to cryocuts, i.e., around 98% of the RNA did not yield appropriate cDNAs in spite of reversal of formalin crosslinking [5, 12]. Since formaldehyde crosslinking preferentially attacks A and G residues and is only partially reversible [12], an unbiased representation of messages by use of an oligo dT primer cannot be expected and has not yet been convincingly shown. Progress in this field has to await protocols that can provide representative amplification of fragmented RNA and hybridization to appropriate DNA microarrays. In addition, owing to the slow tissue penetration rate of formalin ([asymptotically =] 0.5 mm/hour), differential gene expression of cells along the axes towards the center of the tissue might represent fixation-dependent changes in the transcriptome rather than physiologically relevant differences in gene expression. An alternative would consist in switching to newer fixation protocols (which may also penetrate the tissue faster), such as Ethanol fixation or the newly developed HOPE fixative [14], which appear to preserve the morphology sufficiently, and leave most RNA intact. However, extraction of small RNAs etc. and translocation would remain a concern.

Has LCM a future regarding the analysis of subcellular organelles? A theme not yet thoroughly understood is the situation- dependent control over polyadenylation, splicing, and transport of selected messages from the nucleus, in which many pre-messages are retained and degraded [13], to specific locations inside cells as discussed for neuronal cells [7]. Addressing this problem in cell types devoid of obvious and elongated structures would require isolation of intact nuclei and specific sub-cellular regions by LCM, a demanding task in view of the fact that even large organelles, such as single mitochondria, cannot be easily distinguished by light microscopy.

Another aspect of LCM is that it opens a window to high resolution proteomic and metabolistic studies [19]. As shown by Emmert-Buck et al. [3], microdissected areas from cryocuts can be solubilized in efficient tissue lysis buffers, followed by spotting of nanoliter amounts of the Iysates onto glass slides. Specific proteins are subsequently quantitatively determined by appropriate antibodies using the tyramide signal amplification (TSA) staining procedure. As a calibration series of authentic proteins, or the peptides used in immunization, is spotted on the same slide, staining is performed under identical conditions, leading to linear calibration curves of a number of proteins whose slopes depend on the affinities of the antibodies [3,15]. The success of protocols for the determination of small molecular weight (MW) compounds from laser-dissected cryocuts would, in our view, mainly depend on the stability of the compounds, low volatility, and sensitive determination methods. The membranes used in LCM become a concern here: in response to their chemical nature, a number of compounds may adhere firmly to, or partition into the membranes and be lost from analysis. However, it is this property of specifically modified membranes that may be turned into an advantage.

In conclusion, some fruits of laser-assisted microdissection already await harvesting while others still need maturation. At any rate, LCM has already provided scores of new insights into cell- specific expression patterns that may lead to the validation of new diagnostic markers and therapeutic targets.

Acknowledgement. Our studies have been supported by the DFG through SFB 405, B10, and Graduiertenkolleg "Expression Analysis".

PATHOLOGY

RESEARCH AND PRACTICE

(C) Urban & Fischer Verlag

http://www.urbanfischer.de/journals/prp

Pathol. Res. Pract. 199: 419-423 (2003) 0344-0338/03/199/06-419 $15.00/0

References

1. Baugh LR, et al. (2001) Quantitative analysis of mRNA amplification by in vitro transcription. Nucleic Acids Res 29: e29

2. Bissell MJ, Radisky D (2001) Putting tumours in context. Nat Rev Cancer 1: 46-54

3. Chuaqui RF, et al. (2002) Post-analysis follow-up and validation of microarray experiments. Nature Genetics 32 Supplement: 509-514

4. Chung CH, Bernard PS, Perou CM (2002) Molecular portraits and the family tree of cancer. Nature Genetics 32 Supplement: 533-540

5. Cohen CD, et al. (2002) Laser microdissection and gene expression analysis on formaldehyde-fixed archival tissue. Kidney Int 61: 125-132

6. Eberwine J, et al. (1992) Analysis of gene expression in single live neurons. Proc Natl Acad Sci USA 89: 3010-3014

7. Eberwine J, et al. (2001) Local translation of classes of mRNAs that are targeted to neuronal dendrites. Proc Natl Acad Sci USA 98: 7080-7085

8. Elowitz MB, et al. (2002) Stochastic gene expression in a single cell. Science 297: 1183-1186

9. Ernst T, et al. (2002) Decrease and gain of gene expression are equally discriminatory markers for prostate carcinoma: a gene expression analysis on total and microdissected prostate tissue. Am J Pathol 160: 2169-2180

10. Fink L, et al. (1998) Real-time quantitative RT-PCR after laser-assisted cell picking. Nat Med 4: 1329-1333

11. Iscove NN, et al. (2002) Representation is faithfully preserved in global cDNA amplified exponentially from subpicogram quantities of mRNA. Nature Biotechnology 20: 940-943

12. Masuda N, et al. (1999) Analysis of chemical modification of RNA from formalin-fixed samples and optimization of molecular biology applications for such samples. Nucleic Acids Res 27: 4436- 4443

13. Moore MJ (2002) Nuclear RNA turnover. Cell 108: 431-434

14. Olert J, et al. (2001) HOPE fixation: a novel fixing method and paraffin-embedding technique for human soft tissues. Pathol Res Pract 197: 823-826

15. Paweletz CP, et al. (2001) Reverse phase protein microarrays which capture disease progression show activation of pro-survival pathways at the cancer invasion front. Oncogene 20: 1981-1989

16. Rao CV, Wolf DM, Arkin AP (2002) Control, exploitation and tolerance of intracellular noise. Nature 420: 231-237

17. Rosenwald A et al. (2002) The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. New Engl J Med 346: 1937-1947

18. Savagner P (2001) Leaving the neighborhood: molecula\r mechanisms involved during epithelial-mesenchymal transition. Bioessays 23: 912-923

19. Simone NL, et al. (2000) Laser capture microdissection: beyond functional genomics to proteomics. Mol Diagn 5: 301-307

20. Van de Vijver MJ, et al. (2002) A gene expression signature as a predictor of survival in breast cancer. New Engl J Med 347: 1999-2009

21. Van Gelder RN, et al. (1990) Amplified RNA synthesized from limited quantities of heterogenous cDNA. Proc Natl Acad Sci USA 87: 1663-1667

22. Wang E, et al. (2000) High-fidelity mRNA amplification for gene profiling. Nature Biotechnology 18: 457-459

23. Wong YC, Wang YZ (2000) Growth factors and epithelial- stromal interactions in prostate cancer development. Int Rev Cytol 199: 65-116

Received: May 25, 2003

Accepted: June 4, 2003

Manfred Hergenhahn1, Marc Kenzelmann2, Hermann-Josef Grone3

1 Division C040: Genetic Alterations in Carcinogenesis,

2 Division A020: Molecular Biology of the Cell I,

3 Division E090: Department of Cellular and Molecular Pathology, German Cancer Research Center, Heidelberg, Germany

Address for correspondence: Hermann-Josef Grone, German Cancer Research Center, Division E090, Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany.

Phone: +49 6221 42 4350; Fax: +49 6221 42 4352.

E-mail: h.-j.groene@dkfz.de

Copyright Urban & Fischer Verlag 2003

More News in this Category