When Santiago Ramón y Cajal was awarded the Nobel Prize for Physiology or Medicine in 1906 for his work on the anatomy of the nervous system, he shared the prize with Camillo Golgi. Working in a simple laboratory set up in his kitchen, Golgi developed in 1873 the “black reaction,” the hardening of tissue in potassium dichromate followed by permeation of the nervous components by silver nitrate. This Golgi stain greatly facilitated his own studies on nerve cells in the brain and the work of others such as Ramón y Cajal, who developed the concept of the neuron, a single cell body with multiple branching dendrites and an axon that can stretch for up to 1 meter in humans, all visible with the Golgi stain. This specific black staining of single neurons was the cradle of neuroanatomy and a very early example of a chemical-biological interaction. The chemical staining of tissues for microscopic examination, either directly or first by targeting with an antibody, remains the central pillar of histopathology and owes much to the pioneering works of Golgi and Ramón y Cajal.
Technological advances in the field of mass spectrometry in the past 15 years now offer a completely new way of examining tissues, as did Golgi's great advance 140 years ago. Unlike histochemistry and immunohistochemistry (IHC), this new technology does not target molecules in tissues. It is a discovery tool that can be adapted to screening once novel tissue targets have been defined and is known by the forbidding name of matrix-assisted laser desorption/ionization (MALDI) imaging mass spectrometry (MALDI-IMS). Modern mass spectrometers can measure low numbers of ions with accurate mass, meaning that attomole levels of proteins and peptides can be detected as charged species whose mass can be determined with high resolution, facilitating their identification from databases. The secret is not the mass spectrometer, but rather how to persuade molecules such as proteins, peptides, and lipids within tissues to ionize without decomposition into hundreds of smaller fragments. First, a laser is used and rastered across the surface of a mounted tissue section. The principle employed is “soft ionization,” the addition of a matrix to the sample, which absorbs the energy of the laser, and “gently” transfers it to biological molecules causing them to form singly charged ions that can be attracted into the mass spectrometer. The term “matrix-assisted laser desorption” was coined after the discovery that in a mixture of tryptophan and alanine, tryptophan could assist alanine to ionize at a laser energy only 10% of that needed for alanine alone. Now there are several aromatic molecules like tryptophan that are used to assist soft ionization of molecules in biological samples. After spraying with a matrix solution, the thin matrix coating extracts biological molecules from the underlying tissue which co-crystallize with the matrix. When the tissue is then painted with a narrow laser beam, the matrix absorbs the laser energy and facilitates the desorption and ionization of the biological molecules. The ions formed are acquired by the mass spectrometer at defined geometric coordinates across the whole tissue section and result in a large dataset of ions acquired at hundreds to thousands of discrete and defined pixels across the 2D surface of the tissue. Software then displays the pixels in terms of the ions detected at each of those points and distributions of the ion intensities can then be produced. Using their accurate masses or their mass spectra produced when the ion is permitted to fragment prior to entry to the mass spectrometer, ions can be solved in terms of the molecules from which they were derived[3, 4] and then density maps of these molecules are produced across the tissue section under examination.[5, 6] This, in essence, is MALDI-IMS (Fig. 1).
How does MALDI-IMS analysis of tissue samples differ from classical histology or immunohistochemistry? First and importantly, as stated above, IHC, for example, is a targeted analysis using specific antibodies as detection tools. In contrast, MALDI-IMS can be used as a discovery tool and the finding of novel molecules confirmed by IHC. This was recently accomplished with the finding that monomeric ubiquitin distinguished hepatocellular carcinoma (HCC) from adjacent cirrhotic tissue. Second, traditional histology and IHC view light transmitted through the stained tissue section and the trained observer can record the architecture of the tissue, its organelles, and the distribution within the section of the target molecule, in the case of IHC. With MALDI-IMS, the distribution across the tissue section of a large number of different molecules can be visualized simultaneously. This permitted, for example, the understanding that specific phosphocholine species were distributed in zones within hepatocytes from healthy obese patients and those with simple steatosis, but that this zonation was lost in nonalcoholic steatohepatitis (NASH). These observations led to further investigations and to conclusions regarding the etiopathogenesis of NASH.
The hepatic metabolism of drugs and the occurrence of drug-induced hepatocellular damage both have a long research history. The tyrosine kinase inhibitor lapatinib, used for the treatment of breast cancer, was repeat-administered to dogs and their livers examined by MALDI-IMS. Parent drug and 22 metabolites were visible in liver sections at a special resolution of 50 μm. Two of these metabolites generated ions of 473.1045 and 473.1175, that is, different by <0.002%, but were completely resolved by MALDI-IMS displaying differential tissue distributions. The ability to distinguish the parent drug from its metabolites in tissue sections without the use of radioisotopes makes MALDI-IMS a unique methodology. As the authors state, “The ability to map pharmacological receptors to drug or metabolite distribution is transformational in drug development. New insights into the mechanisms of pharmacology and toxicology will be possible.”
MALDI-IMS gives powerful new insights into tissues such as the liver, not only for the spatial distribution of proteins and peptides, but also for lipid species, drugs, and their metabolites. However, it should be recognized that MALDI-IMS is not without its limitations. Specifically, it is not itself a proteomic or metabolomic tool, meaning that global analysis of proteins, lipids, and small molecules is not possible with MALDI-IMS alone. MALDI-IMS images can be generated only for the most prevalent ions, representing perhaps 1,000 proteins, which is a relatively small fraction of the complete proteome for most tissues. Nevertheless, compared with histology or IHC, it represents a major advance in tissue imaging. In addition, the signals obtained by MALDI-IMS are only semiquantitative, but more quantitative than IHC. To obtain further quantitative data on the amounts of a protein, lipid, or small molecule in a tissue, triple quadrupole mass spectrometry (TQMS) with multiple reactions monitoring (MRM) would be required. TQMS with MRM is the basic workhorse of proteomics and lipidomics.
In the current issue, Turtoi et al. seize on the issue of tumor heterogeneity and the major obstacle that this presents to cancer treatment. They employed MALDI-IMS to examine the heterogeneity of the proteome of colorectal carcinoma liver metastases. They accumulated data on over 1,000 proteins and their spatial distribution in the membrane and peritumoral region for eight such metastases. To their surprise, they found that these regions harbored a pattern of protein biomarkers that was reproducible and occurred in defined tissue zones. Their MALDI-IMS findings can be summarized as enhanced protein and lipid synthesis in the peritumoral zone, increased carbohydrate metabolism and DNA repair at the heart of the metastases, and finally, elevated markers for proliferation, motility, and drug metabolism on the metastatic boundary. In addition, two novel biomarkers, latent transforming growth factor beta-binding protein (LTBP2) and transforming growth factor, beta-induced (TGFBI), were consistently expressed in these eight metastases. LTBP2 was found mostly on the rim of the metastasis while TGFBI accumulated mostly at the core of the metastasis. The authors eliminated the possibility that the expression of these two antigens was caused by inflammation, by IHC in 10 cases of cirrhosis due to either alcohol or viral hepatitis. Inflammatory cells were clearly present but the tissues were negative for LTBP2 and TGFBI. The authors were keen to establish that this new information might be useful for in vivo targeting liver metastases using antibodies against LTBP2 and TGFBI. To this end, human colorectal carcinoma cells were xenografted onto the chorioallantoic membrane of fertilized chicken eggs, taking care to select tumor cells (SW1222) that expressed LTBP2 and TGFBI. After 7 days, fluorescence tagged anti-LTBP2 and anti-TGFBI polyclonal antibodies were injected and, after a further day, the xenografts were examined and found to be positively labeled. This work holds out early hopes that antibody-drug conjugates targeted against these two novel therapeutic targets may display sufficient specificity and efficacy to be added to the cancer chemotherapeutic armamentarium.
The work of Turtoi et al. demonstrates amply how the emergent technology of MALDI-IMS can furnish new and important insights into tumors and into liver pathology and also lead rapidly to further investigations that directly seek to define novel therapeutic targets. But to what extent will MALDI-IMS revolutionize the practice of liver histology and replace or enhance current protocols that employ mostly IHC? The best lessons perhaps can be drawn from bacteriology. Back in 1975, it was first demonstrated that inserting lyophilized bacteria on a probe directly into a mass spectrometer and heating the probe to pyrolyze the bacteria under vacuum would produce mass spectra that were characteristic of each bacterial species under investigation. In the case of gram-negative bacteria, these spectra were shown to result from the pyrolysis of phospholipids and ubiquinones in the bacterial samples. Today, it is anticipated that MALDI-TOFMS analysis of bacterial spots may replace Gram staining and biochemical identification of bacterial species in the near future. Why is this? In a study of 1,660 bacterial isolates, 95.4% were correctly identified by MALDI-TOFMS. In most cases, test failure or erroneous identification was due to incorrect database entries. The authors calculated that identification of a single isolate took 6 minutes at a cost of 22%-32% of the cost of traditional bacteriological methods. The high degree of accuracy, high-throughput, low-cost procedures should be welcomed and we predict that they must also come in histopathology in the guise of MALDI-IMS. The scientific case has been made and is obvious, but there is at least one other important practical consideration.
Mass spectrometer manufacturers need to recognize that huge floor-standing instruments costing upwards of $1 million are not going to proliferate in routine pathology labs. A much greater effort is required to reduce the size and unit cost, together with development of software that would be regarded as “friendly” by a pathologist, offset by the promise of proliferation of such instruments. History teaches us some lessons in this regard. Those who invented, developed, and named the polymerase chain reaction in 1986 probably could not have predicted the wildfire spread of PCR instruments across the globe. Rendering the PCR instrument easy to use has been one of the great endowments of the manufacturing community. Doubtless, commercial competition has played a role. Although MALDI-IMS appears considerably more complex than PCR, this should not deter the quest for the proteomic and metabolomic equivalent of PCR.
Diren BeyoĂlu and Jeffrey R. Idle
Hepatology Research Group, Department of Clinical Research, University of Bern, Bern, Switzerland