Knowledge Bank

Frozen tissue sections of $25-50\mu$ thickness were placed on silver chloride plates of 1-square-inch area and 1-mm thickness, and dried at low temperature. As a check on the exact location of areas selected for spectroscopic study and to rule out complicating diseases, histologic preparations were made from the block remaining after cutting the sections for spectroscopy. For recording of the spectra with the Beckman IR-2 infrared spectrophotometer, the silver chloride plate was placed in a cell holder, and the desired part of the section was centered and then covered with an accurately designed mask so that no infrared energy would pass through any other area of the tissue or part of the plate not covered by the section. Most of our tissue spectra were obtained from an area of $8\times 5mm$. Areas as small as $3\times 3mm$, however, could be studied without increasing the proper signal-to-noise ratio of our instrument. Infrared spectra of visceral tissues (liver, heart, medulla, cortex of the kidney, adrenal, thymus, spleen, and lymph node) from rabbits and neural tissue (different parts of the brain, sciatic nerve, and optic nerve) from several species were recorded from $2.5-15\mu$. The infrared spectra of those tissues showed two different types of vibrations: 1. In the region below $8.10\mu$, all tissues showed a number of more or less strong bands at almost identical locations. These bands were very similar to known vibrations ($3.04\mu$ NH stretching, $3.4\mu$ CH bending, $6\mu$ CO stretching, the deformation vibration of NH at $6.44\mu$, and the deformation vibration of CH at $6.88\mu$) found in proteins (Sutherland). An additional medium-strong band which is still unassigned could be seen regularly at about $8.04-8.08\mu$. 2. In the region above $8.10\mu$ ---more toward the real infrared, e.g., between 8.10 and $11\mu$ ---the tissue spectra showed a fingerprint region which is so characteristic for the individual organ that it was frequently possible to identify the organ from the infrared spectrum. Spleen, thymus, and lymph node showed some similarity of fingerprint bands; other viscera could be distinguished by their infrared spectrum without difficulty. It was noteworthy that neural tissue from various locations (different parts of the brain, optic nerve, sciatic nerve) and from different species (fresh human-postmortem material, rabbit, and dog) always showed similar locations of the fingerprint bands. The constancy and reproducibility of tissue spectra were surprisingly high considering the general difficulties and sources of error of infrared spectroscopy of solids and films. For tentative assignments of the characteristic fingerprint regions, a large number of spectra of nucleic-acid metabolites, nucleic acids, adenylic acids, ATP, purine and pyrimidine derivatives, d-ribose, and d-desoxyribose were evaluated. It can be stated with certainty only that none of the strong bands of the tissue fingerprint spectra could be caused by free purines or pyrimidines. D-ribose and d-desoxyribose esters such as occur in nucleic acids, nucleotides, and nucleosides, however, showed a number of infrared absorption bands comparable with those found in the fingerprint region of various tissues. Extraction with a mixture of chloroform and ether had different effects on the fingerprint spectra of different tissues. The characteristic bands of liver tissue were not noticeably influenced by the extraction. In neural tissue, on the other hand, the extraction completely removed the fingerprint bands, such as the strong band at $9.31-9.35\mu$, leaving most of the protein vibrations, such as the deformation vibration of NH at $6.44\mu$, unaffected. It has been pointed out that these results do not necessarily mean that the substances, which were removed by the extraction of neural tissue, must be simple lipids. The disappearance of the strong band at $9.31-9.35\mu$ is also consistent with the removal of -C-O- bonds such as occur in esters. Quantitative estimations in tissue sections, as in films, depend upon relation of optical densities of bands since thickness of either material cannot be measured easily with accuracy. For estimations in the fingerprint region of neural tissue the deformation vibration of NH at $6.44\mu$ was used as an internal standard. The ratios between the optical densities of the internal standard and the fingerprint bands were determined in sections from different parts of the brain in healthy rabbits and in rabbits in insulin shock. Sections from similar material but different thickness gave almost identical values, thus proving the accuracy of our estimations. Slight variations were found in different parts of the brain. Greater differences existed between isolated grey and white matter of the brain. Significant changes, exceeding by far the normal variations, were observed in brain sections from animals in insulin shock. In reviewing the potential usefulness of infrared spectroscopy of tissues, it is felt that although a great many more studies will be necessary for the assignments of tissue spectra, the data presented indicate the value of this technique for physiologic and pathologic investigations.