Detection of heritable mutations as quantitative changes in protein expression.

A computerized search for the appearance of heritable mutations (as indicated by changes in protein expression) was conducted on three sets of mice, whose sires had been either untreated, exposed to 3 gray units of gamma radiation, or treated with 150 mg/kg ethylnitrosourea. Proteins from the livers of approximately 800 mice were separated by two-dimensional electrophoresis, and abundances were measured by using image analysis techniques. Heritable mutations were detected by the appearance of new proteins or by the quantitative decrease in abundance of normally occurring proteins. Measurements of the variability of the protein abundance indicate that at least 48 proteins are consistent enough to be used in searches when mutations are expected to result in a 50% reduction in the normal amount of protein. New proteins were found in four offspring from ethylnitrosourea-treated mice, and in each case a nearby spot was found to be significantly diminished. These mutations were confirmed in subsequent generations. A computer-assisted search detected three of these mutations on the basis of the abundance decrease alone. These results indicate that two-dimensional electrophoresis can be used to detect mutations reflected as quantitative changes in protein expression, provided that the proteins to be monitored are quantitatively stable when samples from different individuals are compared.

A computerized search for the ,appearance of heritable mutations (as indicated by changes in protein expression) was conducted on three sets of mice, whose sires had been either untreated, exposed to 3 gray units of gamma radiation, or treated with 150 mg/kg ethylnitrosourea. Proteins from the livers of approximately 800 mice were separated by two-dimensional electrophoresis, and abundances were measured by using image analysis techniques. Heritable mutations were detected by the appearance of new proteins or by the quantitative decrease in abundance of normally occurring proteins. Measurements of the variability of the protein abundance indicate that at least 48 proteins are consistent enough to be used in searches when mutations are expected to result in a 50% reduction in the normal amount of protein. New proteins were found in four offspring from ethylnitrosourea-treated mice, and in each case a nearby spot was found to be significantly diminished. These mutations were confirmed in subsequent generations. A computer-assisted search detected three of these mutations on the basis of the abundance decrease alone. These results indicate that two-dimensional electrophoresis can be used to detect mutations reflected as quantitative changes in protein expression, provided that the proteins to be monitored are quantitatively stable when samples from different individuals are compared.
Exposure to a mutagen can cause both point mutations and small chromosomal deletions. The combination of a gamete carrying a point mutation with a gamete carrying the unaltered gene could result in an offspring that expresses an altered protein together with the normal protein a t 50% of its normal abundance. A gamete in which a structural gene has been deleted could, if combined with a gamete carrying the normal gene, result in an offspring that expresses the corresponding gene product at 50% of its normal abundance. Therefore, detection of quantitative alterations in protein expression could, theoretically, be used to measure genetic changes that can be tested for heritability and to provide data for estimation of mutation rates.
Two-dimensional electrophoresis has been used successfully to detect qualitative protein changes indicative of point mutations (1, 2) and gene deletions   chemicals (1,2) or ionizing radiation (3). Detection of quantitative protein changes that reflect either point mutations or gene deletions, however, has been hampered by the inability to obtain quantitative measurements from the large number of two-dimensional electrophoresis patterns required for mutation screening (4). Anderson et al. (5) have shown that twodimensional electrophoresis, coupled with computerized data analysis, can detect a 50% reduction in protein amount, provided that the background quantitative variations are small. However, the contribution of individual sample variability, both experimental and biological, to the overall quantitative data dispersion is currently unknown. The magnitude of such variability may well determine the feasibility of ultimately using two-dimensional electrophoresis protein separations to screen human samples for the occurrence of induced mutations following exposure to known or suspected mutagens.
We report here the results of a mutagenesis study in which heritable mutations, represented as altered protein expression, were detected by computer-assisted screening of twodimensional electrophoresis protein patterns. This study was designed to assess the quantitative capabilities of two-dimensional electrophoresis and to evaluate the possible application of this technique to mutation studies in humans. To minimize quantitative variability due to genetic heterogeneity and thus concentrate on quantitative variability introduced by sampling and nongenetic biological factors (e.g. age, diet), we chose to use inbred strains of mice for our initial study. Thus, the results presented here represent the simplest case for the application of two-dimensional electrophoresis to screening for mutations that cause quantitative protein changes and serve as a foundation for human studies in which genetic heterogeneity will contribute additional quantitative variability (5-7).
This study included 797 offspring from untreated male mice or male mice treated with either ethylnitrosourea or gamma radiation. Ethylnitrosourea-induced mutations, previously shown to cause the appearance of new protein spots in twodimensional electrophoresis patterns of mouse liver proteins with a corresponding decrease in the intensity of an adjacent spot ( 2 ) , allowed the detection of rare quantitative protein alterations in two-dimensional electrophoresis patterns to be validated. The ability to detect radiation-induced mutations could then be realistically assessed.

EXPERIMENTAL PROCEDURES
A N D RESULTS'

DISCUSSION
The results of this study demonstrate that quantitative two-dimensional electrophoresis can be used to detect muta-' Portions of this paper (including "Experimental Procedures," "Results," Figs. 1-5, Table I, and additional references) are presented in miniprint at the end of this paper. The abbreviations used are: CV,

Heritu~le ~~t u t i o~
Detected as ~~~t i t~t~u e Protein Changes 12765 tions that cause an altered gene resulting in the expression of a variant protein together with a 50% reduction in the abundance of the normal protein.
The detection of such mutations as quantitative changes in protein expression is, however, limited by the background quantitative variation in each protein monitored. The detection of three out of four ethylnitrosourea-induced mutations based on quantitative changes in normal liver proteins demonstrates this limitation and sets the present detection threshold of the two-dimensiona~ electrophoresis system.
The use of two-dimensional electrophoresis to detect mutations that cause the total loss of one gene copy must still be validated. For this experiment, in order to simplify the mutation search protocol, the assumption was made that the toss of one gene copy would result in a 50% reduction in the synthesis of the amount of the corresponding protein. The possibility exists, however, that intracellular regulatory mechanisms may cause compensatory synthesis of proteins in order to maintain normal concentrations. Given the constraint of quantitative reproducibility defined by our present data, such compensatory mechanisms must be investigated, since quantitative changes of less than 50% that could be significant indicators of mutation might otherwise be ignored. The analysis of protein expression in tissues from heterozygous carriers of known gene inversion or deletion mutations (available as mouse stocks) or in cultured cell lines with induced gene deletions should demonstrate whether or not such mutations are detectable by two-dimensional electrophoresis.
The absence of significant changes in liver protein expression among 369 offspring from irradiated males may be a reflection of (a) the influence of cellular c o m p e n s a t o~ mechanisms that masked gene deletions or (b) the limited number of proteins that had the quantitative stability required for detection of a 50% decrease in protein abundance. Assuming that monitoring the 48 protein spots with coefficient of variation values of no more than 15% would have p e~i t t e d detection of a 50% reduction in expression and that each of the 48 protein spots represents an independent gene locus, and given a specific locus mutation rate of 2.7 X per gray unit per locus as representative of the response to single doses of gamma radiation (8), the expected mutation yield in this study would have been about one event in the 369 gametes screened following exposure to 3 gray units. Thus, zero events is well within the limits of expectation for the number of individuals screened. If more protein spots with low levels of normal variability could be monitored, t,he probability of detecting a quantitative protein variant in a sample of 400 individuals would obviously become more feasible.
In the mouse model system, optimization of the number of protein spots in a two-dimensional electrophoresis pattern that can be monitored for quantitative changes may only require stricter control of quantitative variability introduced by both technical inconsistencies and nongenetic biological factors. Biological factors, on the other hand, contribute significantly to the normal quantitative variability seen in the liver proteins of inbred mice. A subpopulation of liver proteins in male mice, for example, has been found to fluctuate in abundance as a function of sexual maturity (data not shown). Some of the observed quantitative variability may be a function of the liver itself, being a tissue in which protein metabolism is responsive to hormonal controls, diet, andlor circadian rhythm. Another tissue or cell type may, therefore, be better suited to two-dimensional electrophoresis mutagenesis studies. Careful evaluation and modification of sampling protocol should, however, produce an increase in the number of proteins that can be monitored for quantitative protein changes in animal studies.
The applicability of the two-dimensional electrophoresis approach to mutation detection in humans remains to be determined. One consideration is that, unlike the mouse system, nongenetic, or biological variables are not easily controlled among human subjects. Normal genetic differences are also expected to introduce additional background quantitative variation since, when different mouse strains are compared, more genetically regulated quantitative than qualitative protein differences are found (6,7). Although estimates of the occurrence of qualitative genetic variants l i e . protein polymorphisms) in human samples have been made (9)(10)(11), no similar studies have been done to evaluate genetically influenced quantitative protein variability. Finally, human samples for genetic studies are limited to those tissues or body fluids that can be obtained by relatively noninvasive methods, Le. serum, urine, peripheral blood cells, or skin fibroblasts. Of these, only the blood cells or fibroblasts produce two-dimensional electrophoresis patterns comparable in simplicity and resolution to those of the mouse liver pattern (12)(13)(14)(15)(16). The applicabilit~ of two-dimensional electrophoresis to mutation studies utilizing human material should, therefore, be based on the results of studies that measure the quantitative variability of proteins expressed in human cells (e.g. platelets or skin fibroblasts) as a function of both nongenetic (intraindividual variability) and genetic (interindividua~ variability) factors. Such studies would define a subpopulation of proteins that have the quantitative stability required for the detection of heritable mutations and allow a more realistic assessment of the feasibility of using two-dimensional electrophoresis for human mutation studies.
Two-dimensional electrophoresis, together with computerassisted data analysis, can be used to detect mutations as quantitative alterations in protein expression. The use of twodimensional electrophoresis for detecting mutations has an important advantage over other technologies such as the emerging DNA methods. By analyzing proteins expressed in the offspring of exposed individuals, survivable mutations are being monitored. In animal models, the impact of such mutations on the well-being of carriers in several succeeding generations can be assessed, including the consequences of carrying the mutation as a homozygous trait. In addition, rather than identifying DNA damage and having no correlation between the damage and metabolic functions in the organism, the identification of alterations in protein expression allows the identification of the specific lesion, via amino acid sequencing back to the DNA level. Thus, two-dimensional electrophoresis measures damage to functional DNA rather than total DNA. Methods are now being developed for the identification of the altered peptide in the proteins discussed in this paper (17, 18) with a view toward amino acid

Heritable Mutations Detected
as Quantitative Protein Changes sequencing and characterization of the mutation at the gene level.