Molecular Vision 2006; 12:704-711 <>
Received 31 March 2006 | Accepted 20 June 2006 | Published 21 June 2006

Quantitative measurement of young human eye lens crystallins by direct injection Fourier transform ion cyclotron resonance mass spectrometry

Noah E. Robinson,1 Kirsten J. Lampi,2 J. Paul Speir,3 Gary Kruppa,3 Michael Easterling,3 Arthur B. Robinson1

1Oregon Institute of Science and Medicine, Cave Junction, OR; 2Oregon Health and Science University, Portland, OR; 3Bruker Daltonics Corporation, Billerica, MA

Correspondence to: Noah E. Robinson, Oregon Institute of Science and Medicine, 2251 Dick George Road, Cave Junction, OR, 97523; Phone: (541) 592-4142; FAX: (541) 592-2597; email:


Purpose: Human eye lenses at birth are primarily constructed of 12 distinct crystallins and two truncated crystallins. The molecular weights of these 14 proteins vary between about 20,000 and 30,000 Da. The relative amounts of these molecules and their post-synthetic changes with age are of substantial interest in the study of lens biochemistry and lens pathology. Fourier transform mass spectrometry of unfractionated lens homogenates now permits precise quantitative measurement of the relative amounts of lens crystallins. We report herein the measurement of the 14 crystallins in 10 pairs of lenses from humans between the ages of 2 and 300 days.

Methods: Eye lenses were obtained from human donors of various ages in the first year of life. These lenses were homogenized in 0.02 M phosphate buffer at pH 7.0 with 0.001 M EDTA, desalted by washing over a 3,000 Da filter, and injected directly into the nanospray source of a hybrid Fourier transform ion cyclotron resonance mass spectrometer, Qq-FT(ICR)MS, equipped with a 12 Tesla magnet. The crystallins were quantitatively ionized and mass analyzed in the ICR cell of the mass spectrometer. The detected signals of all of the isotopic and charge state species for each crystallin were normalized and summed to determine the protein quantities.

Results: The relative amounts of the 14 crystallins are found to be quite similar from individual to individual at birth. These amounts are in integer ratios to one another that suggest important structural relations within the lens. In two cases, the relative amounts of αA- and βB2-crystallin change proportionally to the logarithm of age during the first year, with αA- decreasing and βB2-crystallin increasing. The changes in αA- and βB2-crystallin are mutually offsetting, with αA-crystallin decreasing from 30% to 18% and βB2-increasing from 12% to 24%.

Conclusions: These observations suggest that the human eye lens at birth is constructed of crystallins in which the numbers of crystallin molecules have regular integral relationships to each other. As the lens develops during the first year, some of these relationships change. While the functional significance of the reciprocal decrease in αA- and increase in βB2-crystallin is not known, βB2-crystallin may substitute for αA-crystallin in the lens structures synthesized during the year after birth. Direct injection FT(ICR)MS of unfractionated lens was found to be an excellent method for the quantitative measurement of lens crystallins.


The lens of the human eye is constructed primarily of an ordered array of lens fiber cells [1]. The principal components within these cells are lens crystallins, of which there are 14 known types in the young, 12 primary gene products and two NH2-terminally truncated products of two of the primary 12 [2,3]. These proteins are packed closely together at very high concentrations in such a way as to provide the lens with its unique optical properties. The abundances of the individual crystallins vary over a range of about 20 fold in the lenses of newborn children [2]. It is to be expected that the relative amounts of the different crystallin subunits correspond to their arrangement in the lens and reflect the specific lens supermolecular structures in which they participate.

As the lens ages, the crystallins also age and change their properties [4-8]. Sometimes these changes in properties cause pathological alterations leading to lens opacity, as in cataract [9]. In order to understand these alterations and to understand the fundamental structure of the lens, it is of value to know the nature of the ordered associations of crystallin subunits and the changes that occur in these ordered arrays with age. This understanding has been impeded, however, by the lack of convenient and reliable means for quantitatively measuring the amounts of the individual crystallins.

We report, herein, the development of a new method for quantitative measurement of lens crystallins and the application of this method to the measurement of the relative amounts of the 14 crystallins in human eye lenses at various ages during the first year of life.

This method is based upon direct nanospray injection of lens homogenates into a 12 Tesla ion cyclotron resonance Fourier transform mass spectrometer. Each lens measurement requires about 20 min of mass spectrometer operation and provides simultaneous and remarkably reliable quantitative values for all 14 crystallins.


Preparation of lens crystallins

Eye lenses from 10 human donors of ages 2 (4), 23, 25, 30, 90, 150, and 300 days were obtained from the Lions Eye Bank of Oregon and processed as previously described [6]. The lenses were homogenized in 1.0 ml of 0.02 M phosphate buffer at pH 7.0 with 0.001 M EDTA, centrifuged to remove any insoluble material, and desalted by repeated H2O centrifugal washing over 3,000 Da (Millipore, Billerica, MA) filters. As a result of the young ages of the donors, the amount of insoluble material in these samples was negligible. Other investigators have similarly reported that, up to 1 year of life, 97% of the total human lens protein is soluble [10]. These eye lens protein solutions were diluted with 50:50:0.1 H2O:acetonitrile:acetic acid for mass spectrometry. For lenses from donors older than studied here where insoluble lens crystallin fractions are present, a different nanospray solvent must be used.

Fourier transform mass spectrometry

The lens protein solutions were measured in a Bruker Daltonics 12.0 Tesla Apex Qe FTMS mass spectrometer manufactured by Bruker Daltonics, Inc. (Billerica, MA). Each sample was flow injected into a Bruker on-line nanospray source at a flow rate of 5 μl/h and a scan rate of 20 scans/min for a period of 20 min with all scans cumulatively summed during measurement and before Fourier transformation. The average resolution was about 200,000, where resolution=(mass)/(peak width at half height). The FTMS utilizes ion image current detection wherein coulomb forces between the circulating ions and the collector plates in the ICR cell result in an electric current flow between the plates. This current is directly proportional to the number of ions and to the charge on each ion.

The superb linear correspondence between detector signal and number of ions present has been previously demonstrated [11]. Moreover, demonstrating this further, the samples measured herein varied in total protein concentrations over a more than 10 fold range, while the measured ratios of crystallins remained remarkably constant, as reported in Table 1.

Calculation of relative percentages of crystallins

Fourier transformation of the collected data produces about 300 mass spectral peaks of significant intensity for each individual protein. Figure 1 illustrates these peaks for βB2-crystallin in a 2-day-old human lens sample. The βB2-crystallin molecules become charged to varying extents in the nanospray source. Figure 1A shows the isotopic distribution for those with a charge of +22. These 20 isotopic peaks arise from the presence, at various positions in the protein, of elemental isotopes that have masses greater than 1H, 12C, 14N, 16O, and 32S. The peaks seen in the distribution in Figure 1A are the sixth through the twenty-fifth peaks. Isotope peaks 1 through 5 are present in negligible quantities. Figure 1B shows the distribution of charge states, each of which has an isotopic distribution. These states range from the least charged at +13 to the most charged at +30. βB2-crystallin has 20 significant isotopic peaks for each charge state. With 18 charge states, the total number of βB2-peaks in the mass spectrum is 360.

The relative quantities in Figure 1B are the summed areas of the peaks in the isotopic envelopes as shown in Figure 1A, after correction for charge. Since the ion image current is proportional to charge, molecules that carry a greater charge produce a higher current. A linear correction has been made so that this difference is removed. These quantities have been normalized to 100 for the most abundant charge state. The distribution of charge states is approximately Gaussian as is illustrated by the fitted Gaussian in Figure 1B.

Figure 1A,B illustrates that essentially all of the injected βB2-crystallin molecules have been measured. The measured amounts of proteins with different isotopic compositions and charge clearly approach zero on either side of the distribution functions, so there are no significant amounts of unmeasured types of proteins. This differs from the usual situation with peptide ionization, wherein the distribution functions usually include molecules with very low numbers of ions, thus allowing the possibility of unionized molecules.

These experimentally measured curves were similar for all 14 crystallins. Therefore, this method of quantitative analysis is exceptionally reliable. A correctable sigmoidal variation in detector response with quantity of protein has been reported [11] for long scan times in 7 Tesla mass spectrometers. This variation was not observed in the isotopic abundances of the 12 Tesla measurements reported herein.

The mass spectrometer measures the ratio of mass/charge for each ion. For instance, the eleventh isotope peak of the +22 ions of βB2-crystallin appears at a mass/charge of 1059.45=[23,276+(22)(1.008)+10]/22. The eleventh isotopic peak has mass increased by 10 Da through isotopic composition. The addition of 22 in the equation arises from the increase in mass by 1 Da from the addition of a proton with each positive charge.

Multiplication of the masses of the observed 360 ions of βB2-crystallin and those of the other crystallins by their individual charges results in a spectrum of mass alone as is illustrated for a 2-day-old human lens sample in Figure 2. The peaks shown appear to be unimodal, but they are actually isotopic envelopes as is shown for αA-crystallin in the inset.

The amounts of each protein in each lens sample were determined by summing the areas of all approximately 300 charge-corrected isotopic peaks for each protein. Within each lens sample, the initial percentage protein values were then concentration normalized by division by the sum of the amounts of αA-, αB-, γD-, γC-, and βB2-crystallin, which constitute about two-thirds of the total protein. This corrects for variations in the overall concentrations of the lens samples. All of the protein values were then multiplied by 66.67 to give value sums of approximately 100%. These five peaks were chosen for the normalization basis because of their large and relatively narrow distributions of amounts. These normalization operations were performed once on the entire data set before any further evaluation of the data was carried out. The values were not subsequently changed in any way.


Normalized molar percentages of the crystallins

Table 1 lists the median, mean, and percentage standard deviation of the mean of the normalized molar percentages of the 14 crystallins for three data sets: the four 2-day-old lenses; the set of 23, 25, 30, 150, and 300-day-old lenses; and the complete set of nine lenses. Listed separately are the values for the 90-day-old lens. This lens showed a sharply lower amount of βB2-crystallin and markedly higher amounts of αB-, γC-, and γS-crystallin and is, therefore quite different from the other nine lens samples. This may be the result of unusual biological variation. The 90-day-old sample was, therefore, omitted from further data analysis. This donor died of sudden infant death syndrome and was on a ventilator prior to death.

The percentages of the individual crystallins are remarkably narrowly distributed. Omitting crystallins with percentages under 4% where experimental error is expected to be higher, the average percentage standard deviation of the means of αA-, αB-, γD-, γC-, γS-, βA4-, βB2-, and βB1-crystallin for the four 2-day-old lenses is 16%. For the 11, 23, 25, 30, 150, and 300-day-old lenses and for the entire set of nine lenses, these values are 22% and 22%, respectively.

These percentages include the sum of all experimental errors of the measurement technique and all biological variation between the individual lenses. If experimental errors are estimated at least at 5%, average biological variation from individual to individual in the relative amounts of the crystallins at birth results in a standard deviation of the mean of no more than 10%, as calculated from the four two-day-old samples.

Table 1 also lists the values from an earlier quantitative study [2] in which 0-, 3-, 4-, and 7-day-old human lenses were evaluated by replicated commasie blue staining and densitometric scanning of 2-D gels. These values have been normalized in the same manner as the current data. Considering the lesser accuracy of the gel, staining, and scanning method, our results are in acceptable agreement with the earlier study.

Age-dependent trends for αA- and βB2-crystallins

Figure 3 and Figure 4 show significant percentage composition trends with age that are present for the αA- and βB2-crystallins. αA-crystallin decreases with age, and βB2-crystallin increases with age. The trends are approximately logarithmic with age. The squared correlation coefficients for the age logarithmic plots are 0.79 for αA-crystallin and 0.76 for βB2-crystallin. In comparison, these coefficients for age linear plots are 0.49 for αA- and 0.65 for βB2-crystallin, so the log plots are a better fit to the data than the linear plots. The percentage changes for the two proteins are offsetting, identical, and comparable to the increase of lens size expected between 2 and 300 days, which is also nonlinear [12]. These changes suggest that βB2-crystallin may be substituting for αA-crystallin in lens fiber cells synthesized after birth.

The decrease in αA-crystallin between 2 days and 300 days is not primarily caused by postsynthetic modifications. Many postsynthetically modified crystallins were measured in these mass spectra, but these measurements are beyond the scope of this report. In the case of αA-crystallin, the sum of all higher mass addition products such as oxidized, phosphorylated, and oxidized and phosphorylated αA-crystallin increased by 2% relative to αA-crystallin between 2 and 300 days, while the sum of lower mass deletion products such as truncated serine αA-crystallin increased by 3% relative to αA-crystallin. Simultaneously, between 2 days and 300 days, the molar percentage of αA-crystallin decreased by 40%.

Integer relationships between lens crystallins

Figure 5 shows remarkable integral relationships between the median normalized molar percentages of the 14 crystallins in the 2-day-old lenses. Figure 5 summarizes these findings of integer relationships on a quantitative axis, so that the closeness of linear fit is graphically illustrated. These relationships were discovered after the data calculations were completed. No additional calculations or adjustments of calculations were thereafter performed. Also shown in Figure 5 are the αA- and βB2-crystallin values for the 300-day-old lens.

It is evident that there are six groups of crystallins that have quantitative integral relationships to each other in the ratios of 1/2, 1, 2, 3, 4, and 10 in the 2-day-old lens. The groups include βA2-, βA3-22-, and βB3-crystallin; βA1-, βA3-, and βB1-15-crystallin; αB-, γD-, and βA4-crystallin; γS- and βB1-crystallin; γC- and βB2-crystallin; and αA-crystallin, respectively. Between 2 days and 300 days, αA-crystallin diminishes from 10 to 6, while βB2-crystallin increases from 4 to 8, thus quantitatively offsetting αA-crystallin. The means and standard deviations of the measured ratios in the 2-day-old crystallin groups are shown in Table 2.

Since βA3-22- and βB1-15-crystallins are post-translational modifications of βA3- and βB1-crystallin, Figure 5 could be constructed with these pairs combined. In that case, the βA3-crystallin combination would form a new group at integer 1.5, and the βB1-crystallin combination would join γC- and βB2-crystallin at integer 4.

It is interesting to note that the summed α-, β-, and γ-crystallin protein forms have ratios to one another of 12:13.5:9. If β-crystallins at 1 or less after normalization and αB-crystallin at 2 were involved in separate structures, this ratio for the remaining array would be 10:9:9. Furthermore, as summarized in Table 3, the β- and γ-crystallins seem to be more closely similar to one another than with the α-crystallins. The α-crystallins include one 10 and one 2, while the β- and γ-crystallin each include one each of 2, 3, and 4 and the β-crystallin several smaller components. This is summarized in Table 3. Therefore, there are approximately equal quantities of α-, β-, and γ-crystallins in the newborn human lens.


The relative in vivo normalized molar percentages of the 14 human lens crystallins measured herein are remarkable in three ways. First, in 2-day-old lenses, these molar percentages are very narrowly distributed for each crystallin. The mean percentage standard deviation for the 8 most abundant crystallins is 16%. This includes all experimental measurement errors and all biological individuality combined. This percentage rises to 22% when lenses of all ages between 2 and 300 days are pooled. Second, the percentages of αA- and βB2-crystallin change during the first 300 days after birth by offsetting amounts that are comparable to the amount of lens growth expected during this time. We have previously observed a decrease in αA-crystallin during aging [6]. These offsetting changes could arise in several ways. A simple way would be for βB2-crystallin to be synthesized instead of αA-crystallin, thereby replacing it in lens crystallin arrays synthesized after birth.

Third, the molar percentages of the 14 crystallins in the 2-day-old lens are found to have integral relationships to one another in the ratios of 1/2, 1, 2, 3, 4, and 10 that are far too pronounced to have arisen by chance. Further, the molar percentages of the α-, β-, and γ-crystallin groups appear to have integral relationships with one another. These percentages probably reflect an ordered structure that exists at birth and is composed of these 14 different proteins. These ratios may be specific for the human lens. The rat lens, for example, has been reported to have quite different ratios [13]. Two of the 14 proteins are truncated β-crystallins that are detected even in the newborn and may reflect the age of the proteins in the older fiber cells found in the nucleus of the lens.

Figure 5 shows the molar percentages of the 14 crystallins that were measured in these experiments in the eye lenses of 2-day-old human subjects. Values for the additional age ranges up to 300-day-old humans are listed in Table 1. The molar percentages in the lenses of 2-day-old subjects for all 14 crystallins were found to lie very near to one of the seven values of 1.5, 3, 6, 9, 12, or 30. In the 300-day-old lens, the molar percentage of βB2-crystallin has shifted from 12 to 24, while αA-crystallin has shifted from 30 to 18, exactly offsetting the βB2-crystallin shift of 12. Dividing each of the molar percentages by 3 (Figure 5 and Table 2), shows that the molar ratios of the crystallins to one another are 0.5:1:2:3:4:10. Table 3 summarizes these by α-, β-, and γ-crystallin types.

So, the 14 crystallins of the newborn human eye lens are found to be present in molar ratios that bear the simple integer relationship of 0.5 to 1 to 2 to 3 to 4 to 10 with one another. These molecules are probably packed very closely with one another in the lens in ordered 3-dimensional arrays controlled by multiple specific mutual binding regions on each protein. The discovery of these integer relationships suggests both that these arrays are highly ordered and that their structures are such as to require molar amounts of each crystallin in specific ratios to the others.

Fourier transform ion cyclotron resonance mass spectrometry was first applied to lens crystallin studies for the purpose of measuring crystallin deamidation [11]. In that application, direct injection of lens homogenates followed by quantitative measurement was proved feasible.

This technique has now been applied herein to the quantitative measurement of the amounts of crystallins in human lenses as a function of age during the first year after birth. This technique has three principal advantages. First, it utilizes lens homogenates that are directly injected into the mass spectrometer with no prior sample preparation or fractionation. This avoids procedures that may compromise quantitation due to absorptive or other losses. Second, this technique entirely resolves the crystallins from one another, which avoids loss of quantitation from cross contamination. Third, virtually every injected crystallin molecule is ionized and quantitatively measured by ion image current in such a way that the resulting measurements are correct and based upon physical principles that are inherently linear and exact.

Thus, this technique far surpasses, in convenience and accuracy, the various methods that depend upon wet chemistry separations of the crystallins. These advantages have permitted, in this initial application, the discovery of quantitative relationships between the crystallins that have been heretofore unknown.


We thank Dr. Christoph Borchers and the University of North Carolina for the use of the 12 Tesla FTMS instrument, National Institutes of Health grant EY 12239 to K. J. Lampi, and grants to the Oregon Institute of Science and Medicine by the Kinsman Foundation and the Morse Foundation for support of this work.


1. Bloemendal H. The vertebrate eye lens. Science 1977; 197:127-38.

2. Lampi KJ, Ma Z, Shih M, Shearer TR, Smith JB, Smith DL, David LL. Sequence analysis of betaA3, betaB3, and betaA4 crystallins completes the identification of the major proteins in young human lens. J Biol Chem 1997; 272:2268-75.

3. Lapko VN, Smith DL, Smith JB. Expression of betaA2-crystallin in human lenses. Exp Eye Res 2003; 77:383-5.

4. Hanson SR, Hasan A, Smith DL, Smith JB. The major in vivo modifications of the human water-insoluble lens crystallins are disulfide bonds, deamidation, methionine oxidation and backbone cleavage. Exp Eye Res 2000; 71:195-207.

5. Kim YH, Kapfer DM, Boekhorst J, Lubsen NH, Bachinger HP, Shearer TR, David LL, Feix JB, Lampi KJ. Deamidation, but not truncation, decreases the urea stability of a lens structural protein, betaB1-crystallin. Biochemistry 2002; 41:14076-84.

6. Lampi KJ, Ma Z, Hanson SR, Azuma M, Shih M, Shearer TR, Smith DL, Smith JB, David LL. Age-related changes in human lens crystallins identified by two-dimensional electrophoresis and mass spectrometry. Exp Eye Res 1998; 67:31-43.

7. Lund AL, Smith JB, Smith DL. Modifications of the water-insoluble human lens alpha-crystallins. Exp Eye Res 1996; 63:661-72.

8. Zhang Z, Smith DL, Smith JB. Human beta-crystallins modified by backbone cleavage, deamidation and oxidation are prone to associate. Exp Eye Res 2003; 77:259-72.

9. Takemoto LJ. Disulfide bond formation of cysteine-37 and cysteine-66 of beta B2 crystallin during cataractogenesis of the human lens. Exp Eye Res 1997; 64:609-14.

10. Thomson JA, Augusteyn RC. Ontogeny of human lens crystallins. Exp Eye Res 1985; 40:393-410.

11. Robinson NE, Lampi KJ, McIver RT, Williams RH, Muster WC, Kruppa G, Robinson AB. Quantitative measurement of deamidation in lens betaB2-crystallin and peptides by direct electrospray injection and fragmentation in a Fourier transform mass spectrometer. Mol Vis 2005; 11:1211-9 <>.

12. Harding J. Cataract: biochemistry, epidemiology, and pharmacology. London: Chapman and Hall; 1991. p. 3.

13. Lampi KJ, Shih M, Ueda Y, Shearer TR, David LL. Lens proteomics: analysis of rat crystallin sequences and two-dimensional electrophoresis map. Invest Ophthalmol Vis Sci 2002; 43:216-24.

Typographical corrections

Robinson, Mol Vis 2006; 12:704-711 <>
©2006 Molecular Vision <>
ISSN 1090-0535