Molecular Vision 2004; 10:490-511 <>
Received 12 April 2004 | Accepted 12 July 2004 | Published 20 July 2004

Gene expression changes during cataract progression in Sparc null mice: Differential regulation of mouse globins in the lens

Fiona C. Mansergh,1 Michael A. Wride,2,3 Veronica E. Walker,1 Steffan Adams,1 Susan M. Hunter,1 Martin J. Evans1

1School of Biosciences and 2School of Optometry and Vision Sciences, Cardiff University, Cardiff, Wales, UK; 3Cardiff Institute of Tissue Engineering and Repair, Cardiff Medicenter, Cardiff, Wales, UK

Correspondence to: Michael A. Wride, School of Optometry and Vision Sciences, Cardiff University, Room 1.80A, Redwood Building, King Edward VII Avenue, Cathays Park, Cardiff CF10 3NB, Wales, UK; Phone: +44 (0)29 2087 0054; FAX: +44 (0)29 2087 4859; email:


Purpose: Sparc/osteonectin is a hydroxyapatite, calcium and, collagen binding protein, implicated in tissue morphogenesis, cell proliferation, and repair. Sparc null mice develop sub-cortical posterior cataract with eventual rupture of the lens. We wished to correlate genotype with phenotype in these mice via analysis of gene expression pattern changes leading to disease.

Methods: We carried out microarray analysis of adult lenses from Sparctm1cam knockout mice on two strain backgrounds of varying phenotypic severity at two time points, 4 and 9 months. Labelled cDNA from Sparctm1cam knockout and age, strain, and sex matched control lenses was hybridized with HGMP NIA 15,000 clone set arrays. Differential expression was confirmed using semi-quantitative RT-PCR.

Results: We have confirmed differential expression of 54 genes. Most notably, 5 of the mouse globin genes, Hbb-b1, Hbb-b2, Hba, Hba-x, and Hbb-y and an EST, C79876, were significantly downregulated in 9-month old Sparc null mice from two genetic backgrounds at different stages of disease. Another downregulated gene, EraF, is involved in folding of globin proteins. Immune response components, including various members of the complement cascade, were upregulated in lenses with advanced cataract.

Conclusions: Five mouse globins show persistent downregulation as a result of Sparc loss. We speculate as to possible roles of this phenomenon on pathogenesis of cataract in these mice. Other confirmed genes allow extension of previous models of cataract development in Sparc null mice.


Lens development exhibits a seamless progression from embryogenesis through ageing [1,2]. Moreover, lens maturation and growth proceed throughout life. Lens fiber cells differentiate from lens epithelial cells at the equator of the lens, migrate inwards at the bow region and mature around pre-existing fiber cells at the center of the lens and lens fiber cells lose their nuclei via an apoptosis-like mechanism as they mature [3,4]. The lens is surrounded and protected by a thick acellular basement membrane, known as the lens capsule, which is composed of collagen, laminin, fibronectin, nidogen, heparan sulphate proteoglycan, and other components that may be important for interactions between the extracellular matrix (ECM) of the capsule and lens epithelial cells [5,6]. Defects in lens function and/or maturation result in clouding of the lens, or cataract. Age related cataract is the most common form of visual impairment worldwide, while congenital cataract is the most common treatable cause of childhood blindness [7]. Cataract is currently treatable in humans only via surgical intervention. Such treatment is beyond the means of many in the developing world, while in the UK there is a tremendous backlog of cataract patients awaiting surgery, a problem unlikely to dissipate considering ageing Western population profiles [8]. Development of pharmaceutical treatments for cataract would therefore be of huge economic and social benefit.

Much progress has been made recently in uncovering the molecular causes of cataract. This area has benefited greatly from advances in genetics, genomics and proteomics in recent years. The discovery of predisposing human mutations, and phenotypes induced in genetically engineered mouse models show that genetic factors are involved in predisposition to cataract. Posttranslational modifications of protein constituents of the lens core during ageing also undoubtedly contribute to the etiology of nuclear cataract in particular [9,10].

Sparc (secreted protein acidic and rich in cysteine, also known as osteonectin) is a 32 kDa secreted glycoprotein that belongs to the matricellular group of proteins [11-13]. Sparc is known to bind calcium, hydroxyapatite, vitronectin, and collagen and to regulate extracellular matrix production, cell adhesion, proliferation, and migration [12,13]. Sparc is expressed widely during embryogenesis, but in adult mice expression is more restricted. It is highest in remodelling tissue and is also noted in tumorigenesis. Sparc is also known to regulate the activity of various growth factors, including fibroblast growth factor (FGF-2), vascular endothelial growth factor (VEGF), and platelet derived growth factor (PDGF) [14].

Sparc has been disrupted via targeted mutagenesis on two separate occasions, with broadly similar results. Both lines of mice develop cataract and osteopenia as they age [15-18]. Sparc is expressed in embryonic lenses from embryonic day 12 (E12), but is restricted in expression in adult lenses. It is found in large quantities in lens epithelial and newly differentiating fiber cells, but not in the lens capsule [13]. This pattern is largely reflected in human lenses. Sparc expression is noted generally in the lens epithelium, but is highest in peripheral epithelial regions [19]. Notably, Sparc may be involved in human as well as murine cataract since expression levels of Sparc are shown to rise in both nuclear and posterior subcapsular cataract [19,20].

Sparctm1cam knockouts (those studied in this article) are mutated in exon 6 of the Sparc gene and were initially bred on a 129Sv/Ev/Mf1Gpi-bb mixed background [15]. We have also bred this mutation onto an inbred 129Sv/Ev background. Targeted disruption of the Sparc locus in this line results in the replacement of exon 6 with a splice accepting marker and a neomycin resistance cassette [15]. This mutation results in premature termination of transcription and translation, deleting half of the follistatin module and resulting in the removal of downstream domains required for correct folding and secretion of Sparc [15]. The mutation has been shown to result in reductions in RNA levels to the extent that they are undetectable by northern blot [15] (although we can still detect residual RNA expression via PCR). Western blot analysis has demonstrated the complete absence of Sparc protein in this line [15].

Detailed morphological analysis has been carried out on the Sparc null line derived by Norose and colleagues [16]. Analyses of our Sparctm1cam knockout line [15] would indicate that the disease process in both lines is similar. While the cataract phenotype is fully penetrant, there are reported differences in the ages at which the two independently derived knockouts develop cataract. These mice show a variable age of cataract onset. Cataracts become obvious between 5 months and over 1 year of age. In contrast, Sparctm1cam knockouts we have generated on an inbred 129Sv/Ev background start to develop signs of lens opacity between 9 and 11 months of age, with mature cataract appearing by 18 months (unpublished data). The other Sparc knockout line was mutated in exon 4 of the Sparc gene and bred on a 129Sv/C57Bl/6J mixed background [16,17]. These mice showed demonstrable opacity via slit lamp by 1.5-3 months and mature cataract by 7-8 months [16,17]. Heterozygote mice of this knockout line on a C57Bl6/J background also develop cataract at 11 months [17]. Sparc protein has been shown to be absent in mice of both knockout lines, indicating that both represent full 'null' mutations. Differences in the age of cataract onset are most likely, therefore, to result from variations in the differing genetic backgrounds of the mice, rather than from the precise location of the knockout mutation [15-17]. Wild type C57Bl/6J mice develop spontaneous lenticular anomalies with a frequency of 5-15% [21], while strains of 129 origin are natural knockouts of CP49, a lens specific member of the intermediate filament superfamily [22,23]. This mutation causes increased light scatter, but not cataract, in lenses of 129 strain. These observations underscore the importance of assessing knockout phenotypes on different genetic backgrounds.

The study of mouse models of disease with defined genetic defects has led to great breakthroughs in the understanding of the mechanisms of human disease [24]. Study of changes in transcription as a result of gene ablation gives an opportunity to uncover the molecular cascade that leads to the onset of symptoms. The invention of microarray technology allows the investigation of transcriptional changes on a global scale. Microarrays have huge potential in many fields, including ophthalmic research [25]. Following array studies looking at normal mouse lenses during maturation [26], we decided to use microarrays to compare lenticular gene expression in abnormal lenses with controls. We have used the NIA 15K array set [27] to investigate changes in gene expression in the lenses of Sparc knockout mice on two genetic backgrounds with different phenotypic severity at various stages during the process of cataract formation. We have confirmed the differential expression of 55 genes at various stages of cataractogenesis using semi-quantitative RT-PCR. Most notably, 5 mouse globin genes and an uncharacterized EST, C79876, were significantly downregulated in 9-month old Sparc null mice from both genetic backgrounds, at different stages of disease. Overall, the studies described here provide a profile of gene expression, which both supports and extends previous cellular and molecular observations of age-onset cataract progression in Sparc knockout mice.


Mouse nomenclature and strain choice

Sparctm1cam refers to Sparc, targeted mutation 1. University of Cambridge 129Sv/Ev refers to mice of inbred strain 129 and substrain Sv/Ev Mf1Gpi-bb refers to the outbred mouse strain MF1. The strain as obtained from the breeders has a polymorphism in GPI. We have selected and maintained homozygosity for the variant 1b (hence Gpi-bb). This allows chimera analysis of any tissue.

We chose the Sparctm1cam knockouts for investigation. These mice have been maintained on both a mixed background (Sparctm1cam129Sv/Ev/Mf1Gpi-bb) and an inbred background (Sparctm1cam129Sv/Ev). These strains have been maintained by the Evans laboratory (along with parent substrains 129Sv/Ev and Mf1Gpi-bb) since the generation of the initial targeted mutation. The availability of the exact control substrain upon which the initial mutation was generated was extremely important as strain differences between controls and experimental mice can otherwise result in spurious expression differences when studied via array. Notably, the ES lines used to generate this mutation were derived (in this laboratory) from 129Sv/Ev mice, meaning that our 129Sv/Ev control mice also matched the genetic background of the mutated ES cells that contribute to the genetic makeup of the mutated strain. This should avoid error arising from linkage disequilibrium (e.g. differences in expression between experimentals and controls that arise because of promoter differences in genes closely linked to the introduced mutation in Sparc). If the ES cells are genetically different from mouse strains into which the mutation is bred, not only the mutation, but also any closely linked polymorphisms, are fixed into the resulting mutant strain and can become the source of significant differences between control and experimental animals. Controls for Sparctm1cam129Sv/Ev mice were age, strain, and sex matched. Controls for Sparctm1cam129Sv/Ev/Mf1Gpi-bb mice were age and sex matched. Mf1Gpi-bb mice were used as controls for the Sparctm1cam129Sv/Ev/Mf1Gpi-bb arrays, while both Mf1Gpi-bb and 129Sv/Ev controls were used for PCR confirmations of these arrays.

Animals, tissue extraction, and photography

Mice were maintained and sacrificed under Home Office license in accordance with British law (comparable with U.S. Public Health Service Policy on Humane Care and Use of Laboratory Animals). Whole eyes were removed and lenses were dissected via posterior incision using number 5 Swiss watchmaker's forceps in phosphate buffered saline under a Leica MZ125 dissecting microscope (Leica Microsystems Ltd., Milton Keynes, UK). Lenses used in photography were dissected leaving a little iris tissue in order to facilitate lens orientation (useful at advanced stages of cataract where the lens was very misshapen and fragile). Dark field photographs were taken using an Olympus CAMEDIA C4040 zoom digital camera mounted on an Olympus SZX12 dissecting microscope (Olympus UK Ltd., London, UK), then downloaded and trimmed using Adobe Photoshop 6 (see Figure 1). Lenses were assessed via post dissection photography as opposed to slit lamp procedures as we have preliminary evidence that the corneas are abnormal in these mice (unpublished data).

Preparation of target RNA

No differences were noted between the sexes in terms of speed of cataract progression. Nevertheless, in order to minimize biological variability, we used male mice for all array and RT PCR experiments. Lenses used in arrays were dissected free of any contaminating tissue and homogenized in TRIzol reagent (Invitrogen Corp., Paisley, UK), using a hand held homogenizer (Wheaton Science Products, Millville, NJ). RNA extraction was carried out according to the manufacturer's protocol. Concentration and RNA quality were assessed via spectrophotometry and formaldehyde gel electrophoresis. Typically we obtained 30 μl of RNA per pair of lenses at an average concentration of 50 ng/μl. One round of RNA amplification was then used per 500 ng lens sample (Arcturus RNA amplification kit, Arcturus Bioscience Inc., Mt. View, CA), using the manufacturer's protocol. Amplified mRNA (1 μg) was then labeled with Cy3 or Cy5 using the CyScribe labeling system (Amersham, Buckinghamshire, UK) according to the manufacturer's protocol. Labeled probe (1 μl) was added to 2 μl glycerol, then run on a "John gel" (a microscope slide sized thin agarose gel). This was scanned using a GeneTac LS IV scanner (Genomic Solutions, Cambridgeshire, UK) in order to assess probe integrity and successful incorporation of label. Successfully labeled control and experimental targets were then combined and prepared for hybridization.


Array slides were obtained from the Human Genome Mapping Project (HGMP; NIA15000 set, printed on two slides; Hinxton, Cambridge, UK). This array set contains approximately 15,000 ESTs obtained from whole embryos and various embryonic tissues. Up to 50% of these ESTs were derived from novel genes. We should note at this point that the ESTs printed on these arrays do not necessarily represent 15,000 genes, we have found evidence of significant redundancy. For example (see Results), we found 15 different ESTs downregulated, all of which represented the same gene, Hba-a1. We have therefore used the term EST throughout this text to refer to the EST clones printed on the array. Multiple ESTs on the array have, in some cases, been shown to represent the same gene.

Array hybridization and scanning

Array slides were incubated in prehybridization solution for 1 h at 42 °C (50% formamide, 5X SSC, 0.1% SDS, and 1% BSA). Targets were dried via vacuum centrifugation then resuspended in 50 μl hybridization solution (49.9% deionized formamide, 49.9% 20X SSC, and 0.2% SDS) with added 1 μl Cot1 DNA and 1 μl poly A oligonucleotide as blocking agents, heated to 95 °C for 5 min and then added to the face of one slide. The printed face of the second slide of the pair was then placed face to face with the first, using the same probe. Slide pairs were then placed on a level plastic cover above some 1X SSC moistened tissue in a slide box. The slide box was sealed with Nescofilm (Karlan Research Products Corporation, Santa Rosa, CA), placed floating in a water bath and hybridized for 24-48 h at 42 °C. Following hybridization, slides were washed once in wash solution 1 (1X SSC, 2% SDS in filtered autoclaved distilled demonized H2O [ddH2O]) for 20 min, then twice in wash solution 2 (0.1X SSC, 0.2% SDS in ddH2O) for 20 min each. Slides were dipped in nuclease free filtered water, then spray dried. Finally, the backs of the slides were cleaned with ddH2O, wiped with 100% ethanol, then wiped dry and scanned at 12.5 μm using a GeneTac LSIV scanner, (Genomic Solutions). Arrays were repeated 5 times with fluor switching, in order to counteract any issues of dye bias that may have arisen from direct labeling. Repetition sets used biological samples derived from at least four control and four experimental animals per set in order to control for biological variation between individual animals. Scanned images were stored and filtered, then analysed using the GeneTac Analyser spot finding software (Genomic Solutions).

Data analysis

Output files from the GeneTac analyser were saved in MS Excel spreadsheet format. MS Excel was used for all further data manipulation. We normalized each channel via total array methods by calculation of the mean intensity value. Normalized intensities were then analyzed using two different methods. First, we transferred normalized intensity values for each of 5 experimental repetitions into a separate Excel file, ensuring a standard order of samples. These data were then formatted for Significance Analysis of Microarrays (SAM). SAM was carried out and genes that showed a fold change of 2 or above and that appeared in a SAM analysis as statistically significant above a delta value of 0.5 (which denotes an error rate of 5%) were selected for further appraisal. SAM, however, has some disadvantages. All replicates have to be in the same order to use SAM, precluding some methods of data filtration. In addition, each cDNA probe (represented by a single accession number) is spotted on the HGMP NIA array slides twice. Using SAM, it is impossible to assess both duplicate spots together. Hence, we supplemented SAM analysis with a second analysis method as follows. Following normalization, we used approximately 700 blank spots per slide to calculate a mean background value plus 2 standard deviations of that background value for each channel. Genes that fell below this cutoff in BOTH control and experimental channels were removed. Genes that failed to show a fold change of 2 were also removed. Remaining data (above background plus 2 standard deviation values, fold change of >2) were retained, combined with similar data from other replicates and sorted via accession number. Genes that appeared above background plus 2 standard deviation values with a fold change >2, in at least 8 out of 10 replicates (this analysis combined values from duplicate spots, giving 10 replicates per EST), were retained for further analysis. Finally, accession numbers obtained from this method and from SAM were compared. Genes appearing as differentially regulated using BOTH methods were deemed significant. In other words, significant ESTs are changed in expression by at least two fold, are above background plus 2 standard deviations in at least one channel and have a delta value of 0.5 using SAM. These ESTs were subjected to bioinformatic analysis and were all assessed via RT-PCR in order to confirm array results.

Arrays and MIAME standards

Data presented here are in full compliance with MIAME standards [28]. Images, raw data, processed data, and a "ReadMe" file are available from the authors.

Bioinformatic analysis of ESTs

ESTs that were significantly differentially regulated as determined above were compiled. Accession numbers were then used to comprehensively search the NCBI databases. Some data compiled from various NCBI databases (UniGene, Homologene, OMIM, LocusLink, PubMed) are shown in Table 1, Table 2, Table 3, Table 4, and Table 5. These data are also available at the Cardiff Array Database.

RT-PCR array confirmations

We tested all genes arising from the four experiments using semi-quantitative RT-PCR. This method was chosen as we had 108 genes to test. We were able to do this in a more cost-effective and high-throughput manner than via the use of quantitative PCR methods. Semi-quantitative RT-PCR was carried out as follows. RNAs were quantified as described above. Even quantities of control and experimental RNA (usually 2-4 μg) were treated with DNAfree (Ambion Inc., Cambridgeshire, UK) according to the manufacturer's protocol. Reverse transcription (RT) reactions were carried out using the Superscript II First Strand Synthesis System for RT PCR (Invitrogen), according to the manufacturer's instructions, except that due to the low concentration of RNA per sample, we tripled the amount of RNA used and therefore the reaction volumes per sample. Unamplified RNA was always used for RT-PCR confirmations in order to correct for bias introduced to the array results via the amplification process. A "no RT" control corresponding to each sample was also produced. These were treated in exactly the same way as the samples except that Superscript II reverse transcriptase (Invitrogen) was not added.

Primer design

Standard primers were used for housekeeping genes (β-actin, GAPDH, and βA1-crystallin, see Table 1, Table 2, and Table 3 for primer sequences). We used 3 housekeeping genes in conjunction in order to ensure that biased expression of one would not adversely affect results. Moreover, GAPDH and β-actin were present on the NIA arrays used and were not significantly altered in expression. Expression of crystallins has been shown to be unperturbed in Sparc null mouse lenses [6]. Primers for ESTs were designed from the EST sequences. However, EST sequences are usually derived from high throughput single pass sequences. In order to check the sequence quality of the ESTs and avoid areas of poor quality sequence when designing primers, accession numbers of NIA ESTs were checked against the NCBI databases. EST sequences that were assigned to Unigene clusters were checked via bl2seq (BLAST 2 sequences) against a consensus mRNA/cDNA sequence corresponding to that gene, if one was given for that cluster. ESTs without a corresponding full length gene sequence were checked via mouse genome BLAST or BLASTnr. Primers were designed with reference to BLAST results. Where possible, areas of mismatch arising from probable EST sequence error (such as those within long runs of the same base or those at the beginning or end of the EST sequence) were avoided. Primers were designed using Primer3 (version 0.1 beta 1a) from FASTA EST sequence (with areas of unreliable sequence excluded). Primers for the mouse hemoglobin genes were designed by hand. Due to the close similarity between different beta isoforms (Hbb-b1, Hbb-b2, Hbb-Y, Hbb-Z) and different alpha isoforms (Hba-a1, Hba-X), primers specific to each isoform were required. Therefore, we generated separate alignments of the alpha and beta globins using DNASIS version 2.5 and designed primers in such a way that each primer was 3' mismatched for every isoform except the one they were intended to amplify. Primer sequences for all ESTs are shown in Table 1, Table 2, and Table 3.


PCRs were initially carried out at the optimum Tm indicated for the primer pair designed. No RT controls were always included. Conditions were varied later if necessary in order that only single bands appeared and the number of cycles for minimum visibility on a gel were always identified. We ensured that PCR bands obtained for housekeeping controls were even at minimum visibility cycles before testing other genes. In order to minimize false positive results, we used at least two sets of Sparc null and 2 sets of control cDNAs from different animals. Control cDNAs were both 129Sv/Ev when carrying out confirmations from the Sparctm1cam129Sv/Ev array experiments. Sparctm1cam129Sv/Ev/Mf1Gpi-bb experiments were confirmed using one 129Sv/Ev control and one Mf1Gpi-bb control. This was done in order to ensure that the controls represented gene expression patterns from both parental strains. Variation was observed in both sets in order for a gene to be confirmed as variably expressed. PCRs were carried out in 20 μl volumes using 0.025 μmol concentrations of each dNTP (100 mM dNTP set, Invitrogen) and 1 unit Taq DNA polymerase (Sigma) per reaction. PCRs were usually carried out using PCR buffer (Sigma) containing 15 mM MgCl2, however, we varied magnesium concentrations where necessary in order to optimize results. Annealing temperature and the number of cycles were varied where necessary, up to a maximum of 50 cycles, and finally some primers were also redesigned. Final PCR conditions are indicated for each gene in Figure 2 and Figure 3.

Clustering analysis

We generated clusters from gene expression profiles of all ESTs corresponding to genes confirmed in any one of our experimental sets. We merged data from all SAM files for all data points. We then extracted the normalized data profiles for all repetitions for the accession numbers of interest. Mean values were obtained for repetitions of the same experiment, from duplicate spots and, where applicable, for different ESTs representing the same gene. We then calculated mean fold changes per gene for each of the four experiments we carried out. These results were then used in clustering analysis (see Figure 4 for fold changes and clusters). Datafiles were imported into GeneSight 3.5 (BioDiscovery, El Segundo, CA). Hierarchical clustering was performed under default conditions to give an idea of the absolute numbers of credible clusters. Clustering analysis was then repeated using 2D self organizing maps and the results are summarized in Figure 4.


Progression of cataract in different strain backgrounds

Progression of cataract in Sparctm1cam129Sv/Ev mice is shown in Figure 1 via darkfield photography of whole lenses. Photography of whole dissected lenses was used in preference to sections here as this also demonstrates how we assessed lenses prior to RNA extraction for arraying. Sparctm1cam129Sv/Ev (4 month, Figure 1C) and Sparctm1cam129Sv/Ev/Mf1Gpi-bb lenses (4 month, not shown) were indistinguishable from 129Sv/Ev and Mf1Gpi-bb controls at 4 months (not shown) and 9 months (Figure 1A and Figure 1B). Sparctm1cam129Sv/Ev lenses at 9 months showed no significant degree of opacity, (Figure 1D). Obvious signs of developing cataract become visible under the dissecting microscope at approximately 11 months in Sparctm1cam129Sv/Ev lenses, particularly in the posterior region (Figure 1E) and became steadily worse over time (e.g. 16 months, Figure 1F), with lens capsule rupture occurring after 18 months (Figure 1G). Sparctm1cam129Sv/Ev/Mf1Gpi-bb cataractous lenses at 9 months used for arrays were similar to the Sparctm1cam129Sv/Ev lens at 16 months, shown in Figure 1F. Sparctm1cam129Sv/Ev/Mf1Gpi-bb mice that had cataract at 9 months tended to progress to lens capsule rupture at about 12-16 months of age as presented here (Figure 1H) and as previously noted [15]. Other Sparctm1cam129Sv/Ev/Mf1Gpi-bb mice showed a pattern of cataract progression more similar to that shown by Sparctm1cam129Sv/Ev mice.

Array results

Four different array comparisons were carried out. These were:

(1) Sparctm1cam129Sv/Ev versus 129Sv/Ev at 4 months of age. Sparctm1cam129Sv/Ev lenses were arrayed against those from 129Sv/Ev mice of the same age and sex at 4 months and 9 months. Four months of age was comfortably pre-symptomatic, while the 9 month time point was chosen to show gene expression patterns just prior to the onset of obvious morphological changes associated with cataract in Sparc knockout mice. Analysis of array results at 4 months showed significant downregulation of only 2 ESTs. Bioinformatic investigation revealed that both of these ESTs represented the Sparc gene. RT PCR, not unsurprisingly, showed that Sparc was significantly downregulated in Sparctm1cam129Sv/Ev knockout mice in comparison to controls (our knockout mice had previously been shown to express small quantities of RNA while Sparc protein was entirely absent [15]).

(2) Sparctm1cam129Sv/Ev versus 129Sv/Ev at 9 months of age. The 9 month Sparctm1cam129Sv/Ev versus 129Sv/Ev array set showed a significant downregulation of 70 ESTs (listed in Table 1). Bioinformatic analysis showed that the 70 accession numbers for these ESTs represented 44 genes, showing that there was significant redundancy within the NIA 15K set. Two ESTs each represented Sparc, Igf2, H2afz, and Sfpq, while H19 was represented by 6 ESTs. The most significant redundancies were 15 ESTs corresponding to Hba-a1, the major mouse alpha globin and 4 ESTs representing Hbb-b2, one of the major mouse beta globins. We considered the high level of representation of these genes (19 ESTs out of a total of 70) highly significant.

Given high sequence similarities within the beta globin and alpha globin gene clusters we considered that cross-hybridization on the array between different mouse globin species was likely. For example, the two major mouse beta globins, Hbb-b1 and Hbb-b2 show 13 coding sequence mismatches at the DNA level. The embryonic globins, Hbb-Y, Hbb-Z, and Hbb-Z are less similar than this but still share a significant level of DNA homology to either the major alpha or major beta globins. We therefore designed primers specific to each of Hba-a1, Hba-x, Hbb-b1, Hbb-b2, Hbb-Y, and Hbb-Z (primers contained 3' terminal mismatches to each of the other globin DNA sequences). Like humans, mice also have two major adult alpha globin genes, giving them a total of 7 globin genes per haploid genome. The coding sequence of these two genes is completely identical at the DNA (and therefore RNA) level, moreover, their expression is co-ordinated. We could not distinguish between these by PCR. Therefore, we refer to both as Hba-a1.

RT-PCR confirmed the downregulation of 22 out of 44 genes, representing 47 out of 70 ESTs (see Figure 2 and Table 4). The success rate of these arrays was therefore variously 67% (per EST/accession number) or 50% (per gene). In addition to the 22 genes detected via array and then confirmed, we also confirmed the downregulation of 3 out of 4 remaining mouse globins (not identified via array but checked because of high sequence homology). Hbb-y, Hbb-b1, and Hba-X were confirmed. Hbb-Z was not amplifiable from our lens cDNAs. Our Hbb-Z primers do amplify Hbb-Z from E14.5 mouse whole eye cDNA, indicating that this globin is simply not expressed at detectable levels in adult lens (data not shown).

(3) Sparctm1cam129Sv/Ev/Mf1Gpi-bb versus Mf1Gpi-bb at 4 months of age. Sparctm1cam129Sv/Ev/Mf1Gpi-bb lenses were arrayed against those from Mf1Gpi-bb mice of the same age and sex at 4 months and 9 months of age. Four month lenses used for arrays resembled control lenses in clarity, while 9 month lenses were at an advanced stage of cataract (similar to Figure 1F). The 4 month Sparctm1cam129Sv/Ev/Mf1Gpi-bb versus Mf1Gpi-bb showed significant downregulation of 14 ESTs and significant upregulation of 3 ESTs. Three downregulated ESTs represented Sparc, the others were unique. When testing Sparctm1cam129Sv/Ev/Mf1Gpi-bb versus Mf1Gpi-bb samples by PCR, differential expression of 8 genes was confirmed. 4 were not confirmed, 2 were unamplifiable and finally, the sequence for 3 ESTs in the database was of insufficient quality to design primers (see Table 2). These data give a success rate for these arrays of 47% per EST or 40% per gene. However, owing to the mixed background origins of the Sparctm1cam129Sv/Ev/Mf1Gpi-bb mice, we repeated the PCRs using 129Sv/Ev controls instead of Mf1Gpi-bb. We considered that this should significantly reduce the possibility of strain differences contributing to variation between controls and experimental mice, as we compared the experimental mice against BOTH parental strains of origin at the PCR level. We found that Sparc was the only gene that maintained the pattern of variability seen on the arrays.

(4) Sparctm1cam129Sv/Ev/Mf1Gpi-bb versus Mf1Gpi-bb at 9 months of age. The 9 month Sparctm1cam129Sv/Ev/Mf1Gpi-bb versus Mf1Gpi-bb arrays showed significant downregulation of 39 ESTs and significant upregulation of 24 ESTs. 3 ESTs represented Sparc, 7 represented Hba-a1, two represented CtsD, and two Smt3h1. The others appeared only once. This dataset therefore contained 53 genes (Table 3). We tested these against both 129Sv/Ev and Mf1Gpi-bb control strains at the PCR level. We were able to confirm 32 genes as significantly differentially expressed with reference to both control strains. This gives a success rate of 65% per EST or 60% per gene for this array set. Given the multiple reappearance of Hba-a1 in this array set, we re-tested all of the mouse globin genes. Again, these were all, apart from Hbb-Z, seen to be downregulated (Figure 3 and Table 5).

Clustering analysis

We performed clustering analysis of all genes that were confirmed as differentially regulated. Results show that the clusters divide broadly into two groups; those genes that are downregulated at 9 months during the onset of cataract and those that are upregulated at later stages of disease, thus strengthening the conclusions reached as a result of pairwise comparisons of both strains at both time points. A third subgroup comprises Sparc, the globins, and C79876 that were downregulated at both stages. There are a few genes (e.g. Fn1, Tgm2) that may be downregulated at early stages of cataractogenesis and upregulated at the late stage studies. It is worth noting, however, that this was not noted with more stringent filtering when the experiments were analyzed in a pair-wise fashion.

Analysis of success rates for array experiments

We identified 152 ESTs as differentially expressed according to the 4 sets of array experiments we carried out. Using the same mouse strains as were used for the array experiments (i.e., replicating the array conditions for the PCRs) we were able to confirm genes corresponding to 101 of these (66%). This is probably a good estimate of array methodology but is not a reflection of biological meaningfulness. We then removed genetic redundancy, both within and between array datasets, and also those genes not replicably variable when 129Sv/Ev controls were substituted for Mf1Gpi-bb controls used on the 4 and 9 month Sparctm1cam129Sv/Ev/Mf1Gpi-bb versus Mf1Gpi-bb arrays. Using this method of calculation, we have confirmed 51 out of a total of 108 genes tested (47%). Fifty of these genes were not known to be variable prior to these experiments. In addition to genes identified by the arrays, we tested 4 extra globin genes as a result of concerns with regard to error arising from cross-hybridization of closely related sequences. Three of these four were also confirmed. The presence of Sparc probes in this array set provided a valuable positive control (Sparc was found to be variable in all 4 experimental sets).

Reasons for our failure to confirm some genes may have been due to circumstances beyond our control. Twenty-five of 51 unconfirmed ESTs were unamplifiable for various reasons, including withdrawal of one record and noticeably poor single pass EST sequence in a number of cases. Some ESTs were unassigned to Unigene clusters and/or unmatched to genomic sequence. It is therefore likely that poor single pass EST sequence, chimeric clones, cloning of genomic DNA contaminants, or cloning of incompletely spliced nuclear RNA may result in inability to design working PCR primers that correspond to these sequences. Other genes may not be expressed in the lens in sufficient quantities to be detectable via PCR from unamplified cDNA. Their appearance in our array results may be due to amplification bias, cross-hybridization of a related sequence or other unspecified error. In summary, approximately 50% of error was due to the fact that we could not test the gene concerned owing to inability to amplify the sequence. This may be due to factors associated with the NIA arrays we used. A large proportion of the ESTs spotted in the arrays are novel and poorly characterized. A positive aspect of this is that these arrays are good for gene discovery. However, the downside is that clones are more likely to include sequencing or other error. The remaining 50% of error (where sequences were amplifiable but not confirmed) may largely be due to amplification bias. Amplification bias has been shown to contribute a large amount of error to array experiments. As we used unamplified RNA to generate cDNA for RT-PCR, any error introduced by amplification would not be confirmed and would be noted as a larger than normal difference between the number of genes picked up on the array and the number confirmed.


We assessed Sparc knockout mouse lenses on two genetic backgrounds that showed varying severity of phenotype. We have examined each of these at two different ages, 4 and 9 months. Both sets of arrays at 4 months showed that Sparc was the only gene confirmed as significantly downregulated. Given that overt signs of cataract appear later than 4 months in both genetic backgrounds, this result was not entirely surprising. These results imply that absence of Sparc does not have an immediate or direct effect upon the expression of other genes in the lens, which correlates with the fact that the lens, even in the most severely affected Sparc null animals, develops normally [6]. Cascades of gene expression change that lead to opacity may not have been initiated at this stage. Alternatively, some genes may be altered in expression post-transcriptionally or may not be represented on the array set used. It is also possible that gene expression changes noted later are a secondary consequence of initial lens damage (such as abnormal protein modifications) that occur at earlier stages in response to the absence of Sparc.

Both sets of 9 month arrays identified different subsets of genes, expressed at different stages in the process of cataract formation. Genes identified from the Sparctm1cam129Sv/Ev arrays represent genes differentially expressed at an earlier stage in the disease process, as these lenses were mostly presymptomatic and were derived from a late-onset phenotype. In contrast, genes identified as misregulated from the Sparctm1cam129Sv/Ev/Mf1Gpi-bb 9 month array set represent differential expression at an advanced stage of disease, in an earlier onset phenotype. Clustering analysis reinforces the broad grouping of confirmed genes into two broad groupings, those downregulated at an early stage of cataract formation and those upregulated at later stages (as we had previously noted via pairwise analysis, only the hemoglobins, Sparc and C79876 appeared downregulated in both 9 month datasets). The composition of both gene sets derived from our array results sheds further light on the process of cataract progression in Sparc null mice.

Early stage gene expression changes

Previous studies would suggest that absence of Sparc has an initial effect on control of invasiveness in posterior lens epithelial and fiber cells [5,6,13]. Sparc can both positively and negatively regulate cell proliferation and has previously been implicated in malignancy. Expression is increased in malignant tumors [14]. Inhibition of epithelial cell proliferation has been shown to occur via the TGF-beta pathway [29]. The absence of Sparc may allow the posterior lens epithelial and fiber cells a greater degree of cell adhesion and invasiveness than is normal. This may cause these cells to extrude the observed actin containing invadopodia that compromise the integrity of the lens capsule [5]. Downregulation of Osf2 (a putative adhesion molecule) and Marcks (an actin crosslinking regulatory protein) could contribute to loss of normal morphology in lens epithelial and fiber cells and the increasing extrusion of invadopodia.

Studies in skin and bone [30,31] have shown that ECM collagen structure and turnover are also affected by the absence of Sparc. Collagen fibrils are smaller, less abundant, and turned over less frequently. We noted the downregulation of Col3a1 and Col1a2. Downregulation of these genes may contribute to structural destabilization of a weakening lens capsule. These changes may result from loss of transcriptional control over posterior lens epithelial and fiber cells that normally produce the correct amounts of proteins to maintain the capsular ECM. Such changes may arise from downregulation of direct and indirect modulators of gene expression such as those identified here (Cbx5, Phtfr, Sfpq, Kpna2). The downregulation of growth factors (Igf2, H19, Gpc3, Mest, Tmpo, Grb10) and cell cycle/apoptosis genes (Birc5) may result in, or arise from, changes in control of cell invasiveness and/or morphology.

Loss of lens capsule integrity may have knock-on effects on the osmotic balance of posterior lens epithelial and fiber cells. These swell in response to the greater permeability of a weakened lens capsule [6]. Swelling of these cells leads to disorganized cell structure, spreading from the posterior region to other areas of the lens (which in normal mice show lower Sparc expression levels) and are presumably therefore affected less rapidly in knockouts [13]. Progressive structural disruption of the outer layers of the lens may result in reductions (or inappropriate increases) in the transport efficiency of metabolites within the lens.

Changes in lens globin concentrations (and changes in globin folding, as a consequence of EraF downregulation) could contribute to opacification. We have proposed various non-mutually exclusive roles for globins in normal lenses [26]. They may play roles in lens iron transport and/or metabolism, may act as oxygen transporters and/or oxygen sinks, or may act to promote apoptosis like phenomena in differentiating lens fiber cells. Globin downregulation may begin to disrupt the ongoing maturation of migrating lens fiber cells during early cataractogenesis, either via disruption in the oxygen gradient or via a decrease in their ability to promote the apoptosis like process that results in loss of the lens fiber cell nucleus. Abnormal retention of the nucleus by mature lens fiber cells has been implicated in cataract in other animal models [32]. Alternatively, opacity could result from inappropriate posttranslational modification of proteins, such as crystallins, leading from disruption of metabolite transfer processes. If globins normally act as oxygen or iron 'sinks', lens capsule permeability and globin downregulation could result in increased concentrations of oxygen and reactive oxygen species (ROS, or iron) in posterior lens epithelial and fiber cells. This could in turn, result in opacity. Downregulation of Mt2 may also contribute to disruption of oxygen or metal metabolism in the lens.

Late stage gene expression changes

Incremental disruption of normal lens structure and function leads to increasing stress in the lens cells. Cell lysis occurs at later stages of cataract in the Sparc null mouse [6]. The upregulation of stress response proteins, such as heat shock proteins and those genes identified as being involved in degradative pathways (Tgm2, Hspb1, Ugt1A, CtsD, Rnf128, Psap), indicate a detrimental effect on cellular viability. Increasing opacity, cell swelling, changes in gross lens structure, cell death, and progressive weakening of the lens capsule may result in substantial lens capsule porosity, allowing small amounts of lens proteins such as the crystallins (or other cellular debris) to escape the confines of the lens. Upregulation of ion channel related proteins (Ms4a6d, P2rx4) may imply an attempt by the cells to control abnormal osmotic balance or abnormal cell-cell communication. Further perturbations in cell adhesion, migration, and morphology are indicated by the upregulation of Fn1, plectin, and CD9.

Leakage of crystallin proteins outside the lens, or inappropriate release of cell debris into the aqueous or vitreous humors probably results in attraction of cellular components of the immune system. Lens rupture or other causes of 'leakage' from the lens capsule (such as hypermature cataract) can result in an acute immune response and inflammation of the eye, known as lens induced uveitis [33]. Phacolytic uveitis has been used to describe uveitis that may occur as a result of lens protein release through an intact lens capsule. This is usually milder than uveitis induced as a result of full lens capsule rupture [33]. The progressive weakening of the lens capsule in Sparc null mice may allow increasing contact between lens proteins and immune system components or signaling molecules. Involvement of an immune response in some form is shown via upregulation of complement components, Fcgr3, Lcn2, Lysz, Fgls, Serping, and Spp1. Increasing lens capsule permeability, progressive swelling of lens epithelial and fiber cells, cell rupture, and a strengthening immune response could then collaborate in effecting lens capsule rupture, followed by the tissue destruction seen in the last stages of cataract in Sparc knockout mice. Notably, neutrophils, lymphocytes, and macrophages have been shown to be present in the vitreous in Sparc null eyes at an advanced stage of cataract [16].

We have followed the progression of cataract in Sparc null mice over time, via tracking of gene expression changes. These data provide excellent snapshots of pathogenic processes occurring in Sparc null lenses and allow elaboration of previously existing models of cataract progression in these mice. Functions already ascribed to the 'known genes' that are differentially regulated have allowed us to build on previous studies allowing us to form hypotheses as to the mechanics of developing opacity. These data, and associated studies, will allow us to identify and functionally characterize more genes involved in normal and pathogenic lens biology, perhaps leading to the identification of potential therapeutic targets and the development of non-surgically based therapies for cataract.


We would like to thank The Human Genome Mapping Project Resource Centre for provision of the NIA microarray slides used in this study. We would also like to thank Dr. Anna Hurley, Steve Turner, and Vicky Workman for assistance with array protocols, scanning, image analysis, data storage, and bioinformatics. We would also like to thank Prof. Alan Clarke and colleagues for help with use of their photographic equipment. This work was funded by the Wellcome Trust and BBSRC (Evans laboratory) and by the National Eye Research Council (UK) and the Royal Society (UK; Wride laboratory).


1. Wride MA. Cellular and molecular features of lens differentiation: a review of recent advances. Differentiation 1996; 61:77-93.

2. Piatigorsky J. Lens differentiation in vertebrates. A review of cellular and molecular features. Differentiation 1981; 19:134-53.

3. Wride MA. Minireview: apoptosis as seen through a lens. Apoptosis 2000; 5:203-9.

4. Bassnett S. Lens organelle degradation. Exp Eye Res 2002; 74:1-6.

5. Norose K, Lo WK, Clark JI, Sage EH, Howe CC. Lenses of SPARC-null mice exhibit an abnormal cell surface-basement membrane interface. Exp Eye Res 2000; 71:295-307.

6. Yan Q, Clark JI, Wight TN, Sage EH. Alterations in the lens capsule contribute to cataractogenesis in SPARC-null mice. J Cell Sci 2002; 115:2747-56.

7. Francis PJ, Moore AT. Genetics of childhood cataract. Curr Opin Ophthalmol 2004; 15:10-5.

8. Minassian DC, Reidy A, Desai P, Farrow S, Vafidis G, Minassian A. The deficit in cataract surgery in England and Wales and the escalating problem of visual impairment: epidemiological modelling of the population dynamics of cataract. Br J Ophthalmol 2000; 84:4-8.

9. Harding J. Cataract: biochemistry, epidemiology, and pharmacology. 1st ed. London: Chapman and Hall; 1991.

10. Derham BK, Harding JJ. Effects of modifications of alpha-crystallin on its chaperone and other properties. Biochem J 2002; 364:711-7.

11. Mason IJ, Taylor A, Williams JG, Sage H, Hogan BL. Evidence from molecular cloning that SPARC, a major product of mouse embryo parietal endoderm, is related to an endothelial cell 'culture shock' glycoprotein of Mr 43,000. EMBO J 1986; 5:1465-72.

12. Lane TF, Sage EH. The biology of SPARC, a protein that modulates cell-matrix interactions. FASEB J 1994; 8:163-73.

13. Yan Q, Blake D, Clark JI, Sage EH. Expression of the matricellular protein SPARC in murine lens: SPARC is necessary for the structural integrity of the capsular basement membrane. J Histochem Cytochem 2003; 51:503-11.

14. Brekken RA, Sage EH. SPARC, a matricellular protein: at the crossroads of cell-matrix communication. Matrix Biol 2001; 19:816-27.

15. Gilmour DT, Lyon GJ, Carlton MB, Sanes JR, Cunningham JM, Anderson JR, Hogan BL, Evans MJ, Colledge WH. Mice deficient for the secreted glycoprotein SPARC/osteonectin/BM40 develop normally but show severe age-onset cataract formation and disruption of the lens. EMBO J 1998; 17:1860-70.

16. Norose K, Clark JI, Syed NA, Basu A, Heber-Katz E, Sage EH, Howe CC. SPARC deficiency leads to early-onset cataractogenesis. Invest Ophthalmol Vis Sci 1998; 39:2674-80.

17. Bassuk JA, Birkebak T, Rothmier JD, Clark JM, Bradshaw A, Muchowski PJ, Howe CC, Clark JI, Sage EH. Disruption of the Sparc locus in mice alters the differentiation of lenticular epithelial cells and leads to cataract formation. Exp Eye Res 1999; 68:321-31.

18. Delany AM, Amling M, Priemel M, Howe C, Baron R, Canalis E. Osteopenia and decreased bone formation in osteonectin-deficient mice. J Clin Invest 2000; 105:915-23. Erratum in: J Clin Invest 2000; 105:1325.

19. Kantorow M, Huang Q, Yang XJ, Sage EH, Magabo KS, Miller KM, Horwitz J. Increased expression of osteonectin/SPARC mRNA and protein in age-related human cataracts and spatial expression in the normal human lens. Mol Vis 2000; 6:24-9 <>.

20. Kantorow M, Horwitz J, Carper D. Up-regulation of osteonectin/SPARC in age-related cataractous human lens epithelia. Mol Vis 1998; 4:17 <>.

21. Robinson ML, Holmgren A, Dewey MJ. Genetic control of ocular morphogenesis: defective lens development associated with ocular anomalies in C57BL/6 mice. Exp Eye Res 1993; 56:7-16.

22. Sandilands A, Wang X, Hutcheson AM, James J, Prescott AR, Wegener A, Pekny M, Gong X, Quinlan RA. Bfsp2 mutation found in mouse 129 strains causes the loss of CP49 and induces vimentin-dependent changes in the lens fibre cell cytoskeleton. Exp Eye Res 2004; 78:109-23.

23. Alizadeh A, Clark J, Seeberger T, Hess J, Blankenship T, FitzGerald PG. Characterization of a mutation in the lens-specific CP49 in the 129 strain of mouse. Invest Ophthalmol Vis Sci 2004; 45:884-91.

24. Graw J. The genetic and molecular basis of congenital eye defects. Nat Rev Genet 2003; 4:876-88.

25. Wilson AS, Hobbs BG, Speed TP, Rakoczy PE. The microarray: potential applications for ophthalmic research. Mol Vis 2002; 8:259-70 <>.

26. Wride MA, Mansergh FC, Adams S, Everitt R, Minnema SE, Rancourt DE, Evans MJ. Expression profiling and gene discovery in the mouse lens. Mol Vis 2003; 9:360-96 <>.

27. Tanaka TS, Jaradat SA, Lim MK, Kargul GJ, Wang X, Grahovac MJ, Pantano S, Sano Y, Piao Y, Nagaraja R, Doi H, Wood WH 3rd, Becker KG, Ko MS. Genome-wide expression profiling of mid-gestation placenta and embryo using a 15,000 mouse developmental cDNA microarray. Proc Natl Acad Sci U S A 2000; 97:9127-32.

28. Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P, Stoeckert C, Aach J, Ansorge W, Ball CA, Causton HC, Gaasterland T, Glenisson P, Holstege FC, Kim IF, Markowitz V, Matese JC, Parkinson H, Robinson A, Sarkans U, Schulze-Kremer S, Stewart J, Taylor R, Vilo J, Vingron M. Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nat Genet 2001; 29:365-71.

29. Schiemann BJ, Neil JR, Schiemann WP. SPARC inhibits epithelial cell proliferation in part through stimulation of the transforming growth factor-beta-signaling system. Mol Biol Cell 2003; 14:3977-88.

30. Bradshaw AD, Puolakkainen P, Dasgupta J, Davidson JM, Wight TN, Helene Sage E. SPARC-null mice display abnormalities in the dermis characterized by decreased collagen fibril diameter and reduced tensile strength. J Invest Dermatol 2003; 120:949-55.

31. Boskey AL, Moore DJ, Amling M, Canalis E, Delany AM. Infrared analysis of the mineral and matrix in bones of osteonectin-null mice and their wildtype controls. J Bone Miner Res 2003; 18:1005-11.

32. Nishimoto S, Kawane K, Watanabe-Fukunaga R, Fukuyama H, Ohsawa Y, Uchiyama Y, Hashida N, Ohguro N, Tano Y, Morimoto T, Fukuda Y, Nagata S. Nuclear cataract caused by a lack of DNA degradation in the mouse eye lens. Nature 2003; 424:1071-4.

33. Van Der Woerdt A. Lens-induced uveitis. Vet Ophthalmol 2000; 3:227-234.

34. Taneda S, Pippin JW, Sage EH, Hudkins KL, Takeuchi Y, Couser WG, Alpers CE. Amelioration of diabetic nephropathy in SPARC-null mice. J Am Soc Nephrol 2003; 14:968-80.

35. Kihm AJ, Kong Y, Hong W, Russell JE, Rouda S, Adachi K, Simon MC, Blobel GA, Weiss MJ. An abundant erythroid protein that stabilizes free alpha-haemoglobin. Nature 2002; 417:758-63.

36. Hawse JR, Hejtmancik JF, Huang Q, Sheets NL, Hosack DA, Lempicki RA, Horwitz J, Kantorow M. Identification and functional clustering of global gene expression differences between human age-related cataract and clear lenses. Mol Vis 2003; 9:515-37 <>.

37. Babalola OE, Danboyi P, Abiose AA. Hereditary congenital cataracts associated with sickle cell anaemia in a Nigerian family. Trop Doct 2000; 30:12-4.

38. Bloomfield SE, Markenson AL, Miller DR, Peterson CM. Lens opacities in thalassemia. J Pediatr Ophthalmol Strabismus 1978; 15:154-6.

39. Konkel DA, Maizel JV Jr, Leder P. The evolution and sequence comparison of two recently diverged mouse chromosomal beta--globin genes. Cell 1979; 18:865-73.

40. Drewell RA, Arney KL, Arima T, Barton SC, Brenton JD, Surani MA. Novel conserved elements upstream of the H19 gene are transcribed and act as mesodermal enhancers. Development 2002; 129:1205-13.

41. Oppermann B, Zhang W, Magabo K, Kantorow M. Identification and spatial analysis of metallothioneins expressed by the adult human lens. Invest Ophthalmol Vis Sci 2001; 42:188-93.

42. Hawse JR, Cumming JR, Oppermann B, Sheets NL, Reddy VN, Kantorow M. Activation of metallothioneins and alpha-crystallin/sHSPs in human lens epithelial cells by specific metals and the metal content of aging clear human lenses. Invest Ophthalmol Vis Sci 2003; 44:672-9.

43. Spector A. Review: Oxidative stress and disease. J Ocul Pharmacol Ther 2000; 16:193-201.

44. Ihara Y, Suzuki YJ, Kitta K, Jones LR, Ikeda T. Modulation of gene expression in transgenic mouse hearts overexpressing calsequestrin. Cell Calcium 2002; 32:21-9.

45. MacKinney AA. Hematology for Students. 1st ed. New York: Taylor and Francis; 2002.

46. Sano H, Hsu DK, Apgar JR, Yu L, Sharma BB, Kuwabara I, Izui S, Liu FT. Critical role of galectin-3 in phagocytosis by macrophages. J Clin Invest 2003; 112:389-97.

47. Lefrancois S, May T, Knight C, Bourbeau D, Morales CR. The lysosomal transport of prosaposin requires the conditional interaction of its highly conserved d domain with sphingomyelin. J Biol Chem 2002; 277:17188-99.

48. Golembieski WA, Rempel SA. cDNA array analysis of SPARC-modulated changes in glioma gene expression. J Neurooncol 2002; 60:213-26.

49. Hohenadl C, Mann K, Mayer U, Timpl R, Paulsson M, Aeschlimann D. Two adjacent N-terminal glutamines of BM-40 (osteonectin, SPARC) act as amine acceptor sites in transglutaminaseC-catalyzed modification. J Biol Chem 1995; 270:23415-20.

Typographical corrections

Mansergh, Mol Vis 2004; 10:490-511 <>
©2004 Molecular Vision <>
ISSN 1090-0535