Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision 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

Purpose: Sparc/osteonectin is a hydroxyapatite, calcium and, collagen binding protein, implicated in tissue morphogen- esis, 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 files [8]. Development of pharmaceutical treatments for cata- embryogenesis through ageing [1,2]. Moreover, lens matura- ract would therefore be of huge economic and social benefit. tion and growth proceed throughout life. Lens fiber cells dif- Much progress has been made recently in uncovering the ferentiate from lens epithelial cells at the equator of the lens, molecular causes of cataract. This area has benefited greatly migrate inwards at the bow region and mature around pre- from advances in genetics, genomics and proteomics in re- existing fiber cells at the center of the lens and lens fiber cells cent years. The discovery of predisposing human mutations, lose their nuclei via an apoptosis-like mechanism as they ma- and phenotypes induced in genetically engineered mouse ture [3,4]. The lens is surrounded and protected by a thick models show that genetic factors are involved in predisposi- acellular basement membrane, known as the lens capsule, tion to cataract. Posttranslational modifications of protein con- which is composed of collagen, laminin, fibronectin, nidogen, stituents of the lens core during ageing also undoubtedly con- heparan sulphate proteoglycan, and other components that may tribute to the etiology of nuclear cataract in particular [9,10]. be important for interactions between the extracellular matrix Sparc (secreted protein acidic and rich in cysteine, also (ECM) of the capsule and lens epithelial cells [5,6]. Defects known as osteonectin) is a 32 kDa secreted glycoprotein that in lens function and/or maturation result in clouding of the belongs to the matricellular group of proteins [11-13]. Sparc lens, or cataract. Age related cataract is the most common form is known to bind calcium, hydroxyapatite, vitronectin, and of visual impairment worldwide, while congenital cataract is collagen and to regulate extracellular matrix production, cell the most common treatable cause of childhood blindness [7]. adhesion, proliferation, and migration [12,13]. Sparc is ex- Cataract is currently treatable in humans only via surgical in- pressed widely during embryogenesis, but in adult mice ex- tervention. Such treatment is beyond the means of many in pression is more restricted. It is highest in remodelling tissue the developing world, while in the UK there is a tremendous and is also noted in tumorigenesis. Sparc is also known to backlog of cataract patients awaiting surgery, a problem un- regulate the activity of various growth factors, including fi- likely to dissipate considering ageing Western population pro- broblast growth factor (FGF-2), vascular endothelial growth factor (VEGF), and platelet derived growth factor (PDGF) [14]. Correspondence to: Michael A. Wride, School of Optometry and Vi- Sparc has been disrupted via targeted mutagenesis on two sion Sciences, Cardiff University, Room 1.80A, Redwood Building, separate occasions, with broadly similar results. Both lines of King Edward VII Avenue, Cathays Park, Cardiff CF10 3NB, Wales, mice develop cataract and osteopenia as they age [15-18]. UK; Phone: +44 (0)29 2087 0054; FAX: +44 (0)29 2087 4859; email: [email protected] Sparc is expressed in embryonic lenses from embryonic day 490 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

12 (E12), but is restricted in expression in adult lenses. It is is largely reflected in human lenses. Sparc expression is noted found in large quantities in lens epithelial and newly differen- generally in the lens epithelium, but is highest in peripheral tiating fiber cells, but not in the lens capsule [13]. This pattern 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 subcap- sular cataract [19,20]. Sparctm1cam knockouts (those studied in this article) are mu- tated 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 replace- ment of exon 6 with a splice accepting marker and a neomy- cin resistance cassette [15]. This mutation results in prema- ture termination of transcription and translation, deleting half of the follistatin module and resulting in the removal of down- stream domains required for correct folding and secretion of Sparc [15]. The mutation has been shown to result in reduc- tions 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 cata- ract 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 gen- erated 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]. Het- erozygote 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 Figure 1. Darkfield photography of whole lenses from control and [15-17]. Wild type C57Bl/6J mice develop spontaneous len- Sparctm1cam mice. Darkfield photography of whole lenses from con- ticular anomalies with a frequency of 5-15% [21], while strains trol and Sparctm1cam mice illustrating the extent of cataract observed of 129 origin are natural knockouts of CP49, a lens specific at varying ages in the different strains used in the microarray analy- member of the intermediate filament superfamily [22,23]. This ses presented herein. Strains and ages are presented in abbreviated mutation causes increased light scatter, but not cataract, in form in the figure. Full details are as follows: A: 129Sv/Ev (129) at 9 lenses of 129 strain. These observations underscore the im- months (m); B: Mf1Gpi-bb (Mf1) at 9 months; C: Sparctm1cam129Sv/Ev portance of assessing knockout phenotypes on different ge- tm1cam tm1cam at 4 months; D: Sparc 129Sv/Ev at 9 months; E: Sparc 129Sv/ netic backgrounds. Ev at 11 months; F: Sparctm1cam129Sv/Ev at 16 months; G: The study of mouse models of disease with defined ge- Sparctm1cam129Sv/Ev at 22 months; H: Sparctm1cam129Sv/Ev/Mf1Gpi-bb at 16 months. Lenses B and D are dissected in the same manner as netic defects has led to great breakthroughs in the understand- lenses used in array experiments. Some iris material was left on the ing of the mechanisms of human disease [24]. Study of changes others presented here in order to facilitate orientation of the lens for in transcription as a result of gene ablation gives an opportu- photography. nity to uncover the molecular cascade that leads to the onset 491 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision of symptoms. The invention of microarray technology allows Ev/Mf1Gpi-bb arrays, while both Mf1Gpi-bb and 129Sv/Ev con- the investigation of transcriptional changes on a global scale. trols were used for PCR confirmations of these arrays. Microarrays have huge potential in many fields, including Animals, tissue extraction, and photography: Mice were ophthalmic research [25]. Following array studies looking at maintained and sacrificed under Home Office license in ac- normal mouse lenses during maturation [26], we decided to cordance with British law (comparable with U.S. Public Health use microarrays to compare lenticular gene expression in ab- Service Policy on Humane Care and Use of Laboratory Ani- normal lenses with controls. We have used the NIA 15K array mals). Whole eyes were removed and lenses were dissected set [27] to investigate changes in gene expression in the lenses via posterior incision using number 5 Swiss watchmaker’s of Sparc knockout mice on two genetic backgrounds with dif- forceps in phosphate buffered saline under a Leica MZ125 ferent phenotypic severity at various stages during the pro- dissecting microscope (Leica Microsystems Ltd., Milton cess of cataract formation. We have confirmed the differential Keynes, UK). Lenses used in photography were dissected leav- expression of 55 genes at various stages of cataractogenesis ing a little iris tissue in order to facilitate lens orientation (use- using semi-quantitative RT-PCR. Most notably, 5 mouse globin ful at advanced stages of cataract where the lens was very genes and an uncharacterized EST, C79876, were significantly misshapen and fragile). Dark field photographs were taken downregulated in 9-month old Sparc null mice from both ge- using an Olympus CAMEDIA C4040 zoom digital camera netic backgrounds, at different stages of disease. Overall, the mounted on an Olympus SZX12 dissecting microscope studies described here provide a profile of gene expression, (Olympus UK Ltd., London, UK), then downloaded and which both supports and extends previous cellular and mo- trimmed using Adobe Photoshop 6 (see Figure 1). Lenses were lecular observations of age-onset cataract progression in Sparc assessed via post dissection photography as opposed to slit knockout mice. lamp procedures as we have preliminary evidence that the corneas are abnormal in these mice (unpublished data). METHODS Preparation of target RNA: No differences were noted Mouse nomenclature and strain choice: Sparctm1cam refers to between the sexes in terms of speed of cataract progression. Sparc, targeted mutation 1. University of Cambridge 129Sv/ Nevertheless, in order to minimize biological variability, we Ev refers to mice of inbred strain 129 and substrain Sv/Ev used male mice for all array and RT PCR experiments. Lenses Mf1Gpi-bb refers to the outbred mouse strain MF1. The strain as used in arrays were dissected free of any contaminating tissue obtained from the breeders has a polymorphism in GPI. We and homogenized in TRIzol reagent (Invitrogen Corp., Pais- have selected and maintained homozygosity for the variant ley, UK), using a hand held homogenizer (Wheaton Science 1b (hence Gpi-bb). This allows chimera analysis of any tis- Products, Millville, NJ). RNA extraction was carried out ac- sue. cording to the manufacturer’s protocol. Concentration and We chose the Sparctm1cam knockouts for investigation. RNA quality were assessed via spectrophotometry and form- These mice have been maintained on both a mixed background aldehyde gel electrophoresis. Typically we obtained 30 µl of (Sparctm1cam129Sv/Ev/Mf1Gpi-bb) and an inbred background RNA per pair of lenses at an average concentration of 50 ng/ (Sparctm1cam129Sv/Ev). These strains have been maintained by µl. One round of RNA amplification was then used per 500 ng the Evans laboratory (along with parent substrains 129Sv/Ev lens sample (Arcturus RNA amplification kit, Arcturus Bio- and Mf1Gpi-bb) since the generation of the initial targeted muta- science Inc., Mt. View, CA), using the manufacturer’s proto- tion. The availability of the exact control substrain upon which col. Amplified mRNA (1 µg) was then labeled with Cy3 or the initial mutation was generated was extremely important Cy5 using the CyScribe labeling system (Amersham, as strain differences between controls and experimental mice Buckinghamshire, UK) according to the manufacturer’s pro- can otherwise result in spurious expression differences when tocol. Labeled probe (1 µl) was added to 2 µl glycerol, then studied via array. Notably, the ES lines used to generate this run on a “John gel” (a microscope slide sized thin agarose mutation were derived (in this laboratory) from 129Sv/Ev mice, gel). This was scanned using a GeneTac LS IV scanner (Ge- meaning that our 129Sv/Ev control mice also matched the nomic Solutions, Cambridgeshire, UK) in order to assess probe genetic background of the mutated ES cells that contribute to integrity and successful incorporation of label. Successfully the genetic makeup of the mutated strain. This should avoid labeled control and experimental targets were then combined error arising from linkage disequilibrium (e.g. differences in and prepared for hybridization. expression between experimentals and controls that arise be- Arrays: Array slides were obtained from the Human Ge- cause of promoter differences in genes closely linked to the nome Mapping Project (HGMP; NIA15000 set, printed on two introduced mutation in Sparc). If the ES cells are genetically slides; Hinxton, Cambridge, UK). This array set contains ap- different from mouse strains into which the mutation is bred, proximately 15,000 ESTs obtained from whole embryos and not only the mutation, but also any closely linked polymor- various embryonic tissues. Up to 50% of these ESTs were phisms, are fixed into the resulting mutant strain and can be- derived from novel genes. We should note at this point that come the source of significant differences between control and the ESTs printed on these arrays do not necessarily represent experimental animals. Controls for Sparctm1cam129Sv/Ev mice 15,000 genes, we have found evidence of significant redun- were age, strain, and sex matched. Controls for dancy. For example (see Results), we found 15 different ESTs Sparctm1cam129Sv/Ev/Mf1Gpi-bb mice were age and sex matched. downregulated, all of which represented the same gene, Hba- Mf1Gpi-bb mice were used as controls for the Sparctm1cam129Sv/ a1. We have therefore used the term EST throughout this text 492 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision to refer to the EST clones printed on the array. Multiple ESTs and experimental channels were removed. Genes that failed on the array have, in some cases, been shown to represent the to show a fold change of 2 were also removed. Remaining same gene. data (above background plus 2 standard deviation values, fold Array hybridization and scanning: Array slides were incu- change of >2) were retained, combined with similar data from bated in prehybridization solution for 1 h at 42 °C (50% other replicates and sorted via accession number. Genes that formamide, 5X SSC, 0.1% SDS, and 1% BSA). Targets were appeared above background plus 2 standard deviation values dried via vacuum centrifugation then resuspended in 50 µl with a fold change >2, in at least 8 out of 10 replicates (this hybridization solution (49.9% deionized formamide, 49.9% analysis combined values from duplicate spots, giving 10 rep- 20X SSC, and 0.2% SDS) with added 1 µl Cot1 DNA and 1 µl licates per EST), were retained for further analysis. Finally, poly A oligonucleotide as blocking agents, heated to 95 °C for accession numbers obtained from this method and from SAM 5 min and then added to the face of one slide. The printed face were compared. Genes appearing as differentially regulated of the second slide of the pair was then placed face to face using BOTH methods were deemed significant. In other words, with the first, using the same probe. Slide pairs were then significant ESTs are changed in expression by at least two placed on a level plastic cover above some 1X SSC moist- fold, are above background plus 2 standard deviations in at ened tissue in a slide box. The slide box was sealed with least one channel and have a delta value of 0.5 using SAM. Nescofilm (Karlan Research Products Corporation, Santa These ESTs were subjected to bioinformatic analysis and were Rosa, CA), placed floating in a water bath and hybridized for all assessed via RT-PCR in order to confirm array results. 24-48 h at 42 °C. Following hybridization, slides were washed Arrays and MIAME standards: Data presented here are once in wash solution 1 (1X SSC, 2% SDS in filtered auto- in full compliance with MIAME standards [28]. Images, raw claved distilled demonized H2O [ddH2O]) for 20 min, then data, processed data, and a “ReadMe” file are available from twice in wash solution 2 (0.1X SSC, 0.2% SDS in ddH2O) for the authors. 20 min each. Slides were dipped in nuclease free filtered wa- Bioinformatic analysis of ESTs: ESTs that were signifi- ter, then spray dried. Finally, the backs of the slides were cantly differentially regulated as determined above were com- cleaned with ddH2O, wiped with 100% ethanol, then wiped piled. Accession numbers were then used to comprehensively dry and scanned at 12.5 µm using a GeneTac LSIV scanner, search the NCBI databases. Some data compiled from various (Genomic Solutions). Arrays were repeated 5 times with fluor NCBI databases (UniGene, Homologene, OMIM, LocusLink, switching, in order to counteract any issues of dye bias that PubMed) are shown in Table 1, Table 2, Table 3, Table 4, and may have arisen from direct labeling. Repetition sets used bio- Table 5. These data are also available at the Cardiff Array logical samples derived from at least four control and four Database. experimental animals per set in order to control for biological RT-PCR array confirmations: We tested all genes aris- variation between individual animals. Scanned images were ing from the four experiments using semi-quantitative RT-PCR. stored and filtered, then analysed using the GeneTac Analyser This method was chosen as we had 108 genes to test. We were spot finding software (Genomic Solutions). able to do this in a more cost-effective and high-throughput Data analysis: Output files from the GeneTac analyser manner than via the use of quantitative PCR methods. Semi- were saved in MS Excel spreadsheet format. MS Excel was quantitative RT-PCR was carried out as follows. RNAs were used for all further data manipulation. We normalized each quantified as described above. Even quantities of control and channel via total array methods by calculation of the mean experimental RNA (usually 2-4 µg) were treated with DNAfree intensity value. Normalized intensities were then analyzed (Ambion Inc., Cambridgeshire, UK) according to the using two different methods. First, we transferred normalized manufacturer’s protocol. Reverse transcription (RT) reactions intensity values for each of 5 experimental repetitions into a were carried out using the Superscript II First Strand Synthe- separate Excel file, ensuring a standard order of samples. These sis System for RT PCR (Invitrogen), according to the data were then formatted for Significance Analysis of manufacturer’s instructions, except that due to the low con- Microarrays (SAM). SAM was carried out and genes that centration of RNA per sample, we tripled the amount of RNA showed a fold change of 2 or above and that appeared in a used and therefore the reaction volumes per sample. SAM analysis as statistically significant above a delta value Unamplified RNA was always used for RT-PCR confirma- of 0.5 (which denotes an error rate of 5%) were selected for tions in order to correct for bias introduced to the array results further appraisal. SAM, however, has some disadvantages. All via the amplification process. A “no RT” control correspond- replicates have to be in the same order to use SAM, preclud- ing to each sample was also produced. These were treated in ing some methods of data filtration. In addition, each cDNA exactly the same way as the samples except that Superscript probe (represented by a single accession number) is spotted II reverse transcriptase (Invitrogen) was not added. on the HGMP NIA array slides twice. Using SAM, it is im- Primer design: Standard primers were used for house- possible to assess both duplicate spots together. Hence, we keeping genes (β-actin, GAPDH, and βA1-crystallin, see Table supplemented SAM analysis with a second analysis method 1, Table 2, and Table 3 for primer sequences). We used 3 house- as follows. Following normalization, we used approximately keeping genes in conjunction in order to ensure that biased 700 blank spots per slide to calculate a mean background value expression of one would not adversely affect results. More- plus 2 standard deviations of that background value for each over, GAPDH and β-actin were present on the NIA arrays used channel. Genes that fell below this cutoff in BOTH control and were not significantly altered in expression. Expression 493 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision of crystallins has been shown to be unperturbed in Sparc null firmed in any one of our experimental sets. We merged data mouse lenses [6]. Primers for ESTs were designed from the from all SAM files for all data points. We then extracted the EST sequences. However, EST sequences are usually derived normalized data profiles for all repetitions for the accession from high throughput single pass sequences. In order to check numbers of interest. Mean values were obtained for repeti- the sequence quality of the ESTs and avoid areas of poor quality tions of the same experiment, from duplicate spots and, where sequence when designing primers, accession numbers of NIA applicable, for different ESTs representing the same gene. We ESTs were checked against the NCBI databases. EST se- then calculated mean fold changes per gene for each of the quences that were assigned to Unigene clusters were checked four experiments we carried out. These results were then used via bl2seq (BLAST 2 sequences) against a consensus mRNA/ in clustering analysis (see Figure 4 for fold changes and clus- cDNA sequence corresponding to that gene, if one was given ters). Datafiles were imported into GeneSight 3.5 for that cluster. ESTs without a corresponding full length gene (BioDiscovery, El Segundo, CA). Hierarchical clustering was sequence were checked via mouse genome BLAST or performed under default conditions to give an idea of the ab- BLASTnr. Primers were designed with reference to BLAST solute numbers of credible clusters. Clustering analysis was results. Where possible, areas of mismatch arising from prob- then repeated using 2D self organizing maps and the results able EST sequence error (such as those within long runs of are summarized in Figure 4. the same base or those at the beginning or end of the EST sequence) were avoided. Primers were designed using Primer3 RESULTS (version 0.1 beta 1a) from FASTA EST sequence (with areas Progression of cataract in different strain backgrounds: of unreliable sequence excluded). Primers for the mouse he- Progression of cataract in Sparctm1cam129Sv/Ev mice is shown moglobin genes were designed by hand. Due to the close simi- in Figure 1 via darkfield photography of whole lenses. Pho- larity between different beta isoforms (Hbb-b1, Hbb-b2, Hbb- tography of whole dissected lenses was used in preference to Y, Hbb-Z) and different alpha isoforms (Hba-a1, Hba-X), prim- sections here as this also demonstrates how we assessed lenses ers specific to each isoform were required. Therefore, we gen- prior to RNA extraction for arraying. Sparctm1cam129Sv/Ev (4 erated separate alignments of the alpha and beta globins using month, Figure 1C) and Sparctm1cam129Sv/Ev/Mf1Gpi-bb lenses DNASIS version 2.5 and designed primers in such a way that (4 month, not shown) were indistinguishable from 129Sv/Ev each primer was 3' mismatched for every isoform except the and Mf1Gpi-bb controls at 4 months (not shown) and 9 months one they were intended to amplify. Primer sequences for all (Figure 1A and Figure 1B). Sparctm1cam129Sv/Ev lenses at 9 ESTs are shown in Table 1, Table 2, and Table 3. months showed no significant degree of opacity, (Figure 1D). PCR: PCRs were initially carried out at the optimum Tm Obvious signs of developing cataract become visible under indicated for the primer pair designed. No RT controls were the dissecting microscope at approximately 11 months in always included. Conditions were varied later if necessary in Sparctm1cam129Sv/Ev lenses, particularly in the posterior re- order that only single bands appeared and the number of cycles gion (Figure 1E) and became steadily worse over time (e.g. for minimum visibility on a gel were always identified. We 16 months, Figure 1F), with lens capsule rupture occurring ensured that PCR bands obtained for housekeeping controls after 18 months (Figure 1G). Sparctm1cam129Sv/Ev/Mf1Gpi-bb were even at minimum visibility cycles before testing other cataractous lenses at 9 months used for arrays were similar to genes. In order to minimize false positive results, we used at the Sparctm1cam129Sv/Ev lens at 16 months, shown in Figure least two sets of Sparc null and 2 sets of control cDNAs from 1F. Sparctm1cam129Sv/Ev/Mf1Gpi-bb mice that had cataract at 9 different animals. Control cDNAs were both 129Sv/Ev when months tended to progress to lens capsule rupture at about 12- carrying out confirmations from the Sparctm1cam129Sv/Ev ar- 16 months of age as presented here (Figure 1H) and as previ- ray experiments. Sparctm1cam129Sv/Ev/Mf1Gpi-bb experiments ously noted [15]. Other Sparctm1cam129Sv/Ev/Mf1Gpi-bb mice were confirmed using one 129Sv/Ev control and one Mf1Gpi-bb showed a pattern of cataract progression more similar to that control. This was done in order to ensure that the controls shown by Sparctm1cam129Sv/Ev mice. represented gene expression patterns from both parental strains. Array results: Four different array comparisons were car- Variation was observed in both sets in order for a gene to be ried out. These were: confirmed as variably expressed. PCRs were carried out in 20 (1) Sparctm1cam129Sv/Ev versus 129Sv/Ev at 4 months of µl volumes using 0.025 µmol concentrations of each dNTP age. Sparctm1cam129Sv/Ev lenses were arrayed against those (100 mM dNTP set, Invitrogen) and 1 unit Taq DNA poly- from 129Sv/Ev mice of the same age and sex at 4 months and merase (Sigma) per reaction. PCRs were usually carried out 9 months. Four months of age was comfortably pre-symp- using PCR buffer (Sigma) containing 15 mM MgCl2, how- tomatic, while the 9 month time point was chosen to show ever, we varied magnesium concentrations where necessary gene expression patterns just prior to the onset of obvious mor- in order to optimize results. Annealing temperature and the phological changes associated with cataract in Sparc knock- number of cycles were varied where necessary, up to a maxi- out mice. Analysis of array results at 4 months showed sig- mum of 50 cycles, and finally some primers were also rede- nificant downregulation of only 2 ESTs. Bioinformatic inves- signed. Final PCR conditions are indicated for each gene in tigation revealed that both of these ESTs represented the Sparc Figure 2 and Figure 3. gene. RT PCR, not unsurprisingly, showed that Sparc was sig- Clustering analysis: We generated clusters from gene nificantly downregulated in Sparctm1cam129Sv/Ev knockout expression profiles of all ESTs corresponding to genes con- mice in comparison to controls (our knockout mice had previ- 494 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

ously been shown to express small quantities of RNA while TABLE 1. SIGNIFICANT GENES: NINE MONTH SPARCTM1CAM129SV/EV Sparc protein was entirely absent [15]). LENSES VERSUS 129SV/EV CONTROLS (2) Sparctm1cam129Sv/Ev versus 129Sv/Ev at 9 months of tm1cam Accession Primer age. The 9 month Sparc 129Sv/Ev versus 129Sv/Ev array number Name Foward primer Reverse primer ------set showed a significant downregulation of 70 ESTs (listed in BG079989 Alb1 CTACACCCAGAAAGCACCTCA CTCAGCAAGAGCCGTTTGTTT AW546550 CAGGGACAAAGGAAAGGCATAG CTGCTTGCTGGTGATAGTTGG Table 1). Bioinformatic analysis showed that the 70 accession AW546837 AACATACTTCAACTGTAGGCAAAATAG TGGAGTTTCTGGAGGATAATAAGG AW549180 GCTCGCTGGTAGCTCTTTGG CTGGTGCCTTGTCTGCTTTG numbers for these ESTs represented 44 genes, showing that AW550210 CTGTATTGATTGAGCCACCAACT CGTCTGAGGGAAACTGGGTG AW550998 GTGGACTACAGCCTCAAACTGG GGAGACAGAGCAGGGATGGTT there was significant redundancy within the NIA 15K set. Two BG065042 GCCCTGAAGTAAGAACCAGATGTC CGCCACCAGCACCCAAAG BG072975 GCACAAGAGGGAAATAAAGGTG GTGAACCCACATCCACGAGTT ESTs each represented Sparc, Igf2, H2afz, and Sfpq, while BG073080 TAGCTCTTGCCCTGTGGTTC CCATACCATTTGCTGCTTTTCTC BG075008 TTCATGGCAGCTGGTCTGC GCATGGCTTGGTGTATCTG H19 was represented by 6 ESTs. The most significant redun- BG076590 TAGAGCAAGTTCGGATTCAGGTT GAGCGGGCGGATAAACACT BI076713 TTGACTGATTGTGTGGGCTTG CGTGAGGAACTGAGGGTTGAG dancies were 15 ESTs corresponding to Hba-a1, the major BG077309 Birc5 TGAATCCTGCGTTTGAGTCGT GTTCCCAGCCTTCCAATTCCT C79876 C79876 AGAGATGGGTCCTCACAAGCA GGTCAAAGGCTGGCACTGG mouse alpha globin and 4 ESTs representing Hbb-b2, one of C80273 C80273 TTTATGGCAGGGTTAGGCTGGG CACACACCGAGGCTACTCCA C75991 Cbx5 GGTTGGTTGCTTTGGGAGATT TGTTATGGGAGGGTATGTGAGG the major mouse beta globins. We considered the high level of BG073735 Col1a2 TGCTGAATCTAATGAAGAGAGAACC TGACCAAGGCTGACACGAAC BG065049 Col2a1 GTTCAGCCCCTCCAATGTCC CTGTAAGAACAGCATCGCCTACC representation of these genes (19 ESTs out of a total of 70) BG074327 Col3a1 AAGGCTGAAGGAAACAGCAAA TTGGTCACTTGCACTGGTTGA BG085518 EraF TGATCTCCACAGGGATAAAGGAG AGGAGGGCAGTGTATTGCTTG highly significant. BG087506 Glud ATCCAATGCACCCAGAGTCAA TTCCACCATGCTTTCCAAACT BG071068 Gnb1 GTGGCAAAGAGGCAGCAGAG CACACATATCGGAAGCAGCAAG Given high sequence similarities within the beta globin BG078478 Gpc3 CAGGTCCGTTCTTTCTTCCAG TTGAACATGGCGTTGGTGTAG BG065213 Grb10 TGGTTACACGGAGTCGTCATTTAG CCTTGCCGAAGTGAATTGTTG and alpha globin gene clusters we considered that cross-hy- BG076718, H19 GTATGCCCTAACCGCTCAGTC GCTCTTCGAGACACCGATCAC BG086671, bridization on the array between different mouse globin spe- BG087028, BG087529, cies was likely. For example, the two major mouse beta globins, BG087637, BG087638 Hbb-b1 and Hbb-b2 show 13 coding sequence mismatches at BG065110, H2afz TACCATTGCTGGTGGTGGTGT TTTGGTTGGTTGGAAGGCTAAT BG076995 the DNA level. The embryonic globins, Hbb-Y, Hbb-Z, and BG079788, Hba-a1 CCACCACCAAGACCTACTTTCC AGGCAAGGAATTTGTCCAGAGA AW550167, Hbb-Z are less similar than this but still share a significant AW551388, BG072574, level of DNA homology to either the major alpha or major BG072760, BG073467, beta globins. We therefore designed primers specific to each BG073468, BG073469, of Hba-a1, Hba-x, Hbb-b1, Hbb-b2, Hbb-Y, and Hbb-Z (prim- BG073608, BG073617, ers contained 3' terminal mismatches to each of the other globin BG074904, BG085818, DNA sequences). Like humans, mice also have two major adult BG086202, BG086243, alpha globin genes, giving them a total of 7 globin genes per BG086422 Hba-X GCGGTTAAGAGCATCGACAAC TTCTCAGTCAGGATAGAAGACAGGA haploid genome. The coding sequence of these two genes is Hbb-b1 ACCTGACTGATGCTGAGAAGG CCCTTGAGGCTGTCCAAGTG AW548291, Hbb-b2 CACCTGACTGATGCTGAGAAGT CCCTTGAGGTTGTCCAGGTTT completely identical at the DNA (and therefore RNA) level, AW548342, BG072990, moreover, their expression is co-ordinated. We could not dis- BG085433 Hbb-Y TTACTGCTGAGGAAAAGACCCTC AAGTGACTAGCCAAAACAATCA tinguish between these by PCR. Therefore, we refer to both as HbbZ TACCTCACACCATGGTTCAC GTACTTGTGGGACAGAGCAT BG085427 Hmgb2 GCGAGGAGCACAAGAAGAAG TCAGAGCAAAACAGGAAGAAGG Hba-a1. BG064815, Igf2 TGGAACATTGGACAGAAACTCA GACAGTAGGGAAGAGACAAGATGGA BG073613 BG066442 Kpna2 CCCTTTCTTGTTGGTGTCCTC ATCCAAGCCTCCACACTCTTC BG076937 Lmnb1 GACATGGAGATCAGCGCCTAC TCAAGCGAATAAACTTCCCATC TM1CAM BG085378 Marcks TCAGATAGTTGTTTCCACCATTCC TTCCTTTTGACCCACCCATC TABLE 2. SIGNIFICANT GENES: FOUR MONTH SPARC 129SV/EV/ BG087241 est CTCAAAGACGGAGGTGTGCTT TGTGGGTAGTGGCTAATGTGG GPI-BB GPI-BB BG074746 Mscp TGGAGGGTGATGGTTGTGAG CTCGGTCATTGGCATCTTTG MF1 LENSES VERSUS MF1 CONTROLS BG063925 Mt2 CGTGGGCTGTGCGAAGTG GTGGAGAACGAGTCAGGGTTG AW553287 Osf2 GTTTTCTGAATTACCAACTCTG CACAGTTACCTTTCCAGGGG Accession Primer BG086397 P4hb AGCTCTGGCTCCCGAGTATG GATGACCGTCACTTCGCTTG number name Foward primer Reverse primer BG070102 Phtf GCCGAAGCTCCATTCGTTAG AAGGAGCTGGACACACCATTT ------BG074984, Sfpq ATGGGTGGTGGTGGAACAA CCCTTCATACTCTTCTCTCCCTCT BG064385 TTCAATCCTTCATCATCCAACG AAAATGTGTTTCAAGGTGTGCAG BG086821 BG067609 TGGGTCACCTTGCTGAATTG GCGATGTGTAATGCCTGCAC BG064802, Sparc GCTGTGTTGGAAACGGAGTTG CTTGCCATGTGGGTTCTGACT BG072874 BG069035 GAACAGGAATTTGGGGAGTAGG GTGGGAAGGGCTCTCACAAG AW548440 Syn1 CACAGTCCACAAAGGCTTTTTATT CCTCGGCTGGCATTATGGAG BG069578 CyclinG1 TCTCAAATGGCAAGGTGTCG TTGCTGTGAAAGCGACTGAAG BG086002 Tia1 CGTTTCGTGAACTATACCATCTCC GGAATCCTTGAGGCACCTTCT BG076460 Gclc CGATCATCTTCTGGCACAGC ACCTCCATTGGTCGGAACTC BG086372 Tmpo CTTCACAGCATTTTCGTATAGATGG TGGTTTCTTTGGTGTTCTTCTGG BG077218 Glo1 GTTTGCTACTAGGGCAGGTT CAGGGACTTCTTAGGATCCT BG077349 TCCGCCACTTCAACAACAAC TTGCTCCTTTTGGGGCTTAC Housekeeping genes BG077685 KRT2-8 GCACTCAGGAGAAGGAGCAG GCGGAGGTTGTTGATGTAGC BG078111 Mttp TGTGGACGTTGTGTTACTGTGG ACTTTTGCCCTGGAAGATGC GAPDH Gapdh ACCACAGTCCATGCCATCAC TCCACCACCCTGTTGCTGTA AW552998 TGCAAGAACAATGGGCAAAC GAGAGACGCTCACACGAATCC β-Actin β-actin CGTGGGCCGCCCTAGGCACCA TTGGCCTTAGGGTTCAGGGGG BG086572 ACCTAGACCGGCACCTGAAG CCTGGTTCATCTCGCTGTTG Crystallin CryβA1 GCTCCTGCCCAAATGTCTCTG GATGGGTCGGAAGGACATGAG βA1 BG064802, Sparc GCTGTGTTGGAAACGGAGTTG CTTGCCATGTGGGTTCTGACT BG078305, BG072874 Significant genes included in this table are those giving a signifi- AW537378, ESTs Poor quality sequence cance level of delta=0.5 using SAM and that have a fold change of BG064861, more than two in 8/10 repetitions. All significant genes are listed AW555102 here, including those that were not confirmed by PCR. The columns Significant genes included in this table are those giving a signifi- list the GenBank accession number and primer name, and the for- cance level of delta=0.5 using SAM and that have a fold change of ward and reverse primer sequences. The primer name is not listed more than two in 8/10 repetitions. All significant genes are listed where the EST is uncharacterized, where ESTs correspond to known here, including those that were not confirmed by PCR. The columns genes, the gene name is used. Primer sequences for housekeeping list the GenBank accession number and primer name, and the for- genes are listed at the bottom; these are standard house keeping gene ward and reverse primer sequences. The primer name is not listed primers used by us before and are not directly associated with any where the EST is uncharacterized, where ESTs correspond to known ESTs on the array, therefore no accession numbers are given. genes, the gene name is used. 495 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

RT-PCR confirmed the downregulation of 22 out of 44 whole eye cDNA, indicating that this globin is simply not genes, representing 47 out of 70 ESTs (see Figure 2 and Table expressed at detectable levels in adult lens (data not shown). 4). The success rate of these arrays was therefore variously (3) Sparctm1cam129Sv/Ev/Mf1Gpi-bb versus Mf1Gpi-bb at 4 67% (per EST/accession number) or 50% (per gene). In addi- months of age. Sparctm1cam129Sv/Ev/Mf1Gpi-bb lenses were ar- tion to the 22 genes detected via array and then confirmed, we rayed against those from Mf1Gpi-bb mice of the same age and also confirmed the downregulation of 3 out of 4 remaining sex at 4 months and 9 months of age. Four month lenses used mouse globins (not identified via array but checked because for arrays resembled control lenses in clarity, while 9 month of high sequence homology). Hbb-y, Hbb-b1, and Hba-X were lenses were at an advanced stage of cataract (similar to Figure confirmed. Hbb-Z was not amplifiable from our lens cDNAs. 1F). The 4 month Sparctm1cam129Sv/Ev/Mf1Gpi-bb versus Mf1Gpi- Our Hbb-Z primers do amplify Hbb-Z from E14.5 mouse bb showed significant downregulation of 14 ESTs and signifi- cant upregulation of 3 ESTs. Three downregulated ESTs rep- TM1CAM TABLE 3. SIGNIFICANT GENES: NINE MONTH SPARC 129SV/EV/ resented Sparc, the others were unique. When testing GPI-BB GPI-BB MF1 LENSES VERSUS MF1 CONTROLS Sparctm1cam129Sv/Ev/Mf1Gpi-bb versus Mf1Gpi-bb samples by PCR, Accession Primer number name Foward primer Reverse primer differential expression of 8 genes was confirmed. 4 were not ------BG075211 Adfp GAGGGGTTTACTGAGCTTTGACC GATGGCAGGCGACATCTACTC confirmed, 2 were unamplifiable and finally, the sequence for BG086439 Anxa2 CCTGGAGACGGTGATTTTGG CTCTGCTCGTCTGCCCTTTG AU043642 TGTTCTGTCCTCCGTAATCACTC TTTGTAATTGAGTCAGCGCATC 3 ESTs in the database was of insufficient quality to design BG066868 TTGCCCCTAAACTCAGAGAACG GCATTGATATTCATAAGGGAATTTGG BG066913 GAGAACTTCAGACCAATTTCCC GGGTGTTGGATTTTGTCAAATGC primers (see Table 2). These data give a success rate for these BG067621 GCTCCCAAGATCCAACTACGAG AACGGCTACCACATCCAAGG BG069287 TGAGCAGAAATGAACAGCATGA TTACTTTGAACTACATTGGCTTGG arrays of 47% per EST or 40% per gene. However, owing to BG069482 TGTGGTACAAGGCTCCGTC CAGTTGGACACGCAGCACAT tm1cam BG080548 TTTAAGCATATTAGTCAGCGGAGGA GGGCCTCGATCAGAAGGACTTG the mixed background origins of the Sparc 129Sv/Ev/ BG085421 GTGGCAAAGGAGGATGAAGG CCAGGCACCTCAGAAACAAAG Gpi-bb BG074814 C1qa CACAGACGGGGATCGTTTATTC CAGCGGATTCCTCATTTTCC Mf1 mice, we repeated the PCRs using 129Sv/Ev con- BG087868 C1qb GTCCGGCCTAGAAGCATCAC ATCCCTGGGTCCCCTTTCTC Gpi-bb AW547306 C1qg TCAGGAACCAGGGTGGACTT CAAGCAGGTCAGCTCCGGAG trols instead of Mf1 . We considered that this should sig- BG087171 C1r TCCACTGCCACAATCCTTCC GCCAGAAGACCAGCTTCACC BG087429 C3 CCAGCAGGTCATCAAGTCAGG AGATTCTGTGAATGCCCCAAG nificantly reduce the possibility of strain differences contrib- C79876 C79876 AGAGATGGGTCCTCACAAGCA GGTCAAAGGCTGGCACTGG BG078093 Cbx3 GGACCGTCGTGTAGTGAATGG AACTCTCCGCTGCTGTCTGTG uting to variation between controls and experimental mice, as BG069465 Cd63 TCTGCATCTGCTGCTGGAAG GCAGGCCATTACCCATGAGA BG087410 Cd9 CTCTTGTCCCACGCAACTCC CCAGGAAACCAACCAGCATC we compared the experimental mice against BOTH parental BG082965 Cdk8 CGTTGGTGTTGCTGATTTGTG GCTCTCCTACTCCCTCTCCTTTG BG087382 CtsD CCTGCCCCACTCACACATAG TACTGCCCACACAGCAAAGG strains of origin at the PCR level. We found that Sparc was the BG087124 Fcgr3 CCAATGGCTACTTCCACCAC GGTCTGCTCCATTTGACACC BG072156 Fgls GTGTAGGGAGGGGCAAGGAG TCGGAAGATGGTGGGTGATG only gene that maintained the pattern of variability seen on BG085662 Fn1 TTCCATCACCCTCACCAACC GTGGCTGTGGACTTGCTTCC BG088898 Gp49a GGATTGATGGGATGAGCTGTG GCCAGAGGTTGATGGAGAGATG the arrays. AW550167, Hba-a1 CCACCACCAAGACCTACTTTCC AGGCAAGGAATTTGTCCAGAGA tm1cam Gpi-bb Gpi-bb AW551388, (4) Sparc 129Sv/Ev/Mf1 versus Mf1 at 9 BG073468, tm1cam Gpi-bb BG073469, months of age. The 9 month Sparc 129Sv/Ev/Mf1 BG074904, Gpi-bb BG086243, versus Mf1 arrays showed significant downregulation of BG086422 Hba-X GCGGTTAAGAGCATCGACAAC TTCTCAGTCAGGATAGAAGACAGGA 39 ESTs and significant upregulation of 24 ESTs. 3 ESTs rep- Hbb-b1 ACCTGACTGATGCTGAGAAGG CCCTTGAGGCTGTCCAAGTG Hbb-b2 CACCTGACTGATGCTGAGAAGT CCCTTGAGGTTGTCCAGGTTT resented Sparc, 7 represented Hba-a1, two represented CtsD, Hbb-Y TTACTGCTGAGGAAAAGACCCTC AAGTGACTAGCCAAAACAATCA HbbZ TACCTCACACCATGGTTCAC GTACTTGTGGGACAGAGCAT and two Smt3h1. The others appeared only once. This dataset BG070071 Hspb1 GCCTCGAAAGTAACCGGAATG CTGGCAAGCACGAAGAAAGG BG079631 Hspcb CCCAGCTCATGTCCCTCATC GTGCCAGACTTAGCAATGGTTC therefore contained 53 genes (Table 3). We tested these against BG070106 Lcn2 GCCACACTCACCACCCATTC GGCCACCACGGACTACAACC Gpi-bb BG064176 Lgals3 TCCCACTCCTAAGGCACACA CAAGTCCTGGTTGAAGCTGACC both 129Sv/Ev and Mf1 control strains at the PCR level. BG067456 Lyzs CGTGACACTGGGAACATCCT TGGGAAACAGCAGTCGTGTG BG073671 Mfge8 CCACAGCCAACCTGTCAACC TCCCTCTCTGCACACCACAC We were able to confirm 32 genes as significantly differen- BG087418 Ms4a6d TGGAATCAGCTTTCCCCAAAC GGGCTAGGCTGCTGTGAACC BG087762 P2rx4 GGAATTGGGACTGGAAGGTG ACGGAATATGGGGCAGAAGG tially expressed with reference to both control strains. This BG085244 plectin TAGCTGCTGAGGGGAACTGG GCTGGAAGGCTGAAGCAGAG BG072730 Ppgb GCTACCTGTTCTCCGCTCTCC TTCGGAAGGCTCTCCACATC gives a success rate of 65% per EST or 60% per gene for this C75970 Ppp2cb GGTCACATGAGGCTCTCCAC TTTAACCATGCCAACGGTCTC BG088310 Psap TGCTGGCTGGCTTCTGTAATG CTCCTGGCACTTGGTTCTGG array set. Given the multiple reappearance of Hba-a1 in this BG084583 Rgs10 GCCACACCCTCTGATGTTCC TCCTAAGAGTCCATCCCTCTCC BG074704 Rnf128 CACAAGCCACACAAAGTAGAAATAAAG ACTAAATCGTAGAACATTGGTAACTGG array set, we re-tested all of the mouse globin genes. Again, BG075648 Serping1 CAGTTCCGTCTGCCAGTTCC TGCAGCGACCTTTCCTCTTC BG064722, Smt3h1 GCCATCAAGCAGCAATGGA CACGTTTTCAGCCATGTTGTG these were all, apart from Hbb-Z, seen to be downregulated BG064723 BG064802, Sparc GCTGTGTTGGAAACGGAGTTG CTTGCCATGTGGGTTCTGACT (Figure 3 and Table 5). BG078305, BG072874 Clustering analysis: We performed clustering analysis BG070062 Spp1 AGCTTGTCCTTGTGGCTGTG TGAAGAGCGGTGAGTCTAAGGAG BG072750 Tgfbi TTTCTCCCCTTAACCCCTTCC GCGACTTGCCCCTGTCTATC of all genes that were confirmed as differentially regulated. BG074617 Tgm2C CACAGGCTTCCACGCTTTTC CACTGCACCAGACCTGTACCC BG075073 Tmsb4x CATGCAAGTTCTTTTCCCTCTC ACGAGCATTGCCTTCTTATTTTAC Results show that the clusters divide broadly into two groups; BG076072 Ugt1a1 AACCTTCCCTGATGGGTGTGG GCCTATGGCTGCCGGAAATG BG068728 Poor quality sequence those genes that are downregulated at 9 months during the AW546550 CAGGGACAAAGGAAAGGCATAG CTGCTTGCTGGTGATAGTTGG AW558227 Poor quality sequence onset of cataract and those that are upregulated at later stages BG072494 Poor quality sequence BG086297 Record removed at of disease, thus strengthening the conclusions reached as a submitter’s request result of pairwise comparisons of both strains at both time Significant genes included in this table are those giving a signifi- points. A third subgroup comprises Sparc, the globins, and cance level of delta=0.5 using SAM and that have a fold change of C79876 that were downregulated at both stages. There are a more than two in 8/10 repetitions. All significant genes are listed few genes (e.g. Fn1, Tgm2) that may be downregulated at early here, including those that were not confirmed by PCR. The columns stages of cataractogenesis and upregulated at the late stage list the GenBank accession number and primer name, and the for- studies. It is worth noting, however, that this was not noted ward and reverse primer sequences. The primer name is not listed with more stringent filtering when the experiments were ana- where the EST is uncharacterized, where ESTs correspond to known genes, the gene name is used. lyzed in a pair-wise fashion. 496 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

TABLE 4. CONFIRMED GENES: NINE MONTH SPARCTM1CAM129SV/EV LENSES VERSUS 129SV/EV CONTROLS

Unigene full length Accession accession c/s Fold Possible role in number Identity number mouse change Comments Sparc null cataract ------BG079989 Alb1 albumin 1, NM_009654.1 5 -14.08 Soluble, monomeric protein Downregulation of Mm.16773 which comprises about Alb1 may disrupt one-half of the blood serum cell-cell protein. Functions as a communication in the carrier protein for lens. Sparc is known steroids, fatty acids, and to bind albumin, thyroid hormones, plays a expression of Alb1 role in stabilizing reduced in Sparc extracellular fluid volume null mice with 103600 diabetic nephropathy [34] BG077309 Birc5 baculoviral NM_009689.1 11 -5.46 BIRC5 is an inhibitor of Lens fiber cells IAP apoptosis (IAP), or lose their nuclei as repeat-containing 5 programmed cell death, that they mature via a Mm.8552 is selectively overexpressed tightly controlled in common human cancers process of ‘arrested 603352 apoptosis’ [3,4] Downregulation could affect migration and maturation patterns of maturing cells in the lens C75991 Cbx5 chromobox NM_007626.2 15 -215.70 Heterochromatin protein-1 Downregulation may homolog 5 (HP1) is a methyl-lysine alter gene (Drosophila HP1a) binding protein localized at silencing, thereby (aka HP1) heterochromatin sites, where changing patterns of Mm.28003 it mediates gene silencing gene expression Involved in the repressive functions of the retinoblastoma protein 604478 BG073735 Col1a2 procollagen, NM_007743.1 6 -36.37 Major constituent of bone, Collagen fibrils in type I, alpha 2 extracellular matrix Sparc null skin are Mm.168712 structural constituent smaller and fewer in conferring tensile strength number [30] The activity. Mutations and lens or lens capsule polymorphisms associated may show a reduction with OI, Ehlers-Danlos in tensile strength syndrome, Marfan syndrome as a result of and osteoporosis collagen 120160 downregulation BG074327 Col3a1 procollagen, NM_009930.1 1 -66.77 Fetal (and blood vessel) Sparc is known to type III, alpha 1 collagen is also called interact with Mm.234850 collagen III Its synthesis collagens type I, is defective in III and IV, also see Ehlers-Danlos syndrome type Col1a2 above IV 120280 BG085518 Eraf erythroid NM_133245.1 position -22.61 ERAF forms a stable complex EraF may be associated factor unknown with free alpha-hemoglobin important for the (aka EDRF or but not beta- or hemoglobin correct folding and AHSP) Mm.218857 A (alpha2-beta2). Knockout stability of Hba-a1 mice show shorter [35] May be erythrocyte halflives. downregulated with Downregulated in globins in a transmissible spongiform co-ordinated fashion encephalopathies (TSEs) (see below) 605821 downregulation may result in incorrect Hba-a1 folding C79876 EST none 3 -13.88 also down in 9 month Function unknown Sparctm1Cam129Sv/Ev/Mf1Gpibb lenses. BG078478 Gpc3 glypican 3 NM_016697.2 X -7.79 The glypicans are 1 of the Loss of this gene Mm.22515 2 major families of is already cell-surface proteoglycans. implicated in GPC3 is mutated in cataract Simpson-Golabi-Behmel downregulation may syndrome, an overgrowth therefore encourage syndrome that can include the development of cataract as a feature opacity 312870

497 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision TABLE 4. CONTINUED. Unigene full length Accession accession c/s Fold Possible role in number Identity number mouse change Comments Sparc null cataract ------BG065213 Grb10 growth factor NM_010345.1 11 -20.82 Imprinted, but acts Downregulation of receptor bound independently of Igf2. Grb10 growth modulators protein 10 Mm.24640 as a potent growth may have effects on inhibitor the control of lens 601523 epithelial and fiber cell homeostasis and maturation BG076718, H19 H19 fetal liver NM_023123.1 7 -29.09 H19 is a developmentally Downregulation of BG086671, mRNA Mm.14802 regulated gene with putative growth modulators BG087028, tumor suppressor activity; may have effects on BG087529, loss of H19 expression may the control of lens BG087637, be involved in Wilms epithelial and fiber BG087638 tumorigenesis 103280 cell homeostasis and Imprinted, see Igf2, below. maturation BG079788, Hba-a1 hemoglobin NM_008218.1 11 -52.43 Major mouse alpha Downregulation may AW550167, alpha, adult chain hemoglobin (2 identical disrupt any or all AW551388, 1 Mm.196110 genes) MGI:85168 Possible normal functions in BG072574, roles in normal lenses: the lens, leading to BG072760, structural (may act as a opacity BG073467, crystallin), may have a role Downregulation also BG073468, in lens iron or oxygen noted in human BG073469, metabolism, or may be cataract [36] BG073608, pro-apoptotic [26] BG073617, BG074904, BG085818, BG086202, BG086243, BG086422 Hba-x: Hemoglobin NM_010405.2 11 n/a Mouse embryonic globin, not detected on X, alpha-like homologous to hemoglobin array, but checked embryonic chain in zeta in man as other globins Hba complex 142310 downregulated Mm.141758 Expressed in normal lenses [26], see Hba-a1 Hbb-b1 hemoglobin NM_008220.2 7 n/a not detected on array, but Expressed in normal beta, major chain checked as other globins lenses [26], see GeneID: 15129 downregulated. Homologous to Hba-a1. Cataracts hemoglobin delta in man have been associated with sickle cell anemia and thalassemia [37,38] AW548291, Hbb-b2 hemoglobin, NM_016956.2 7 -146.01 The alpha and beta loci Expressed in normal AW548342, beta adult minor determine the structure of lenses [26],see BG072990, chain Mm.233825 the 2 types of polypeptide Hba-a1 BG085433 chains in adult hemoglobin, Hb A. Hbb-b2 differs by nine amino acids from beta adult major chain (Hbb-b1) [39], 141900 Hbb-y hemoglobin Y, NM_008221.2 7 n/a not detected on array, but Expressed in normal beta-like embryonic checked as other globins lenses [26], see chain Mm.35830 downregulated. Hba-a1 BG064815, Igf2 insulin-like NM_010514.1 7 -37.40 IGF-II is a major modulator Downregulation of BG073613 growth factor 2 of placental and fetal growth modulators Mm.3862 growth. H19 and Igf2 are may have effects on reciprocally imprinted and the control of lens form a co-ordinately epithelial and fiber regulated 130 kb unit in the cell homeostasis and mouse (controlled by maturation enhancers, silencers and an imprinting control region (ICR)) [40] BG066442 Kpna2 karyopherin NM_010655.2 11 -2.76 maps to same region as Downregulation may (importin) alpha 2 Grb10, involved in result in altered Mm.12508 recognition of nuclear nuclear protein localization signal and import may affect import of proteins to the signal transduction nucleus 600685 may change access of newly translated trancription factors to target DNA. This may have knock-on effects on downstream transcription patterns 498 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

TABLE 4. CONTINUED. Unigene full length Accession accession c/s Fold Possible role in number Identity number mouse change Comments Sparc null cataract ------BG085378 Marcks myristoylated NM_008538.1 10 -2.94 MARCKS is a filamentous Downregulation of alanine rich protein actin crosslinking protein, Marcks may result in kinase C substrate MARCKS may be a regulated loss of control of Mm.30059 crossbridge between actin cell invasiveness and the plasma membrane, resulting in the also key regulatory appearance of molecule. Important in invadopodia that neural and retinal extend from the development epithelial cells 177061 into the lens capsule ECM BG087241 Mest mesoderm NM_008590.1 6 -8.81 Imprinted, paternally Downregulation of specific transcript expressed, heterozygosity growth modulators (aka Peg1) Mm.1089 causes aberrant maternal may have effects on behavior, inheritance of the control of lens defective allele from the epithelial and fiber father causes growth cell homeostasis and retardation 601029 maturation BG063925 Mt2 metallothionein NM_008630.1 8 -4.25 Metallothioneins: possible Expressed in lens 2, LOC330827 roles in zinc and copper epithelium, hypothetical gene homeostasis, metal upregulated in the supported by detoxification, regulation presence of cadmium AK046375. Mm.147226 or synthesis of or zinc ions [41] metalloproteins, protection [42]. Downregulation against reactive oxygen may affect metal species (ROS) metabolism or protection against ROS Cataract is promoted by ROS [43] AW553287 Osf2-pending NM_015784.1 3 -14.59 may function as a may also act as an osteoblast specific homophilic adhesion molecule adhesion molecule in factor 2 (fasciclin in bone formation lens, downregulation I-like) Mm.10681 MGI:1355212 may affect cell adhesivity (see Marcks, above) Osf2 and Sparc are both upregulated in a mouse model of cardiomyopathy co-ordinated regulation [44] BG070102 Phtf putative NM_013629.1 3 -41.45 isolated from subtracted Downregulation may homeodomain cDNA library using mRNAs have knock-on transcription factor isolated from human effects on Mm.25877 erythroleukemic cells downstream 604950 transcription patterns BG074984, Sfpq splicing factor NM_023603.1 4 -5.53 copurified and associated Downregulation may BG086821 proline/glutamine with polypyrimidine affect splicing rich (polypyrimidine tract-binding protein, binds patterns and/or tract binding the polypyrimidine tract of efficiency of protein associated) mammalian introns, essential transcription, Mm.140 pre-mRNA splicing factor therefore may have required early in knock-on effects on spliceosome formation downstream 605199 transcription patterns BG064802, Sparc secreted NM_009242.1 11 -20.17 gene knocked out in these Positive control. BG072874 acidic cysteine rich mice, positive control glycoprotein Mm.35439 BG086372 Tmpo; Thymopoietin AK011988.1 10 -5.00 polypeptide thymus hormone, Downregulation may (gene encodes 3 pleiotropic actions on be involved in different isoforms) prothymocytes, mature T effects on immune Mm.124 cells, nicotinic system components acetylcholine receptor, and seen at a later pituitary corticotrophs May stage of cataract bind lamin B and development (see chromosomes, involved in Figure 4) nuclear architecture and cell cycle control 188380 This table lists significant genes confirmed by PCR defined by accession number and identity. The mean fold change values from the arrays are given as are the mouse chromosomal locations (c/s mouse) where they are known. Where genes of known function have been confirmed by PCR an overview of gene function is provided (comments with OMIM number) and possible roles for these gene in the lens and cataractogenesis are suggested (possible role in Sparc null cataract). 499 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

TABLE 5. CONFIRMED GENES: NINE MONTH SPARCTM1CAM129SV/EV/MF1GPI-BB LENSES VERSUS MF1GPI-BB CONTROLS

Unigene full length Accession accession c/s Fold Possible role in number Identity number mouse change Comments Sparc null cataract ------BG075211 Adfp adipose NM_007408.2 4 6.66 novel 50-kD Role unknown differentiation related membrane-associated protein Mm.381 protein whose mRNA levels are induced rapidly and maximally after triggering adipocyte differentiation 103195 BG086439 Anxa2 annexin A2 Mm.584 NM_007585.2 9 2.63 ANX2 is an autocrine Role unknown factor that may enhance could be involved osteoclast formation in resorption of and bone resorption abnormal tissue 151740 by cellular components of the immune system BG074814 C1qa; Complement NM_007572.1 4 7.42 C1q is composed of 3 Complement can component 1, q different species of mediate clotting, subcomponent, alpha chains, called A, B, chemotaxis, polypeptide Mm.370 and C (see above for vascular other 2) C1q binds to permeability, immunoglobulin phagocytosis, and complexes with cell lysis [45] resulting serial Cell lysis is activation of C1r noted at later (enzyme), C1s stages of Sparc (proenzyme) and the null cataract other 8 components of [5,6,13,16] complement 120550 BG087868 C1qb complement NM_009777.1 4 8.80 C1q, the first See above, C component 1, q subcomponent of C1, has Complement can subcomponent, beta a complicated 18-chain mediate clotting, polypeptide Mm.2570 structure: 6 A, 6 B, chemotaxis, and 6 C chains vascular 120570 permeability, phagocytosis, and cell lysis [45]. Cell lysis is noted at later stages of Sparc null cataract [5,6,13,16] AW547306 C1qg complement NM_007574.1 4 5.27 Complement deficiency See below for component 1, q is associated with complement B subcomponent, gamma systematic lupus chain Complement polypeptide (aka erythromatosis can mediate complement component 120575 clotting, C1q, C chain) Mm.3453 chemotaxis, vascular permeability, phagocytosis and cell lysis [45]. Cell lysis is noted at later stages of Sparc null cataract [5,6,13,16] BG087171 C1r complement NM_023143.1 6 5.29 C1 is a complex of Complement can component 1, r C1q, C1r, and C1s, mediate clotting, subcomponent Mm.24276 activated by complexing chemotaxis, with antigen or an vascular acidic macromolecule permeability, (e.g.DNA) C1r phagocytosis, and deficiency leads to cell lysis [45]. lupus like Cell lysis is symptoms/reduced noted at later resistance to stages of Sparc infection null cataract 216950 [5,6,13,16]

500 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

TABLE 5. CONTINUED. Unigene full length Accession accession c/s Fold Possible role in number Identity number mouse change Comments Sparc null cataract ------BG087429 C3; Complement component NM_009778.1 17 3.99 C3 is an acute phase Complement can 3 Mm.19131 reactant; increased mediate clotting, synthesis of C3 is chemotaxis, induced during acute vascular inflammation permeability, 120700 phagocytosis, and cell lysis [45]. Cell lysis is noted at later stages of Sparc null cataract [5,6,13,16] BG087410 Cd9 CD9 antigen Mm.2956 NM_007657.2 6 2.77 CD9, a member of the Upregulation may transmembrane-4 play a role in superfamily, interacts abnormal tissue with integrins and remodelling other membrane during proteins, and may cataractogenesis participate in cell migration and adhesion. CD9 is important in the gamete fusion process at fertilization 143030 BG087382, Ctsd; Cathepsin D NM_009983.2 7 9.95 Cathepsin D is one of May be BG087383 Mm.231395 the lysosomal upregulated as a proteinases result of 116840 increased degradative requirements of stressed cells C79876 EST none 3 -5.78 also downregulated in 9 Function unknown month Sparctm1Cam129Sv/Ev lenses BG087124 Fcgr3 Fc receptor, IgG, NM_010188.2 1 9.32 CD16 is included in the Upregulation of low affinity III (aka zeta natural killer components of the CD16) Mm.22119 cell receptor complex. immune system Mutations cause could contribute increased to the ocular susceptibility to inflammation, infection 146740 tissue destruction, and re-organization that occurs in Sparc null lenses at advanced stages of cataract BG072156 Fgls-pending fragilis NM_025378.1 7 7.62 interferon induced Upregulation of Mm.141021 transmembrane protein, components of the 605579 immune system could contribute to the ocular inflammation, tissue destruction, and re-organization that occurs in Sparc null lenses at advanced stages of cataract

501 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

TABLE 5. CONTINUED. Unigene full length Accession accession c/s Fold Possible role in number Identity number mouse change Comments Sparc null cataract ------BG085662 Fn1 fibronectin 1 BC004724.1 1 22.28 Roles in cell Fn1 is a Mm.193099 adhesion, morphology, component of the and surface lens capsule[6] architecture. Absence TGF-beta has may cause loss of effects on contact inhibition of expression levels movement in transformed of Sparc and Fn1, cells Most cells suggesting some depend on fibronectin degree of for binding to co-ordinated collagen 135600 regulation. Degraded fibronectin is a potent monocyte chemoattractant [45] BG088898 Gp49a; Glycoprotein 49 A NM_008147.1 10 22.95 membrane protein, low Role unknown, Mm.196617 homology to human homology may leukocyte suggest immune immunoglobulin-like system function receptor AW550167, Hba-a1 hemoglobin NM_008218.1 11 -7.00 Major mouse alpha Downregulation AW551388, alpha, adult chain 1 hemoglobin (2 identical may disrupt any BG073468, Mm.196110 genes) MGI:85168. or all normal BG073469, Possible roles in functions in the BG074904, normal lenses lens, leading to BG086243, structural (may act as opacity BG086422 a crystallin), may have Downregulation a role in lens iron or also noted in oxygen metabolism, or human cataract may be pro-apoptotic [36] [26] Hba-x: Hemoglobin X, NM_010405.2 11 n/a Mouse embryonic globin Expressed in alpha-like embryonic homologous to normal lenses chain in Hba complex hemoglobin zeta in man [26], see Hba-a1 Mm.141758 142310. not detected on array, but checked as other globins downregulated Hbb-b1 hemoglobin beta, NM_008220.2 7 n/a not detected on array, Expressed in major chain GeneID: but checked as other normal lenses 15129 globins downregulated. [26], see Hba-a1 Homologous to Cataracts have hemoglobin delta in been associated man with sickle cell anemia and thalassemia [37,38] Hbb-b2 hemoglobin, beta NM_016956.2 7 -10.81 The alpha and beta Expressed in adult minor chain loci determine the normal lenses Mm.233825 structure of the 2 [26], see Hba-a1 types of polypeptide chains in adult hemoglobin, Hb A. Hbb-b2 differs by nine amino acids from beta adult major chain (Hbb-b1) [39] 141900 Hbb-y hemoglobin Y, NM_008221.2 7 n/a not detected on array Expressed in beta-like embryonic but checked as other normal lenses chain Mm.35830 globins downregulated [26] see Hba-a1 BG070071 Hspb1; Heat shock protein NM_013560.1 5 5.56 Heat-shock proteins Involved in the 1 (aka HEAT-SHOCK (HSPs) belong to the stress response PROTEIN 27; HSP27) stress protein family, probably Mm.13849 induced in response to upregulated as a environmental result of challenges/ increased cell developmental stress transitions HSP27 protein shows sequence similarity to mammalian alpha-crystallins (known chaperones) 602195 502 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

TABLE 5. CONTINUED.

Unigene full length Accession accession c/s Fold Possible role in number Identity number mouse change Comments Sparc null cataract ------

BG070106 Lcn2 lipocalin 2 Mm.9537 AK002932.1 2 9.60 associated with Upregulation of neutrophil gelatinase, this gene may be may function as a involved in modulator of cataract induced inflammation, induces immune reaction, apoptosis in a wide increased variety of leukocytes, apoptosis, or participates in the abnormal iron antibacterial iron metabolism depletion strategy of the innate immune system. 600181 BG064176 Lgals3: Lectin, NM_010705.1 14 5.18 may play an important Upregulation of galactose binding, role in both innate and components of the soluble 3 Mm.2970 adaptive immunity by immune system contributing to could contribute phagocytic clearance of to the ocular microorganisms and inflammation, apoptotic cells [46] tissue destruction and re-organization that occurs in Sparc null lenses at advanced stages of cataract BG067456 Lyzs lysozyme Mm.45436 NM_017372.2 10 13.38 catalyzes the Upregulation of hydrolysis of certain components of the mucopolysaccharides of immune system bacterial cell walls. could contribute Mutations associated to the ocular with increased inflammation, resistance to infection tissue or renal amyloidosis destruction and 153450 re-organization that occurs in Sparc null lenses at advanced stages of cataract BG073671 Mfge8 milk fat NM_008594.1 7 7.67 Lactadherin is Role unknown, if globule-EGF factor 8 expressed by mammary degraded, may protein. (aka epithelial cells as a contribute to lactadherin). Mm.1451 cell surface protein, plaque formation secreted as part of the in abnormal milk fat globule lenses membrane Main constituent of aortic medial amyloid is a 50 aa peptide derived from a domain of lactadherin 602281 BG087418 Ms4a6d NM_026835.1 19 5.84 Family of proteins Increasing membrane-spanning related to the porosity of the 4-domains, subfamily A, B-cell-specific antigen lens capsule may member 6D Mm.170657 CD20 They may result in represent proteins alterations in functioning either osmotic status, directly as resulting in ligand-gated ion observed cell channels or as swelling [6] essential components of Upregulation of such channels ion channel MGI:1931900 components may represent an attempt to compensate for this

503 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

TABLE 5. CONTINUED.

Unigene full length Accession accession c/s Fold Possible role in number Identity number mouse change Comments Sparc null cataract ------BG085244 Musculus IMAGE: 1064756 AY032898.1 7 3.09 intermediate filament Upregulation may plectin mRNA, 3’UTR binding protein and may result in further Mm.234912 provide mechanical perturbations in strength to cells and cell adhesion, tissues by acting as a migration and cross linking element morphology of the cytoskeleton Mutations cause muscular dystrophy with epidermolysis bullosa simplex 601282 BG087762 P2rx4 purinergic NM_011026.1 5 5.25 Activation of P2X Increasing receptor P2X, receptors leads to the porosity of the ligand-gated ion channel opening of nonselective lens capsule may 4 Mm.27861 cation channels that result in are permeable to Na+, alterations in K+, and, in some cases, osmotic status, Ca(2+) 600846 resulting in observed cell swelling [6] Upregulation of ion channel regulators may represent an attempt to compensate for this BG072730 Ppgb protective protein NM_008906.2 2 3.12 mutations cause Role unknown for beta-galactosidase neuraminidase Mm.7046 deficiency with beta-galactosidase deficiency 256540 BG088310 Psap prosaposin NM_011179.1 10 3.27 Lysosomal prosaposin Upregulation may Mm.233010 (65 kDa) is a reflect increased nonenzymic protein that lysosomal is transported to the degradation lysosomes in a mannose requirements by 6-phosphate-independent stressed cells manner [47] BG084583 Rgs10 regulator of NM_026418.1 7 4.76 RGS proteins negatively Upreguation G-protein signalling 10 regulate signaling reflects changes Mm.18635 pathways involving in signalling 7-transmembrane patterns brought receptors and about by heterotrimeric G increasing proteins 602856 abnormality BG074704 Rnf128 ring finger NM_023270.3 X 5.23 type I transmembrane Upregulation protein 128 Mm.27764 zinc RING finger implies higher protein in the levels of endocytic pathway that ubiquitinated functions as an E3 proteins targeted ubiquitin ligase for degradation, Expression of RNF128 implies cells are limits stressed activation-induced IL2 and IL4 production by T lymphocytes 300439 BG075648 Serping1 serine (or NM_0097760.1 2 4.68 C1 inhibitor is a Upregulation of cysteine) proteinase plasma protein involved components of the inhibitor, clade G, in the regulation of immune system member 1 Mm.38888 the complement cascade could contribute C1 inhibitor regulates to the ocular the first component of inflammation, complement (C1) by tissue inhibition of the destruction and proteolytic activity of re-organization its subcomponents C1r that occurs in and C1s 606860 Sparc null lenses at advanced stages of cataract 504 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

TABLE 5. CONTINUED.

Unigene full length Accession accession c/s Fold Possible role in number Identity number mouse change Comments Sparc null cataract ------

BG064802, Sparc secreted acidic NM_009242.1 11 -12.60 knocked out in Sp mix Positive control BG078305, cysteine rich mice, positive control BG072874 glycoprotein Mm.35439 for these arrays BG070062 Spp1: Secreted NM_009263.1 3 40.65 aka ostepontin. Role Upregulation of phosphoprotein 1 in immunosurveillance, components of the Mm.285918 role of colocalization immune system with CD44 in migration, could contribute cell fusion, and to the ocular resorption in inflammation, osteoclasts plays an tissue important role in tumor destruction and growth through the re-organization enhancement of that occurs in angiogenesis in vivo Sparc null lenses LocusID: 20750 at advanced stages of cataract BG072750 Tgfbi transforming NM_009369.1 13 13.05 mutations cause corneal Sparc and Tgfbi growth factor, beta dystrophy 601692 are known to induced, also known as induce each beta-IG-H3; BIGH3 other’s Mm.14455 expression under different circumstances [29] Array studies using glioma cell lines suggest Tgfbi may be induced by Sparc [48] BG074617 Tgm2 transglutaminase NM_009373.2 2 3.52 Tgm2 leads to formation Upregulation 2, C polypeptide of a cross-linked implies the Mm.18843 protein scaffold in presence of a cells undergoing greater than apoptosis. This usual number of stabilizes cell dying cells membranes (prevent Sparc in release of harmful differentiating intracellular cartilage is components). TGM2 is known to be a also the autoantigen in major target for celiac disease transglutaminase 190196 crosslinking [49] BG076072 Ugt1a1; UDP-glucuronosyl NM_013701.1 1 6.55 induced in transgenic Upregulation transferase 1 family, mice, functions to implies the need member 1 Mm.42472 eliminate steroids, to eliminate heme metabolites and greater than environmental toxins normal levels of 191740 toxins or toxic metabolites, suggesting increased cell stress

This table lists significant genes confirmed by PCR defined by accession number and identity. The mean fold change values from the arrays are given as are the mouse chromosomal locations (c/s mouse) where they are known. Where genes of known function have been confirmed by PCR an overview of gene function is provided (comments with OMIM number) and possible roles for these gene in the lens and cataractogenesis are suggested (possible role in Sparc null cataract).

Analysis of success rates for array experiments: We iden- tion of biological meaningfulness. We then removed genetic tified 152 ESTs as differentially expressed according to the 4 redundancy, both within and between array datasets, and also sets of array experiments we carried out. Using the same mouse those genes not replicably variable when 129Sv/Ev controls strains as were used for the array experiments (i.e., replicat- were substituted for Mf1Gpi-bb controls used on the 4 and 9 ing the array conditions for the PCRs) we were able to con- month Sparctm1cam129Sv/Ev/Mf1Gpi-bb versus Mf1Gpi-bb arrays. firm genes corresponding to 101 of these (66%). This is prob- Using this method of calculation, we have confirmed 51 out ably a good estimate of array methodology but is not a reflec- of a total of 108 genes tested (47%). Fifty of these genes were 505 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

not known to be variable prior to these experiments. In addi- tion 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 rea- sons, 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 ge- nomic sequence. It is therefore likely that poor single pass EST sequence, chimeric clones, cloning of genomic DNA con- taminants, 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 ex- pressed 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. How- ever, 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 be- tween the number of genes picked up on the array and the number confirmed.

DISCUSSION 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 sur-

Figure 2. RT-PCR confirmation of genes differentially regulated in 9 month Sparctm1cam129Sv/Ev lenses. Gene name, number of cycles and annealing temperature used in the PCR reactions corresponding to each gene are given to the right. The lane labels (duplicated at the top and bottom) identify the control 129Sv/Ev 9 month male lenses (C); Sparctm1cam129Sv/Ev 9 month male lenses (Sp); control, no RT added to the reaction (NC); and Sparctm1cam129Sv/Ev, no RT added to the reaction (Nsp). Two sets of samples from different control and experimental animals were used. 506 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

prising. 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, devel- ops normally [6]. Cascades of gene expression change that lead to opacity may not have been initiated at this stage. Alter- natively, some genes may be altered in expression post-tran- scriptionally 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 ab- normal 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 ar- rays 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 pheno- type. In contrast, genes identified as misregulated from the Sparctm1cam129Sv/Ev/Mf1Gpi-bb 9 month array set represent dif- ferential expression at an advanced stage of disease, in an ear- lier 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 com- position 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 regu- late 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.

Figure 3. RT-PCR confirmation of genes differentially regulated in 9 month Sparctm1cam129Sv/Ev/Mf1Gpi-bb lenses. Data were compared for 129Sv/Ev and Mf1Gpi-bb mice. The gene name, number of cycles, and annealing temperature used in the PCR reactions corresponding to each gene are given to the right. The lane labels (duplicated at the top and bottom) identify control 9 month male lenses (C); Sparctm1cam129Sv/Ev/Mf1Gpi-bb, 9 month male lenses (Sp); controls with no RT added to the reaction (NC); and Sparctm1cam129Sv/Ev/Mf1Gpi- bb, no RT added to the reaction (NSp). Two sets of samples from different control and experimental animals were used. 507 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

Studies in skin and bone [30,31] have shown that ECM Tmpo, Grb10) and cell cycle/apoptosis genes (Birc5) may re- collagen structure and turnover are also affected by the ab- sult in, or arise from, changes in control of cell invasiveness sence of Sparc. Collagen fibrils are smaller, less abundant, and/or morphology. and turned over less frequently. We noted the downregulation Loss of lens capsule integrity may have knock-on effects of Col3a1 and Col1a2. Downregulation of these genes may on the osmotic balance of posterior lens epithelial and fiber contribute to structural destabilization of a weakening lens cells. These swell in response to the greater permeability of a capsule. These changes may result from loss of transcriptional weakened lens capsule [6]. Swelling of these cells leads to control over posterior lens epithelial and fiber cells that nor- disorganized cell structure, spreading from the posterior re- mally produce the correct amounts of proteins to maintain the gion to other areas of the lens (which in normal mice show capsular ECM. Such changes may arise from downregulation lower Sparc expression levels) and are presumably therefore of direct and indirect modulators of gene expression such as affected less rapidly in knockouts [13]. Progressive structural those identified here (Cbx5, Phtfr, Sfpq, Kpna2). The disruption of the outer layers of the lens may result in reduc- downregulation of growth factors (Igf2, H19, Gpc3, Mest, tions (or inappropriate increases) in the transport efficiency

Figure 4. Cluster analysis from gene expression profiles of all ESTs corresponding to genes confirmed in any one of our experimental sets. Cluster analysis using 2D self organizing maps (GeneSight version 3.5) from gene ex- pression profiles. Mean ratios for confirmed genes across all experiments. Figures in red represent those identified as statistically significant in in- dividual array experiments (see Methods for significance criteria).

508 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision of metabolites within the lens. shown via upregulation of complement components, Fcgr3, Changes in lens globin concentrations (and changes in Lcn2, Lysz, Fgls, Serping, and Spp1. Increasing lens capsule globin folding, as a consequence of EraF downregulation) permeability, progressive swelling of lens epithelial and fiber could contribute to opacification. We have proposed various cells, cell rupture, and a strengthening immune response could non-mutually exclusive roles for globins in normal lenses [26]. then collaborate in effecting lens capsule rupture, followed by They may play roles in lens iron transport and/or metabolism, the tissue destruction seen in the last stages of cataract in Sparc may act as oxygen transporters and/or oxygen sinks, or may knockout mice. Notably, neutrophils, lymphocytes, and mac- act to promote apoptosis like phenomena in differentiating lens rophages have been shown to be present in the vitreous in fiber cells. Globin downregulation may begin to disrupt the Sparc null eyes at an advanced stage of cataract [16]. ongoing maturation of migrating lens fiber cells during early We have followed the progression of cataract in Sparc cataractogenesis, either via disruption in the oxygen gradient null mice over time, via tracking of gene expression changes. or via a decrease in their ability to promote the apoptosis like These data provide excellent snapshots of pathogenic processes process that results in loss of the lens fiber cell nucleus. Ab- occurring in Sparc null lenses and allow elaboration of previ- normal retention of the nucleus by mature lens fiber cells has ously existing models of cataract progression in these mice. been implicated in cataract in other animal models [32]. Al- Functions already ascribed to the ‘known genes’ that are dif- ternatively, opacity could result from inappropriate posttrans- ferentially regulated have allowed us to build on previous stud- lational modification of proteins, such as crystallins, leading ies allowing us to form hypotheses as to the mechanics of from disruption of metabolite transfer processes. If globins developing opacity. These data, and associated studies, will normally act as oxygen or iron ‘sinks’, lens capsule perme- allow us to identify and functionally characterize more genes ability and globin downregulation could result in increased involved in normal and pathogenic lens biology, perhaps lead- concentrations of oxygen and reactive oxygen species (ROS, ing to the identification of potential therapeutic targets and or iron) in posterior lens epithelial and fiber cells. This could the development of non-surgically based therapies for cata- in turn, result in opacity. Downregulation of Mt2 may also ract. contribute to disruption of oxygen or metal metabolism in the lens. ACKNOWLEDGEMENTS Late stage gene expression changes: Incremental dis- We would like to thank The Human Genome Mapping Project ruption of normal lens structure and function leads to increas- Resource Centre for provision of the NIA microarray slides ing stress in the lens cells. Cell lysis occurs at later stages of used in this study. We would also like to thank Dr. Anna Hurley, cataract in the Sparc null mouse [6]. The upregulation of stress Steve Turner, and Vicky Workman for assistance with array response proteins, such as heat shock proteins and those genes protocols, scanning, image analysis, data storage, and identified as being involved in degradative pathways (Tgm2, bioinformatics. We would also like to thank Prof. Alan Clarke Hspb1, Ugt1A, CtsD, Rnf128, Psap), indicate a detrimental and colleagues for help with use of their photographic equip- effect on cellular viability. Increasing opacity, cell swelling, ment. This work was funded by the Wellcome Trust and changes in gross lens structure, cell death, and progressive BBSRC (Evans laboratory) and by the National Eye Research weakening of the lens capsule may result in substantial lens Council (UK) and the Royal Society (UK; Wride laboratory). capsule porosity, allowing small amounts of lens proteins such as the crystallins (or other cellular debris) to escape the con- REFERENCES fines of the lens. Upregulation of ion channel related proteins 1. Wride MA. Cellular and molecular features of lens differentia- (Ms4a6d, P2rx4) may imply an attempt by the cells to control tion: a review of recent advances. Differentiation 1996; 61:77- abnormal osmotic balance or abnormal cell-cell communica- 93. tion. Further perturbations in cell adhesion, migration, and 2. Piatigorsky J. Lens differentiation in vertebrates. A review of cel- lular and molecular features. Differentiation 1981; 19:134-53. morphology are indicated by the upregulation of Fn1, plectin, 3. Wride MA. Minireview: apoptosis as seen through a lens. Apoptosis and CD9. 2000; 5:203-9. Leakage of crystallin proteins outside the lens, or inap- 4. Bassnett S. Lens organelle degradation. Exp Eye Res 2002; 74:1- propriate release of cell debris into the aqueous or vitreous 6. humors probably results in attraction of cellular components 5. Norose K, Lo WK, Clark JI, Sage EH, Howe CC. Lenses of SPARC- of the immune system. Lens rupture or other causes of ‘leak- null mice exhibit an abnormal cell surface-basement membrane age’ from the lens capsule (such as hypermature cataract) can interface. Exp Eye Res 2000; 71:295-307. result in an acute immune response and inflammation of the 6. Yan Q, Clark JI, Wight TN, Sage EH. Alterations in the lens cap- eye, known as lens induced uveitis [33]. Phacolytic uveitis sule contribute to cataractogenesis in SPARC-null mice. J Cell Sci 2002; 115:2747-56. has been used to describe uveitis that may occur as a result of 7. Francis PJ, Moore AT. Genetics of childhood cataract. Curr Opin lens protein release through an intact lens capsule. This is usu- Ophthalmol 2004; 15:10-5. ally milder than uveitis induced as a result of full lens capsule 8. Minassian DC, Reidy A, Desai P, Farrow S, Vafidis G, Minassian rupture [33]. The progressive weakening of the lens capsule A. The deficit in cataract surgery in England and Wales and the in Sparc null mice may allow increasing contact between lens escalating problem of visual impairment: epidemiological mod- proteins and immune system components or signaling mol- elling of the population dynamics of cataract. Br J Ophthalmol ecules. Involvement of an immune response in some form is 2000; 84:4-8. 509 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

9. Harding J. Cataract: biochemistry, epidemiology, and pharmacol- ery in the mouse lens. Mol Vis 2003; 9:360-96 . ogy. 1st ed. London: Chapman and Hall; 1991. 27. Tanaka TS, Jaradat SA, Lim MK, Kargul GJ, Wang X, Grahovac 10. Derham BK, Harding JJ. Effects of modifications of alpha-crys- MJ, Pantano S, Sano Y, Piao Y, Nagaraja R, Doi H, Wood WH tallin on its chaperone and other properties. Biochem J 2002; 3rd, Becker KG, Ko MS. Genome-wide expression profiling of 364:711-7. mid-gestation placenta and embryo using a 15,000 mouse de- 11. Mason IJ, Taylor A, Williams JG, Sage H, Hogan BL. Evidence velopmental cDNA microarray. Proc Natl Acad Sci U S A 2000; from molecular cloning that SPARC, a major product of mouse 97:9127-32. embryo parietal endoderm, is related to an endothelial cell ‘cul- 28. Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P, ture shock’ glycoprotein of Mr 43,000. EMBO J 1986; 5:1465- Stoeckert C, Aach J, Ansorge W, Ball CA, Causton HC, 72. Gaasterland T, Glenisson P, Holstege FC, Kim IF, Markowitz 12. Lane TF, Sage EH. The biology of SPARC, a protein that modu- V, Matese JC, Parkinson H, Robinson A, Sarkans U, Schulze- lates cell-matrix interactions. FASEB J 1994; 8:163-73. Kremer S, Stewart J, Taylor R, Vilo J, Vingron M. Minimum 13. Yan Q, Blake D, Clark JI, Sage EH. Expression of the matricellular information about a microarray experiment (MIAME)-toward protein SPARC in murine lens: SPARC is necessary for the struc- standards for microarray data. Nat Genet 2001; 29:365-71. tural integrity of the capsular basement membrane. J Histochem 29. Schiemann BJ, Neil JR, Schiemann WP. SPARC inhibits epithe- Cytochem 2003; 51:503-11. lial cell proliferation in part through stimulation of the trans- 14. Brekken RA, Sage EH. SPARC, a matricellular protein: at the forming growth factor-beta-signaling system. Mol Biol Cell crossroads of cell-matrix communication. Matrix Biol 2001; 2003; 14:3977-88. 19:816-27. 30. Bradshaw AD, Puolakkainen P, Dasgupta J, Davidson JM, Wight 15. Gilmour DT, Lyon GJ, Carlton MB, Sanes JR, Cunningham JM, TN, Helene Sage E. SPARC-null mice display abnormalities in Anderson JR, Hogan BL, Evans MJ, Colledge WH. Mice defi- the dermis characterized by decreased collagen fibril diameter cient for the secreted glycoprotein SPARC/osteonectin/BM40 and reduced tensile strength. J Invest Dermatol 2003; 120:949- develop normally but show severe age-onset cataract formation 55. and disruption of the lens. EMBO J 1998; 17:1860-70. 31. Boskey AL, Moore DJ, Amling M, Canalis E, Delany AM. Infra- 16. Norose K, Clark JI, Syed NA, Basu A, Heber-Katz E, Sage EH, red analysis of the mineral and matrix in bones of osteonectin- Howe CC. SPARC deficiency leads to early-onset null mice and their wildtype controls. J Bone Miner Res 2003; cataractogenesis. Invest Ophthalmol Vis Sci 1998; 39:2674-80. 18:1005-11. 17. Bassuk JA, Birkebak T, Rothmier JD, Clark JM, Bradshaw A, 32. Nishimoto S, Kawane K, Watanabe-Fukunaga R, Fukuyama H, Muchowski PJ, Howe CC, Clark JI, Sage EH. Disruption of the Ohsawa Y, Uchiyama Y, Hashida N, Ohguro N, Tano Y, Sparc locus in mice alters the differentiation of lenticular epi- Morimoto T, Fukuda Y, Nagata S. Nuclear cataract caused by a thelial cells and leads to cataract formation. Exp Eye Res 1999; lack of DNA degradation in the mouse eye lens. Nature 2003; 68:321-31. 424:1071-4. 18. Delany AM, Amling M, Priemel M, Howe C, Baron R, Canalis 33. Van Der Woerdt A. Lens-induced uveitis. Vet Ophthalmol 2000; E. Osteopenia and decreased bone formation in osteonectin-de- 3:227-234. ficient mice. J Clin Invest 2000; 105:915-23. Erratum in: J Clin 34. Taneda S, Pippin JW, Sage EH, Hudkins KL, Takeuchi Y, Couser Invest 2000; 105:1325. WG, Alpers CE. Amelioration of diabetic nephropathy in 19. Kantorow M, Huang Q, Yang XJ, Sage EH, Magabo KS, Miller SPARC-null mice. J Am Soc Nephrol 2003; 14:968-80. KM, Horwitz J. Increased expression of osteonectin/SPARC 35. Kihm AJ, Kong Y, Hong W, Russell JE, Rouda S, Adachi K, mRNA and protein in age-related human cataracts and spatial Simon MC, Blobel GA, Weiss MJ. An abundant erythroid pro- expression in the normal human lens. Mol Vis 2000; 6:24-9 . tein that stabilizes free alpha-haemoglobin. Nature 2002; 20. Kantorow M, Horwitz J, Carper D. Up-regulation of osteonectin/ 417:758-63. SPARC in age-related cataractous human lens epithelia. Mol 36. Hawse JR, Hejtmancik JF, Huang Q, Sheets NL, Hosack DA, Vis 1998; 4:17 . Lempicki RA, Horwitz J, Kantorow M. Identification and func- 21. Robinson ML, Holmgren A, Dewey MJ. Genetic control of ocu- tional clustering of global gene expression differences between lar morphogenesis: defective lens development associated with human age-related cataract and clear lenses. Mol Vis 2003; ocular anomalies in C57BL/6 mice. Exp Eye Res 1993; 56:7- 9:515-37 . 16. 37. Babalola OE, Danboyi P, Abiose AA. Hereditary congenital cata- 22. Sandilands A, Wang X, Hutcheson AM, James J, Prescott AR, racts associated with sickle cell anaemia in a Nigerian family. Wegener A, Pekny M, Gong X, Quinlan RA. Bfsp2 mutation Trop Doct 2000; 30:12-4. found in mouse 129 strains causes the loss of CP49 and induces 38. Bloomfield SE, Markenson AL, Miller DR, Peterson CM. Lens vimentin-dependent changes in the lens fibre cell cytoskeleton. opacities in thalassemia. J Pediatr Ophthalmol Strabismus 1978; Exp Eye Res 2004; 78:109-23. 15:154-6. 23. Alizadeh A, Clark J, Seeberger T, Hess J, Blankenship T, 39. Konkel DA, Maizel JV Jr, Leder P. The evolution and sequence FitzGerald PG. Characterization of a mutation in the lens-spe- comparison of two recently diverged mouse chromosomal beta— cific CP49 in the 129 strain of mouse. Invest Ophthalmol Vis globin genes. Cell 1979; 18:865-73. Sci 2004; 45:884-91. 40. Drewell RA, Arney KL, Arima T, Barton SC, Brenton JD, Surani 24. Graw J. The genetic and molecular basis of congenital eye de- MA. Novel conserved elements upstream of the H19 gene are fects. Nat Rev Genet 2003; 4:876-88. transcribed and act as mesodermal enhancers. Development 25. Wilson AS, Hobbs BG, Speed TP, Rakoczy PE. The microarray: 2002; 129:1205-13. potential applications for ophthalmic research. Mol Vis 2002; 41. Oppermann B, Zhang W, Magabo K, Kantorow M. Identification 8:259-70 . and spatial analysis of metallothioneins expressed by the adult 26. Wride MA, Mansergh FC, Adams S, Everitt R, Minnema SE, human lens. Invest Ophthalmol Vis Sci 2001; 42:188-93. Rancourt DE, Evans MJ. Expression profiling and gene discov- 42. Hawse JR, Cumming JR, Oppermann B, Sheets NL, Reddy VN, 510 Molecular Vision 2004; 10:490-511 ©2004 Molecular Vision

Kantorow M. Activation of metallothioneins and alpha-crystal- 47. Lefrancois S, May T, Knight C, Bourbeau D, Morales CR. The lin/sHSPs in human lens epithelial cells by specific metals and lysosomal transport of prosaposin requires the conditional in- the metal content of aging clear human lenses. Invest Ophthalmol teraction of its highly conserved d domain with sphingomyelin. Vis Sci 2003; 44:672-9. J Biol Chem 2002; 277:17188-99. 43. Spector A. Review: Oxidative stress and disease. J Ocul Pharmacol 48. Golembieski WA, Rempel SA. cDNA array analysis of SPARC- Ther 2000; 16:193-201. modulated changes in glioma gene expression. J Neurooncol 44. Ihara Y, Suzuki YJ, Kitta K, Jones LR, Ikeda T. Modulation of 2002; 60:213-26. gene expression in transgenic mouse hearts overexpressing 49. Hohenadl C, Mann K, Mayer U, Timpl R, Paulsson M, calsequestrin. Cell Calcium 2002; 32:21-9. Aeschlimann D. Two adjacent N-terminal glutamines of BM- 45. MacKinney AA. Hematology for Students. 1st ed. New York: 40 (osteonectin, SPARC) act as amine acceptor sites in Taylor and Francis; 2002. transglutaminaseC-catalyzed modification. J Biol Chem 1995; 46. Sano H, Hsu DK, Apgar JR, Yu L, Sharma BB, Kuwabara I, Izui 270:23415-20. S, Liu FT. Critical role of galectin-3 in phagocytosis by mac- rophages. J Clin Invest 2003; 112:389-97.

The print version of this article was created on 9 Aug 2004. This reflects all typographical corrections and errata to the article through that date. Details of any changes may be found in the online version of the article. 511