Complex Antibody Profile Changes in an Experimental Autoimmune Glaucoma Animal Model

Abstract

purpose. Increased serum antibodies against heat shock protein 27 (HSP27) have been identified in patients with glaucoma. Immunization with HSP27 caused retinal ganglion cell (RGC) loss in animals. The authors analyzed whether HSP27 immunization not only causes RGC loss but also affects systemic antibody patterns.

methods. Rats were immunized with HSP27 and were surveyed for 4, 5, and 6 weeks (groups 1–3). Control animals were humanely killed after 6 weeks (group 4). Intraocular pressure was measured before and 2 and 4 weeks after immunization. Fundus images were taken at the same time. Retinal flatmounts were prepared, and Brn-3a labeled RGCs were counted. Serum was collected during the study to detect antibody patterns against retinal antigens through Western blot analysis and mass spectrometry techniques. Patterns were analyzed by multivariate statistical techniques, and biomarkers were identified with the use of mass spectrometry.

results. No significant changes in intraocular pressure were observed, and no fundus abnormalities were noted. The animals immunized with HSP27 showed lower RGC density than controls (P < 0.05). Two and 4 weeks after immunization, we detected a significant difference in antibody profiles between groups 1 and 4 (P < 0.05) and groups 3 and 4 (P < 0.05). Proteins with different antibody level expression after immunization included heat shock protein 90, α-enolase, and glyceraldehyde-3-phosphate dehydrogenase.

conclusions. After immunization with HSP27, animals showed IOP-independent RGC loss and changes in serum antibody patterns. Thus, this model might be a beneficial approach to study the development and effects of anti–retinal antibodies and their involvement in RGC loss.

Glaucoma, a main cause of blindness, ¹ is responsible for the progressive loss of retinal ganglion cells (RGCs), atrophy of the optic nerve, and a gradual loss of visual field. ^{2 3} In several studies antibody patterns in the sera of patients with glaucoma were analyzed. Antibodies against, for example, heat shock proteins (HSPs), ⁴ γ-enolase, ⁵ glutathione-S-transferase, ⁶ glycosaminoglycans, ⁷ and fodrin ⁸ have been identified, most notably in normal pressure glaucoma. Furthermore, comparing antibody profiles of patients with glaucoma and healthy subjects, ^{8 9 10} we could show significantly increased and decreased antibody reactivity against ocular antigens. These findings suggest that autoimmunity may play a significant role in disease progression. 

Currently all “glaucoma” animal models are based on direct damage of RGCs or optic nerve. RGC loss is caused by vein cautery or saline injection models that induce chronic ocular hypertension ^{11 12 13 14} or ischemia/reperfusion conditions. ^{15 16} Other models use optic nerve crush to cause damage. ^{17 18} However, these models are not suitable to analyze possible immunologic involvement in the pathogenesis of glaucoma. 

Heat shock proteins are intracellular molecules that play an important role in cell survival. HSP27 is a small HSP, a molecular chaperone, ¹⁹ that was detected in rat RGCs. ²⁰ However, the regular expression rate of HSP27 in the retina of rats is controversial. ^{21 22} In addition, elevated levels of HSP27 have been reported in nerve axotomy models ²³ and after optic nerve transection. ²¹ Immunohistochemical examination on human postmortem eyes demonstrated increased retinal staining of HSP27 in glaucomatous eyes. ⁴ Studies on patients with glaucoma revealed increased reactivity against HSP27 in patients with normal tension glaucoma. ^{10 24} Tezel et al. ²⁵ showed that HSP27 antibodies enter RGCs endocytically and trigger apoptosis. 

Additional evidence for the involvement of specific antibodies in a disease is usually gained through animal models, eliciting the disease by selective immunogenes. Some models use immunization with an antigen to trigger a disease; examples are models for experimental autoimmune encephalitis (EAE), ^{26 27} experimental autoimmune uveitis (EAU), ^{28 29 30} experimental autoimmune arthritis, ^{31 32} and Alzheimer-like disease. ³³  

Initially Wax et al. ³⁴ immunized rats with human HSP27 or HSP60 and detected significant RGC loss 1 and 4 months after immunization. The most prominent RGC loss was observed in the region immediately adjacent to the area centralis. 

Based on that study, our study aimed to confirm RGC loss in animals after immunization with HSP27 and to broaden insights into the mechanisms of RGC loss, particularly analyzing potential changes in the antibody profiles of rat serum after immunization. 

Methods

Animals

Male Lewis rats were obtained from Charles River (Portage, MI). All animal experiments were performed in conformity with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and the study was approved by the local Animal Care Committee. 

Detailed observations and health checks, including eye examinations, were preformed daily. All injections were prepared immediately before immunization, as follows: 100 μL incomplete Freund adjuvant and 1 μg pertussis toxin (both Sigma-Aldrich, St. Louis, MO) were mixed with 100 μg purified human HSP27 (Stressgen Bioreagents, Victoria, BC, Canada) and injected intraperitoneally. 

Animals were humanely killed after 4 (group 1, n = 6), 5 (group 2, n = 6), and 6 weeks (group 3, n = 9). Group 4 (n = 10) served as control animals, received no immunization, and were humanely killed after 6 weeks. 

Intraocular Pressure Measurements

IOP was measured with a tonometer (TonoPen; Medtronic, Jacksonville, FL) in all animals in groups 3 and 4, as described elsewhere. ^{35 36} During the study, 10 measurements were carried out per eye after the administration of topical anesthesia with 0.5% proparacaine (Alcon, Fort Worth, TX). IOP was measured before and 2 and 4 weeks after immunization. 

Fundus Images

Fundus images were taken 1 week before and 2 and 4 weeks after immunization of all animals in groups 3 and 4. 

The eyes of the rats were dilated with atropine and phenylephrine (both Alcon), and animals were sedated with ketamine (50 mg/kg) and xylazine (5 mg/kg) beforehand. Animals were placed on a tray in front of a retinal camera (TRC-50X; Topcon, Paramus, NJ). The camera was modified with an additional lens for imaging animals. A digital color camera (Exwave HAD 3CCD; Sony, Park Ridge, NJ) was connected to the camera, and appropriate software (IMAGEnet 2000, version 2.0; Topcon) was used to take three photographs per eye for further evaluation. 

Blood Collection

Blood from group 1 and 2 animals was taken at the end of the study (4 and 5 weeks after immunization) through heart puncture. Blood from group 3 and the control group was collected through tail vein puncture before immunization and 2 and 4 weeks after immunization and through heart puncture at the end of the study (6 weeks after immunization). 

All blood samples were immediately transferred to reaction tubes. After clotting for 20 minutes, samples were centrifuged at 9000 rpm for 20 minutes (Spectrafuge 7M; E&K Scientific, Santa Clara, CA). All serum samples were stored at −80°C for further analysis. 

Retinal Flatmounts and RGC Counts

After euthanatization, animal eyes were enucleated and fixed in 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA). Retinal ganglion cells were labeled with Brn-3a, a specific marker for RGCs, ³⁷ according to standard protocols. ³⁴ Brn-3a, also known as Brn3.0 and Pou4f1, plays an important role in RGC development. ^{38 39}  

Retinal flatmounts were incubated with primary anti–rat Brn-3a antibody (1:50; Chemicon, Temecula, CA) followed by a biotin-conjugated IgG secondary antibody (1:200; Vector, Burlingame, CA). After incubation with extravidin (1:200; Sigma-Aldrich), flatmounts were incubated with Tris-HCl buffer and afterward with a buffer solution of 0.15% diaminobenzidine tetrahydrochloride (Sigma-Aldrich) in 0.05 M Tris-HCl buffer. This buffer was exchanged with 0.15% DAB 1% CoCl2-Tris buffer with 0.03% H₂O₂ (Sigma-Aldrich). When the desired staining of RGC nuclei was achieved the reaction was stopped with Tris. Retinas were placed on covered glass slides. 

Analysis of Brn-3a labeled retinas was performed with the use of imaging software (NIS-Elements; Nikon, Tokyo, Japan) according to the manufacturer’s instructions. Briefly, images were obtained with a digital camera (DS-Fi1; Nikon) connected to a histologic microscope (Eclipse 80i; Nikon) equipped with an automatic scanning table. One large image assembled of 100 to 120 images was taken of each retinal flatmount with 100× magnification at a resolution of 2500 × 1200 pixels. For each flatmount, Brn-3a labeled RGCs in four eccentricities from the optic nerve head (distance approximately 0.4–3 mm) were counted (NIS-Elements; Nikon). The same grid was used for all flatmounts in this study. Data were analyzed by ANOVA and Tukey post hoc test. 

Western Blot Analysis

The serum samples were analyzed with immunodetection. Western blot analysis was performed as described previously. ^{8 9 40} Pooled homogenized bovine retinas were used for 13.5% SDS-PAGE, and gels were transferred onto nitrocellulose membranes. Blots were cut into strips and incubated with rat serum (1:10). Peroxidase-conjugated anti–rat-IgG secondary antibody was used (1:500) before visualization with 0.05% 4-chloro-1-naphthol (both Sigma-Aldrich, Munich, Germany). 

Data were acquired with a color flatbed scanner (CanoScan 8400F; Canon, Krefeld, Germany). Digital image analysis and evaluation of Western blot analysis were performed (BioDocAnalyze; Biometra, Goettingen, Germany). Based on the densitogram of each Western blot, multivariate statistical techniques were used to detect differences in the distribution of antibodies against retinal antigens. Densitographic data, such as peak height, localization, and area under the curve, were exported to statistical analysis software (Statistica, version 8.0; Statsoft, Tulsa, OK), and statistical calculations were performed. The profiles were compared by an analysis of discriminance. 

Protein G Bead Approach

Antibody profiles from sera collected at the end of the study were also analyzed with protein chips (Bio-Rad, Hercules, CA) techniques. 

Samples were prepared as follows: Protein G magnetic beads (Dynabead; Dynal Biotech, Oslo, Norway) were repeatedly washed with 0.1 M NaAc (pH 5). Then beads were simultaneously incubated with 70 μL rat serum and 120 μL homogenized bovine retina (preparation as described). After washing steps, samples were eluted with 20 μL 0.1 M glycine (pH 3; MP Biomedicals, Eschwege, Germany). Elutes were separated from beads and concentrated. 

A SELDI-TOF (surface-enhanced laser desorption/ionization time of flight) protein chip system (Bio-Rad) was used for protein profiling of elutes on two different chromatographic surfaces: a weak cation exchange surface (CM10) and a hydrophobic chip (H50). All chips were pretreated according to the manufacturer’s standard protocols. A 4-μL sample was placed on each chip, followed by 1 μL matrix consisting of 0.02 g sinapinic acid (Fluka, Steinheim, Germany) with 750 μL acetonitrile, 750 μL H₂O, and 15 μL trifluoroacetic acid (Merck, Darmstadt, Germany). 

Protein chips were analyzed (PBS-IIc ProteinChip Reader with ProteinChip software, version 3.2; Bio-Rad). Each array was read at two laser intensities, with focus mass 8000 or 20,000 Da. Data were transferred to database software (CiphergenExpress, version 2.1; Bio-Rad) to automatically detect peaks, normalize spectra, and create a peak cluster list. Exclusively normalized peak intensities were used for statistical analyses. 

Cluster lists were exported to statistical analysis software (Statistica, version 8.0; Statsoft) containing peak intensity values for each group. Based on these normalized peak intensities, multivariate analysis of discriminance was calculated. Not only does this analysis test the zero hypothesis (namely, that mean data vectors of different groups derive from a multivariate normally distributed population), it also shows which groups are statistically different. Based on the findings, discriminant function analysis can be used to determine which peaks at a particular molecular weight caused the mean value comparison to become significant and which variables can discriminate between groups. 

Protein Identification through Mass Spectrometry

Protein identification was performed as described previously. ⁴¹ Briefly, corresponding immunoreactive bands were cut from SDS gels and digested tryptically. Samples were spotted on an anchor chip target (Bruker, Bremen, Germany) between two layers of cinnamic acid matrix (0.02 g cinnamic acid, 5 mL water, 5 mL acetonitrile, 10 μL trifluoroacetic acid). All peptide identifications were performed with a Maldi-TOFTOF mass spectrometer (Ultraflex Maldi-TOF/TOF; Bruker Daltonics, Bremen, Germany). 

The MS/MS spectra were exported (MASCOT; Matrix Science, Boston, MA; available at www.matrixscience.com) for database search using NCBI (National Institutes of Health, Bethesda, MD) and SwissProt (Swiss Institute of Bioinformatics, Geneva, Switzerland) databases. Only identifications with significant mascot scores from two independent gel bands were used. 

Results

Intraocular Pressure Data

Figure 1Ashows the intraocular pressure variation between immunized and nonimmunized animals. There was no significant difference in IOP between group 3 (immunized with HSP27) and group 4 (controls) before immunization (P = 1.0). Two and 4 weeks after immunization, no significant difference in IOP was observed between the immunized and the nonimmunized group (P = 0.99 and P = 1.0). 

Fundus Images

Fundus images were taken of animals in groups 3 and 4 at three time points: before immunization and at 2 and 4 weeks after immunization. No distinctive occurrences such as abnormalities of retinal blood vessels or retinal bleeding could be seen on any fundus images taken of animals during the study. The optic disc and the blood vessels were clearly visible in all images at all time points, and no differences were noted among groups (Fig. 1B) . 

RCG Counts

Brn-3a–labeled RGCs were counted in four eccentricities from the optic nerve on all flatmounts. RGC loss was observed in all animals after immunization. Five and 6 weeks after immunization, animals expressed significantly lower RGC density (P < 0.05; 35% loss after 5 weeks); it had not been significant 4 weeks after immunization (6% loss). Mean RGC densities of all four groups are shown in Figure 1C . 

To gain more information about possible RGC density decreased in certain regions of the retina, a more detailed analysis is shown in Figure 2A . Here the RGC density of four eccentricities from the optic nerve head can be seen. In this study control animals had a peak RGC density greater than 3000 cells/mm², whereas 5 weeks after immunization no RGC density higher than 2228 cells/mm² was detectable. Five and 6 weeks after immunization, we observed the greatest decrease in RGC density in eccentricities 2 and 3, approximately 2 to 3 mm from the optic nerve head. Differences in RGC density are clearly visible in images taken of the area approximately 2 mm from the optic nerve in all groups (Fig. 2B) . 

Antibody Profiles through Western Blot Analysis

Complex antibody profiles were detected in all serum samples. All animals had elevated anti–HSP27 antibody levels after immunization with HSP27 (Fig. 3A) . 

Figure 3Bshows complex IgG antibody profiles of all four groups. Upregulation of antibody reactivity in the immunized groups (4, 5, 6 weeks) in comparison with the control group could be seen (e.g., 27 and 87 kDa), as well as downregulations (e.g., 15 and 48 kDa). The antigen at 27 kDa was identified as HSP27, and the one at approximately 87 kDa was identified as HSP90 (HSP90; mascot score, 451). Antigen bands corresponding to antibody reactivity at approximately 15 and 47 kDa were identified as hemoglobin subunit alpha (mascot score, 459) and α-enolase (mascot score, 371). 

Figure 4Bshows antibody profiles only for groups 3 at 4 at different time points: before immunization (baseline) and 2, 4, and 6 weeks after immunization. Group 3 showed upregulation at different time points compared with baseline, such as at 27 (HSP27) and 58 kDa. Decreased reactivity after immunization could be seen at 19 and 34 kDa. The antibody bands at approximately 19 and 34 kDa were identified as protein DJ 1 (mascot score, 90) and glyceraldehyde-3-phosphate dehydrogenase (mascot score, 91). 

Through multivariate statistical methods significant differences could be observed for group 1 and controls (P = 0.0063) and at 2 weeks after immunization in group 3 (P = 0.026). Table 1shows the Mahalanobis distances calculated from Western blot analyses for three different time points of group 3; the greatest Mahalanobis distances in group 3 were detected 2 weeks after immunization. These distances are calculated from the analysis of discriminance and provide a measure of how different the two groups were from each other. ⁴² In this study Mahalanobis distances were based on complex antibody profiles detected with Western blot analysis. In terms of Mahalanobis (statistical) distance, group 1 was revealed to be different from the others (distance of group 1 to control group: 6.79). The distance of group 2 to the control group was 2.31, and the distance of group 3 to the control group was 1.57. 

Antibody Profiles through Protein G Bead Approach

Complex antibody profiles from protein chip (Bio-Rad) measurement can be seen in Figure 5 . We confirmed the significant differences in IgG antibody profiles between group 1 and controls (P = 0.09). In accordance with findings on Western blot analysis, group 1 showed a greater Mahalanobis distance from controls (1.55) than group 3 (0.85). 

Figure 4Ashows the antibody profiles of group 3 before and 2, 4, and 6 weeks after immunization. In terms of statistical difference, the antibody profile 4 weeks after immunization showed the greatest distance from the baseline (10.79; P < 0.05; Table 1 ). There was also a significant difference between the antibody profiles 2 weeks after immunization (P = 0.05). 

Canonical roots of the different time points of group 3 can be seen in Figure 6 . Canonical roots are based on discriminance factors. ⁴³ Each spot represents one sample. The closer two points were to each other, the more similar were the Seldi-TOF antibody profiles of these samples. A separation can be seen in antibody profiles detected 2 and 4 weeks after immunization compared to preimmunization samples. 

Discussion

In previous studies we detected specific antibody profiles against retinal antigens in the sera of patients with glaucoma. ^{9 10 40} It is still unclear, however, whether these antibodies play a role in the pathogenesis of the disease. Conceivably they are just a secondary phenomenon rather than the cause of glaucoma. Hence, we used an experimental autoimmune animal model to examine how HSP27-mediated RGC loss affects the systemic antibody pattern. A decrease in RGC density after HSP27 immunization was recently shown by Wax et al. ³⁴ We could confirm that RGC loss can be caused by immunization: 5 and 6 weeks after immunization with HSP27, animals showed a significantly lower density of Brn-3a–labeled RGCs (Figs. 1C 2A) , with the most significant loss in eccentricities approximately 2 to 3 mm from the optic nerve head. Given the stability of IOPs throughout this study and the lack of significant variation, we conclude that RGC loss observed in this animal model is IOP independent. 

In correlation with other studies, immunized animals developed antibodies against HSP27. In guinea pigs with chronic EAE, increased IgG serum antibody levels against myelin basic protein, after immunization with spinal cord homogenate, were detected 21 days after immunization up to a maximum on day 36. ⁴⁴ After 124 days, IgG levels returned to normal. In our animal model, we surveyed the IgG patterns only up to 42 days after immunization. It would be interesting to perform a comparably longer study to determine whether similar phenomena can be observed. 

The observed changes in IgG antibody profiles against retinal antigens in this study were validated with the use of two different techniques: Western blot analysis and Seldi-TOF. Both methods revealed that changes in antibody reactivity were not limited to HSP27. We could identify some antigens with modified antibody reactivity after HSP27 immunization; antibodies against HSP90, a molecular chaperone such as HSP27, were upregulated in immunized animals. Other types of antibody reactivity—such as against glyceraldehyde-3-phosphate dehydrogenase and α-enolase (both glycolytic enzymes), hemoglobin subunit alpha (involved in oxygen transport), and protein DJ-1 (protects against endogenous oxidative stress in neurons ^{45 46} )—were downregulated after immunization. Interestingly, lower antibody reactivity against α-enolase and glyceraldehyde-3-phosphate dehydrogenase was previously detected in the aqueous humor of patients with primary open-angle glaucoma. ⁴⁷ Mutations of protein DJ-1 are associated with Parkinson disease, and loss of protein DJ-1 causes neurodegeneration. ⁴⁸  

Similar to findings in our study, various serum antibody changes were observed in a mouse EAE model. Zephir et al. ⁴⁹ showed changes in antibody patterns against brain homogenate after immunization. This study supports the hypothesis that autoimmune diseases are a shift of the regular autoreactivity in immunized animals. ⁴⁹ Furthermore, antibodies to a myelin/oligodendrocyte glycoprotein could amplify demyelination and modify clinical expression of acute EAE in rats. ⁵⁰ The authors suggested that antibodies directly meet their target structure throughout the CNS, can be presented systemically, and become pathogenic as soon as the blood-brain barrier is opened. ⁵⁰ Our model might be based on a similar mechanism: in addition to the HSP27 antibodies, which are expressed as a consequence of immunization, other types of immunoreactivity have developed (e.g., an increased level of antibodies against HSP90). They might damage the RGCs further or amplify RGC loss and, hence, the severity of the disease. Further studies on the experimental autoimmune glaucoma model are needed to show whether this can be proven. 

Most studies on autoimmune animal models, such as EAE and EAU, are focused on cellular immunity. The role of antibodies in these models has been controversial. Several studies suggest that antibodies are essential for certain aspects of this disease, ^{49 51} whereas some authors propose a minor role for antibodies and a major role for cell-mediated immunity. ⁵² We cannot exclude that the cellular immune component might also play a role in the pathogenesis of our autoimmune glaucoma model, but we propose that disease onset and progression in our model are based on antibody changes. Still, the cellular aspect must be explored in further studies. 

In this study, group 3 showed significant differences in antibody profiles 2 and 4 weeks after immunization (P < 0.5) but not after 6 weeks (Table 1 ; Fig. 6 ). Our findings suggest that most global antibody pattern changes occurred shortly after immunization. However, thus far it is unclear whether this resulted from fading of the antigen injection. Further studies must show how antibody profiles change when antigens are injected more than once to mimic a repeated exposure. Repeated antigen injections might cause progressive change of antibody patterns and more severe damage in animals. Accordingly, we do not know whether patients with glaucoma are constantly exposed to antigens that cause changes of their antibody patterns. Again, this must be tested in our animal model. 

Many open questions are still to be answered. How do anti–HSP27 antibodies enter RGCs and cause apoptosis? How is the cellular component involved in the pathogenesis? This autoimmune glaucoma animal model could help us gain more information on the role of antibodies in glaucoma and how they affect RGC loss. 

 Supported by the Boehringer Ingelheim Foundation and by Grant MAIFOR07 from Johannes Gutenberg University.

 Presented at the annual meetings of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, April 2007, and the Society for Neuroscience, San Diego, California, November 2007.

 Submitted for publication November 12, 2008; revised April 17, 2009; accepted August 3, 2009.

 Disclosure: S.C. Joachim, None; F.H. Grus, None; D. Kraft, None; K. White-Farrar, Alcon Research Ltd. (E); G. Barnes, Alcon Research Ltd. (E); M. Barbeck, None; S. Ghanaati, None; S. Cao, Alcon Research Ltd. (E); B. Li, Alcon Research Ltd. (E); M.B. Wax, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

(A) Intraocular pressure variation in groups 3 and 4: animals immunized with HSP27 (columns) and control animals (zero line). IOP was measured before immunization (baseline) and at 2 and 4 weeks after immunization. (B) Fundus images of the left eye of one animal from group 3. Left: baseline. Middle: two weeks after immunization. Right: four weeks after immunization. (C) Quantification of RGC density (±SE) in all four groups. Control group and animals immunized with HSP27 after 4, 5, and 6 weeks. The difference between the mean cell counts of control and HSP27-immunized groups was significant 5 and 6 weeks after immunization. P = 0.05.

Figure 2.

 

(A) Retinal ganglion cell density of immunized animals after 4, 5, and 6 weeks and control animals were plotted for four different eccentricities from the optic nerve head (1 = closest to optic nerve head). (B) Brn-3a–labeled retinal flatmounts from a control animal (left) and from animals 4, 5, and 6 weeks after immunization with HSP27. Scale bar, 10 μm.

Figure 3.

 

(A) Box plot shows the antibody reactivity at 27 kDa in animals from group 3 at four time points: before immunization (baseline) and 2, 4, and 6 weeks after immunization. (B) Mean antigen-antibody reactivity of all four groups was plotted against the corresponding molecular weight of the retinal antigen. The following groups were included: group 1, killed 4 weeks after immunization with HSP27; group 2, killed 5 weeks after immunization; group 3, killed 6 weeks after immunization; and group 4, control group (no immunization). The x-axis shows the molecular weight (kilodaltons), and the y-axis shows the density of the antigen-antibody-reactivity (U) against the retinal antigens.

Figure 4.

 

(A) Examples of Western blot analyses of animals from group 3. The first two exemplary blots shown were incubated with serum collected before immunization, and the other three were incubated with samples 2 weeks after immunization with HSP27. (B) Mean antigen-antibody reactivity of four different time points of group 3: before immunization with HSP27 and 2, 4, and 6 weeks after immunization. The x-axis shows the molecular weight in kilodaltons, and the y-axis shows the density of the antigen-antibody reactivity (U) from the Western blot analysis. (C) Antibody profile generated by protein array methods. Again, profiles at four different time points of group 3 can be seen. The x-axis shows the molecular weight in kilodaltons and the chip surface, and the y-axis shows the density of the antigen-antibody reactivity (U).

Table 1.

 

P Values and Mahalanobis Distances of Group 3

	Western Blot Analysis			Protein Chip Analysis
Weeks after Immunization	P	Mahalanobis Distance	P	Mahalanobis Distance
2	0.03	3.39	0.01	6.36
4	0.12	2.33	0.0004	10.79
6	0.29	1.57	0.20	2.70

 All values are compared against baseline (before immunization). Significant P values are marked in bold.

Figure 5.

 

Seldi-TOF-MS spectra of several elutes. Top: samples were prepared on CM10 surface chips. Bottom: samples were prepared on H50 surface chips. Left: grayscale images of both surface types. The x-axis shows the molecular weight in daltons, and the y-axis shows the intensity. Right: Seldi-TOF spectra.

Figure 6.

 

Canonical roots of the four different time points of group 3: before immunization with HSP27 (baseline) and 2, 4, and 6 weeks after immunization. These canonical roots were derived from analysis of discriminance and were plotted for each animal. The two-dimensional graph shows the similarity of the groups: the closer the points are to each other, the more similar the antibody patterns were.

The authors thank Nora Wirtz for her assistance in the preparation of this manuscript.