Functional immunomics: Microarray analysis of IgG autoantibody repertoires predicts the future response of mice to induced diabetes
Authors:
Francisco J. Quintana, Peter H. Hagedorn, Gad Elizur, Yifat Merbl, Eytan Domany, Irun R. Cohen
Can be found in: www.pnas.org/content/101/suppl.2/14615.full
INTRODUCTION
One's repertoire of antibodies encodes the history of one's past immunological experience. Can the present autoantibody repertoire be consulted to predict susceptibility to a future disease? In this article, an antigen microarray chip is developed and bioinformatic analysis is used to study a model of type 1 diabetes developing in nonobese diabetic male mice in which the disease was accelerated and synchronized injecting cyclophosphamide at 4 weeks of age. Sera was obtained, the mice were treated to induce cyclophosphamide-accelerated diabetes (CAD), here 9 mice became severely diabetic, whereas 10 mice permanently resisted diabetes.
Male mice of the nonobese diabetic (NOD) group develop type 1 diabetes at a low incidence and late age. However, the start of diabetes can be significantly accelerated and synchronized by exposing NOD mice to cyclophosphamide. So, the CAD model of type 1 diabetes gives an opportunity to test whether the global autoantibody repertoire might reflect susceptibility to CAD in still healthy mice, before the cyclophosphamide insult is administered.
Recently, microarray antigen chips have been used to detect high-titer autoantibodies to known antigens in autoimmune diseases. However, rather than focusing only on known antigens, here some individual immune systems are profiled by their global patterns of autoantibodies free of bias for high-titer reactivities to particular antigens.
MATERIALS AND METHODS
Mice. Male NOD (non-obese diabetic) mice were raised an maintained under pathogen-free conditions. The mice were 4 weeks old at the start of the experiments. Nineteen mice were studied individually.
CAD. Diabetes was accelerated and synchronized by two injections of cyclophosphamide at 4 weeks of age, and again, 1 week later. In the colony, this treatment of NOD males leads of an incidence of diabetes of approximately 50%. The mice developing diabetes go on to die unless they are treated with insulin; those males that do not develop diabetes within 1 month after two injections of cyclophosphamide do not become diabetic later in life.
The next image is a schematic repersentation of the protocol.
The experimental protocol is as follows:
- Numbers refer to age of the mice (in weeks)
- Black vertical lines are serum sample collection (weeks 4 and 9)
- Grey vertical lines are Cyclophosphamide injection (weeks 4 and 5)
- Grey box (at week 6) shows when mice developed diabetes.
- Grey box (between week 11 and 13) shows time of death of untreated diabetic mice.
Diabetes
Blood glucose was measured weekly. A mouse was considered diabetic when
its blood glucose concentration was >13 mM on two
consecutive examinations. Of
the 19 mice treated,
9 developed diabetes, and 10 remained healthy throughout a 2-month
period.
Serum
Serum samples were collected 1 day before the first injection of
cyclophosphamide and 1 month after the second injection.
Antigens
The 266 antigens spotted on the microarray chips include proteins, synthetic peptides, nucleotides, and
phospholipid.
Image and Data Processing.
The pixels that comprised
each spot in the TIFF files and the local background were identified by
using histogram segmentation.
The intensity of each spot and its local
background were calculated as the mean of the corresponding pixel
intensities. None
of the spots containing antigens showed
saturation. Technically faulty spots were identified by visual
inspection and were
removed from the data set. For each spot,
the local background intensity was subtracted from the spot intensity.
Spots with
negative intensities were removed from the
data set. A log-base-2 transformation of the intensities resulted in
reasonably
constant variability at all intensity
levels. The log intensity of each antigen was calculated as the mean of
the log intensities
of the replicates on each slide. The
coefficient of variability between replicates on each array was ≈30%.
RESULTS
Selection of Informative Antigens for Pre-CAD Mice
There have been reported previously that coupled two-way clustering (CTWC) could be used to successfully separate human subjects
already diabetic from healthy persons.
How ever different
approach was taken. Based on the sera taken before
cyclophosphamide treatment, there have been listed the 27 antigens that separated best between the sera
of the 10 mice that later resisted the induction of
CAD, and of the 9 mice that later
developed CAD.
Figure A |
Above figure Left is the two-way SPC of
these antigens. The filled rectangles
at the top of the box are the mice susceptible to CAD induction; empty rectangles are mice
resistant to CAD induction.
The 27 antigens
are clustered at the rows and are
identified by number.
It can be seen that all 9 mice that were found later to be susceptible
to CAD could be separated from 8 of the 10 mice
that were later found to resist CAD.
Figure B |
Figure C |
Selection of Informative Antigens for Post-CAD Mice.
Then, 27 antigens were used effective in pre-CAD clustering to analyze
the patterns of IgG antibodies developing in the
diabetic and healthy mice post-CAD. These 27 antigens failed to discriminate between the two
groups of mice. A third set of 27
antigens was generated by performing the
Wilcoxon rank-sum test on the ratios by which each antigen changed
post-CAD/pre-CAD.
The ratios provide information on
reactivity changes toward the antigen.
Figures B and C show that previously mentioned antigens could indeed separate between the healthy and diabetic mice post-CAD. Thus, the IgG repertoires of the pre- and post-CAD groups of
healthy and sick mice could be clustered, but the
informative patterns of reactivity
required modified sets of antigens to develop discriminating patterns.
It can be seen that some of the antigens from the set of pre-CAD antigens were also present in the post-CAD set, or in the set of antigens determined from the pre-CAD/post-CAD ratios. For example, three of the pre-CAD antigen reactivities were also prominent post-CAD. The shared and distinct antigens are shown as a Venn diagram (figure above) for the overlap between the three lists of antigens mentioned previously.
DISCUSSION
The antigen microarray chip described in this paper
required much preliminary work to obtain consistent results.
Patterns of IgM antibodies were analyzed before and after CAD.
The article show that the patterns
of IgG antibodies expressed pre-CAD in male NOD mice can mark
susceptibility or resistance
to CAD induced later. There were also found
patterns of IgG antibodies characteristic of healthy or diabetic mice
post-CAD, but these
patterns required sets of antigens that
differed from the informative pre-CAD set. Thus, IgG reactivities to some antigens may mark future susceptibility
to CAD, but not CAD itself, once the disease emerges,
and conversely, some IgG reactivities may
mark the disease but not the susceptibility. Hence, prediction of future
disease and diagnosis of present
disease can depend on different data sets of information, at least in the CAD
model.
Another notable finding was that health, both pre- and post-CAD, was associated with relatively high IgG autoreactivity to
self-antigens, to which the susceptible mice were low IgG responders. Thus, some types of active autoimmunity may actually protect against autoimmune disease.
However, environmental factors can also prevent the development of type 1 diabetes. Stimulation of the NOD mouse immune system
by infection, by vaccination with microbial antigens, or by treatment with ligands that activate innate immune receptors, can prevent diabetes. Type 1 diabetes appears
in very young people,
so critical aspects of autoimmune organization must occur fairly early
in one's lifetime. The results of this bioinformatic
study would suggest that, some organized patterns of
IgG autoantibodies
are shared by groups of individuals, at
least among NOD mice.
The bioinformatic analysis described here relates to two separate, but linked issues: predictive medicine by means of functional
immunomics and the biological meaning of the autoimmune repertoire.
The present study investigated
patterns of antibodies, but not their function in the disease process.
Nevertheless, the list
of informative antigens may be connected
to other observations regarding the pathophysiology of type 1 diabetes.
The mammalian immune system, in addition to its well studied role in defending the body against foreign invaders, is now understood to be heavily involved in maintaining the integrity of the body from within; immune system cells and molecules, which comprise the inflammatory response, are key factors in wound healing, neuroprotection, connective tissue formation, angiogenesis, tissue morphology and regeneration, and waste disposal. To dispense reparative inflammation at the right sites and occasions, the immune system has to assess the state of the body on the fly. In this respect, the immune system acts as it were the body's onboard bioinformatic computer. If so, predictive medicine would do well to mine this immune information, as this article suggests it might.
Referencias
Functional immunomics: Microarray analysis of IgG autoantibody repertoires predicts the future response of mice to induced
diabetes;
Francisco J. Quintana, Peter H. Hagedorn, Gad Elizur, Yifat Merbl, Eytan Domany, Irun R. Cohen;
"This paper results from the Arthur M. Sackler Colloquium of the National
Academy of Sciences, “Therapeutic Vaccines: Realities
of Today and Hopes for Tomorrow,”
held April 1–3, 2004, at the National Academy of Sciences in Washington,
DC."
Link to the article: http://www.pnas.org/content/101/suppl.2/14615.full
[Consultado el 27 de Mayo de 2013]
El formato de la referencia les sigue dando broncas :P Van 4 pts extra en la última tarea.
ResponderEliminar