Neurology -- Correspondence for Singer et al., 65 (11) 1701-1707

Correspondence published:

Reply from the authors: Harvey S Singer, John J. Hong, Dustin Y.Yoon, Phillip N. Williams (13 February 2006)

Serum autoantibodies do not differentiate PANDAS and Tourette syndrome from controls: Russell C Dale, Andrew J Church, Paul M Candler, Miles Chapman, Davide Martino, and Gavin Giovannoni (13 February 2006)

Reply from the authors 13 February 2006

Harvey S Singer,
The Johns Hopkins Hospital
600 N. Wolfe Street, Jefferson 124, Baltimore, MD 21287,
John J. Hong, Dustin Y.Yoon, Phillip N. Williams
Send Correspondence to journal:
Re: Reply from the authors

hsinger{at}jhmi.edu Harvey S Singer, et al.

We appreciate the comments of Dale et al and fully recognize their pioneering efforts in this field. We are pleased that no issue is raised about the failure of our ELISA and immunoblotting assays using fresh human postmortem caudate, putamen, and BA10, to distinguish the PANDAS or Tourette syndrome cohorts from controls. [1]
Dale et al have successfully identified several neuronal surface glycolytic enzymes as possible autoantigen targets in post- streptococcal disease including aldolase C, neuron-specific enolase, non- neuronal enolase, and pyruvate kinase M1. [4] As we describe, we used commercially prepared preparations for each of these specific epitopes. Both neuron-specific enolase and non-neuronal enolase were derived from human brain. Since no human preparations are available for pyruvate kinase M1 and aldolase C, we used antigens prepared from rabbit skeletal muscle. These preparations have a 93% and 79% homology, respectively, with their human brain isoforms. If provided, we would be pleased to replicate our studies using recombinant human isoforms as epitopes.
A second issue raised by Dale et al relates to the quantity of commercial protein used in our Western immunoblotting assays. Their concern was that only 1 �g of commercial protein was loaded per gel, an amount substantially less than that used by their group. For the record, we loaded 1 �g per lane not per gel. Preliminary studies using serial dilutions of the candidate antigens indicated that 1 �g per lane was optimal for the assays. Additionally, we used a detection system (electrochemiluminescence) that has a high degree of sensitivity. [5]
Finally, Dale et al noted the high degree of positivity of aldolase C reactivity, at least 91% among patients and controls. In order to determine the quality of signal to all specific antigens, each gel was stripped and reprobed with a commercially purchased monoclonal antibody specific to that antigen. Only bands from serum that overlapped with those from the monoclonal antibody were considered positive.
We believe that there is ongoing controversy over the presence and role of autoantibodies, perhaps due to methodological variations. [5] Before accepting a proposed autoimmune hypothesis, future studies must establish a correlation among antineuronal antibodies, clinical symptomatology, and a streptococcal infection.
Additionally, proponents must confirm the preabsorption of antibodies by Group A -(beta)hemolytic streptococci and show antibody binding to membrane bound glycolytic enzymes on intact neurons. We look forward to possible collaborative efforts with our friends across the pond.
References

References
1. Singer HS, Hong JJ, Yoon DY, Williams PN. Serum autoantibodies do not differentiate PANDAS and Tourette syndrome from controls. Neurology 2005;65:1701-1707
2. Church AJ, Dale RC, Lees AJ, Giovannoni G, Robertson MM. Tourette's syndrome: a cross sectional study to examine the PANDAS hypothesis. J Neurol Neurosurg Psychiatry. 2003 May;74:602-607.
3. Church AJ, Dale RC, Giovannoni G. Anti-basal ganglia antibodies: a possible diagnostic utility in idiopathic movement disorders? Arch Dis Child. 2004 Jul;89:611-614.
4. Dale RC, Candler PM, Church AJ, Wait R, Pocock JM, Giovannoni G. Neuronal surface glycolytic enzymes are candidate autoantigens in post- streptococcal autoimmune CNS disease. J Neuroimmunol 2006 (in press).
5. Rippel CA, Hong JJ, Yoon DY, Williams PN, Singer HS. Methodological factors affect the measurement of anti-basal ganglia antibodies. Ann Clin Lab Sci 2005;35(2).
Disclosure: The authors report no conflicts of interest.

Serum autoantibodies do not differentiate PANDAS and Tourette syndrome from controls 13 February 2006

Russell C Dale,
Neuroinflammation Unit, Institute of Neurology
Ninth Floor, London WC1N 3BG, UNITED KINGDOM,
Andrew J Church, Paul M Candler, Miles Chapman, Davide Martino, and Gavin Giovannoni
Send Correspondence to journal:
Re: Serum autoantibodies do not differentiate PANDAS and Tourette syndrome from controls

r.dale{at}ion.ucl.ac.uk Russell C Dale, et al.

We read the article by Singer et al with interest and would like to raise a number of methodological issues and concerns. [1] The authors found no autoantibody binding against brain homogenates and candidate neuronal autoantigens in PANDAS and Tourette syndrome sera.
Our group has demonstrated that auto-antibodies in PANDAS and Tourette syndrome sera bind to auto-antigens of molecular weight 40, 45 (doublet) and 60kDa. [2,3] We have identified these autoantigens as neuronal isoforms of the glycolytic enzymes aldolase C, neuron-specific enolase and pyruvate kinase M1, and the ubiquitous glycolytic enzyme non-neuronal enolase. We found autoantibodies against these neuronal glycolytic enzymes (NGE) more commonly in patients compared to controls. [4]
We would like to point out the following methodological issues. Singer et al claim to use the human brain autoantigens pyruvate kinase M1 and aldolase C from Santa Cruz Biotehcnology Inc. [1] The antigens produced by this company are rabbit muscle pyruvate kinase (not the human brain M1 isoform), and rabbit muscle aldolase (not the human brain C isoform). An NCBI BLAST search shows that the pyruvate kinase and aldolase available from Santa Cruz Biotechnology have only 93% and 79% homology respectively, with the human brain isoforms. We have also acknowledged this problem and therefore created a recombinant form of human aldolase C and pyruvate kinase M1.
Another methodological concern relates to the quantity of commercial antigen used. Singer et al used 1��g of autoantigen per 2D gel in all of their Western blotting assays. [1] During screening experiments we found small quantities (such as 1��g) were inadequate to derive any meaningful signal in any patients or controls. We subsequently used 20 ��g of pyruvate kinase per 2D gel, and 100��g of aldolase C per 2D gel. [4]
Singer et al found antibody binding against both neuron-specific enolase and non- neuronal enolase in only 0-3% of patients and controls. We believe this may be due to inadequate signal in any patient or control due to inadequate protein load. Likewise Singer et al found positive binding to aldolase in 91-98% of patients and controls possibly related to non- specific background binding. Related to this methodological point, Singer et al did not use a positive or negative control to demonstrate significant or specific signal. [1]
We hope these comments lend some balance to the continuing controversy regarding the role of autoimmunity in post-streptococcal CNS disease.
Disclosure: The authors report no conflicts of interest.