- Email: [email protected]
HuR, a novel target of anti-Hu antibodies, is expressed in non-neural tissues
Journal of Neuroimmunology 92 Ž1998. 152–159
HuR, a novel target of anti-Hu antibodies, is expressed in non-neural tissues L. Burt Nabors a , Henry M. Furneaux b, Peter H. King b
a,)
a Department of Neurology, UniÕersity of Alabama, Birmingham, AL 35295, USA Program in Molecular Pharmacology and Therapeutics, Memorial Sloan-Kettering Cancer Center, New York, NY 10021, USA
Received 13 May 1998; revised 4 August 1998; accepted 11 August 1998
Abstract Paraneoplastic encephalomyelitis ŽPEM. is characterized by a diverse set of clinical signs that are limited to the nervous system. The serologic hallmark of PEM is the presence of circulating autoantibodies, collectively referred to as ‘anti-Hu,’ which immunoreact specifically with members of the Elav protein family. Until recently, the ELAV antigens were only detected in neurons, thus strongly supporting a role for anti-Hu antibodies in the selective neural tissue injury in PEM. The identification of HuR, however, a new member with a broad, non-neural pattern of RNA expression, raises several fundamental questions regarding PEM. First, why are non-neural tissues spared in PEM? Second, why is PEM predominantly associated with neuroendocrine tumors? To begin addressing these questions, we sought to determine whether the antibody response to HuR differs from the neural-specific counterparts in patients with PEM, and to characterize the protein expression pattern of this novel antigen in peripheral tissues and tumors. Using sera from 11 patients with Hu-positive PEM, we found that the majority of samples Ž73%. were weakly or non-reactive for recombinant HuR on Western blot, in contrast to consistently strong immunoreactivity with the neural-specific members HuD and Hel-N1. We also demonstrate that HuR is expressed at the protein level in both non-neural tissues and non-neuroendocrine tumors. These findings suggest that immunoreactive differences among Elav family members may contribute to the neural-restrictive pattern of tissue injury in patients with PEM. q 1998 Elsevier Science B.V. All rights reserved. Keywords: Paraneoplastic disease; Autoantibodies; RNA-binding proteins
1. Introduction Paraneoplastic encephalomyelitis ŽPEM. is a remote effect of cancer and is characterized by a wide range of neurological signs and symptoms, including sensory neuropathy, ataxia, dementia, and deafness ŽDalmau et al., 1992b; Lucchinetti et al., 1998.. The disease is associated with the production of high-titer anti-Hu autoantibodies Žalso referred to as type I anti-nuclear autoantibodies. which are directed against the Elav family of RNA-binding proteins ŽSzabo et al., 1991; Kim and Baker, 1993; Dropcho and King, 1994; Lucchinetti et al., 1998.. This family is highly conserved over evolution, and human members that have been characterized include HuC ŽSzabo et al., 1991., HuD ŽSzabo et al., 1991., and Hel-N1 ŽDropcho ) Corresponding author. 1235 Jefferson Tower, 625 S. 19th St., Birmingham, AL 35294-0007, USA. Tel.: q1 205 9758116; fax: q1 205 9340928; e-mail: [email protected]
and King, 1994; King et al., 1994.. These proteins are closely linked at two levels. First, they possess three RNA-binding domains that share a high degree of sequence homology to each other and to the Drosophila prototype, ELAV ŽRobinow et al., 1988; Yao et al., 1993.. Second, family members are normally restricted to neurons ŽRobinow et al., 1988; Szabo et al., 1991; King et al., 1994.. The aberrant expression of this family in small-cell lung cancer ŽSCLC. ŽDalmau et al., 1992a; King and Dropcho, 1996; King, 1997., which is the most common tumor in patients with PEM, has been postulated to trigger the anti-Hu immunological response, resulting in selective neural tissue injury ŽDalmau et al., 1992a,b; Dropcho, 1995.. Recently, another family member, HuR, was cloned and found to have a more ubiquitous pattern of RNA expression, including many non-neural tissues ŽMa et al., 1996.. This member also contains three RNA-binding domains which are highly homologous to the neural-restricted counterparts. The absence of non-neural tissue
0165-5728r98r$ - see front matter q 1998 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 5 - 5 7 2 8 Ž 9 8 . 0 0 1 9 6 - 9
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
injury in patients with PEM, and the consistent association of PEM with neurally differentiated tumors, however, suggests that HuR does not participate in the immunopathogenesis of PEM. HuR, for example, may be expressed only at the RNA level in peripheral tissues, or it may be less antigenic than its neural counterparts, despite the overall sequence homology. As an initial step in addressing these possibilities, we have analyzed the immunoreactive patterns of patients with PEM to HuR, Hel-N1, and HuD. We have also utilized the serum of one patient with high-titer anti-HuR antibodies to characterize the protein expression pattern of HuR in normal peripheral tissue as well as lung tumors that are not associated with PEM.
2. Materials and methods 2.1. Serum and tissue samples Serum samples were obtained from 11 patients with anti-Hu PEM who were previously diagnosed by established techniques. Controls consisted of cancer patients without PEM. Human tissues Žneoplastic and normal. were
153
obtained from the University of Alabama Tissue Procurement Office. A HeLa Žepitheloid cancer. cell line was obtained from the American Type Culture Collection ŽRockville, MD.. 2.2. Western blotting Recombinant GST-HuR, GST-HuD, and g10-Hel-N1 fusion proteins were cloned and expressed as previously described ŽDropcho and King, 1994; King et al., 1994; Chung et al., 1996; Ma et al., 1996.. Cellular proteins were extracted from tissues using standard techniques ŽSambrook et al., 1989; Andrews and Faller, 1991.. All proteins were quantitated using a commercial kit ŽBio-Rad, Richmond, CA.. Recombinant proteins Ž200 ng. and cellular proteins Ž60 mg. were electrophoresed, blotted and probed as described previously except that blotto was used as the blocking reagent ŽSambrook et al., 1989; Dropcho and King, 1994.. All sera were diluted at 1:2000 in 1 = blotto. The degree of immunoreactivity was qualitatively determined by visual inspection of band intensity. Preabsorption studies were done by incubating the serum with excess recombinant HuD for 12 h at 48 Ž2 mg of HuD in a 1:2000
Fig. 1. ŽA. Western blot of recombinant Hu antigens HuR, HuD, and Hel-N1 Ž200 ng. using sera Ždiluted 1:2000. from three patients with Hu-positive PEM. ŽB. Summary of the immunoreactivity of 11 Hu-positive PEM patients against HuR, Hel-N1, and HuD Ž1:2000 dilution.. The degree of immunoreactivity was estimated by visual inspection as shown in the Western blot.
154
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
dilution of serum. prior to immunoblot assay. To assess tissue distribution of HuR, IgG was purified from patient 1 using a protein A-sepharose column ŽBio-Rad. and biotinylated by a standard technique ŽHarlow and Lane, 1988.. Blots were probed with this purified IgG Ž1:100 dilution. and were developed with the Vectastain ABC kit using a DAB substrate ŽVector, Burlingame, CA..
A 10-mm frozen section of human frontal cortex tissue was thaw-mounted onto a subbed slide and fixed in cold acetone for 10 min. The section was blocked in 1 = blotto and then probed overnight at 48C with the serum of patient 1 Ždiluted at 1:2000.. The section was washed, and then probed with alkaline phosphatase-conjugated streptavidin according to the manufacturer’s specification ŽPierce, Rockford, IL.. The blot was developed with chromogen fast red ŽBiogenex, San Raymond, CA..
This fragment represents nucleotides a159–416 of the GenBank sequence U38175. The glyceraldehyde-3 phosphate dehydrogenase ŽGAPDH. positive control and HelN1 templates were used as described previously ŽKing, 1997.. All templates were linearized with appropriate restriction enzymes to generate antisense riboprobes. Templates were transcribed with 20 units of either T7 or T3 RNA polymerase ŽPromega, Madison, WI.. All transcription reactions were carried out in buffer provided by the manufacturer along with 60 mCi of w a-32 Px UTP ŽAmersham, Arlington Heights, IL., CTP, ATP, GTP, 10 mM DTT, and 20 units of RNasin ŽPromega.. Reactions were carried out at 378C for 1.5 h, and probes were subsequently purified on a polyacrylamide gel. Sizes of protected probe are: HuR Ž257 nt., Hel-N1 Ž197 nt. and GAPDH Ž130 nt.. RNA hybridizations and RNase digestions were done as described previously ŽBall and King, 1997; King, 1997.. Ten micrograms of total RNA was used in each hybridization.
2.4. RNA protection assay
2.5. Case history
The HuR template was obtained by subcloning a Pst1rSac1 digestion fragment of the HuR cDNA ŽMa et al., 1996. into Bluescript SK ŽStratagene, La Jolla, CA..
Patient 1 was a 70-year-old white female who presented with a two-month history of rapidly progressive dementia, gait ataxia, and lower extremity weakness. On mental
2.3. Immunocytochemistry
Fig. 2. ŽA. MRI results for patient 1 indicating proton-density lesions in the basal ganglia bilaterally Žleft panel. and diffuse enhancement of the cerebellum with gadolinium contrast Žright panel.. ŽB. Immunocytochemistry of human frontal cortex using biotinylated IgG from patient 1 Ždiluted 1:100.. Typical nuclear staining is seen in cortical neurons Žarrows.. Magnification: 20 = .
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
155
status evaluation, the patient was initially alert, but over the hospital course became progressively unresponsive. Cranial nerve examination showed dysarthric speech, bilateral facial weakness, diminished vertical gaze, and impaired saccades. Motor examination showed diffuse weakness and hypotonia in the lower extremities, cogwheel rigidity, and bradykinesia. There was dysmetria in the upper limbs bilaterally. Coordination and gait could not be tested because of lower extremity weakness. Laboratory evaluation was significant for CSF abnormalities including elevated protein at 158 mgrdL Žnormal, 12–60 mgrdL. and 420 WBCsrml Žnormal, 0–5 WBCsrml. which were all lymphocytes. A brain MRI showed T1, T2, and proton density lesions in the striatum bilaterally ŽFig. 2A, left panel. as well as diffuse enhancement of the cerebellum with gadolinium contrast ŽFig. 2A, right panel.. A chest CT scan revealed a 2-cm subcarinal mass. Needle biopsy of this lesion was non-diagnostic on two occasions.
indicated a range of immunoreactivity, from strong ŽFig. 1A, patient 1., to weak ŽFig. 1A, patient 2. to absent ŽFig. 1A, patient 3.. In the majority Ž73%. of samples, however, the immunoreactivity was either weak or absent ŽFig. 1B.. One of seven control patients with cancer, but without Hu-positive PEM, showed weak immunoreactivity to HuR Žnot shown.. With the neural-restricted homologue, HuD, on the other hand, there was strong immunoreactivity in 9 of 11 patients, and moderate in the other two. Although Hel-N1 also had overall strong immunoreactive patterns, there was some variability with 3 of 11 patients being weakly reactive ŽFig. 1B, patients 1, 2, and 8.. Those same patients, however, demonstrated moderate-to-strong patterns with HuD. Despite the overall homology, these findings indicate significant immunoreactive differences of individual Elav family members in patients with PEM.
3. Results
With the serum of patient 1, a band for HuR could be detected by Western blot even at a dilution of 1:20,000 Žnot shown.. This patient presented with a clinical syndrome of rapidly progressive dementia, cerebellar ataxia, and extrapyramidal signs in conjunction with neural imaging demonstrating signal abnormalities in the basal ganglia and cerebellum ŽFig. 2A.. The imaging abnormalities in-
3.1. ImmunoreactiÕity patterns of HuR with PEM sera The sera from 11 patients with Hu-positive PEM were analyzed by Western blot for immunoreactivity to recombinant HuR, HuD, and Hel-N1. The results for HuR
3.2. Subspecificity of antibodies to HuR in a patient with PEM
Fig. 3. ŽA. Western blot analysis of recombinant Hu antigens using serum from patient 1 ŽFig. 1.. The left panel represents untreated serum and the right panel represents serum which was preabsorbed with excess recombinant HuD protein. ŽB. Western blot of various tissues using biotinylated IgG from patient 1. Lane 1, HeLa nuclear extract; 2, squamous cell carcinoma Žlung.; 3, adenocarcinoma Žlung.; 4, adenocarcinoma Žlung.; 5, squamous cell carcinoma Žlung.; 6, muscle; 7, liver; 8, kidney; 9, heart.
156
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
cluded increased proton density signal, predominantly in the basal ganglia and contrast enhancement in the cerebellum. This pattern is consistent with an inflammatory response and has been reported by others in patients with limbic encephalitis ŽDirr et al., 1990; Dalmau et al., 1992b.. The serum demonstrated typical anti-Hu patterns of immunoreactivity by immunocytochemistry. Fig. 2A shows a section of human frontal cortex which was incubated with IgG purified from this patient’s serum. Strong nuclear staining can be seen, and to a lesser extent, staining of the cell body. The nucleoli were not stained which is typical for anti-Hu immunoreactivity ŽGraus et al., 1985; Anderson et al., 1988.. This patient’s serum was further analyzed by preabsorbing the serum with excess HuD ŽFig. 3A.. The panel on the left shows the baseline immunoreactivity with the three recombinant proteins. The panel on the right shows loss of signal for HuD and Hel-N1 after preabsorbing the serum with HuD. The band for HuR, however, remained strong, indicating the presence of a subset of antibodies specific for this protein. Since both recombinant HuR and HuD were synthesized as fusion proteins with glutathione-S-transferase, it is highly likely that the anti-
body subset was directed toward HuR rather than the fusion tag. 3.3. Tissue distribution of HuR Although the HuR mRNA has been detected in nonneural tissues by reverse-transcription PCR, the protein expression pattern of HuR has not been characterized. One potential explanation for the absence of peripheral tissue injury in PEM, for example, could be post-transcriptional regulation of HuR mRNA such that there is minimal or no expression at the protein level. To address that possibility, we analyzed a panel of non-neural tissue Žincluding liver, heart, kidney, and muscle. and four non-neuroendocrine lung tumor extracts with purified IgG from patient 1 and found two immunoreactive bands, approximately 32–35 kDa in size in all tissues ŽFig. 3B.. One of these bands is most likely HuR as this was the size recently observed by Myer et al. Ž1997. in Hela cells. The identity of the second band is unknown; however, it may represent either a splice variant of HuR or possibly another related ELAV family member. Included in this blot was a nuclear extract sample
Fig. 4. RNase protection assay with HuR and Hel-N1 riboprobes using total RNA from non-neuroendocrine lung tumor Ža. and non-neural tissues Žb.. The upper bands represent the protected bands for HuR and Hel-N1, and the lower band in all panels represents the internal GAPDH control Žas indicated by arrows.. The first lane in each panel shows the unprotected probes for the test construct Župper band. and GAPDH Žlower band.. RNA samples are as follows: ŽA. lane 1, squamous cell carcinoma Žlung.; 2, adenocarcinoma Žlung.; 3, adenocarcinoma Žlung.; 4, squamous cell carcinoma Žlung., P, riboprobes, and ŽB. lane 1, HeLa; 2, muscle; 3, liver; 4, kidney; 5, heart; 6, brain; q, NCI N417, P, riboprobes.
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
157
Fig. 5. Comparison of amino acid identity among HuR, Hel-N1, and HuD. Above is a schematic diagram which depicts the highly conserved structure of the Elav protein family. Below is a comparison of amino acid identity of HuR and the neural-restricted homologues Hel-N1 and HuD within the identified regions in the diagram. RRM, RNA recognition motif; aa, amino acids.
from HeLa cells as a positive control ŽFig. 3B, lane 1.. We have since expressed HuR without the large fusion tag, and have confirmed this smaller size Žnot shown.. Expression of HuR could also be seen in all four non-neuroendocrine lung tumor samples tested ŽFig. 3B, lanes 2–5.. Although the immunoblot was not quantitative, no amplification of signal was necessary to visualize the bands, indicating that the protein was not rare. The non-neuroendocrine lung tumors, such as the ones analyzed here, included squamous cell carcinoma and adenocarcinoma, both of which are not associated with PEM ŽDalmau et al., 1992a.. 3.4. RNA expression patterns of HuR To support the conclusion that the band visualized on the Western blot in Fig. 3 represents HuR, the normal and neoplastic tissues were analyzed for RNA expression of elaÕ family members. The previous RNA study with HuR was based on reverse-transcription PCR which is not quantitative ŽMa et al., 1996.. Here, we utilized a sensitive RNase protection assay to analyze these tissues for HuR expression. As shown in Fig. 4, protected bands for HuR were detected with all tissues including the brain. The lower band in these blots represents the protected band for GAPDH which was used as a control for the RNA. The HuR probe included nucleotide sequences from the Nterminus, where there is no homology with other family members Žsee Fig. 5., thereby eliminating the possibility of cross-hybridization. With the exception of the brain and the small-cell lung carcinoma line NCI N417, these same tissues did not express Hel-N1 ŽFig. 4, right panels., or as previously published, HuD ŽSzabo et al., 1991; King,
1997.. This RNA data therefore supports the conclusion that the immunoreactive bands seen on Western blot represent HuR and not the neural-restricted ELAV members.
4. Discussion Although not proven, the pathogenic role of anti-Hu antibodies in PEM is supported by several lines of evidence. Firstly, patients with PEM consistently have high titers of anti-Hu antibodies ŽDalmau et al., 1990.. Secondly, until recently, the Elav family was considered restricted to neurons which parallels the pattern of selective neural tissue injury in PEM ŽSzabo et al., 1991; Dalmau et al., 1992a,b; King et al., 1994.. Moreover, the majority of tumors such as small-cell lung carcinoma ŽSCLC., which are linked to PEM, have neuroendocrine differentiation and express these antigens ŽDalmau et al., 1992a.. The recent cloning of HuR in humans, however, indicates that the ELAV family is not restricted to neurons as previously thought ŽMa et al., 1996.. Here we have confirmed the presence of HuR transcripts in non-neural tissues including lung, muscle, liver, heart, and kidney. The RNase protection assay, unlike reverse-transcription PCR Žas used by Ma et al., 1996., also permits quantitation of transcripts and indicated that HuR mRNA is relatively abundant in these tissues as protected bands could be detected with only 10 mg of total RNA. We also analyzed four non-SCLC lung tumors that are not associated with PEM, and found universal expression of HuR as well ŽFig. 4A.. The neural-restricted Hel-N1 transcripts were not detected. The
158
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
absence of HuD mRNA in non-SCLC tumors has been previously described ŽKing, 1997.. The lack of association of PEM with non-SCLC tumors, and the lack of peripheral tissue damage in PEM, suggest that HuR does not participate in the immunopathogenesis of PEM. In this study, we have addressed two possible explanations. The first possibility is that there may be immunoreactive or antigenic differences between HuR and the neural-restricted counterparts in patients with PEM. We therefore analyzed and compared the immunoreactive patterns of HuR, Hel-N1, and HuD in patients with Hupositive PEM. While HuD consistently showed strong immunoreactive patterns, HuR reactivity was weak to absent in 73% of patients. Hel-N1 reactivity, although more variable, was also moderate-to-strong overall. These findings indicate that there were substantial differences in immunoreactive patterns between HuR and the neural-restricted counterparts in patients with PEM. The homology of this protein family resides in the RNA-binding domains Žsee Fig. 5., and these regions may account for the observed cross-reactivity among family members ŽKim and Baker, 1993.. Significant divergence of sequence, however, occurs in the N-terminus and the bridging regions where there is less than 20% and 50% amino acid identity, respectively. Epitope mapping of HuD indicates that the immunodominant regions reside upstream from the bridging region and the third RRM ŽManley et al., 1995; Graus et al., 1998., suggesting that the N-terminus may contribute to the immunoreactive differences among family members ŽFig. 5.. Interestingly, we identified one patient Žpatient 1. with typical clinical and serological signs of PEM who had a high titer of antibodies against HuR. We were able to abrogate the immunoreactivity of that patient’s serum to HuD and Hel-N1 by preabsorbing the serum with recombinant HuD. The immunoreactivity with HuR persisted, however, indicating that the patient harbored a subset of antibodies specific for this member. Even between Hel-N1 and HuD, there were differences in immunoreactive patterns among several patients ŽFig. 1B, patients 1, 2, and 8.. These immunoreactive patterns may contribute to the neural-specific tissue injury observed in PEM. To address the possibility that the HuR protein product is not expressed at the protein level in peripheral tissues and non-neuroendocrine tumors, we utilized purified IgG from patient 1 in an immunoblot analysis of these tissues. We detected two immunoreactive bands of 32–35 kDa in the same tissues in which we demonstrated HuR mRNA expression. This molecular mass is consistent with that of HuR described by Myer et al. Ž1997. in HeLa cells. Since the RNase protection assay was negative for Hel-N1 and HuD ŽKing, 1997., this band very likely represents HuR. The weaker reactivity of anti-Hu serum against HuR could potentially explain why previous studies have not detected HuR expression in non-neural tissues ŽDalmau et al., 1992a; Okano and Darnell, 1997.. The absence of peripheral
tissue injury in PEM, and the lack of association of PEM with non-neuroendocrine tumors, therefore, cannot be explained by post-transcriptional down-regulation of HuR protein expression. The cellular localization of HuR, on the other hand, may differ from the neural-restricted homologues, and therefore not be ‘presented’ to the immune system. There is now evidence, for example, that some Hu antigens are expressed on the cell surface of neuroendocrine tumors such as SCLC or neuroblastoma ŽTora et al., 1997.. We did not address that possibility in this study. 5. Conclusion In summary, we have found a substantially weaker pattern of HuR immunoreactivity in patients with Hu-positive PEM compared to that of the neural-restricted homologues HuD and Hel-N1. This finding, coupled with the finding of ubiquitous HuR protein expression, suggests that immunoreactive differences among family members may in part explain the neural-restrictive pattern of tissue injury in patients with PEM. Acknowledgements This study was supported by NIH grant 1 K08NS01587-05 ŽPHK., RPG-97-111-01-CCE from the American Cancer Society ŽPHK. and NS29682 ŽHMF.. References Anderson, N.E., Rosenblum, M.K., Graus, F., Wiley, R.G., Posner, J.B., 1988. Autoantibodies in paraneoplastic syndromes associated with small-cell lung cancer. Neurology 38, 1391–1398. Andrews, N.C., Faller, D.V., 1991. A rapid micropreparation technique for extraction of DNA-binding proteins from limiting numbers of mammalian cells. Nucleic Acids Res. 19, 2499. Ball, N.S., King, P.H., 1997. Neuron-specific Hel-N1 and HuD as novel molecular markers of neuroblastoma: a correlation of HuD messenger RNA levels with favorable prognosis. Clin. Cancer Res. 3, 1859–1865. Chung, S., Jiang, L., Cheng, S., Furneaux, H., 1996. Purification and properties of HuD, a neuronal RNA-binding protein. J. Biol. Chem. 271, 11518–11524. Dalmau, J., Furneaux, H.M., Gralla, R.J., Kris, M.G., Posner, J.B., 1990. Detection of the anti-Hu antibody in the serum of patients with small cell lung cancer—a quantitative Western blot analysis. Ann. Neurol. 27, 544–552. Dalmau, J., Furneaux, H.M., Cordon-Cardo, C., Posner, J.B., 1992a. The expression of the Hu Žparaneoplastic encephalomyelitisrsensory neuronopathy. antigen in human normal and tumor tissues. Am. J. Pathol. 141, 881–886. Dalmau, J., Graus, F., Rosenblum, M.K., Posner, J.B., 1992b. Anti-Huassociated paraneoplastic encephalomyelitisrsensory neuronopathy. A clinical study of 71 patients. Medicine 71, 59–72. Dirr, L., Lester, A., Donofrio, P., Smith, M., 1990. Evolution of brain MRI abnormalities in limbic encephalitis. Neurology 40, 1304–1306. Dropcho, E.J., King, P.H., 1994. Autoantibodies against the Hel-N1 RNA-binding protein among patients with lung carcinoma: an association with type I anti-neuronal nuclear antibodies. Ann. Neurol. 36, 200–205.
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159 Dropcho, E.J., 1995. Autoimmune central nervous system paraneoplastic disorders: mechanisms, diagnosis, and therapeutic options. Ann. Neurol. 37 ŽS1., S102–S113. Graus, F., Cordon-Cardo, C., Posner, J.B., 1985. Neuronal antinuclear antibody in sensory neuronopathy from lung cancer. Neurology 35, 538–543. Graus, Y.F., Verschuuren, J.J., Degenhardt, A., van Breda Vriesman, P.J., De Baets, M.H., Posner, J.B., Burton, D.R., Dalmau, J., 1998. Selection of recombinant anti-HuD Fab fragments from a phage display antibody library of a lung cancer patient with paraneoplastic encephalomyelitis. J. Neuroimmunol. 82, 200–209. Harlow, E., Lane, D., 1988. Antibodies: A laboratory manual. Cold Spring Harbor: Cold Spring Harbor Laboratory. Kim, Y.S., Baker, B.S., 1993. The Drosophila gene rbp9 encodes a protein that is a member of a conserved group of putative RNA binding proteins that are nervous system-specific in both flies and humans. J. Neurosci. 13, 1045–1056. King, P.H., Levine, T.D., Fremeau, R.T., Keene, J.D., 1994. Mammalian homologs of Drosophila ELAV localized to a neuronal subset can X bind in vitro to the 3 UTR of mRNA encoding the Id transcriptional repressor. J. Neurosci. 14, 1943–1952. King, P.H., Dropcho, E.J., 1996. Expression of Hel-N1 and Hel-N2 in small-cell lung carcinoma. Ann. Neurol. 39, 679–681. King, P.H., 1997. Differential expression of the neuroendocrine genes Hel-N1 and HuD in small-cell lung carcinoma: evidence for downregulation of HuD in the variant phenotype. Int. J. Cancer 74, 378–382. Lucchinetti, C.F., Kimmel, D.W., Lennon, V., 1998. Paraneoplastic oncologic profiles of patients seropositive for type I anti-neuronal nuclear autoantibodies. Neurology 50, 652–657.
159
Ma, W., Cheng, S., Campbell, C., Wright, A., Furneaux, H., 1996. Cloning and characterization of HuR, an ubiquitously expressed Elav-like protein. J. Biol. Chem. 271, 8144–8151. Manley, G.T., Smitt, P.S., Dalmau, J., Posner, J.B., 1995. Hu antigens: reactivity with Hu antibodies, tumor expression, and major immunogenic sites. Ann. Neurol. 38, 102–110. Myer, V.E., Fan, X.C., Steitz, J.A., 1997. Identification of HuR as a protein implicated in AUUUA-mediated mRNA decay. EMBO J. 16, 2130–2139. Okano, H.J., Darnell, R.B., 1997. A hierarchy of Hu RNA-binding proteins in developing and adult neurons. J. Neurosci. 17, 3024–3037. Robinow, S., Campos, A., Yao, K., White, K., 1988. The elaÕ gene product of Drosophila, required in neurons, has three RNP consensus motifs. Science 242, 1570–1572. Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular cloning: A laboratory manual Ž2nd Edn... Cold Spring Harbor: Cold Harbor Laboratory. Szabo, A., Dalmau, J., Manley, G., Rosenfeld, M., Wong, E., Henson, J., Posner, J.B., Furneaux, H.M., 1991. HuD, a paraneoplastic encephalomyelitis antigen contains RNA binding domains and is homologous to elaÕ and sex-lethal. Cell 67, 325–333. Tora, M., Graus, F., de Bolos, C., Real, F.X., 1997. Cell surface expression of paraneoplastic encephalomyelitisrsensory neuronopathy-associated Hu antigens in small-cell lung cancers and neuroblastomas. Neurology 48, 735–741. Yao, K., Samson, M., Reeves, R., White, K., 1993. Gene elaÕ of Drosophila melanogaster: a prototype for neuronal-specific RNA binding protein gene family that is conserved in flies and humans. J. Neurobiol. 24, 723–739.
HuR, a novel target of anti-Hu antibodies, is expressed in non-neural tissues L. Burt Nabors a , Henry M. Furneaux b, Peter H. King b
a,)
a Department of Neurology, UniÕersity of Alabama, Birmingham, AL 35295, USA Program in Molecular Pharmacology and Therapeutics, Memorial Sloan-Kettering Cancer Center, New York, NY 10021, USA
Received 13 May 1998; revised 4 August 1998; accepted 11 August 1998
Abstract Paraneoplastic encephalomyelitis ŽPEM. is characterized by a diverse set of clinical signs that are limited to the nervous system. The serologic hallmark of PEM is the presence of circulating autoantibodies, collectively referred to as ‘anti-Hu,’ which immunoreact specifically with members of the Elav protein family. Until recently, the ELAV antigens were only detected in neurons, thus strongly supporting a role for anti-Hu antibodies in the selective neural tissue injury in PEM. The identification of HuR, however, a new member with a broad, non-neural pattern of RNA expression, raises several fundamental questions regarding PEM. First, why are non-neural tissues spared in PEM? Second, why is PEM predominantly associated with neuroendocrine tumors? To begin addressing these questions, we sought to determine whether the antibody response to HuR differs from the neural-specific counterparts in patients with PEM, and to characterize the protein expression pattern of this novel antigen in peripheral tissues and tumors. Using sera from 11 patients with Hu-positive PEM, we found that the majority of samples Ž73%. were weakly or non-reactive for recombinant HuR on Western blot, in contrast to consistently strong immunoreactivity with the neural-specific members HuD and Hel-N1. We also demonstrate that HuR is expressed at the protein level in both non-neural tissues and non-neuroendocrine tumors. These findings suggest that immunoreactive differences among Elav family members may contribute to the neural-restrictive pattern of tissue injury in patients with PEM. q 1998 Elsevier Science B.V. All rights reserved. Keywords: Paraneoplastic disease; Autoantibodies; RNA-binding proteins
1. Introduction Paraneoplastic encephalomyelitis ŽPEM. is a remote effect of cancer and is characterized by a wide range of neurological signs and symptoms, including sensory neuropathy, ataxia, dementia, and deafness ŽDalmau et al., 1992b; Lucchinetti et al., 1998.. The disease is associated with the production of high-titer anti-Hu autoantibodies Žalso referred to as type I anti-nuclear autoantibodies. which are directed against the Elav family of RNA-binding proteins ŽSzabo et al., 1991; Kim and Baker, 1993; Dropcho and King, 1994; Lucchinetti et al., 1998.. This family is highly conserved over evolution, and human members that have been characterized include HuC ŽSzabo et al., 1991., HuD ŽSzabo et al., 1991., and Hel-N1 ŽDropcho ) Corresponding author. 1235 Jefferson Tower, 625 S. 19th St., Birmingham, AL 35294-0007, USA. Tel.: q1 205 9758116; fax: q1 205 9340928; e-mail: [email protected]
and King, 1994; King et al., 1994.. These proteins are closely linked at two levels. First, they possess three RNA-binding domains that share a high degree of sequence homology to each other and to the Drosophila prototype, ELAV ŽRobinow et al., 1988; Yao et al., 1993.. Second, family members are normally restricted to neurons ŽRobinow et al., 1988; Szabo et al., 1991; King et al., 1994.. The aberrant expression of this family in small-cell lung cancer ŽSCLC. ŽDalmau et al., 1992a; King and Dropcho, 1996; King, 1997., which is the most common tumor in patients with PEM, has been postulated to trigger the anti-Hu immunological response, resulting in selective neural tissue injury ŽDalmau et al., 1992a,b; Dropcho, 1995.. Recently, another family member, HuR, was cloned and found to have a more ubiquitous pattern of RNA expression, including many non-neural tissues ŽMa et al., 1996.. This member also contains three RNA-binding domains which are highly homologous to the neural-restricted counterparts. The absence of non-neural tissue
0165-5728r98r$ - see front matter q 1998 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 5 - 5 7 2 8 Ž 9 8 . 0 0 1 9 6 - 9
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
injury in patients with PEM, and the consistent association of PEM with neurally differentiated tumors, however, suggests that HuR does not participate in the immunopathogenesis of PEM. HuR, for example, may be expressed only at the RNA level in peripheral tissues, or it may be less antigenic than its neural counterparts, despite the overall sequence homology. As an initial step in addressing these possibilities, we have analyzed the immunoreactive patterns of patients with PEM to HuR, Hel-N1, and HuD. We have also utilized the serum of one patient with high-titer anti-HuR antibodies to characterize the protein expression pattern of HuR in normal peripheral tissue as well as lung tumors that are not associated with PEM.
2. Materials and methods 2.1. Serum and tissue samples Serum samples were obtained from 11 patients with anti-Hu PEM who were previously diagnosed by established techniques. Controls consisted of cancer patients without PEM. Human tissues Žneoplastic and normal. were
153
obtained from the University of Alabama Tissue Procurement Office. A HeLa Žepitheloid cancer. cell line was obtained from the American Type Culture Collection ŽRockville, MD.. 2.2. Western blotting Recombinant GST-HuR, GST-HuD, and g10-Hel-N1 fusion proteins were cloned and expressed as previously described ŽDropcho and King, 1994; King et al., 1994; Chung et al., 1996; Ma et al., 1996.. Cellular proteins were extracted from tissues using standard techniques ŽSambrook et al., 1989; Andrews and Faller, 1991.. All proteins were quantitated using a commercial kit ŽBio-Rad, Richmond, CA.. Recombinant proteins Ž200 ng. and cellular proteins Ž60 mg. were electrophoresed, blotted and probed as described previously except that blotto was used as the blocking reagent ŽSambrook et al., 1989; Dropcho and King, 1994.. All sera were diluted at 1:2000 in 1 = blotto. The degree of immunoreactivity was qualitatively determined by visual inspection of band intensity. Preabsorption studies were done by incubating the serum with excess recombinant HuD for 12 h at 48 Ž2 mg of HuD in a 1:2000
Fig. 1. ŽA. Western blot of recombinant Hu antigens HuR, HuD, and Hel-N1 Ž200 ng. using sera Ždiluted 1:2000. from three patients with Hu-positive PEM. ŽB. Summary of the immunoreactivity of 11 Hu-positive PEM patients against HuR, Hel-N1, and HuD Ž1:2000 dilution.. The degree of immunoreactivity was estimated by visual inspection as shown in the Western blot.
154
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
dilution of serum. prior to immunoblot assay. To assess tissue distribution of HuR, IgG was purified from patient 1 using a protein A-sepharose column ŽBio-Rad. and biotinylated by a standard technique ŽHarlow and Lane, 1988.. Blots were probed with this purified IgG Ž1:100 dilution. and were developed with the Vectastain ABC kit using a DAB substrate ŽVector, Burlingame, CA..
A 10-mm frozen section of human frontal cortex tissue was thaw-mounted onto a subbed slide and fixed in cold acetone for 10 min. The section was blocked in 1 = blotto and then probed overnight at 48C with the serum of patient 1 Ždiluted at 1:2000.. The section was washed, and then probed with alkaline phosphatase-conjugated streptavidin according to the manufacturer’s specification ŽPierce, Rockford, IL.. The blot was developed with chromogen fast red ŽBiogenex, San Raymond, CA..
This fragment represents nucleotides a159–416 of the GenBank sequence U38175. The glyceraldehyde-3 phosphate dehydrogenase ŽGAPDH. positive control and HelN1 templates were used as described previously ŽKing, 1997.. All templates were linearized with appropriate restriction enzymes to generate antisense riboprobes. Templates were transcribed with 20 units of either T7 or T3 RNA polymerase ŽPromega, Madison, WI.. All transcription reactions were carried out in buffer provided by the manufacturer along with 60 mCi of w a-32 Px UTP ŽAmersham, Arlington Heights, IL., CTP, ATP, GTP, 10 mM DTT, and 20 units of RNasin ŽPromega.. Reactions were carried out at 378C for 1.5 h, and probes were subsequently purified on a polyacrylamide gel. Sizes of protected probe are: HuR Ž257 nt., Hel-N1 Ž197 nt. and GAPDH Ž130 nt.. RNA hybridizations and RNase digestions were done as described previously ŽBall and King, 1997; King, 1997.. Ten micrograms of total RNA was used in each hybridization.
2.4. RNA protection assay
2.5. Case history
The HuR template was obtained by subcloning a Pst1rSac1 digestion fragment of the HuR cDNA ŽMa et al., 1996. into Bluescript SK ŽStratagene, La Jolla, CA..
Patient 1 was a 70-year-old white female who presented with a two-month history of rapidly progressive dementia, gait ataxia, and lower extremity weakness. On mental
2.3. Immunocytochemistry
Fig. 2. ŽA. MRI results for patient 1 indicating proton-density lesions in the basal ganglia bilaterally Žleft panel. and diffuse enhancement of the cerebellum with gadolinium contrast Žright panel.. ŽB. Immunocytochemistry of human frontal cortex using biotinylated IgG from patient 1 Ždiluted 1:100.. Typical nuclear staining is seen in cortical neurons Žarrows.. Magnification: 20 = .
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
155
status evaluation, the patient was initially alert, but over the hospital course became progressively unresponsive. Cranial nerve examination showed dysarthric speech, bilateral facial weakness, diminished vertical gaze, and impaired saccades. Motor examination showed diffuse weakness and hypotonia in the lower extremities, cogwheel rigidity, and bradykinesia. There was dysmetria in the upper limbs bilaterally. Coordination and gait could not be tested because of lower extremity weakness. Laboratory evaluation was significant for CSF abnormalities including elevated protein at 158 mgrdL Žnormal, 12–60 mgrdL. and 420 WBCsrml Žnormal, 0–5 WBCsrml. which were all lymphocytes. A brain MRI showed T1, T2, and proton density lesions in the striatum bilaterally ŽFig. 2A, left panel. as well as diffuse enhancement of the cerebellum with gadolinium contrast ŽFig. 2A, right panel.. A chest CT scan revealed a 2-cm subcarinal mass. Needle biopsy of this lesion was non-diagnostic on two occasions.
indicated a range of immunoreactivity, from strong ŽFig. 1A, patient 1., to weak ŽFig. 1A, patient 2. to absent ŽFig. 1A, patient 3.. In the majority Ž73%. of samples, however, the immunoreactivity was either weak or absent ŽFig. 1B.. One of seven control patients with cancer, but without Hu-positive PEM, showed weak immunoreactivity to HuR Žnot shown.. With the neural-restricted homologue, HuD, on the other hand, there was strong immunoreactivity in 9 of 11 patients, and moderate in the other two. Although Hel-N1 also had overall strong immunoreactive patterns, there was some variability with 3 of 11 patients being weakly reactive ŽFig. 1B, patients 1, 2, and 8.. Those same patients, however, demonstrated moderate-to-strong patterns with HuD. Despite the overall homology, these findings indicate significant immunoreactive differences of individual Elav family members in patients with PEM.
3. Results
With the serum of patient 1, a band for HuR could be detected by Western blot even at a dilution of 1:20,000 Žnot shown.. This patient presented with a clinical syndrome of rapidly progressive dementia, cerebellar ataxia, and extrapyramidal signs in conjunction with neural imaging demonstrating signal abnormalities in the basal ganglia and cerebellum ŽFig. 2A.. The imaging abnormalities in-
3.1. ImmunoreactiÕity patterns of HuR with PEM sera The sera from 11 patients with Hu-positive PEM were analyzed by Western blot for immunoreactivity to recombinant HuR, HuD, and Hel-N1. The results for HuR
3.2. Subspecificity of antibodies to HuR in a patient with PEM
Fig. 3. ŽA. Western blot analysis of recombinant Hu antigens using serum from patient 1 ŽFig. 1.. The left panel represents untreated serum and the right panel represents serum which was preabsorbed with excess recombinant HuD protein. ŽB. Western blot of various tissues using biotinylated IgG from patient 1. Lane 1, HeLa nuclear extract; 2, squamous cell carcinoma Žlung.; 3, adenocarcinoma Žlung.; 4, adenocarcinoma Žlung.; 5, squamous cell carcinoma Žlung.; 6, muscle; 7, liver; 8, kidney; 9, heart.
156
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
cluded increased proton density signal, predominantly in the basal ganglia and contrast enhancement in the cerebellum. This pattern is consistent with an inflammatory response and has been reported by others in patients with limbic encephalitis ŽDirr et al., 1990; Dalmau et al., 1992b.. The serum demonstrated typical anti-Hu patterns of immunoreactivity by immunocytochemistry. Fig. 2A shows a section of human frontal cortex which was incubated with IgG purified from this patient’s serum. Strong nuclear staining can be seen, and to a lesser extent, staining of the cell body. The nucleoli were not stained which is typical for anti-Hu immunoreactivity ŽGraus et al., 1985; Anderson et al., 1988.. This patient’s serum was further analyzed by preabsorbing the serum with excess HuD ŽFig. 3A.. The panel on the left shows the baseline immunoreactivity with the three recombinant proteins. The panel on the right shows loss of signal for HuD and Hel-N1 after preabsorbing the serum with HuD. The band for HuR, however, remained strong, indicating the presence of a subset of antibodies specific for this protein. Since both recombinant HuR and HuD were synthesized as fusion proteins with glutathione-S-transferase, it is highly likely that the anti-
body subset was directed toward HuR rather than the fusion tag. 3.3. Tissue distribution of HuR Although the HuR mRNA has been detected in nonneural tissues by reverse-transcription PCR, the protein expression pattern of HuR has not been characterized. One potential explanation for the absence of peripheral tissue injury in PEM, for example, could be post-transcriptional regulation of HuR mRNA such that there is minimal or no expression at the protein level. To address that possibility, we analyzed a panel of non-neural tissue Žincluding liver, heart, kidney, and muscle. and four non-neuroendocrine lung tumor extracts with purified IgG from patient 1 and found two immunoreactive bands, approximately 32–35 kDa in size in all tissues ŽFig. 3B.. One of these bands is most likely HuR as this was the size recently observed by Myer et al. Ž1997. in Hela cells. The identity of the second band is unknown; however, it may represent either a splice variant of HuR or possibly another related ELAV family member. Included in this blot was a nuclear extract sample
Fig. 4. RNase protection assay with HuR and Hel-N1 riboprobes using total RNA from non-neuroendocrine lung tumor Ža. and non-neural tissues Žb.. The upper bands represent the protected bands for HuR and Hel-N1, and the lower band in all panels represents the internal GAPDH control Žas indicated by arrows.. The first lane in each panel shows the unprotected probes for the test construct Župper band. and GAPDH Žlower band.. RNA samples are as follows: ŽA. lane 1, squamous cell carcinoma Žlung.; 2, adenocarcinoma Žlung.; 3, adenocarcinoma Žlung.; 4, squamous cell carcinoma Žlung., P, riboprobes, and ŽB. lane 1, HeLa; 2, muscle; 3, liver; 4, kidney; 5, heart; 6, brain; q, NCI N417, P, riboprobes.
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
157
Fig. 5. Comparison of amino acid identity among HuR, Hel-N1, and HuD. Above is a schematic diagram which depicts the highly conserved structure of the Elav protein family. Below is a comparison of amino acid identity of HuR and the neural-restricted homologues Hel-N1 and HuD within the identified regions in the diagram. RRM, RNA recognition motif; aa, amino acids.
from HeLa cells as a positive control ŽFig. 3B, lane 1.. We have since expressed HuR without the large fusion tag, and have confirmed this smaller size Žnot shown.. Expression of HuR could also be seen in all four non-neuroendocrine lung tumor samples tested ŽFig. 3B, lanes 2–5.. Although the immunoblot was not quantitative, no amplification of signal was necessary to visualize the bands, indicating that the protein was not rare. The non-neuroendocrine lung tumors, such as the ones analyzed here, included squamous cell carcinoma and adenocarcinoma, both of which are not associated with PEM ŽDalmau et al., 1992a.. 3.4. RNA expression patterns of HuR To support the conclusion that the band visualized on the Western blot in Fig. 3 represents HuR, the normal and neoplastic tissues were analyzed for RNA expression of elaÕ family members. The previous RNA study with HuR was based on reverse-transcription PCR which is not quantitative ŽMa et al., 1996.. Here, we utilized a sensitive RNase protection assay to analyze these tissues for HuR expression. As shown in Fig. 4, protected bands for HuR were detected with all tissues including the brain. The lower band in these blots represents the protected band for GAPDH which was used as a control for the RNA. The HuR probe included nucleotide sequences from the Nterminus, where there is no homology with other family members Žsee Fig. 5., thereby eliminating the possibility of cross-hybridization. With the exception of the brain and the small-cell lung carcinoma line NCI N417, these same tissues did not express Hel-N1 ŽFig. 4, right panels., or as previously published, HuD ŽSzabo et al., 1991; King,
1997.. This RNA data therefore supports the conclusion that the immunoreactive bands seen on Western blot represent HuR and not the neural-restricted ELAV members.
4. Discussion Although not proven, the pathogenic role of anti-Hu antibodies in PEM is supported by several lines of evidence. Firstly, patients with PEM consistently have high titers of anti-Hu antibodies ŽDalmau et al., 1990.. Secondly, until recently, the Elav family was considered restricted to neurons which parallels the pattern of selective neural tissue injury in PEM ŽSzabo et al., 1991; Dalmau et al., 1992a,b; King et al., 1994.. Moreover, the majority of tumors such as small-cell lung carcinoma ŽSCLC., which are linked to PEM, have neuroendocrine differentiation and express these antigens ŽDalmau et al., 1992a.. The recent cloning of HuR in humans, however, indicates that the ELAV family is not restricted to neurons as previously thought ŽMa et al., 1996.. Here we have confirmed the presence of HuR transcripts in non-neural tissues including lung, muscle, liver, heart, and kidney. The RNase protection assay, unlike reverse-transcription PCR Žas used by Ma et al., 1996., also permits quantitation of transcripts and indicated that HuR mRNA is relatively abundant in these tissues as protected bands could be detected with only 10 mg of total RNA. We also analyzed four non-SCLC lung tumors that are not associated with PEM, and found universal expression of HuR as well ŽFig. 4A.. The neural-restricted Hel-N1 transcripts were not detected. The
158
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159
absence of HuD mRNA in non-SCLC tumors has been previously described ŽKing, 1997.. The lack of association of PEM with non-SCLC tumors, and the lack of peripheral tissue damage in PEM, suggest that HuR does not participate in the immunopathogenesis of PEM. In this study, we have addressed two possible explanations. The first possibility is that there may be immunoreactive or antigenic differences between HuR and the neural-restricted counterparts in patients with PEM. We therefore analyzed and compared the immunoreactive patterns of HuR, Hel-N1, and HuD in patients with Hupositive PEM. While HuD consistently showed strong immunoreactive patterns, HuR reactivity was weak to absent in 73% of patients. Hel-N1 reactivity, although more variable, was also moderate-to-strong overall. These findings indicate that there were substantial differences in immunoreactive patterns between HuR and the neural-restricted counterparts in patients with PEM. The homology of this protein family resides in the RNA-binding domains Žsee Fig. 5., and these regions may account for the observed cross-reactivity among family members ŽKim and Baker, 1993.. Significant divergence of sequence, however, occurs in the N-terminus and the bridging regions where there is less than 20% and 50% amino acid identity, respectively. Epitope mapping of HuD indicates that the immunodominant regions reside upstream from the bridging region and the third RRM ŽManley et al., 1995; Graus et al., 1998., suggesting that the N-terminus may contribute to the immunoreactive differences among family members ŽFig. 5.. Interestingly, we identified one patient Žpatient 1. with typical clinical and serological signs of PEM who had a high titer of antibodies against HuR. We were able to abrogate the immunoreactivity of that patient’s serum to HuD and Hel-N1 by preabsorbing the serum with recombinant HuD. The immunoreactivity with HuR persisted, however, indicating that the patient harbored a subset of antibodies specific for this member. Even between Hel-N1 and HuD, there were differences in immunoreactive patterns among several patients ŽFig. 1B, patients 1, 2, and 8.. These immunoreactive patterns may contribute to the neural-specific tissue injury observed in PEM. To address the possibility that the HuR protein product is not expressed at the protein level in peripheral tissues and non-neuroendocrine tumors, we utilized purified IgG from patient 1 in an immunoblot analysis of these tissues. We detected two immunoreactive bands of 32–35 kDa in the same tissues in which we demonstrated HuR mRNA expression. This molecular mass is consistent with that of HuR described by Myer et al. Ž1997. in HeLa cells. Since the RNase protection assay was negative for Hel-N1 and HuD ŽKing, 1997., this band very likely represents HuR. The weaker reactivity of anti-Hu serum against HuR could potentially explain why previous studies have not detected HuR expression in non-neural tissues ŽDalmau et al., 1992a; Okano and Darnell, 1997.. The absence of peripheral
tissue injury in PEM, and the lack of association of PEM with non-neuroendocrine tumors, therefore, cannot be explained by post-transcriptional down-regulation of HuR protein expression. The cellular localization of HuR, on the other hand, may differ from the neural-restricted homologues, and therefore not be ‘presented’ to the immune system. There is now evidence, for example, that some Hu antigens are expressed on the cell surface of neuroendocrine tumors such as SCLC or neuroblastoma ŽTora et al., 1997.. We did not address that possibility in this study. 5. Conclusion In summary, we have found a substantially weaker pattern of HuR immunoreactivity in patients with Hu-positive PEM compared to that of the neural-restricted homologues HuD and Hel-N1. This finding, coupled with the finding of ubiquitous HuR protein expression, suggests that immunoreactive differences among family members may in part explain the neural-restrictive pattern of tissue injury in patients with PEM. Acknowledgements This study was supported by NIH grant 1 K08NS01587-05 ŽPHK., RPG-97-111-01-CCE from the American Cancer Society ŽPHK. and NS29682 ŽHMF.. References Anderson, N.E., Rosenblum, M.K., Graus, F., Wiley, R.G., Posner, J.B., 1988. Autoantibodies in paraneoplastic syndromes associated with small-cell lung cancer. Neurology 38, 1391–1398. Andrews, N.C., Faller, D.V., 1991. A rapid micropreparation technique for extraction of DNA-binding proteins from limiting numbers of mammalian cells. Nucleic Acids Res. 19, 2499. Ball, N.S., King, P.H., 1997. Neuron-specific Hel-N1 and HuD as novel molecular markers of neuroblastoma: a correlation of HuD messenger RNA levels with favorable prognosis. Clin. Cancer Res. 3, 1859–1865. Chung, S., Jiang, L., Cheng, S., Furneaux, H., 1996. Purification and properties of HuD, a neuronal RNA-binding protein. J. Biol. Chem. 271, 11518–11524. Dalmau, J., Furneaux, H.M., Gralla, R.J., Kris, M.G., Posner, J.B., 1990. Detection of the anti-Hu antibody in the serum of patients with small cell lung cancer—a quantitative Western blot analysis. Ann. Neurol. 27, 544–552. Dalmau, J., Furneaux, H.M., Cordon-Cardo, C., Posner, J.B., 1992a. The expression of the Hu Žparaneoplastic encephalomyelitisrsensory neuronopathy. antigen in human normal and tumor tissues. Am. J. Pathol. 141, 881–886. Dalmau, J., Graus, F., Rosenblum, M.K., Posner, J.B., 1992b. Anti-Huassociated paraneoplastic encephalomyelitisrsensory neuronopathy. A clinical study of 71 patients. Medicine 71, 59–72. Dirr, L., Lester, A., Donofrio, P., Smith, M., 1990. Evolution of brain MRI abnormalities in limbic encephalitis. Neurology 40, 1304–1306. Dropcho, E.J., King, P.H., 1994. Autoantibodies against the Hel-N1 RNA-binding protein among patients with lung carcinoma: an association with type I anti-neuronal nuclear antibodies. Ann. Neurol. 36, 200–205.
L.B. Nabors et al.r Journal of Neuroimmunology 92 (1998) 152–159 Dropcho, E.J., 1995. Autoimmune central nervous system paraneoplastic disorders: mechanisms, diagnosis, and therapeutic options. Ann. Neurol. 37 ŽS1., S102–S113. Graus, F., Cordon-Cardo, C., Posner, J.B., 1985. Neuronal antinuclear antibody in sensory neuronopathy from lung cancer. Neurology 35, 538–543. Graus, Y.F., Verschuuren, J.J., Degenhardt, A., van Breda Vriesman, P.J., De Baets, M.H., Posner, J.B., Burton, D.R., Dalmau, J., 1998. Selection of recombinant anti-HuD Fab fragments from a phage display antibody library of a lung cancer patient with paraneoplastic encephalomyelitis. J. Neuroimmunol. 82, 200–209. Harlow, E., Lane, D., 1988. Antibodies: A laboratory manual. Cold Spring Harbor: Cold Spring Harbor Laboratory. Kim, Y.S., Baker, B.S., 1993. The Drosophila gene rbp9 encodes a protein that is a member of a conserved group of putative RNA binding proteins that are nervous system-specific in both flies and humans. J. Neurosci. 13, 1045–1056. King, P.H., Levine, T.D., Fremeau, R.T., Keene, J.D., 1994. Mammalian homologs of Drosophila ELAV localized to a neuronal subset can X bind in vitro to the 3 UTR of mRNA encoding the Id transcriptional repressor. J. Neurosci. 14, 1943–1952. King, P.H., Dropcho, E.J., 1996. Expression of Hel-N1 and Hel-N2 in small-cell lung carcinoma. Ann. Neurol. 39, 679–681. King, P.H., 1997. Differential expression of the neuroendocrine genes Hel-N1 and HuD in small-cell lung carcinoma: evidence for downregulation of HuD in the variant phenotype. Int. J. Cancer 74, 378–382. Lucchinetti, C.F., Kimmel, D.W., Lennon, V., 1998. Paraneoplastic oncologic profiles of patients seropositive for type I anti-neuronal nuclear autoantibodies. Neurology 50, 652–657.
159
Ma, W., Cheng, S., Campbell, C., Wright, A., Furneaux, H., 1996. Cloning and characterization of HuR, an ubiquitously expressed Elav-like protein. J. Biol. Chem. 271, 8144–8151. Manley, G.T., Smitt, P.S., Dalmau, J., Posner, J.B., 1995. Hu antigens: reactivity with Hu antibodies, tumor expression, and major immunogenic sites. Ann. Neurol. 38, 102–110. Myer, V.E., Fan, X.C., Steitz, J.A., 1997. Identification of HuR as a protein implicated in AUUUA-mediated mRNA decay. EMBO J. 16, 2130–2139. Okano, H.J., Darnell, R.B., 1997. A hierarchy of Hu RNA-binding proteins in developing and adult neurons. J. Neurosci. 17, 3024–3037. Robinow, S., Campos, A., Yao, K., White, K., 1988. The elaÕ gene product of Drosophila, required in neurons, has three RNP consensus motifs. Science 242, 1570–1572. Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular cloning: A laboratory manual Ž2nd Edn... Cold Spring Harbor: Cold Harbor Laboratory. Szabo, A., Dalmau, J., Manley, G., Rosenfeld, M., Wong, E., Henson, J., Posner, J.B., Furneaux, H.M., 1991. HuD, a paraneoplastic encephalomyelitis antigen contains RNA binding domains and is homologous to elaÕ and sex-lethal. Cell 67, 325–333. Tora, M., Graus, F., de Bolos, C., Real, F.X., 1997. Cell surface expression of paraneoplastic encephalomyelitisrsensory neuronopathy-associated Hu antigens in small-cell lung cancers and neuroblastomas. Neurology 48, 735–741. Yao, K., Samson, M., Reeves, R., White, K., 1993. Gene elaÕ of Drosophila melanogaster: a prototype for neuronal-specific RNA binding protein gene family that is conserved in flies and humans. J. Neurobiol. 24, 723–739.