Bordetella Dermonecrotic Toxin Is a Neurotropic Virulence Factor That Uses CaV3.1 as the Cell Surface Receptor.

Bordetella pertussis, which causes pertussis, a contagious respiratory disease, produces three major protein toxins, pertussis toxin, adenylate cyclase toxin, and dermonecrotic toxin (DNT), for which molecular actions have been elucidated. The former two toxins are known to be involved in the emergence of some clinical symptoms and/or contribute to the establishment of bacterial infection. In contrast, the role of DNT in pertussis remains unclear. Our study shows that DNT affects neural cells through specific binding to the T-type voltage-gated Ca2+ channel that is highly expressed in the central nervous system and leads to neurological disorders in mice after intracerebral injection. These data raise the possibility of DNT as an etiological agent for pertussis encephalopathy, a severe complication of B. pertussis infection.

ered to be related to host immune modulations and some clinical symptoms such as hypoglycemia and leukocytosis (8,9). ACT increases the level of intracellular cAMP in target cells to a supraphysiological level, which results in immunomodulation, including the marked production of cytokines and dysfunction of immunocompetent myeloid cells (8,9,12). Previous studies using animal models reported that these two toxins function in the establishment of bacterial infection by altering the inflammatory responses of hosts (8,9,12).
In contrast, nothing is known about the role of DNT in pertussis. DNT is a singlechain polypeptide of 1,464 amino acids. The N-terminal 30-amino-acid region is responsible for binding to target cells via an unknown receptor, and the ϳ300-aminoacid C terminus carries an enzymatically active domain with transglutaminase activity that activates the small GTPases of the Rho family through deamidation or polyamination (13)(14)(15)(16)(17). DNT of Bordetella bronchiseptica, which is virtually identical to that of B. pertussis (see Fig. S1 in the supplemental material), is known to cause turbinate atrophy by inhibiting osteoblastic differentiation in atrophic rhinitis, B. bronchiseptica infection of pigs (18)(19)(20). However, the role of DNT in B. pertussis infection has not been elucidated. In pertussis, unlike atrophic rhinitis, no pathological abnormality in bone tissues has been reported. Moreover, possible target organs/tissues other than bone tissues have not been explored.
In this study, we aimed at identifying a receptor(s) for DNT and, based on its tissue distribution, searched for potential target organs/tissues. Our results demonstrate that DNT recognizes the T-type voltage-gated Ca 2ϩ channels Ca V 3.1 and Ca V 3.2 as cell surface receptors. According to public databases, Ca V 3.1 is dominantly expressed in the nervous system. Indeed, the toxin affected cultured neural cells that expressed Ca V 3.1. In addition, intracerebral injection of DNT caused flaccid paralysis in mice. We concluded that DNT has aspects of the neurotropic virulence factor of B. pertussis. The possibility of its involvement in pertussis encephalopathy is also discussed.

RESULTS
CRISPR-Cas9 screening for DNT receptors. In order to identify a DNT receptor, we adopted genome-wide screening with the CRISPR-Cas9 system, which is a powerful technique to identify desired genes. However, DNT is unsuitable as a probe for high-throughput screening because the enzyme action of DNT, which does not cause cell death, is not readily detected. We therefore generated DNT-DT A , which consists of the N-terminal fragment of DNT, including the receptor-binding domain and the translocation domain (DNT  ) (21), and the active domain of diphtheria toxin (DT 26 -218 ) ( Fig. 1A and B). DNT-DT A caused the death of DNT-sensitive MC3T3-E1 and Rat-1 cells but not resistant cells (22,23) (Fig. 1C). The cytotoxicity of DNT-DT A was inhibited in the presence of the binding-domain-containing DNT  , suggesting that DNT-DT A intoxicates the cells by binding to the receptor for DNT (Fig. 1D). Using DNT-DT A as the probe, we carried out screening ( Fig. 2A). MC3T3-E1 cells that stably expressed Cas9 were transduced with the lentiviral library of single guide RNAs (sgRNAs) targeting 19,150 genes (5 unique sgRNAs for each gene). After three rounds of screening with DNT-DT A , sgRNA regions integrated into the genomic DNA of DNT-DT A -resistant cells were sequenced, and targeted genes were identified ( Fig. 2B and Table S3). From the identified genes, we picked up Dhx29, Taok1, and Cacna1g, on the basis of the number of unique sgRNAs, and genes encoding membrane proteins, Tmem151b and Nrg2, and generated 7 clones of MC3T3-E1 mutants for each gene for further analyses. Sensitivity to DNT-DT A was markedly reduced or abolished in ΔCacna1g cells, whereas almost all of the other mutant cells were sensitive, but some clones exhibited reduced sensitivity (Fig. 2C). Therefore, we arbitrarily selected 2 of 7 clones for each mutant and further examined their sensitivity to DNT (Fig. 2D). As judged by the deamidation of intracellular Rho (Rho 63E ), ΔCacna1g cells were confirmed to be resistant to DNT, demonstrating that Cacna1g is involved in the intoxication process of the toxin. Ca V 3.1 serves as the DNT receptor. Cacna1g (CACNA1G in humans) encodes the T-type voltage-gated Ca 2ϩ channel Ca V 3.1, which comprises four domains, each containing six transmembrane helices (Fig. 3A). The gene carries at least 12 (for mouse) and 24 (for human) alternative splice variants (24)(25)(26). We succeeded in cloning three distinct cDNAs of the splice variants of Cacna1g, which were tentatively designated variant 1 (v1), v2, and v3 ( Fig. 3A and B). Ectopic expression of v3 but not v1 and v2 restored sensitivity to DNT and DNT-DT A of ΔCacna1g cells and intrinsically DNTresistant Balb3T3 cells ( Fig. 3C and D and Fig. S2A to C). Immunoprecipitation assays revealed the interaction between extracellularly added DNT and Ca V 3.1 on the membrane of MC3T3-E1/ΔCacna1g/ϩCacna1g v3 cells ( Fig. 3E and Fig. S2D). The cytotoxic effects of DNT-DT A on MC3T3-E1 cells were reduced in the presence of ProTx-I (Fig. 3F), a spider toxin that specifically binds to Ca V 3.1 (27). These results demonstrate that certain splice variants of Ca V 3.1 serve as the receptor conferring DNT sensitivity to cells. In the present study, v3, but not v1 and v2, functioned as the receptor. As the alternative splice sites of v1, v2, and v3 were located in the first domain (domain 1 [D1]) (Δ5=E2 and Δ5=E8) and the second domain (D2) (ΔE14 and Δ5=E16) ( Fig. 3A and B), these domains may be involved in the interaction with DNT.
DNT is a neurotropic toxin. According to the public databases of the Genotype-Tissue Expression (GTEx) (https://www.gtexportal.org/home/) project (Table S4) and BioGPS (http://biogps.org/#gotoϭwelcome), CACNA1G (human) and Cacna1g (mouse) are dominantly expressed in the cerebellum and other brain tissues and in female genital organs, suggesting that these tissues are targeted by DNT in Bordetella infection. We therefore examined if neural cells are affected by DNT. P19 murine embryonal carcinoma cells differentiate into neural cells, including neurons and glial cells, after incubation in the presence of retinoic acid (RA) (Fig. S3) (28,29). The cells responded to DNT in both the differentiated and undifferentiated states, as judged by morphological alterations and the deamidation of Rho (Fig. 4A to E). Differentiated P19 cells highly expressed Ca V 3.1, compared to undifferentiated cells, in which Ca V 3.1 expression was barely detected by immunoblotting (Fig. 4E). Accordingly, Rho of the differentiated cells was highly deamidated. The differentiated cells lost neurites and aggregated upon treatment with DNT ( Fig. 4A to D). These results are consistent with previous observations that neurite outgrowth was inhibited by the activation of the Rho signaling pathway (30)(31)(32)(33). NTera2/cl.D1 (NT2) human embryonal carcinoma cells and T98G human glioblastoma cells were also sensitive to DNT, indicating that the toxin intoxicates cells of human origin, similarly to those of mouse origin (Fig. 4F).
In addition to Ca V 3.1, Ca V 3.2 and Ca V 3.3 are known as the isotypes of the T-type voltage-gated Ca 2ϩ channels, encoded by CACNA1H (Cacna1h) and CACNA1I (Cacna1i), respectively. As reverse transcription-PCR (RT-PCR) revealed that P19 cells express all the isotypes (Fig. 5A), we examined the sensitivity of the isotypes to the toxins using these cells ( Fig. 5B and C). Cacna1g and Cacna1h double-knockout P19 cells were resistant to DNT-DT A and DNT. The single and double knockouts of Cacna1g and/or Cacna1i did not abolish sensitivity. Cacna1h knockout cells, with or without Cacna1i knockout, exhibited moderate sensitivity to DNT-DT A and DNT. In addition, Balb3T3 cells ectopically expressing Ca V 3.1 and Ca V 3.2, but not Ca V 3.3, were sensitive to DNT ( Fig. 5D). Taken together, Ca V 3.1 (encoded by Cacna1g) and Ca V 3.2 (Cacna1h), but not Ca V 3.3 (Cacna1i), function as the receptors for DNT. As MC3T3-E1 cells do not express Ca V 3.2 (Fig. 5A), we successfully identified Cacna1g as the receptor gene by the primary screening of sgRNA-introduced MC3T3-E1 cells with the CRISPR-Cas9 system.
Neurological disorders caused by DNT in mice. The above-described results suggest that DNT is a neurotropic toxin. Because Ca V 3.1 is dominantly expressed in the central nervous system (Table S4), we examined if DNT causes any neurological disorders by intracerebrally injecting the toxin into mice. From 1 day after injection, the mice developed neurological symptoms such as flaccidity of tails and hind limbs ( Fig. 6A and B and Movie S1). These symptoms were not observed with 10-fold-larger amounts of PT or ACT (Fig. 6B). An enzymatically inactive mutant of DNT, DNT C1305A (14), did not cause any symptoms, indicating that the Rho-targeting transglutaminase   activity of the toxin is necessary for its neurotoxicity. In the mice injected with DNT, myelin basic protein (MBP) and interleukin-6 (IL-6) levels in the cerebrospinal fluid (CSF) were markedly increased (Fig. 6C), indicating demyelination and inflammation in the central nervous system. These signs were similarly stated in a recent case report of pertussis-associated encephalitis/encephalopathy (3). The levels of both factors correlated with the severity of clinical symptoms. Hemorrhage was noted in brains of mice injected with DNT upon both macroscopic and microscopic examinations (Fig. 6D, E, and G). Magnetic resonance imaging revealed severe inflammation with watery infiltration along the cerebroventricular area in DNT-injected mice (Fig. 6F). From these results, we concluded that DNT, when intracerebrally injected, causes encephalitis in mice.

DISCUSSION
DNT is commonly produced by pathogenic Bordetella species such as B. pertussis, B. parapertussis, and B. bronchiseptica. B. bronchiseptica DNT is known to cause turbinate atrophy in swine atrophic rhinitis, B. bronchiseptica infection, by inhibiting osteoblastic differentiation (18)(19)(20). It was also reported that the toxin directly affects osteoblastic MC3T3-E1 cells, suggesting that osteoblastic cells express the DNT receptor. However, as only a few cell lines were found to be sensitive to DNT (20,21,23), the toxin receptor was considered not to be ubiquitous and instead was considered to be unique to particular cells. In this study, we demonstrated that the T-type voltage-gated Ca 2ϩ channels Ca V 3.1 and Ca V 3.2 serve as DNT receptors and confer sensitivity to the toxin. Ca V 3.1 is reportedly expressed in osteoblastic cells during osteogenesis (34). This is consistent with previous results that DNT affects osteoblastic cells. Indeed, the sensitivity of osteoblastic MC3T3-E1 cells to the toxin was found to be dependent on the expression of Ca V 3.1 (Fig. 5A).
It is also known that the Ca V 3 channels, which are expressed in neural tissues and cardiac and smooth muscles, are involved in neuronal excitability, pacemaker activity in the sinoatrial node, the circadian rhythm of sleep and wakefulness, hormone secretion, and peripheral pain sensation (35)(36)(37). According to public databases, Ca V 3.1 is dominantly expressed in the central nervous system, whereas Ca V 3.2 expression does not exhibit a characteristic tissue distribution (https://www.gtexportal.org/home/gene/ CACNA1H). Being attracted to this point, we examined the neurotoxicity of DNT at the cellular level and found that it affected neural cells in vitro, indicating that DNT has aspects of a neurotropic toxin; this is the first example of a neurotropic virulence factor produced by B. pertussis. These results prompted us to explore if DNT is involved in neurological disorders that are recognized as pertussis encephalitis/encephalopathy; peripheral nervous disorders have not been observed in B. pertussis infection, although the Ca V 3 channels are also known to be distributed in the peripheral nervous system. The intracerebral injection of DNT, but not PT and ACT, caused encephalitis in mice.
In B. pertussis infection, encephalitis/encephalopathy (here, we use the term "pertussis encephalopathy" according to previous reports) has long been recognized as a rare complication, which develops in up to 1% of patients and imposes a significant burden (6,7); studies from 2001 to 2003 in the United States reported that 33 of 28,998 patients (0.11%) developed encephalopathy (38). The pathophysiology and etiological agents of encephalopathy remain unknown. Possible explanations included hemorrhage in the central nervous system resulting from increased venous pressure due to coughing paroxysms, hypoxia, venous stasis attributable to leukocytosis, hypoglycemia, and secondary infections by neurotropic microbes. However, some case reports negated these explanations and instead pointed out unknown toxins or toxin-like components of B. pertussis as the causative agents (3,4,7). DNT, which causes encephalitis in mice, may be the most probable candidate for such a bacterial component. On the other hand, DNT has been considered to play little role in the pathogenesis of pertussis, partly because it is not secreted but remains localized in the bacterial cytoplasm (8,39,40). To bridge this gap, we retrieved case reports of pertussis encephalopathy that clearly stated the course of the disease (1-5, 7). Only a few reports were available, but in 4 of 6 case reports, the patients, in the early stages or before developing neurological disorders, were administered ␤-lactam antibiotics including, cefuroxime (4), amoxicillin (5), cephalexin and ampicillin (7), or piperacillin (3). ␤-Lactam antibiotics inhibit the synthesis of peptidoglycan and may therefore liberate DNT from bacterial cells. Indeed, we confirmed that treatment of B. pertussis infection with ampicillin and piperacillin resulted in the release of active DNT (Fig. 7). Macrolides such as erythromycin, clarithromycin, and azithromycin, which are the first-line antibiotics for pertussis, did not release DNT.
In this study, we raised the possibility of DNT as an etiological agent of pertussis encephalopathy. However, as not all pertussis patients develop encephalopathy, several other factors should be involved in its development. The use of ␤-lactam antibiotics releasing DNT from bacterial cells may be one of the risk factors. Cofactors may also be required to help DNT cross the blood-brain barrier and enter the central nervous system. PT, which was reported to affect the integrity of cerebral barriers (41,42), may function as such an accessory factor. Indeed, PT is known to exacerbate experimental  (43,44). In addition, the anamnestic history, vaccinations, age, and sex of patients and other bacterial components may influence the onset of encephalopathy. Although further studies are required to address many remaining questions, our study provides a clue to understanding the pathophysiology of pertussis encephalopathy.

MATERIALS AND METHODS
Antibodies, immunoprecipitation assays, magnetic resonance imaging (MRI) analyses, and other utilized methods are given in Text S1 in the supplemental material.
Bacterial strains and cultures. B. pertussis strain Tohama I was maintained in the laboratory. A dnt-deficient mutant of B. pertussis Tohama I was generated according to methods described previously (45). B. pertussis was grown in Stainer-Scholte (SS) medium or on Bordet-Gengou agar (Becton, Dickinson, Franklin Lakes, NJ, USA) containing 0.4% (wt/vol) polypeptone or HiPolypeptone (Wako Pure Chemical Industries, Ltd., Japan), 0.8% glycerol, 20% defibrinated horse blood, and 10 g/ml ceftibuten (BG plate). The culture supernatants of B. pertussis were harvested by centrifugation at 6,800 ϫ g for 5 min and filtered through a 0.22-m-pore-size filter.
Recombinant DNT and DNT derivatives. The primers used for plasmid construction are listed in Table S1. The plasmids encoding DNT and DNT derivatives of B. bronchiseptica were constructed as follows. pQEDNTwt and pGEXDNT 1-54 -His were constructed as described previously (22). pDTA1 was provided by E. Mekada, Osaka University (46).
For DNT-DT A /pQE40, DNA fragments were amplified by PCR with a combination of primers, EcoRV-DNT 1009 Fw and DNT 1185 -DTARv, and pQEDNTwt as the template and with another combination of primers, HindIII-DTA 28 Rv and DNT 1185 -DTAFw, and pDTA1 as the template. The resultant fragments were used as the templates for PCR to amplify DNA fragments encoding DNT-DT A using the primers EcoRV-DNT 1009 Fw and HindIII-DTA 28 Rv. The obtained fragments were inserted into the EcoRV-HindIII site of pQEDNTwt by the seamless ligation cloning extract (SLiCE) technique (47).
For DNT-DT A /pColdII, the DNA fragment encoding DNT-DT A was amplified by PCR with a combination of primers, NdeI-DNT 2 Fw and EcoRI-DTA 218 Rv, and DNT-DT A /pQE40 as the template and inserted into the NdeI-EcoRI site of pColdII (TaKaRa) by the SLiCE technique.
The plasmids for DNT and DNT derivatives of B. pertussis were constructed as follows. For BpDNTwt/ pQE40, a DNA fragment covering the DNT gene was amplified by nested PCR using B. pertussis Tohama I genomic DNA as the template with the first primers BpDNT-nested-F and BpDNT-nested-R and the second primers BamHI-BpDNT 2 -F and Bp-DNT 1464 -HindIII-R. The resultant PCR product was inserted into the BamHI-HindIII site of pQE40 (Qiagen) by the SLiCE technique.
For BpDNT C1305A /pQE40, site-directed mutagenesis was used to replace Cys with Ala at position 1305 in BpDNTwt/pQE40 using a QuikChange kit and the primer pair DNT C1350A -F and DNT C1305A -R.
The expression plasmids were introduced into Escherichia coli M15(pREP4) or BL21(DE3), and the recombinant DNT proteins were produced and purified by Ni or glutathione affinity chromatography according to the manufacturer's instructions (Qiagen, TaKaRa, and GE Healthcare). The recombinant proteins of full-length DNT were further purified by anion-exchange chromatography with a MonoQ column (GE Healthcare) in 20 mM Tris-HCl (pH 7.6) containing 1 M urea and eluted with a linear gradient of NaCl from 10 to 500 mM in the same buffer. The purified DNT proteins were dialyzed against and kept in 50 mM phosphate buffer (pH 7.4) containing 0.3 M Na 2 SO 4 and 1 M urea at 4°C.
Cas9-expressing MC3T3-E1 cells. The DNA fragment encoding enhanced green fluorescent protein (EGFP) was obtained from pX330mEGFP (48) by digestion with FseI and inserted into the FseI site of pPB-pgkBSD-CBh-hSpCas9n, which is a derivative of pPB-SA hyg NP21 pA (48,49), yielding pPB-pgkBSD-CBh-hSpCas9n-EGFP. Subsequently, the pgkBSD region of the plasmid was replaced by pgkNeo, a phosphoglycerate kinase 1 (PGK) promoter-driven neomycin resistance gene. The obtained plasmid was designated pPB-pgkNeo-CBh-hSpCas9n-EGFP. MC3T3-E1 cells were cotransfected with pPB-pgkNeo-CBh-hSpCas9n-EGFP and pCMV-hyPBase (49), which carries a piggyBac transposase, using the Neon transfection system (Invitrogen) according to the manufacturer's instructions and were cultured for 24 h and then for 6 days in the presence of 400 g/ml of G418. Independent clones of the surviving cells were isolated by the limiting-dilution method. A clone that highly expressed Cas9-EGFP, as judged by fluorescence microscopy, was selected and designated Cas9-E1.
Generation of the genome-wide Cas9-E1-sgRNA library and screening. We utilized genome-wide mouse lentiviral CRISPR guide RNA (gRNA) library v1 (50) (catalog number 50947; Addgene), which contains five unique sgRNAs for each of 19,150 genes. 293FT cells were transfected with the library plasmids by using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. The medium was replaced 24 h after transfection, and lentiviral vectors were obtained 2 days after transfection by centrifugation of the culture supernatant at 1,700 ϫ g for 15 min at 4°C. At a multiplicity of infection (MOI) of 0.3, 10 7 cells of Cas9-E1 were infected with the lentiviral library on 150-mm dishes in the presence of 8 g/ml of Polybrene (Millipore) such that a single sgRNA was introduced into more than 30 cells. The cells were selected by incubation in the presence of 8 g/ml of puromycin and screened with DNT-DT A , as follows.
For the first round of screening, 5 ϫ 10 7 cells were treated with 2 g/ml of DNT-DT A in 100-mm dishes. After the 36-h treatment, the cells were washed three times with Dulbecco's modified phosphatebuffered saline (D-PBS) to remove dead cells. The remaining cells were reseeded, incubated for 24 h, and subjected to another round of toxin screening. After three rounds of screening, the remaining cells were collected, and their genomic DNA was extracted by isopropanol precipitation and subjected to the following analyses, together with the genomic DNA from untreated original cells. The regions of sgRNA were amplified by PCR from the genomic DNA templates using the primers pST-Cas9-S27 and pST-Cas9-AS28 with Q5 Hot Start high-fidelity DNA polymerase (New England BioLabs). The PCR products were sequenced using Ion PGM (Thermo Fisher Scientific).
Generation of MC3T3-E1 and P19 knockout cells. pX330mEGFP carrying sgRNA for each target gene (Table S2) at the BbsI site was introduced into MC3T3-E1 cells by electroporation or into P19 cells with Lipofectamine 3000 (Invitrogen). After the 48-h incubation, EGFP-expressing cells were sorted and placed into each well of a 96-well plate using the FACSAria II system (BD Biosciences). The knockout cells were used for the following assays after cultivation for at least 14 days.
Establishment of Cacna-expressing cells. cDNA of Cacna1g was obtained by PCR using the cDNA library prepared from MC3T3-E1 cells as the template and primers BS300 and BS297 and inserted into the BamHI-NotI site of pCX4pur, provided by Tsuyoshi Akagi (51). Plat-E cells were transfected with the resultant plasmids, pCX4pur-Cacna1g-v1, -v2, and -v3, using Lipofectamine 3000 according to the manufacturer's instructions. The medium was replaced 16 h after transfection. The culture supernatant containing viral vectors was collected by centrifugation 2 days after transfection and filtered on a 0.45-m-pore-size polyvinylidene difluoride membrane. Balb3T3 or Cacna1g knockout MC3T3-E1 cells (2 ϫ 10 4 cells) were seeded into each well of a 6-well plate. After 2 days, the medium was replaced with that containing the appropriately diluted retroviral supernatant and 8 g/ml of Polybrene. Two days after infection, the puromycin-resistant cells were selected by incubation with 8 g/ml of puromycin for at least 3 days. cDNAs of Cacna1h and Cacna1i were obtained by PCR using Fantom clones (52) (clone identification numbers M5C1001I06 and M5C1084D17, respectively; DNAFORM) as the templates, the primers BamHI-Cacna1h-F1 and Cacna1h-NotI-R1, and the primers BamHI-Cacna1i-F1 and Cacna1i-NotI-R1. The cDNAs were cloned into the vector and introduced into Balb3T3 cells as described above.
Cell culture assay. For the assay of DNT-DT A cytotoxicity, 5 ϫ 10 3 cells were seeded into each well of a 96-well plate and incubated for 24 h. The medium was replaced with that containing DNT-DT A , and the cells were subsequently incubated for 36 h or the indicated periods. DNT-DT A was used at 2 g/ml unless otherwise specified. Cell viability was quantified using Cell Counting kit 8 (catalog number 343-07623; Wako). The absorbance at 450 nm of each well was measured using the Glomax multidetection system (Promega). The rate of cell death was calculated by the equation cell death (%) ϭ (A buf Ϫ A tox )/(A buf Ϫ A trtn ) ϫ 100, where A buf , A tox , and A trtn are the net absorbances of the untreated sample, the toxin-treated sample, and the Triton X-100-treated sample, respectively.
For the detection of cytopathic effects by DNT, cells were seeded into a 24-well plate at 2.0 ϫ 10 4 cells/well and incubated for 24 h. The cells were further incubated without FCS for 24 h and then incubated with or without 50 ng/ml of DNT for an additional 16 h. The cells were fixed with 4% paraformaldehyde in D-PBS, permeabilized with 0.5% Triton X-100 in D-PBS for 5 min, and stained with rhodamine-phalloidin. Images were taken with a FluoView FV10i microscope (Olympus, Tokyo, Japan). Independently, cells were seeded at 2.4 ϫ 10 5 cells/well into a 6-well plate and treated with DNT as described above. After treatment, the cell lysates were obtained by ultrasonic treatment using a Bioruptor sonicator (Cosmo Bio, Tokyo, Japan), followed by centrifugation. Proteins in the supernatants were precipitated with cold 10% trichloroacetic acid and subjected to SDS-PAGE and immunoblotting for deamidated Rho.
P19 cells were differentiated into neural cells according to a method reported previously (28). The differentiated cells were seeded into a 6-well plate at 3.6 ϫ 10 6 cells/well, incubated for 6 days, and treated with DNT preparations at the indicated concentrations. Undifferentiated cells were seeded at 1.2 ϫ 10 5 cells/well and similarly treated with the DNT preparations after a 2-day incubation.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.01 MB.