H-2 histocompatibility region influences the inhibition of arachidonic acid cascade by dexamethasone and phenytoin in mouse embryonic palates

We have reported that susceptibility to glucocorticoid- and phenytoin-induced cleft palate and glucocorticoid receptor levels in mice are influenced by the H-2 histocompatibility complex on chromosome 17. Phenytoin competes with glucocorticoids for the glucocorticoid receptor and inhibits production of prostaglandins and thromboxanes. In this paper, we have investigated whether glucocorticoids and phenytoin inhibit arachidonic acid release and prostaglandin biosynthesis directly in the embryonic palates and whether the H-2 gene complex influences the degree of inhibition. Using congenic strains varying only in the H-2 region, we demonstrate here that both glucocorticoids and phenytoin inhibit the release of 3H-arachidonic acid and prostaglandin biosynthesis from embryonic palatal tissue, prelabeled with 3H-arachidonic acid. The degree of inhibition of arachidonic acid release and of prostaglandin biosynthesis is greater in the strain with H-2a (A/Wy) than in its corresponding congenic partner H-2b (A.BY). Thus, these results provide further evidence for a similar genetic and biochemical pathway for the teratogenic action of both phenytoin and glucocorticoids.     

TheH-2 class II genes and the susceptibility to the development of pulmonary adenoma in mice

To determine the locus in theH-2 complex that affects susceptibility to the development of pulmonary adenomas in mice,H-2 congenic and recombinant strains of mice with A/Wy, BALB/c, C3H, and B10 backgrounds were subjected to treatment with urethane. The average number and the incidence of adenoma foci were recorded five months after the treatment. InH-2 congenic strains on the A/Wy background, the average number of adenoma foci per mouse was significantly higher in mice of the A/Wy, A/J, and A-Tla ^b (H-2 ^a ) strains than in A.BY (H-2 ^b ) mice. In BALB/c and C3H congenic strains, the strains carrying theH-2 ^k haplotype were more susceptible than those carrying theH-2 ^b haplotype. InH-2 congenic strains on the B 10 background, the average number and incidence of foci was also higher in haplotypesa, h2, k, andj than in haplotypesb, s, f, d, r, h4, i3, i5, and4. The average numbers of adenoma foci in (A/J × A.BY)F₁ (H-2 ^a /H-2 ^b ) and (B10 × B10.A)F₁ (H-2 ^b /H-2 ^a ) were intermediate between the numbers in the parental strains. In [B10.A (4R) × B10.A (3R)]F₁ (H-2 ^h4 /H-2 ⁱ³ ) and [B10.A (4R) × B10.A (5R)]F₁ (H-2 ^h4 /H-2 ⁱ⁵ ), the numbers of adenoma foci were higher than in resistant parental recombinants. These patterns of response to urethane matched the patterns of the immune response to lactate dehydrogenase-B (LDH-B) and immunoglobulin gamma 2a (IgG2a) proteins. These differences between mice in their susceptibility to the development of pulmonary adenomas is probably due to the polymorphism of the class II genes in theH-2 complex.     

Influence of H-2-linked genes on glucocorticoid receptors in the fetal mouse palate

The binding of ³H-dexamethasone to cytosolic receptors in fetal jaws and in cytosols and nuclei of primary cell cultures of fetal palates was studied in various congenic strains of mice. The amount of specific binding was greater in palatal tissues from B10.A and BlO.A(2R) mice than in B10 or B10.A(5R) preparations. These differences were not observed in the liver. Since the strains with higher levels of glucocorticoid receptor are known to be more susceptible to cortisone-induced cleft palate than the strains with low receptor levels, it is suggested that quantitative variation in receptor levels may be involved in determining H-2-linked differences in cleft-palate susceptibility. Whether or not this is the case, it appears that an H-2-linked gene affects the quantity of a cytosolic glucocorticoid-binding protein which translocates to the nucleus.     

Phenytoin teratogenicity in the primary and secondary mouse embryonic palate is influenced by the H-2 histocompatibility locus

Inbred and congenic strains of mice have been studied for susceptibility to phenytoin-induced cleft lip with or without cleft palate (CLP) and isolated cleft palate (CP). The role of genes linked to the H-2 complex on chromosome 17 has been confirmed. Congenic strains with the A background have identical levels of spontaneous CLP, whereas those strains having the A background with the H-2a haplotype have significantly higher rates of induced CLP than their congenic partners with the H-2b or H-2s haplotype. No such significant difference in the degree of CLP produced by phenytoin is demonstrable in strains with the B background. Rates of isolated CP produced by phenytoin are significantly higher in strains with H-2a than in their congenic partner strains with either H-2b or H-2s, whether the background is A or B.     

Glucocorticoid-Induced Cleft Palate in the Mouse: Two Major Histocompatibility Complex, H-2, Loci with Different Mechanisms

Isolated cleft palate is induced in the progeny of pregnant mice that are given glucocorticoids. The incidence varies among inbred strains and with dose and stage of gestation when the drug is given. One chromosomal region responsible for strain-associated differences in sensitivity is the major histocompatibility complex, H-2. H-2^a is associated with susceptibility, H-2^b with resistance. There appear to be both maternal and embryonic genetic factors affecting the sensitivity to glucocorticoids. In experiments reported here congenic strains of mice with H-2^a, H-2^d and H-2^k haplotypes on a C57BL/10 genomic background were used. This allowed the determination of the effect on sensitivity by two H-2 subregions; the subregions are H-2K to I-E and I-C to H-2D. Methods included dose-response analysis and reciprocal cross analysis using dexamethasone given on day 12 of pregnancy. Results show that each subregion affects the strain's sensitivity to dexamethasone-induced cleft palate. The regression coefficients for B10.A-H-2^a (45.4 ± 4.13) were different from those for B10.BR-H-2^k (67.2 ± 10.8) and B10.D2-H-2^d (70.5 ± 9.74). The estimated mean arcsine% cleft palate at 160 mg/kg was different for each strain: B10.A- H-2^a, 53.1 ± 2.19; B10.BR-H-2^k, 33.1 ± 2.27; B10.D2-H-2^d, 25.0 ± 2.75. Different patterns of change in sensitivity were observed among the reciprocal crosses. In summary, the H-2K to I-E subregion seemed to influence both maternal and embryonic factors, whereas only embryonic factors were influenced by the I-C to H-2D subregion. These data suggest that the mechanisms affecting glucocorticoid sensitivity which are genetically encoded within each H-2 subregion are different, and there is an interaction between the alleles. The mode of interaction can be either complementation or epistasis.     

Histocompatibility genes (theH-2 complex) and susceptibility to spontaneous lung tumors in mice

The incidence of spontaneous lung tumors in relation toH-2, the major histocompatibility complex, was studied in congenic strains of mice on the B10-, A-, and C3H-backgrounds.The most relevant results were obtained with congenic strains on the B10-background. The strains could be divided into two groups: one with a low frequency of spontaneous lung tumors carrying the haplotypesH-2 ^b ,H-2 ^h4 ,H-2 ^d ,H-2 ⁱ H-2 ^r and one with a higher incidence of lung tumors carrying the haplotypesH-2 ^f ,H-2 ^m ,H-2 ^h2 ,H-2 ^a . The differences between these two groups were highly significant.Analysis of the results obtained with the recombinant strains indicated that genes in theIB region determined the susceptibility to the development of spontaneous lung tumors.The comparison of the results in the B10, B10.A and A strain has shown that the incidence in the B10.A strain carrying the haplotypeH-2 ^a derived from the highly susceptible strain A (H-2 ^a ) on the resistant background strain B10 (H-2 ^b ) is intermediate between these two strains. This shows, that other genes of the background are also involved.The lung tumor incidence in (B10.A × B10)F₁ hybrids was intermediate between the two parental strains.The results obtained in the strains C3H with the haplotypeH-2 ^k , C3H.B10 with the haplotypeH-2 ^b and C3H.NB with the haplotypeH-2 ^p , were inconclusive because of the early mortality which occurred among the animals of these strains. The strains A (H-2^a) and A.SW (H-H-2 ^s ) were both equally susceptible.     

Use ofH-2 mutations to quantitate alloreactivity: Alloreactivity is strongest against H-2 antigens which are closest to self

Lymph-node cells fromH-2 allogeneic, intra-H-2 recombinant andH-2 mutant congenic strains were sensitized in limiting dilution cultures to quantitate the cytotoxic T-lymphocyte precursor frequencies (CTL.Pf) against antigens encoded by different regions of theH-2 complex. When fourH-2K ^b mutants of C57BL/6 (B6) were tested, we observed anti-B6 CTL.Pf that were as high or higher than those of recombinant strains which differ from B6 at theK end of theH-2 complex. Relative to strains completelyH–2 allogeneic to B6, the CTL.Pf inH-2 ^bm1,H-2 ^bm3 andH-2 ^bm5 averaged 40–50 percent, andH-2 ^bm8 averaged 140 percent. Recombinant strains B10.A (4R) and B10.D2 (R103), which differ from B6 at theK end of theH-2 complex, averaged 60 percent of the completelyH-2 allogeneic value. Since the mutant and wild-type gene products have no serological and minimal structural differences relative to other alleles atH-2K, these results indicate that the CTL.Pf does not increase with increasing H-2 antigenic disparity between any two strains. Rather, the data suggests that the T-cell receptor repertoire recognizes those H-2 molecules or determinants closest to self.     

Genetic control of acquired resistance to visceral Leishmaniasis in mice

A series ofH-2 and non-H-2 congenic resistant (CR) strains on a C57BL/10Sn background were infected with 10⁷ amastigotes ofLeishmania donovani. Non-H-2 congenic strains B10.LP-H-3 ^b and B10.CE(30NX) and (B10.LP-H-3 ^b × B10)F₁ hybrids showed a very rapid decrease in liver-parasite burdens beyond day 21. Parasite counts for these strains at day 35 were significantly lower than for all other strains tested. The rapid decrease in parasite numbers, massive lymphocellular infiltration into the liver and strong delayed hypersensitivity reactions to parasite antigens in strains congenic for a portion of chromosome 2 indicated that acquired immunity toL. donovani was controlled by a dominant gene at or near theIr-2 locus. In addition, B10.129(10M) mice, which differ from C57BL/10Sn at theH-11 locus, showed highly significant increases in parasite numbers at day 35. Other observations supporting the absence of acquired immunity in B10.129(10M) included negative delayed hypersensitivity tests to parasite antigens and the absence of lymphocellular infiltrate into the liver. Although the differences were not as pronounced,H-2 CR strains withH-2 ^b ,H-2 ^a , andH-2 ^k haplotypes also showed significantly greater decreases in parasite numbers by day 35 as compared to otherH-2 CR strains.     

Corticosteroid-induced cleft palate in mice and H-2 haplotype: Maternal and embryonic effects

This report confirms and expands on the original preliminary observations made by Bonner and Slavkin that corticosteroid-induced cleft palate in mice is associated with H-2 haplotype. Using three congenic strains, B10, B10.A, and B10.D2, our studies demonstrate that B10.A (H-2 ^b) is most susceptible and B10.D2 (H-2 ^d) is least susceptible, B10 (H-2 ^b) being intermediate. Variation in fetal loss among strains accounts for less than 1 percent of the variation in cleft-palate frequency among strains; variation in H-2 haplotype, however, accounts for more than 60 percent of the variation in cleft-palate frequency. With regard to all possible reciprocal F₁ hybrids, our results indicate that while there is a significant maternal effect, maternal haplotype can account for only 11 percent of the variation in cleft-palate frequency among crosses. Embryonic haplotype accounts for 17 percent of the variation, which is indicative of an important embryonic effect. Finally, our studies suggest that susceptibility to corticosteroid-induced cleft palate is associated with the K end of the H-2 complex.     

The production,testing, and utility of double congenic mouse strains

Two new double congenic strains, B10-H-2 ^a H-7 ^b /Wts and B10-H-2 ^d H-7 ^b /Wts, were selected to differ from B10.A and B10.D2/o, respectively, at theH-7 locus. The survival time ofH-7-incompatible skin grafts is dependent upon theH-2 haplotype of recipient and donor.     

Interacting genetic factors controlling the antibody response against the H-2.2 specificity

The antibody response against the H-2.2 specificity has been studied in three H-2 ^d strains, B10.D2, DBA/2, and BALB/c, and their hybrids (B10.D2 × DBA/2)F₁ and (B10.D2 × BALB/c)F₁. The genetic control of the response appears to be complex: The three pure strains are responders, whereas both hybrids when immunized with C3H-HTG are nonresponders. Individual analysis of N3 offspring is compatible with the idea that, in this combination, an Ea-4 incompatibility between donor and immunized strain is necessary for the anti-H-2.2 response to occur. H-2 ^d /H-2 ^k hybrids (B10.BR × B10.D2)F₁ or (B10.BR × DBA/2)F₁ are responders when immunized with C57BL/10 (H-2 ^b ) but not with B10.A(2R) (H-2 ^h ), indicating that simultaneously recognized H-2 specificities are necessary for the anti-H-2.2 response.     

The role ofH-2 regions in overcoming tissue incompatibility

Neonatal transplantation tolerance to the products of theH-2 ^b complex was induced in B10.A (H-2 ^a ) mice. On the basis of the survival of skin allografts it was found that antigens determined by theD region of theH-2 ^b complex (of the B10.A(2R) strain) were most easily overcome and that tolerance to the products of theD end of theH-2 complex (of the B10.A(4R) strain) was also easy to induce. The antigens produced by theK end ofH-2 (of the B10.A(5R) and B10.A(3R) strains) represented a stronger incompatibility barrier and a difference in the entireH-2 ^b complex caused strongest resistance to tolerance induction. When tolerance to the products of the entireH-2 ^b complex was induced in newborn B10.A mice, and the neonatally treated animals were grafted simultaneously with five different grafts, those disparate at theK end ofH-2 and in the entireH-2 region were rejected in some animals, while the grafts disparate at theD end of H-2 remained intact in the same mice. No dependence on theI-J subregion was observed in this system. Furthermore, tolerance was more easily inducible in male than in female B10.A mice.     

Studies ofH-2K b mutant mice

Three newH-2 ^b mutant strains, B6.C-H-2 ^bm9 , B6.C-H-2 ^bm10 and B6.C-H–2 ^bm11 , are described. The three mutant strains are of the gain and loss type as they reject skin grafts reciprocally with the parental C57BL/6Kh. The mutations, which arose independently, are all allelic at the same locus as 11 other mutant strains already described. By complementation and other studies the mutated gene has been shown to beH-2K ^b . The strains were typed directly and by absorption with antisera specific for H-2K^b and H-2D^b private and public specificities and for Ia^b specificities. Each strain typed differently with these sera. The strain B6.C-H-2 ^bm9 was found to be serologically identical with C57BL/6. The strains B6.C-H-2 ^bm10 and B6.C-H-2 ^bm11 were found to have alterations in the private H-2K^b specificity, H-2.33, and in the public specificity, H-2.5, but to a different extent. B6.C-H- 2^bm10 had a marked decrease in the amount of H-2.33 expressed on the splenic cell surface as compared to C57BL/6 and also has a marked decrease in the expression of H-2.5 on both spleen and red blood cells. In comparison, B6.C-H-2 ^bm11 has a decrease in the expression of H-2.33 but an increase in the expression of H-2.5 on both splenic and red blood cells. The other H-2^b specificities appeared to be unaltered as compared with C57BL/6.     

Cytolytic T cells activated by H-2-controlled E molecules cross-react with A molecules

Cell-mediated lymphocytotoxicity was generated in four strain combinations differing only by the cell-surface expression of the class II E molecule controlled by the H-2 complex. The four combinations were: B10.D2(R107) anti-B10.A(3R), B10.A(4R) anti-B10.A(2R), B10.GD anti-B10.D2(R101), and B10.S(7R) anti-B10.S(9R). In all four of these combinations, the stimulator expresses E molecules on the cell surface, while the responder does not. The cytolytic T lymphocytes generated in the B10.D2(R107) anti-B10.A(3R) and B10.A(4R) anti-B10.A(2R) combinations reacted not only with the stimulator but also with strains that do not express cell-surface E molecules, in particular, strains carrying the H-2 ^f and H-2 ^q haplotypes. The cross-reactivity with E-negative strains could be blocked by monoclonal antibodies specific for the A^f or A^q molecules but not by antibodies recognizing determinants on E or class I (K) molecules. The anti-H-2^f cross-reactivity could be inhibited by H-2 ^q cold targets and, reciprocally, the anti-H-2^q reactivity could be blocked by H-2 ^f cold targets. These findings are interpreted as indicating that the cytolytic T lymphocytes stimulated by E molecules can recognize and lyse cells lacking E molecules but expressing A molecules. The observed E-A cross-reactivity supports the notion of structural and functional relatedness between the A and E molecules and suggests a common evolutionary origin of the A- and E-encoding loci.     

Effects of H-2 complex and non-H-2 background on urethane-induced chromosomal aberrations in mice

The incidence of in vivo urethane-induced chromosomal aberrations was examined in H-2 congenic strains of mice with B10 and A backgrounds. Chromosome analysis of bone-marrow cells could divide 7 lines of A.H-2 congenic strains into 2 groups: one with a higher frequency of chromosomal aberrations such as in A/Wy (haplotype H-2a), A/J (H-2a), A.AL (H-2al) and A.TL (H-2tl), and the other consisting of A.TH (H-2t2), A.CA (H-2f), A.BY (H-2b) and A.SW (H-2s). The same tendency was also observed in the spleen cells. Among B10.H-2 congenic mice, B10.A (H-2a), B10.BR (H-2k), B10.A(3R) (H-2i3), B10.A(5R) (H-2i5) and B10.S(9R) (H-2t4) exhibited significantly higher rates of induced chromosomal aberrations than those in B10 (H-2b), B10.S (H-2s), B10.A(2R) (H-2h2), B10.A(4R) (H-2h4) and B10.S(7R) (H-2t2). To determine the effect on non-H-2 genetic backgrounds on urethane-induced chromosomal aberrations, 4 pairs of strains which have the same H-2 haplotypes, such as in B10 vs. A.BY (H-2b), B10.A vs. A/Wy (H-2a), B10.S vs. A.SW (H-2s), and B10.S(7R) vs. A.TH (H-2t2), were compared. The strains with a B10 background exhibited significantly higher frequencies of deletions and lower frequencies of exchanges than the strains with an A background. These data suggested that at least two genes are involved in the regulation of urethane-induced chromosomal aberrations in mice, one of which is mapped between the S and D regions in the H-2 complex, and another not belonging to H-2.     

Immune response to the Slp alloantigen in the mouse:H-2-linked qualitative and quantitative control

Immunization of inbred mouse strains lacking the Slp allotype results in the production of Slp antibodies in some strains but elicits no detectable response in other strains. Analysis of standard inbred and congenic resistant strains reveals that both the qualitative and quantitative ability to respond to the Slp allotype is associated with theH-2 haplotype of the recipient. Three different response phenotypes can be identified utilizing complement fixation and quantitative immunodiffusion tests. Strains which carry theH-2 ^q haplotype are high responders,H-2 ^k strains are intermediate in response, whileH-2 ^b andH-2 ^v strains produce no detectable antibody. The characteristic response patterns of high and intermediate responders were manifest by day 35 of immunization and continued as discrete response types after a second booster. Quantitative data in the immune response of the intra-H-2 recombinant B10.A(4R) suggest that the recombination event which established theH-2 ^h4 chromosome disturbs the proper function of the genetic determinant controlling response to Slp.     

Serological and complementation studies in four C57BL/6H-2 mutants

C57BL/6 (H-2 ^b ) mice, and four mutants (B6.C-H-2 ^ba , B6-H-2 ^bg1 , B6-H-2 ^bg2 , B6-H-2 ^bh ) derived from this strain after separate mutations had occurred at the same locus within theH-2 complex, were analyzed to determine whether the mutations had led to anyH-2 (or Ia) difference which could be detected serologically. The strains were typed directly with antisera specific for H-2K and H-2D public and private specificities and for the Ia specificities; quantitative absorption studies were also performed for the relevant H-2K^b, H-2D^d and Ia^b specificities. In no case was any quantitative or qualitative difference detected serologically between any of the strains. In addition, by using a variety of techniques to produce and assay for antibody, we failed to produce any antisera between the parental strains and the four mutants. TheH-2 mutations therefore appear to give rise to a type of antigenic specificity which is recognized byT cells and which generateT, but notB cell responses; nor are they recognized by H-2 or Ia alloantisera. The location of the mutating locus within theH-2 complex was shown by the complementation method to be within theK orIA region and not in theIB region, since crosses of the mutant strains with B10.A(4R) or D2.GD failed to complement for a subsequent C57BL/6 skin graft.     

Complementation between a gene of theI-E subregion and a non-H-2 gene in the class-specific suppression of IgG2a antibody to sheep erythrocytes

A low level of IgG2a antibodies is observed in B10 mice after primary immunization with SRBC. Analysis of the response in different H-2^b mice and among B10 animals with differentH-2 haplotypes reveals that this selective isotype deficiency is under the control of at least two genes: a background gene and anH-2-linked gene. Responses ofH-2 recombinant B10 strains map theH-2-linked gene to theI-E subregion. Evidence is presented for complementation betweenH-2 and non-H-2 genes in the determination of the low responder phenotype. Low responsiveness appears to be inherited as a dominant trait. Possible functions of the two series of genes are discussed in relation to suppressor mechanisms.     

Receptor specificity,functional characteristics,and cell-surface phenotype of a highly selected anti-I-Ab-specific,long-term T-cell line

A highly selected alloreactive T-cell line was developed by repeated restimulation of B10.D2/n lymph-node cells with irradiated C57BL/10Sn (BIO) spleen cells in long-term MLC for up to 2 1/2 years. Continuous growth of the line requires restimulation every 2 to 4 weeks with fresh H-2^b stimulator cells. The line proliferates strongly against H-2^b but not againstH-2 ^d ,H-2 ^f ,H-2 ^q ,H-2 ^r , orH-2 ^s stimulators. Analysis of recombinant mouse strains showed that the proliferative response is directed against I-A^b but not K^b or D^b determinants. During the growth period of the line, strong cross-reactivity with H-2^p (B10.P) and weak cross-reactivity with H-2^k strains (e.g., CBA/J and B10.BR) was observed. A clone with exquisite specificity for I-A^b, but with no cross-reactivity with H-2^p or H-2^k was isolated from the line; thus clonal heterogeneity of the line still exists despite the highly selective growth conditions. — The majority of T cells from the line or clone were shown to bind I-A^b but not K^b or D^b determinants either spontaneously during restimulation with fresh B10 stimulator cells or via membrane vesicles expressing I-A^b determinants. No killing activity by the line in either specific or nonspecific cytolytic T-cell assays was observed nor was the T 145 glycoprotein, characteristic of killer T cells, detected.Abbreviations used in this paper B6 C57BL/6J - B10 C57BL/10Sn - Con A Concanavalin A - CTL cytotoxic T lymphocyte - FCS fetal calf serum - FDA fluorescein diacetate - FITC fluorescein isothiocyanate - Ia I-region-associated antigens - LPS lipopolysaccharide fromE. coli - Lyt T-lymphocyte-defined antigen - MLC mixed leukocyte culture - NP-40 nonidet P-40 - PAGE pofyacrylamide gel electrophoresis - PHA phytohemagglutinin fromPhaseolus vulgaris - PM plasma membrane - SDS sodium dodecyl sulfate - TCGF T-cell growth factor(s) - TdR thymidine     

Additional mapping of mouse chromosome 2 genes

The purpose of this work was to elucidate the genetic fine structure of the central portion of mouse chromosome (Chr) 2. Seven Chr 2 congenic mouse strains [B10.PA(L)-pa we un a ^t , B10.PA(L)-pa A ^w , B10.PA(L)-we un a ^t , B10.PA(J)-pa a, B10.FS-we A ^w , B10.C-we A ^w , and B10.YBR-a] were produced. Breeding studies were carried out using strains B10.PA(L)-pa we un a ^t and B10.LP-H-13 ^b to accurately determine the recombination frequencies between marker genes pa and we (1.9%±0.3), we and un (8.8%±0.5), and un and a ^t (4.5%±0.4) of strain B10.PA(L)-pa we un a ^t . These strains and other Chr 2 congenic strains were typed for immunologically defined loci using monoclonal antibody (mAb) C23 reactive with the gene product of B2m ^b T-lymphocyte clone C1 reactive with the gene product of H-3 ^a and H-3 ^c , and lymphocyte clone H1.8 reactive with the gene product of Hd-1 ^a . B2m and H-3 typing located a recombinational event separating [pa B2m H-3] from we (the order of bracketed genes is not known). Hd-1 typing indicated that Hd-1 maps distal to [H-42, H-44] and proximal to un. The gene order [pa, B2m, H-3], we, [H-42, H-45], Hd-1, un, H-13, a ^t , with H-44 mapping centromeric to Hd-1, is indicated by the data. Address correspondence and offprint requests to: R. J. Graff.