Increased production of EGF or TGF-a by the respiratory epi-

Transgenic mice expressing transforming growth factor alpha (TGF-alpha) in type II cells under control of the lung-specific surfactant protein-C (SP-C) promoter develop pulmonary fibrosis and marked airspace hypoplasia. To identify cellular signaling mechanisms involved in lesion formation, we generated transgenic mice expressing a mutant epidermal growth factor receptor lacking a portion of the intracytoplasmic domain (EGF-R-M) under control of the human SP-C promoter. Transcripts of the SP-C-EGF-R-M transgene were detected in distal bronchiolar and type II cells by in situ hybridization. The morphology of lungs from the SP-C-EGF-R-M transgenic mice was normal. Lung fibrosis was not detectable and airspace hypoplasia was significantly corrected in bitransgenic mice derived by breeding SP-C-TGF-alpha and SP-C-EGF-R-M mice. Correction of lung pathology in the bitransgenic mice occurred without altering the level of hTGF-alpha mRNA. To further demonstrate that reversal of TGF-alpha lesions required signaling through the EGF-R, SP-C-TGF-alpha transgenic mice were bred to mice homozygous for the wa-2 mutation which encodes a mutated EGF-R. TGF-alpha-induced lesions were reversed in homozygous wa-2 mice. Amelioration of TGF-alpha-dependent pulmonary lesions in SP-C-EGF-R-M mice or wa-2/wa-2 mice supports the concept that autocrine and paracrine signaling mediate fibrosis and airspace remodeling caused by TGF-alpha.

The A/Japan/57 influenza hemagglutinin (HA) was expressed in BALB/c mice under the transcriptional control of the surfactant protein C (SP-C) promoter, resulting in expression of HA in type II alveolar epithelial cells, as well as low level variable expression in other tissues, including the thymus in some of the founder lines. Transgenic animals were able to recover from infection with A/Japan/57 influenza, and they were able to mount antibody responses to A/ Japan/57 HA in titers similar to wild type. We therefore tested their CD4ϩ T lymphocyte responses to HA and found them to be similar to wild type responses. However, CD8ϩ T cells from A/Japan/57-infected transgenic animals were unable to express cytolytic activity against target cells expressing the A/ Japan/57 HA. The CD8 ϩ T cell tolerance was also extremely specific, since transgenics immunized with an influenza strain containing a single amino acid substitution in a dominant HA epitope were able to mount full cytolytic responses to that epitope, but not the wild-type epitope. Adoptive transfer of CD8 ϩ T cell clones into transgenic animals resulted extensive interstitial pneumonitis that was antigen-specific and associated with significant morbidity and mortality. We conclude that a lung-specific transgene may lead to specific CD8 ϩ T cell tolerance, with CD4 ϩ T cell and B cell reactivity to the antigen, and that CD4 ϩ T cell reactivity may remain intact to an antigen expressed in the thymus, even when CD8 ϩ T cell tolerance exists. This observation may have profound implications concerning immune-mediated lung diseases, particularly those mediated by CD4 ϩ T cells. ( J. Clin. Invest. 1996. 98:914-922.)

We assembled three hybrid beta-globin genes by fusing the mouse beta-major promoter and initial transcribed region to one of three goat beta-like globin gene bodies: beta c (preadult), beta F (fetal), or epsilon II (embryonic). Thymidine kinase (tk)-deficient murine erythroleukemia (MEL) cells were cotransformed with one of these constructs and a separate plasmid bearing the tk gene. Half of the 24 cell lines containing either the mouse beta/goat beta c or mouse beta/goat beta F genes expressed the transferred genes at significant levels; in many cases the hybrid genes were, like the endogenous beta-globin genes, inducible with dimethyl sulfoxide. We obtained 13 cell lines containing the mouse beta/goat epsilon II hybrid gene, 6 of which were cotransfected with a mouse beta/human beta fusion gene known to function in MEL cells. In contrast to the results with the other fusion genes, the mouse beta/goat epsilon II hybrid was very poorly expressed: in two separate experiments, 0 of 13 and 2 of 13 lines showed significant mouse beta/goat epsilon II RNA levels after induction. In all these lines the endogenous mouse beta and cotransfected mouse beta/human beta genes were expressed. As an initial test of possible reasons for the inactivity of the mouse beta/goat epsilon II hybrid, we recloned this fusion gene into a tk-bearing plasmid, adjacent to the tk gene. Of 12 cell lines transformed with this plasmid, 11 produced mouse beta/goat epsilon II RNA; in 6 cases the expression was both strong and dimethyl sulfoxide inducible.

We assembled three hybrid 13-globin (10,12,13,52), and the heat shock-responsive element of the Drosophila hsp-70 gene (9,27 MATERIALS AND METHODSConstruction of fusion genes. We assembled three p-globin fusion genes for this study (Fig. 1) 'n20-kb' 1 jig of HSV-106, enough globin plasmid to give a 10-fold molar excess over HSV-106, and the remainder as highmolecular-weight purified F4-12B2 cell DNA. In the experiments with pUCtk-r, the plasmid was cleaved with ScaI g3C before transformation (Fig. 2); either 1.8 or 9 ,ug of the * pUCtk-e plasmid per plate was used. As in the earlier studies, F4-12B2 DNA was added so that a total of 25 ,ug of DNA was applied to each plate. Regardless of which plasmids were used for transformation, the DNA precipitate was left on the cells for 9 h, when they were fed with nonselective D medium. Twenty hours later, the initial medium was replaced with selective medium (HAT; 15 ,ug of hypoxanthine per ml, 0.2 jiM aminopterin, and 5 jig of thymidine per ml). . DNAs from other lines were dot blotted. Dot blots were or P3F). The mouse prepared by suspending the DNAs in 5 ,ul of 0.1 M NaOH-2 ,B/human 1 construct was that described by Chao et al. (6) and was kindly provided by that laboratory.After our initial round of experiments with the three constructs shown in Fig. 1, we wanted to test the effect of placing the mouse 13/goat Ell gene on the same DNA fragment as the selectable marker gene for thymidine kinase (tk). To accomplish this we first prepared the 1.8-kilobase (kb) PvuII fragment from herpes simplex virus 106 (HSV-106) (24) which contains the entire herpesvirus tk gene; this was cloned by blunt-end ligation into the unique SspI site of pUC18. The EcoRI fragment containing the mouse 1/goat Ell gene was then cloned into the BamHI site in the polylinker of the pUCtk plasmid by blunt-end ligation, creating plasmid pUCtk-r. This construct is diagrammed in Fig. 2.Introduction of globin genes into MEL cells. The F4-12B2