Effects of dominant-negative lac repressor mutations on operator specificity and protein stability

The specific binding of dominant-negative (I-d) lactose (lac) repressors to wild-type (wt) as well as mutant (Oc) lac operators has been examined to explore the sequence-specific interaction of the lac repressor with its target. Mutant lacI genes encoding substitutions in the N-terminal 60 amino acids (aa) were cloned in a derivative of plasmid pBR322. Twelve of these lacI-d missense mutations were transferred from F'lac episomes using general genetic recombination and molecular cloning, and nine lacI missense mutations were recloned from M13-lacI phages [Mott et al., Nucl. Acids Res. 12 (1984) 4139-4152]. The mutant repressors were examined for polypeptide size and stability, for binding the inducer isopropyl-beta-D-thiogalactoside (IPTG), as well as binding to wt operator. The mutant repressors' affinities for wt operator ranged from undetectable to about 1% that of wt repressor, and the mutant repressors varied in transdominance against repressor expressed from a chromosomal lacIq gene. Six of the I-d repressors were partially degraded in vivo. All repressors bound IPTG with approximately the affinity of wt repressor. Repressors having significant affinity for wt operator or with substitutions in the presumed operator recognition helix (aa 17-25) were examined in vivo for their affinities for a series of single site Oc operators. Whereas the Gly-18-, Ser-18- and Leu-18-substituted repressors showed altered specificity for position 7 of the operator [Ebright, Proc. Natl. Acad. Sci. USA 83 (1986) 303-307], the His-18 repressor did not affect specificity. This result may be related to the greater side-chain length of histidine compared to the other amino acid substitutions.     

Affinities of tight-binding lactose repressors for wild-type and pseudo-operators

Five tight-binding (Itb) mutants of the Escherichia coli lactose (lac) repressor have been characterized with regard to their non-specific affinity for DNA and their specific affinity for the wild-type operator and several sequence-altered (pseudo-) operators. Repressor-operator association rates were determined in the presence or absence of competitor DNA, dissociation rates of repressor from various DNA fragments were measured, and equilibrium competition for repressor binding was examined for several pseudo-operator DNAs. The mutant repressors exhibited increased non-specific affinity for DNA, and variable increases in affinity for sequence-altered operators. The known positions of amino acid substitutions for three of these Itb repressors support suggestions that residues 51 to 64 are important for operator recognition in addition to residues 1 to 50.     

Degradation of the DNA-binding domain of wild-type and i-d lac repressors in Escherichia coli.

It has been shown that 28 transdominant mutant lac repressors which have lost operator DNA-binding ability in vivo and in vitro, but still bind inducer and are able to form tetramers (i-d repressors), could be divided into two groups by their capacity or incapacity to bind non-specifically to the phosphate groups of the DNA backbone. All but one of 15 analysed i-d repressors with amino acid substitutions to the C-terminal of residue 52 showed uneffected non-specific DNA binding. All 13 tested i-d repressors with amino acid substitutions to the N-terminal of residue 53 did not bind to double-stranded DNA, and 11 of these repressors derived from missense mutations in the lacI gene were endogenously degraded. The degradation in vivo only affects the amino-terminal 50-60 residues producing a mutant-specific pattern of stable repressor fragments. These fragments are tetrameric and capable of binding inducer in vivo and in vitro. The proteolytic attack presumably takes place during synthesis of the i-d repressors, since the resulting fragments are stable, both in vivo (as shown by a pulse-chase experiment) and in vitro. The proteolysis in vivo depends on the growth conditions of the bacteria and is higher in cells grown in minimal media than in rich media. Wild-type repressor is only susceptible to limited proteolysis in cells grown in minimal media but not in cells grown in rich media. The results suggest that the majority of the sequence alterations before residue 53 in missense mutant i-d lac repressor proteins affect the three-dimensional structure of the amino-terminal DNA-binding domain of the repressor protein, making it susceptible to proteolytic attack by one or several intracellular proteases.     

Isolation of altered specificity mutants of the single-chain 434 repressor that recognize asymmetric DNA sequences containing the TTAA and TTAC subsites.

A novel single-chain (sc) protein framework containing covalently dimerized DNA-binding domains (DBD) of the phage 434 repressor was used to construct combinatorial mutant libraries in order to isolate mutant DBDs with altered specificities. The library members contain one wild-type DBD and one mutant domain with randomized amino acids in the DNA-contacting region. Based on previous studies, the mutant sc derivatives are expected to recognize a general ACAA-6 bp-NNNN sequence, where ACAA is contacted by the wild-type and NNNN by the mutant domain. In principle, any sequence can stand for NNNN and serve as a selection target. Here an in vivo library screening method was used to isolate mutant sc repressors that interact with an asymmetric operator containing the TTAA target. Several mutants showed high affinity in vitro binding to operators containing the target and strong (up to 80-fold) preference for the TTAA target over the wild-type TTGT. Specificity studies revealed that certain mutants bound with substantially higher affinities (K(d) approximately 10(-11)M) to operators containing the TTAC sequence, a close homolog of the TTAA target. Thus, we have fortuitously isolated mutant sc repressors that show up to a several hundred-fold preference for TTAC over TTGT.     

Isolation and characterization of chitinases from Verticillium lecanii

Degenerate PCR primers corresponding to conserved domains of fungal chitinases were designed, and PCR was performed on genomic DNA of the entomogenous fungus Verticillium lecanii (Zimmermann) Viegas. Two distinct PCR fragments, chf1 and chf2, were isolated and used to identify two DNA contigs. Analyses of these two contigs revealed that we had obtained the full-length DNA sequence including the promoter, 5' untranslated region, open reading frame (ORF), and 3' untranslated regions for two distinct chitinase-like genes. These two genomic DNA sequences exhibited 51% identity at the amino acid (aa) level and were designed as acidic (chi1) and basic (chi2) chitinase-like genes. The isolated cDNA for chi1 gene is 1110 bp with a predicted protein of 370 aa and molecular mass of 40.93 kDa, and its ORF was uninterrupted in its corresponding genomic DNA sequence. The cDNA for the chi2 gene is 1269 bp, a predicted ORF of 423 aa and molecular mass of 45.95 kDa. In contrast, the ORF was interrupted by three introns in its corresponding genomic DNA. The basic chitinase gene (chi2) was successfully expressed in the Pichia pastoris system; optimum enzymatic activity was observed at 22 degrees C and at pH 7.5. CHI1 and CHI2 were clustered into two different phylogenetic groups according to their sequence alignments with 28 other fungal chitinases. A chitin-binding domain, comprising two sub-domains that exhibit similarities at the aa level to chitin binding domains in bacteria, was identified in 30 fungal chitinase sequences examined.     

Identification of nonessential regions of the nsp2 replicase protein of porcine reproductive and respiratory syndrome virus strain VR-2332 for replication in cell culture

The nonstructural protein 2 (nsp2) of porcine reproductive and respiratory syndrome virus (PRRSV) is a multidomain protein and has been shown to undergo remarkable genetic variation, primarily in its middle region, while exhibiting high conservation in the N-terminal putative protease domain and the C-terminal predicted transmembrane region. A reverse genetics system of PRRSV North American prototype VR-2332 was developed to explore the importance of different regions of nsp2 for viral replication. A series of mutants with in-frame deletions in the nsp2 coding region were engineered, and infectious viruses were subsequently recovered from transfected cells and further characterized. The results demonstrated that the cysteine protease domain (PL2), the PL2 downstream flanking sequence (amino acids [aa] 181 to 323), and the putative transmembrane domain were critical for replication. In contrast, the segment of nsp2 preceding the PL2 domain (aa 13 to 35) was dispensable for viral replication, and the nsp2 middle hypervariable region (aa 324 to 813) tolerated 100-aa or 200-aa deletions but could not be removed as a whole; the largest deletion was about 400 aa (nsp2Delta324-726). Characterization of the mutants demonstrated that those with small deletions possessed growth kinetics and RNA expression profiles similar to those of the parental virus, while the nsp2Delta324-726 mutant displayed decreased cytolytic activity on MARC-145 cells and did not develop visible plaques. Finally, the utilization of the genetic flexibility of nsp2 to express foreign genes was examined by inserting the gene encoding green fluorescent protein (GFP) in frame into one nsp2 deletion mutant construct. The recombinant virus was viable but impaired and unstable and gradually gained parental growth kinetics by the loss of most of the GFP gene.     

Two domains in the terminal inverted-repeat sequence of transposon Tn3

Tn3 and related transposons have terminal inverted repeats (IR) of about 38 bp that are needed as sites for transposition. We made mini-Tn3 derivatives which had a wild-type IR of Tn3 at one end and either the divergent IR of the Tn3-related transposon, gamma delta or IS101, or a mutant IR of Tn3 at the other end. We then examined both in vivo transposition (cointegration between transposition donor and target molecules) of these mini-Tn3 elements and in vitro binding of Tn3-encoded transposase to their IRs. None of the elements with an IR of gamma delta or IS101 mediated cointegration efficiently. This was due to inefficient binding of transposase to these IR. Most mutant IR also interfered with cointegration, even though transposase bound to some mutant IR as efficiently as it did to wild type. This permitted the Tn3 IR sequence to be divided into two domains, named A and B, with respect to transposase binding. Domain B, at positions 13-38, was involved in transposase binding, whereas domain A, at positions 1-10, was not. The A domain may contain the sequence recognized by some other (e.g., host) factor(s) to precede the actual cointegration event.     

Characterization of the Haemophilus influenzae topA locus: DNA topoisomerase I is required for genetic competence

A gene essential for the development of genetic competence in Haemophilus influenzae (Hi) was identified as a homolog of the Escherichia coli (Ec) topA gene, which encodes DNA topoisomerase I (TopI). The Hi topA locus was initially identified by mini-TnlOkan mutagenesis. Three independent insertion events within 500 bp of each other resulted in mutant strains that shared a similar phenotype. Each was deficient in competence-induced DNA binding, showed increased sensitivity to UV irradiation, and had an increased doubling time as compared to the wild-type (wt) strain. The nucleotide sequence of a 6.6-kb fragment containing the wt allele was determined. The sequence contained an open reading frame (ORF) of 868 amino acids (aa) that was interrupted by each of the mini-Tn10kan mutations. The deduced aa sequence had a molecular mass of 98 155 Da, a pI of 8.59 and showed strong similarity to Ec TopI. Examination of the topoisomer distribution of a test plasmid in an Hi mutant carrying an insertion in this ORF showed an increase in the level of supercoiling, indicating that TopI is necessary to relax supercoiled DNA in Hi. Complementation studies and insertional inactivation of genes downstream from topA indicated that TopI and not some downstream gene product was essential for competence. Four other ORFs were identified and two of these had homology to known genes. ORF1, which was truncated at one end of the sequenced region, shared strong sequence similarity to the C-terminal end of Ec pyridine nucleotide transhydrogenase β subunit. ORF4, which was also truncated, showed strong sequence similarity to the N-terminal end of Ec threonyl-tRNA synthetase.     

Functional analysis of the two domains in the terminal inverted repeat sequence required for transposition of Tn3.

Bacterial transposon Tn3 has a 38-bp terminal inverted repeat (IR) sequence. The IR sequence has been divided into two domains, A and B, of which domain B is bound by transposase, and domain A is not Here, we defined the two domains more precisely by constructing three IR mutants with a 2-bp substitution at relevant sites within the IR sequence, followed by examination of the binding of transposase to the fragments containing these IR mutants: domain A was located at bp 1-11, whereas domain B was at bp 12-38. To see if the two domains in the IR are functionally distinct, we constructed mini-Tn3 derivatives flanked by two IRs with various 2-bp substitutions within domain A or B, and analyzed their ability to mediate cointegration. The mini-Tn3 derivatives flanked by IR(A+ B+) and IR(A- B+) [or IR(A+ B-)] and those flanked by IR(A-B+) and IR(A+ B-) mediate cointegration more efficiently than the mini-Tn3 derivatives flanked by two IR(A- B+)s or by two IR(A+ B-)s. These results and others presented here indicate that the two domains of IR are functionally distinct in promoting cointegration.     

Targeting the Escherichia coli lac repressor to the mammalian cell nucleus

We have previously shown that about 90% of total Escherichia coli lac repressor synthesized in mammalian cells is located in the cytoplasm [Hu and Davidson, Cell 48 (1987) 555-566]. To target a functional lac repressor to the nucleus, we mutated 10 nucleotides at the 3' end of the coding sequence, thus adding the nuclear localization signal of the simian virus 40 large-T antigen to the C terminus of the repressor. The mutant lacI gene and the wild-type (wt) gene, both in standard animal cell expression vectors, driven by the promoter of the Rous sarcoma virus long terminal repeat, were stably transfected into three rodent cell lines. In confirmation of our previous results, only about 10% of the wt repressor, but all of the mutant protein, was localized in the nucleus. DNase I footprint analyses showed that the mutant repressor retained the same operator DNA-binding specificity as wt repressor. Furthermore, both repressor-operator complexes could be dissociated by addition of isopropyl-beta-D-thiogalactopyranoside in vitro. However, the ratio of number of repressor molecules per nucleus that, by in vitro assay, could bind to the operator sequence to the number of monomer repressor polypeptides per nucleus, as determined by Western blotting, was about 1:4 for the wt repressor and about 1:30 for the mutant repressor. This suggests that: (a) the mutant repressor assembles into tetramers inefficiently; and/or (b) it has reduced binding affinity to the operator sequence; and/or (c) it has higher binding affinity to nonspecific DNA.     

FACS-optimized mutants of the green fluorescent protein (GFP)

We have constructed a library in Escherichia coli of mutant gfp genes (encoding green fluorescent protein, GFP) expressed from a tightly regulated inducible promoter. We introduced random amino acid (aa) substitutions in the twenty aa flanking the chromophore Ser-Tyr-Gly sequence at aa 65–67. We then used fluorescence-activated cell sorting (FACS) to select variants of GFP that fluoresce between 20-and 35-fold more intensely than wild type (wt), when excited at 488 nm. Sequence analysis reveals three classes of aa substitutions in GFP. All three classes of mutant proteins have highly shifted excitation maxima. In addition, when produced in E. coli, the folding of the mutant proteins is more efficient than folding of wt GFP. These two properties contribute to a greatly increased (100-fold) fluorescence intensity, making the mutants useful for a number of applications.     

Identification of apocalmodulin and Ca2+-calmodulin regulatory domain in skeletal muscle Ca2+ release channel, ryanodine receptor

Fusion proteins and full-length mutants were generated to identify the Ca(2+)-free (apoCaM) and Ca(2+)-bound (CaCaM) calmodulin binding sites of the skeletal muscle Ca(2+) release channel/ryanodine receptor (RyR1). [(35)S]Calmodulin (CaM) overlays of fusion proteins revealed one potential Ca(2+)-dependent (aa 3553-3662) and one Ca(2+)-independent (aa 4302-4430) CaM binding domain. W3620A or L3624D substitutions almost abolished completely, whereas V3619A or L3624A substitutions reduced [(35)S]CaM binding to fusion protein (aa 3553-3662). Three full-length RyR1 single-site mutants (V3619A,W3620A,L3624D) and one deletion mutant (Delta4274-4535) were generated and expressed in human embryonic kidney 293 cells. L3624D exhibited greatly reduced [(35)S]CaM binding affinity as indicated by a lack of noticeable binding of apoCaM and CaCaM (nanomolar) and the requirement of CaCaM (micromolar) for the inhibition of RyR1 activity. W3620A bound CaM (nanomolar) only in the absence of Ca(2+) and did not show inhibition of RyR1 activity by 3 microm CaCaM. V3619A and the deletion mutant bound apoCaM and CaCaM at levels compared with wild type. V3619A activity was inhibited by CaM with IC(50) approximately 200 nm, as compared with IC(50) approximately 50 nm for wild type and the deletion mutant. [(35)S]CaM binding experiments with sarcoplasmic reticulum vesicles suggested that apoCaM and CaCaM bind to the same region of the native RyR1 channel complex. These results indicate that the intact RyR1 has a single CaM binding domain that is shared by apoCaM and CaCaM.     

Topogenesis of a nucleolar protein: determination of molecular segments directing nucleolar association.

To identify the element(s) in nucleolar proteins which determine nucleolus-specific topogenesis, we have used different kinds of cDNA constructs encoding various chimeric combinations of mutants of the constitutive nucleolar protein NO38 (B23): 1) with an amino terminally placed short "myc tag"; 2) with two different carboxyl terminally attached large alpha-helical coiled coil structures, the lamin A rod domain or the rod domain of vimentin; 3) with the sequence-related nucleoplasmic histone-binding protein nucleo-plasmin; and 4) with the soluble cytoplasmic protein pyruvate kinase. To avoid the problem of formation of complexes with endogenous wild-type (wt) molecules and "piggyback" localization, special care was taken to secure that the mutants and chimeras used did not oligomerize as is typical of protein NO38 (B23). Using microinjection and transfection of cultured cells, we found that the segment comprising the amino-terminal 123 amino acids (aa) alone was sufficient to effect nucleolar accumulation of the construct molecules, including the chimeras with the entire rod domains of lamin A and vimentin. However, when the amino-terminal 109 aa were deleted, the molecules still associated with the nucleolus. The results of further deletion experiments and of domain swaps with nucleoplasmin all point to the topogenic importance of two independent molecular regions located at both the amino- and carboxyl-terminal end. Our definition of dominant elements determining the nucleolar localization of protein NO38 (B23) as well as of diverse nonnucleolar proteins will help to identify its local binding partner(s) and functions, the construction of probes examining other proteins or sequence elements within the nucleolar microenvironment, and the generation of cells with an altered nuclear architecture.     

Human monoclonal antibodies isolated from type I diabetes patients define multiple epitopes in the protein tyrosine phosphatase-like IA-2 antigen

Protein tyrosine phosphatase-like IA-2 autoantigen is one of the major targets of humoral autoimmunity in patients with insulin-dependant diabetes mellitus (IDDM). In an effort to define the epitopes recognized by autoantibodies against IA-2, we generated five human mAbs (hAbs) from peripheral B lymphocytes isolated from patients most of whom had been recently diagnosed for IDDM. Determination and fine mapping of the critical regions for autoantibody binding was performed by RIA using mutant and chimeric constructs of IA-2- and IA-2beta-regions. Four of the five IgG autoantibodies recognized distinct epitopes within the protein tyrosine phosphatase (PTP)-like domain of IA-2. The minimal region required for binding by three of the PTP-like domain-specific hAbs could be located to aa 777-979. Two of these hAbs cross-reacted with the related IA-2beta PTP-like domain (IA-2beta aa 741-1033). A further PTP-like domain specific hAb required the entire PTP-like domain (aa 687-979) for binding, but critical amino acids clustered in the N-terminal region 687-777. An additional epitope could be localized within the juxtamembrane domain (aa 603-779). In competition experiments, the epitope recognized by one of the hAbs was shown to be targeted by 10 of 14 anti-IA-2-positive sera. Nucleotide sequence analysis of this hAb revealed that it used a V(H) germline gene (DP-71) preferably expressed in autoantibodies associated with IDDM. The presence of somatic mutations in both heavy and light chain genes and the high affinity or this Ab suggest that the immune response to IA-2 is Ag driven.