In Vitro Selection of Membrane-spanning Leucine Zipper Protein-Protein Interaction Motifs Using POSSYCCAT
2001; Elsevier BV; Volume: 276; Issue: 49 Linguagem: Inglês
10.1074/jbc.m105362200
ISSN1083-351X
AutoresRolf Gurezka, Dieter Langosch,
Tópico(s)Lipid metabolism and biosynthesis
ResumoA membrane-spanning heptad repeat motif mediates interaction between transmembrane segments. This motif was randomized with three different sets of mostly hydrophobic residues in the context of POSSYCCAT, a modified ToxR transcription activator system. The resulting combinatorial libraries were subjected to different levels of selective pressure to obtain groups of transmembrane segments that are distinguished by their ability to self-interact in bacterial membranes. Upon relating self-interaction to amino acid composition, the following conclusions were made. First, randomization with only Leu, Ile, Val, Met, and Phe resulted in unexpected robust self-interaction with little sequence specificity. Second, with more complex amino acid mixtures that represent natural transmembrane segments more closely, self-interaction critically depended on amino acid composition of the interface. Whereas the contents of Ile and Leu residues increased with the ability to self-interact, the contents of Pro and Arg residues decreased. Third, heptad repeat motifs composed of Leu, Ile, Val, Met, and Phe were ∼40-fold over-represented in transmembrane segments of single-span membrane proteins as compared with motifs composed of the more complex amino acid mixtures. This suggests that heptad motifs composed of the smaller subset of amino acids were enriched in the course of natural single-span membrane protein evolution. A membrane-spanning heptad repeat motif mediates interaction between transmembrane segments. This motif was randomized with three different sets of mostly hydrophobic residues in the context of POSSYCCAT, a modified ToxR transcription activator system. The resulting combinatorial libraries were subjected to different levels of selective pressure to obtain groups of transmembrane segments that are distinguished by their ability to self-interact in bacterial membranes. Upon relating self-interaction to amino acid composition, the following conclusions were made. First, randomization with only Leu, Ile, Val, Met, and Phe resulted in unexpected robust self-interaction with little sequence specificity. Second, with more complex amino acid mixtures that represent natural transmembrane segments more closely, self-interaction critically depended on amino acid composition of the interface. Whereas the contents of Ile and Leu residues increased with the ability to self-interact, the contents of Pro and Arg residues decreased. Third, heptad repeat motifs composed of Leu, Ile, Val, Met, and Phe were ∼40-fold over-represented in transmembrane segments of single-span membrane proteins as compared with motifs composed of the more complex amino acid mixtures. This suggests that heptad motifs composed of the smaller subset of amino acids were enriched in the course of natural single-span membrane protein evolution. positive selection system based on chromosomally integrated CAT (a modified version of the ToxR system previously used to study interactions between transmembrane segments) heptad repeat motif of Leu on an Ala host sequence chloramphenicol acetyltransferase ToxR protein lacking a transmembrane segment maltose binding protein Miller units polymerase chain reaction transmembrane segment Sequence-specific interactions between α-helical transmembrane segments are known to drive homo- and heterotypic assembly of many integral membrane proteins (1Lemmon M.A. Engelman D.M. Q. Rev. Biophys. 1994; 27: 157-218Crossref PubMed Scopus (177) Google Scholar, 2Popot J.-L. Engelman D.M. Annu. Rev. Biochem. 2000; 69: 881-922Crossref PubMed Scopus (529) Google Scholar). Self-assembly of transmembrane segment helices depends on steric complementarity of their characteristically shaped surfaces allowing for multiple van der Waals interactions (1Lemmon M.A. Engelman D.M. Q. Rev. Biophys. 1994; 27: 157-218Crossref PubMed Scopus (177) Google Scholar, 3Dieckmann G.R. DeGrado W.F. Curr. Opin. Struct. Biol. 1997; 7: 486-494Crossref PubMed Scopus (49) Google Scholar, 4Bowie J.U. Nat. Struct. Biol. 2000; 7: 91-94Crossref PubMed Scopus (33) Google Scholar). In addition, hydrogen bonding between membrane-embedded polar residues has been shown to drive interaction of model transmembrane segments (5Zhou F.X. Cocco M.J. Russ W.P. Brunger A.T. Engelman D.M. Nat. Struct. Biol. 2000; 7: 154-160Crossref PubMed Scopus (353) Google Scholar, 6Choma C. Gratkowski H. Lear J.D. DeGrado W.F. Nat. Struct. Biol. 2000; 7: 161-166Crossref PubMed Scopus (341) Google Scholar, 7Zhou F.X. Merianos H.J. Brünger A.T. Engelman D.M. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 2250-2255Crossref PubMed Scopus (314) Google Scholar). Depending on the geometry of amino acid packing, pairs of transmembrane segment helices favor either positive or negative crossing angles (8Langosch D. Heringa J. Proteins Struct. Funct. Genet. 1998; 31: 150-160Crossref PubMed Scopus (119) Google Scholar, 9Bowie J.U. J. Mol. Biol. 1997; 272: 780-789Crossref PubMed Scopus (278) Google Scholar). Previously, we demonstrated that the interfaces of those pairs of transmembrane segments adopting positive crossing angles within bacteriorhodopsin, the photosynthetic reaction center, and cytochrome-coxidase are described by a repeated heptad (a..de.g) pattern of amino acids that is characteristic of soluble leucine zipper interaction domains (8Langosch D. Heringa J. Proteins Struct. Funct. Genet. 1998; 31: 150-160Crossref PubMed Scopus (119) Google Scholar). Apart from these multispan proteins, membrane-spanning leucine zipper domains were also found to drive assembly of certain single-span membrane proteins (10Arkin I.T. Adams P.D. MacKenzie K.R. Lemmon M.A. Brünger A.T. Engelman D.M. EMBO J. 1994; 13: 4757-4764Crossref PubMed Scopus (173) Google Scholar, 11Simmerman H.K.B. Kobayashi Y.M. Autry J.M. Jones L.R. J. Biol. Chem. 1996; 271: 5941-5946Abstract Full Text Full Text PDF PubMed Scopus (222) Google Scholar, 12Pinto L.H. Dieckmann G.R. Gandhi C.S. Papworth C.G. Braman J. Shaugnessy M.A. Lear J.D. Lamb R.A. DeGrado W.F. Proc. Natl. Acad. Sci. U. S. A. 1997; 94: 11301-11306Crossref PubMed Scopus (317) Google Scholar). Furthermore, a ga..de.ga..de.ga heptad repeat motif of Leu residues drives self-assembly of artificial transmembrane segments in natural membranes and in detergent solution (13Gurezka R. Laage R. Brosig B. Langosch D. J. Biol. Chem. 1999; 274: 9265-9270Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar). Interestingly, variants of this Leu-rich heptad motif were found within the transmembrane segments of a diverse set of single-span natural membrane proteins where they appear to be important for oligomeric assembly and function (13Gurezka R. Laage R. Brosig B. Langosch D. J. Biol. Chem. 1999; 274: 9265-9270Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar, 14Kubatzky K.F. Ruan W. Gurezka R. Cohen J. Ketteler R. Watowich S.S. Neumann D. Langosch D. Klingmüller U. Curr. Biol. 2001; 11: 110-115Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar, 15Huber O. Kemmler R. Langosch D. J. Cell Sci. 1999; 112: 4415-4423Crossref PubMed Google Scholar). Here, we investigated the relationship between self-interaction of a membrane-spanning heptad motif and its amino acid composition. To this end, we randomized the ga..de.ga..de.ga motif with three different sets of mostly hydrophobic residues and expressed the resulting sequences in the context of POSSYCCAT1(Positive Selection System based onChromosomally integrated CAT), which is a modified version of the ToxR system previously used to study interactions between transmembrane segments (16Langosch D.L. Brosig B. Kolmar H. Fritz H.-J. J. Mol. Biol. 1996; 263: 525-530Crossref PubMed Scopus (217) Google Scholar, 17Brosig B. Langosch D. Protein Sci. 1998; 7: 1052-1056Crossref PubMed Scopus (192) Google Scholar). ToxR chimeric proteins harboring the random transmembrane segment were then isolated with or without selective pressure. Subsequently, the relationship between self-assembly and amino acid composition of the interfacial positions was established. The ToxR/MalE coding region from vector pToxRΔTM (13Gurezka R. Laage R. Brosig B. Langosch D. J. Biol. Chem. 1999; 274: 9265-9270Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar) was amplified using a sense primer with an XbaI site, the ribosome binding site from pASK111 (18Skerra A. Gene. 1994; 151: 131-135Crossref PubMed Scopus (302) Google Scholar), and an antisense primer, including a sequence encoding the myc antibody tag and anXhoI site. The resulting PCR fragment was inserted into pASK111 cut with XbaI/XhoI. Upon digesting this vector with XbaI/AseI, the fragment harboring the ToxR/MalE coding region, the F1 origin, and the chloramphenicol resistance gene was isolated. This fragment was ligated to a PCR fragment of vector pBAD (Invitrogen) harboring the araBADpromoter, the araC gene, and the ColE1 origin as well as terminal XbaI and AseI sites. In the resulting vector the BamHI site of the araBAD promoter was eliminated by point mutagenesis. Upon digesting this vector withPstI/AseI, the fragment harboring thearaBAD promoter, the ToxR/MalE coding region, and thearaC gene was isolated. This fragment was ligated to a PCR fragment harboring the kanamycin resistance gene and the F1 origin, which was amplified from vector pET24d (Novagen) using a sense primer with the T2 transcription terminator from the cat2 gene fromStaphylococcus aureus plus an NdeI site and an antisense primer with a PstI site. This resulted in vector pToxRIVΔTM. Derivatives of pToxRIV encoding AZ2, L1/6/8/13/15A, or L2/5/9/12/16A transmembrane segments were made by ligating synthetic oligonucleotide cassettes encoding the desired sequences into pToxRIV previously cut with NheI and BamHI. The ctx promoter region of plasmid pLDR10-ctxΔsiglacZ (kindly supplied by H.-J. Fritz, Göttingen) was amplified using a sense primer with anXbaI site and an antisense primer where 34 nucleotides at the 3′-end were homologous to the 5′-end of the cat gene. The cat gene of plasmid pLDR10-ctxΔsiglacZ was amplified using the foregoing PCR fragment as the sense primer and an antisense primer, including the T2 terminator and a BamHI site. The resulting PCR fragment was ligated to a restriction fragment, which resulted from digesting pLDR10 (19Diederich L. Rasmussen L.J. Messer W. Plasmid. 1992; 28: 14-24Crossref PubMed Scopus (119) Google Scholar) withXbaI/BamHI and harbors the attP phage lambda attachment site, a ColE1-like origin, and catand bla genes. The resulting vector is termed pLDR10ctx::cat. The ctx::cat hybrid gene contained in vector pLDR10ctx::cat was integrated into the attB phage lambda attachment site ofEscherichia coli DH5α as described previously (19Diederich L. Rasmussen L.J. Messer W. Plasmid. 1992; 28: 14-24Crossref PubMed Scopus (119) Google Scholar). The resulting strain is termed Chr3. Combinatorial libraries were constructed by oligonucleotide-directed mutagenesis (Bio-Rad MutaGene II kit) using a single-stranded template derived from vector pToxRIVΔTM and the following antisense primers: 5′-ATT GGC TTG GGT TGA TCA GAA TCC CSA NSA NTG CSA NSA NAG CTG CSA NSA NTG CSA NSA NAG CTG CSA NSA NGC TAG CTC GAT TCC CCA AGT (library III), 5′-ATT GGC TTG GGT TGA TCA GAA TCC CAV BAV BTG CAV BAV BAG CTG CAV BAV BTG CAV BAV BAG CTG CAV BAV BGC TAG CTC GAT TCC CCA AGT (library IV), and 5′-ATT GGC TTG GGT TGA TCA GAA TCC CRV HRV HTG CRV HRV HAG CTG CRV HRV HTG CRV HRV HAG CTG CRV HRV HGC TAG CTC GAT TCC CCA AGT (library V). B, H, N, R, S, and V refer to equimolar mixtures of G/T/C, A/C/T, G/A/T/C, A/G, G/C, and G/C/A, respectively. Crude libraries were obtained by transforming the mutagenesis reactions into XL1-blue cells (Stratagene). PD28 libraries were obtained by digesting plasmid preparations (2 μg) isolated from the crude libraries with 24 units of BamHI for 2 h and subsequently with 400 units of exonuclease III for 3 h at 37 °C. This procedure removed un-mutagenized parental pToxRIVΔTM plasmids. To select for ToxR chimeric proteins integrated into the inner membrane, the plasmids were transformed into PD28 cells and grown for 16 h at 37 °C in M9 minimal medium containing 0.4% maltose and 33 μg/ml kanamycin as described previously (17Brosig B. Langosch D. Protein Sci. 1998; 7: 1052-1056Crossref PubMed Scopus (192) Google Scholar). Plasmid preparations isolated from crude or from PD28 libraries were transformed into Chr3 cells and grown overnight at 37 °C in LB medium containing 50 μg/ml ampicillin plus 33 μg/ml kanamycin. The overnight culture was inoculated at a dilution of 1:100 into LB medium containing 50 μg/ml ampicillin, 33 μg/ml kanamycin, 2% (w/v) glucose, and 1% (w/v) l-arabinose. AtA 600 = 0.6 the cultures were diluted 50- to 500-fold and plated onto LB agar plates containing 50 μg/ml ampicillin, 33 μg/ml kanamycin, 2% (w/v) glucose, 1% (w/v)l-arabinose plus 0, 30, 60, 90, 120, or 180 μg/ml chloramphenicol. Upon incubation for 16–24 h at 37 °C, colonies were picked and plasmids were isolated and sequenced. Based on the sequences obtained from crude libraries without chloramphenicol selection, we found that their amino acid compositions closely matched the theoretical compositions defined by codon usage and given in Fig.3 A (data not shown). Plasmid-transformed FHK12 cells were grown for 24 h at 37 °C under shaking in the presence of 2% (w/v) glucose, 0.4 mm isopropyl-1-thio-β-d-galactopyranoside, 33 μg/ml kanamycin, and 0.2% (w/v) l-arabinose unless indicated otherwise. β-Galactosidase activity of cell-free extracts was determined as described previously (16Langosch D.L. Brosig B. Kolmar H. Fritz H.-J. J. Mol. Biol. 1996; 263: 525-530Crossref PubMed Scopus (217) Google Scholar) and is given in Miller units (MU). Western blotting was done as described with an antiserum recognizing the MalE moiety of the constructs (16Langosch D.L. Brosig B. Kolmar H. Fritz H.-J. J. Mol. Biol. 1996; 263: 525-530Crossref PubMed Scopus (217) Google Scholar). The predicted transmembrane segments of single-span or multispan membrane proteins contained in Swiss-Prot version 39+ (total of 94,151 entries) were searched with theXX..XX.XX..XX.XXheptad repeat motif where X corresponds to any amino acid used for transmembrane segment randomization of libraries III, IV, or V; dots correspond to any amino acid except the charged residues lysine, arginine, glutamate, and aspartate. For single-span membrane proteins, a subset of 4348 proteins that contain single predicted transmembrane segments was searched. For multispan membrane proteins, a subset of 58,849 transmembrane segments contained within 8425 proteins was searched. Because prediction of the exact transmembrane segment termini is uncertain and dependent on the prediction method, we included 5 residues at either terminus in the searches. Very similar results were obtained, however, when the searches were restricted to the predicted membrane-spanning sequences. All searches were conducted using the Findpatterns option of the HUSAR sequence analysis package made available by the German Cancer Research Center, Heidelberg. Chimeric ToxR proteins consist of the cytoplasmic ToxR domain, a transmembrane segment of choice, and a periplasmic maltose binding protein (MalE) domain (20Kolmar H. Hennecke F. Götze K. Janzer B. Vogt B. Mayer F. Fritz H.-J. EMBO J. 1995; 14: 3895-3904Crossref PubMed Scopus (63) Google Scholar). Anchored within the inner membrane of expressing E. coli cells, these ToxR chimera are thought to exist in monomer/dimer equilibria depending on the mutual affinity of the respective transmembrane segments (16Langosch D.L. Brosig B. Kolmar H. Fritz H.-J. J. Mol. Biol. 1996; 263: 525-530Crossref PubMed Scopus (217) Google Scholar, 17Brosig B. Langosch D. Protein Sci. 1998; 7: 1052-1056Crossref PubMed Scopus (192) Google Scholar). The ToxR domain of the self-assembled form activates the cholera toxin (ctx) promoter thus driving expression of a downstream lacZreporter gene in FHK12 reporter cells. β-Galactosidase activity is therefore diagnostic of ToxR assembly in the membrane. Because self-assembly of transmembrane segments is expected to exhibit concentration dependence like any reversible protein-protein interaction, we constructed the pToxRIV vector allowing for regulated expression. To this end, we replaced the endogenous ToxR promoter (21Kolmar H. Fritsch C. Kleeman G. Götze K. Stevens F.J. Fritz H.J. Biol. Chem. Hoppe-Seyler. 1994; 375: 61-69Crossref PubMed Scopus (40) Google Scholar) by the araBAD promoter, which is induced by arabinose and repressed by glucose (22Niland P. Hühne R. Müller-Hill B. J. Mol. Biol. 1996; 264: 667-674Crossref PubMed Scopus (56) Google Scholar). In an initial experiment, we compared β-galactosidase activities elicited by different constructs, which differ in their abilities to self-interact. Accordingly, we compared the AZ2 motif, which is a previously established self-interacting heptad repeat motif of Leu residues (13Gurezka R. Laage R. Brosig B. Langosch D. J. Biol. Chem. 1999; 274: 9265-9270Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar), two mutants where Leu ata and d or at e and gpositions was replaced by Ala (L1/6/8/13/15A, L2/5/9/12/16A), and a ToxR protein lacking the transmembrane segment (ΔTM) (Fig.1). As expected, increasing arabinose concentrations strongly increased total β-galactosidase activities while the relative differences in signal strength elicited by the different constructs decreased. That is, experimental discrimination between transmembrane segments of different mutual affinity is optimized by precisely tuning the levels of ToxR protein expression. In vitro selection of rare self-interacting transmembrane segments requires exposure of combinatorial libraries to selective pressure where only cells expressing transcriptionally active ToxR proteins survive. To construct a system allowing for positive selection, we placed the gene encoding chloramphenicol acetyltransferase (CAT) under transcriptional control of the ctx promoter. A single copy of thectx::cat hybrid sequence was stably integrated into the lambda attB site of the E. coli chromosome thus yielding the Chr3 reporter strain (Fig.2). Because self-assembling ToxR proteins confer chloramphenicol resistance to Chr3 cells, this system is termed POSSYCCAT. Transmembrane segments of different mutual affinity can be readily distinguished from each other using POSSYCCAT. By monitoring colony size as an indicator of growth inhibition by 90 or 120 μg/ml chloramphenicol, we found that the presence of 1% (w/v) arabinose plus 2% (w/v) glucose clearly differentiated between AZ2 and the mutants described above. Upon plating at 180 μg/ml CM, only AZ2-expressing cells survived. Thus, these conditions are suited to select for self-interacting transmembrane segments. Pursuing a somewhat different strategy, Russ and Engelman (23Russ W.P. Engelman D.M. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 863-868Crossref PubMed Scopus (321) Google Scholar) have previously constructed the TOXCAT system where the ctx::cat sequence is on a multicopy plasmid. By monitoring colony size, we found that cells harboring a plasmid-borne ctx::catsequence could not reproducibly distinguish between the degree of chloramphenicol resistance conferred by the AZ2 motif, its mutants, and ToxRΔTM. Thus, the presence of only one copy of the catreporter gene, as in POSSYCCAT, significantly increased the power of the selection system to discriminate between interactions of different affinity. Thea, d, e, and g positions of the ga..de.ga..de.ga motif constituting the interface of pairs of transmembrane segments with positive crossing angles (8Langosch D. Heringa J. Proteins Struct. Funct. Genet. 1998; 31: 150-160Crossref PubMed Scopus (119) Google Scholar, 13Gurezka R. Laage R. Brosig B. Langosch D. J. Biol. Chem. 1999; 274: 9265-9270Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar) were randomized. Three different sets of almost exclusively hydrophobic amino acids encoded by degenerate codons were chosen (Fig.3 A). These residues correspond to five (library III) or seven (libraries IV and V) of the nine most abundant amino acids in transmembrane segments (24Arkin I.T. Brünger A.T. Biochim. Biophys. Acta. 1998; 1429: 113-128Crossref PubMed Scopus (163) Google Scholar, 25Senes A. Gerstein M. Engelman D.M. J. Mol. Biol. 2000; 296: 921-936Crossref PubMed Scopus (509) Google Scholar). Ala was chosen for the non-interfacial b, c, and fpositions, because an oligo-Ala sequence does not self-interact (13Gurezka R. Laage R. Brosig B. Langosch D. J. Biol. Chem. 1999; 274: 9265-9270Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar). The corresponding nucleotide sequences were inserted into the pToxRIVΔTM vector by insertion mutagenesis. To assess the average capability of the random transmembrane segment sequences in these libraries to self-interact, we determined β-galactosidase activities of FHK12 cell populations expressing the whole libraries. Library III elicited the highest average β-galactosidase activity followed by libraries V and IV. With libraries III and V, β-galactosidase activity increased substantially upon prior selective digestion (see "Experimental Procedures") of parental pToxRIVΔTM plasmids remaining in these "crude" libraries (Fig. 3 B). Upon elimination of ΔTM plasmids, each library contained ∼104 unique clones. This effect of pre-digestion was compared with the consequences of eliminating ToxR proteins being incapable of membrane integration. The rationale behind this experiment was to obtain a rough measure of the content of non-integrating transmembrane segments in our libraries. To this end, the crude libraries were transformed into and re-isolated from PD28 cells. PD28 is an MalE deletion strain that can survive on minimal media with maltose as the only carbon source only upon expression of membrane-integrated ToxR proteins that present their MalE moiety toward the periplasm (17Brosig B. Langosch D. Protein Sci. 1998; 7: 1052-1056Crossref PubMed Scopus (192) Google Scholar, 26Bedouelle H. Duplay P. Eur. J. Biochem. 1988; 171: 541-549Crossref PubMed Scopus (112) Google Scholar). To test the validity of this approach, PD28 cells transformed with membrane-integrating AZ2 or cytoplasmic ΔTM plasmids (13Gurezka R. Laage R. Brosig B. Langosch D. J. Biol. Chem. 1999; 274: 9265-9270Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar) were grown on rich media and mixed at a ratio of 1:9. Subsequently, this mixture of cells was cultured on maltose-containing minimal medium and plasmids were re-isolated. Restriction digestion showed that 9 out of 10 clones surviving passage through PD28 cells corresponded to AZ2. This indicated that about 99% of non-integrating ΔTM proteins were eliminated by this procedure. Prior passage of library III through PD28 cells increased average β-galactosidase activity similar to pre-digestion of pToxRIVΔTM plasmids. With library V, passage through PD28 cells had a somewhat stronger effect than pre-digestion (Fig. 3 B). With library IV, neither treatment had a strong effect. Accordingly, it appears that the vast majority of our random transmembrane segment sequences, especially those encoded by libraries III and IV, are capable of membrane insertion; only the crude version of library V contained a significant fraction of non-integrating ToxR proteins. In the next step, we tested the ability of the different libraries to confer chloramphenicol resistance to Chr3 cells. Pre-digested libraries were passed through PD28 cells, transformed into Chr3 cells, and plated onto agar plates containing from 0 to 180 μg/ml chloramphenicol. Upon overnight growth, the numbers of surviving colonies were counted. In line with the different average β-galactosidase activities, the libraries displayed different abilities to survive increasing selective pressure. With library III, the number of surviving clones decreased only moderately with increasing chloramphenicol concentration. A similarly modest decrease was observed for cells transformed with pToxRIVAZ2 and is therefore not reflecting successive elimination of non-interacting sequences. In contrast to that, the colony numbers of libraries IV and V decreased much more strongly and as a continuous function of increasing chloramphenicol concentration (Fig.3 C). At 180 μg/ml chloramphenicol, the number of surviving colonies of libraries III, V, and IV was reduced to about 42, 17, and 0.11%, respectively, of the numbers that were originally plated. This suggests a broader distribution of affinities in libraries IV and V than in library III. Taken together, randomizing with only Leu, Ile, Val, Met, and Phe appeared to produce mainly well interacting sequences. However, increasing the complexity of the amino acid complement used for randomization strongly decreased the fractions of self-interacting transmembrane segments. This is reflected by decreased average β-galactosidase activities as well as by decreased fractions of colonies surviving chloramphenicol selection. To obtain transmembrane segments with different mutual affinities, we isolated ToxR proteins from Chr3 cells cultured with or without selective pressure. In the first step, we isolated ToxR plasmids without selective pressure, yet passed through PD28 cells to exclusively obtain proteins capable of membrane insertion. Individual plasmids were then re-transformed into FHK12 cells to determine the β-galactosidase activities elicited by the encoded ToxR proteins. According to this quantitative measure of self-interaction, the sequences were classified into four different groups representing low (<200 MU), medium (200–800 MU), high (800–1500 MU), or very high (1500–2000 MU) β-galactosidase activities (Table I). Surprisingly, ToxR proteins derived from library III under these unselective conditions almost exclusively elicited β-galactosidase activities similar to (800–1500 MU) or above (1500–2000 MU) that of the AZ2 construct, which corresponds to 1128 ± 242 MU (see Fig.1). On the other hand, mostly low affinity transmembrane segments were isolated without selection from libraries IV and V, as expected.Table ITMS sequences pooled according to self-interactionRange of β-galactosidase activities elicited in FHK12 cells<200200–800800–15001500–2000MULibrary IIIFM..IV.VV..II.ILIV..VV.VM..LI.LL*IL..IM.ML..IF.LL*LF..IL.VL..LI.IIIV..IL.IL..IL.MMLL..IL.LV..II.LV*LI..FF.MT..FL.LL*LV..VI.FL..VL.FV*LL..MF.LF..LI.LM*FI..IF.ML..VF.FI*IL..IM.FL..VP.LM*IV..VF.VL..LI.IL*IL..LI.LL..FI.FL*IV..VF.II..VL.LL*II..LI.VL..IV.IM*LV..VF.II..IF.LL*MI..FF.II..LL.LL*IL..VF.FI..VV.VV*LI..LF.LL..MI.ILFL..FV.VI..LV.VL*LV..FV.MF..FL.LIML..VI.LI..LV.VMMV..ML.ML..LV.LVLL..IM.LF..VI.LLVI..VL.MV..MM.LLLL..LF.MV..VL.MMIL..II.II..PI.IILL..MI.FM..VM.VVML..VL.LM..FL.FILL..LL.LL..LI.LI*IL..LL.VI..VM.LLLL..MV.II..ML.IVVL..MV.ML..MI.LAML..FV.MV..PI.IVIM..LF.ML..LI.LLIV..VF.IL..VL.LV*IM..FM.II..IL.MFLI..LV.FI..VL.MLVL..FL.IL..IL.LLLV..LL.VV..IL.LIIL..ML.IL..VF.FILibrary VSF..CS.IV..GS.CFCC..AS.CA..VT.IS*FV..SC.FA..CG.CA*VC..GS.VS..IV.VAAF..SI.TS..TA.IT*TI..AV.CC..AC.CG*VV..VT.AT..GI.TSFF..SI.SA..CS.FC*TG..II.VI..VF.VS*TS..PC.FC..AG.SAGC..CS.CG..IC.FV*VF..II.VA..VV.VA*AG..AC.TC..FS.IAFV..VM.LV..VM.LL*VA..FA.CI..CS.SC*IF..GF.AA..TT.STFA..AC.TS..IS.CI*SA..SI.VS..VG.AV*SV..TS.SA..ST.SATI..GI.SS..VG.FIGC..GT.AA..IG.ICGV..SA.IA..AT.VASS..IG.IG..IT.CTFT..FV.VA..CV.SGTC..SV.GC..GF.AATT..AC.IT..AT.AACF..GS.CS..VF.CVVS..AF.CA..SS.CVTG..GT.VC..FS.VVCS..IV.IG..TA.VI*GT..GT.SS..AG.SGGT..VS.FF..AT.SGPI..GT.II..AF.TSGA..PT.IV..GT.ASTI..GI.SS..VG.FTFG..IS.IF..SA.VS*TC..SA.FV..GT.AVFC..TC.IV..TT.IASS..TT.SF..GV.VT*AS..TA.FA..SA.TVCG..AS.II..AA.FGFF..CV.AS..PG.FC*IC..SI.AF..CG.ASFF..CV.AS..IG.FC*SV..AT.TA..VS.VF*GA..PT.IV..GT.ATFG..IF.GT..GA.FA*SF..AV.SA..CI.AA*VC..FA.TS..AF.AFSF..IV.IS..AG.FS*TS..GA.AI..SS.AGVV..CA.SF..GC.VT*GF..SF.AF..CS.II*SA..FA.TF..TT.VI*Library IVVP..VV.SV..TA.AVPI..PT.TL..LA.IG*RT..AL.LI..LL.GV*VA..AV.IA..GA.TTRV..LP.IT..LA.SA*LP..AI.LT..GL.IG*LS..PS.LI..GV.LSPG..IA.AA..AL.AS*PP..VI.TT..AA.LI*TG..AA.TT..VL.VAGT..II.II..VA.TI*TT..AL.TA..AV.FG*LT..TV.II..ST.GSTL..AV.TT..LA.TA*SS..LV.LG..LT.TL*TP..LT.SV..LT.LTIL..II.VS..TA.ST*IT..IG.LA..TS.LI*PA..VQ.RI..ST.ARRI..VP.IA..AG.LA*PS..IR.IS..AG.LI*TV..PT.PS..PP.VTSI..PT.IT..IA.AV*SA..RA.PL..AL.RL*LG..RP.RA..TP.PTPV..IP.AI..IA.AS*IP..AI.SA..IA.LV*TG..TG.PT..AA.SLPT..IP.TL..AG.LI*QT..LL.VI..TA.TT*TA..IV.LR..TT.RRPP..VQ.VL..IT.AT*VP..VL.SV..TA.AV*GL..VG.AG..TP.RPRR..ST.LI..TT.LT*AA..IT.AT..TT.AI*TA..VS.PT..RP.RTPS..RT.TV..TP.RLVP..VL.TV..TA.AI*SP..RL.AA..AT.TTVT..AA.LA..SL.SP*PA..LA.SS..IT.LA*VS..RP.GP..RT.PTTP..IA.II..IT.VI*LP..AI.LT..GL.IG*GV..RA.PR..TP.TPGP..AL.LG..IT.AA*TL..RT.GT..RT.LSAT..PT.AP..RA.VTTP..TA.PS..GP.IIVT..RI.PA..PP.ILIR..AT.PV..PP.RPRT..PL.LR..EP.VLLS..VR.LG..TA.ITRL..PR.TT..RP.TPRA..SG.TA..SV.APTA..RT.LR..LR.PPIP..RT.GT..TA.ATGP..GP.SR..LP.VPHeptad repeat sequences are shown. Dots represent invariant Ala residues. Asterisks denote heptad repeat sequences that were obtained upon selection by chloramphenicol. All other sequences were obtained without selection. Open table in a new tab Heptad repeat sequences are shown. Dots represent invariant Ala residues. Asterisks denote heptad repeat sequences that were obtained upon selection by chloramphenicol. All other sequences were obtained without selection. In the second step, we applied selective pressure to obtain high affinity transmembrane segments. To this end, colonies were picked upon selection by chloramphenicol concentrations from 60 to 180 μg/ml. Again, the degrees of self-interaction were then determined by analyzing β-galactosidase activity in FHK12 cells. Prior passage through PD28 cells could be omitted here, because cells expressing non-inserting ToxR proteins, as ex
Referência(s)