Identification of Important Amino Acid Residues That Modulate Binding of Escherichia coli GroEL to Its Various Cochaperones

Corresponding author: Costa Georgopoulos, Département de Biochimie Médicale, Centre Médical Universitaire, Université de Genève, 1, rue Michel Servet, 1211 Geneva 4, Switzerland., costa.georgopoulos{at}medecine.unige.ch (E-mail)

Communicating editor: A. L. SONENSHEIN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic experiments have shown that the GroEL/GroES chaperone machine of Escherichia coli is absolutely essential, not only for bacterial growth but also for the propagation of many bacteriophages including . The virulent bacteriophages T4 and RB49 are independent of the host GroES function, because they encode their own cochaperone proteins, Gp31 and CocO, respectively. E. coli groEL44 mutant bacteria do not form colonies above 42° nor do they propagate bacteriophages , T4, or RB49. We found that the vast majority (40/46) of spontaneous groEL44 temperature-resistant colonies at 43° were due to the presence of an intragenic suppressor mutation. These suppressors define 21 different amino acid substitutions in GroEL, each affecting one of 13 different amino acid residues. All of these amino acid residues are located at or near the hinge, which regulates the large en bloc movements of the GroEL apical domain. All of these intragenic suppressors support bacteriophages , T4, and RB49 growth to various extents in the presence of the groEL44 allele. Since it is known that the GroEL44 mutant protein does not interact effectively with Gp31, the suppressor mutations should enhance cochaperone binding. Analogous intragenic suppressor studies were conducted with the groEL673 temperature-sensitive allele.

THE groES and groEL genes of Escherichia coli were originally discovered because mutations in them blocked the propagation of bacteriophage (GEORGOPOULOS et al. 1973 ; STERNBERG 1973 ). Subsequently, it was shown both genetically (TILLY and GEORGOPOULOS 1982 ; ZEILSTRA-RYALLS et al. 1994 ) and biochemically (CHANDRASEKHAR et al. 1986 ) that the GroES and GroEL proteins interact. Studies by many laboratories established that GroEL/GroES constitute a chaperone machine, capable of promoting the proper folding of many proteins both in vitro and in vivo (reviewed by HARTL 1996 ; FENTON and HORWICH 1997 ; LORIMER 1997 ; RICHARDSON et al. 1998 ; SIGLER et al. 1998 ). The groES and groEL genes are the only chaperone-encoding genes that are essential for E. coli growth under all conditions examined (FAYET et al. 1989 ). The exact reason for this essentiality is not known, but is likely related to the finding that some of GroEL's in vivo substrates include the products of essential genes (EWALT et al. 1997 ; HOURY et al. 1999 ).

The heptameric GroES protein interacts with the tetradecameric GroEL protein through its mobile loop, as established by both genetic and biochemical criteria (LANDRY et al. 1993 ; HUNT et al. 1996 ). It was shown that all groES mutations isolated originally as blocking bacteriophage growth map in the DNA segment encoding the GroES mobile loop. The GroES mobile loop, upon binding to GroEL, becomes fixed, assuming a ß-hairpin structure centered around a universally conserved glycine (G) residue and followed by three universally conserved hydrophobic residues (LANDRY et al. 1993 ). The crystal structures of both GroEL alone (BRAIG et al. 1994 ) and the GroEL/ADP/GroES complex (XU et al. 1997 ) were solved in the laboratory of the late Paul Sigler at Yale University. In addition, the crystal structure of E. coli GroES was solved by HUNT et al. 1996 . The only contacts seen between GroES and GroEL in the crystal structure are between the conserved hydrophobic groups I25/V26/L27 of GroES mobile loop and L234, L237, and V264 of GroEL (XU et al. 1997 ).

The GroEL/GroES mechanism of in vitro substrate folding was worked out in substantial detail, with major contributions by various laboratories. Briefly, an unfolded substrate exhibiting hydrophobic groups binds to one ring of an unoccupied GroEL molecule (referred to now as the cis ring). The subsequent binding of ATP to the cis ring subunits leads to massive conformational changes of this ring, which permit substrate release in the central GroEL cavity, and GroES binding. Following ATP hydrolysis in the cis ring of GroEL, the binding of another substrate molecule and ATP to the unoccupied ring (termed the trans ring) leads to destabilization and release of GroES and substrate from the cis ring. If the released substrate has not properly folded, it may rebind to the same or other GroEL ring or even to another chaperone protein. The whole GroEL cycle is thought to take 10–15 sec (reviewed in SIGLER et al. 1998 ; HORWICH et al. 1999 ).

Bacteriophage T4 encodes its own GroEL cochaperone, Gp31, the product of gene 31 (NIVINSKAS and BLACK 1988 ; KEPPEL et al. 1990 ; VAN DER VIES et al. 1994 ). The role of Gp31 as a cochaperone was originally inferred from genetic studies (LAEMMLI et al. 1970 ; GEORGOPOULOS et al. 1972 ; RICHARDSON and GEORGOPOULOS 1999 ; see Table 1). Gp31 is an essential protein of bacteriophage T4, and its primary function is to permit the efficient folding of Gp23, the major capsid protein of the bacteriophage (LAEMMLI et al. 1970 ; DOERMANN and SIMON 1984 ; SIMON and RANDOLPH 1984 ; ANDREADIS and BLACK 1998 ). The crystal structure of the Gp31 cochaperone was solved by HUNT et al. 1997 . Its overall heptameric structure is similar to that of GroES, although there is only 14% identity between GroES and Gp31 (KOONIN and VAN DER VIES 1995 ). Despite the low sequence identity, the Gp31 cochaperone can substitute for GroES in E. coli growth (VAN DER VIES et al. 1994 ; F. KEPPEL and C. GEORGOPOULOS, unpublished results). In a recent study, we demonstrated that the virulent bacteriophage RB49, which is distantly related to T4 (MONOD et al. 1997 ), encodes CocO, also a GroEL cochaperone (ANG et al. 2001 ). The CocO and Gp31 cochaperones are 35% identical to each other, can substitute for the folding of each other's major capsid protein, and substitute for GroES in E. coli growth (ANG et al. 2001 ; F. KEPPEL and C. GEORGOPOULOS, unpublished results).

Despite extensive genetic and biochemical studies, our knowledge of the exact mechanistic details of GroEL/GroES and GroEL/Gp31 interactions is still relatively limited. In this work, we sought clues to such mechanistic details by undertaking a genetic study to search for intragenic suppressors of groEL mutations. We took advantage of the existence of the groEL44 and groEL673 temperature-sensitive mutations, which block the growth of bacteriophages , T4, and RB49, as well as bacterial growth at elevated temperatures (GEORGOPOULOS et al. 1973 ). We obtained temperature-resistant revertants of groEL44 and groEL673, sequenced their groEL genes, and extensively analyzed them in relation to their bacteriophage growth pattern. Our combined studies led us to the identification of specific amino acid residues of GroEL that play key roles in promoting conformational changes in GroEL.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Bacteria, bacteriophage, and plasmids:
All of the bacterial strains used in the course of this work were previously described. The parental strain is B178, a W3110 galE sup⁺ derivative, originally obtained from Dale Kaiser at Stanford (SKB178). The various groES and groEL mutants used were described by GEORGOPOULOS et al. 1973 and ZEILSTRARYALLS et al. 1993 , ZEILSTRARYALLS et al. 1994 . They include B178groEL44, B178groEL673, JZ660-B178groES42, JZ661-B178groES42 groEL (V174F), JZ662-B178groES42 groEL(G375S), JZ648-B178 groES619, JZ560-B178groES619 groEL (V190I), JZ564-B178groES619 groEL (G375C), and JZ558-B178groES619 groEL (G375S). All of these strains carry a Tn10 (Tet^R) transposon, 60% cotransducible with the groESgroEL operon by bacteriophage P1 (FAYET et al. 1989 ).

All of the bacteriophages used in the course of this work were previously described, namely cI, T4, T41, T431 (T31A), RB49, RB4922, and P1L4 (GEORGOPOULOS et al. 1972 , GEORGOPOULOS et al. 1973 ; RICHARDSON and GEORGOPOULOS 1999 ; RICHARDSON et al. 1999 ; ANG et al. 2001 ).

Media and buffers:
Luria-Bertani (LB)-broth contains 10 g/liter of Tryptone, 5 g/liter of yeast extract, 5 g/liter NaCl, pH adjusted to 7.4. LB-agar is LB-broth supplemented with 10 g/liter agar. Soft agar is LB-broth supplemented with 6 g/liter agar. TSG buffer is 10 mM Tris/HCl, pH 7.4, 150 mM NaCl, and 0.03% gelatin; and -dil buffer is 10 mM Tris/HCl, pH 7.5, 10 mM MgSO₄.

Isolation of temperature-resistant suppressors:
Individual colonies of groEL44 or groEL673 mutant bacteria were obtained on LB-agar plates at 30°. Part of the colony was picked up using a sterile toothpick, streaked onto an LB-agar plate, and incubated at 43° for 24 hr. A single temperature-resistant (Tr⁺) revertant colony from each streak was further purified through single colony reisolation at 43° on LB-agar plates.

Bacteriophage growth tests:
Approximately 0.2 ml of an overnight bacterial culture, grown at 30°, was added to 3 ml soft agar, mixed, and used to seed fresh LB-agar plates. The various bacteriophages were diluted 10-fold serially, either in -dil buffer (for ) or in TSG buffer (for T4 and RB49 derivatives). Approximately 5 µl of each bacteriophage dilution was deposited on the LB-agar plates, previously seeded with individual bacterial cultures, allowed to dry, and the LB-agar plates were incubated at 37° or 43° for 18 hr. The efficiency of plating of each individual bacteriophage was determined in comparison to its efficiency of plating on B178 wild type at 37°. For the RB4922 bacteriophage mutant, which does not plaque on B178 at 37°, the efficiency of plating was compared to that on its permissive host, groEL44 mutant bacteria at 37°.

Bacteriophage P1-mediated transduction experiments:
Bacteriophage P1-mediated transduction was carried out essentially as previously described by taking advantage of the presence of a Tn10 (Tet^R) transposon, 60% cotransducible with the groESgroEL operon by bacteriophage P1 (FAYET et al. 1989 ). Briefly, B178 wild-type recipient cells, growing exponentially at 37° (5 x 10⁸ cells/ml), were infected with a bacteriophage P1L4 lysate, which was grown on the appropriate donor host, at an approximate multiplicity of infection of 0.1 plaque-forming bacteriophage per recipient cell. Each culture was supplemented with 5 x 10^-3 M CaCl₂ to ensure efficient bacteriophage P1 adsorption. Infected cultures were shaken at 37° for 30 min to allow for adsorption and expression of the Tet^R phenotype. Cells were collected by centrifugation at 5000 rpm for 10 min, resuspended in 100 µl of LB-broth, and spread on LB-agar plates supplemented with 12.5 µg/ml of tetracycline/HCl and 5 x 10^-3 M sodium citrate (to prevent readsorption and killing by bacteriophage P1). The LB-agar plates were incubated at 30° for 24 hr. Individual Tet^R transductants were subsequently tested for growth at 43°, i.e., possession of the Tr⁺ phenotype.

DNA extraction, PCR amplification, and DNA sequencing:
Chromosomal DNA was isolated using the QIAGEN (Valencia, CA) QIAamp DNA mini kit. The groEL gene was divided into approximately three equal parts, for which we designed the following primers: For the middle segment of the groEL gene 5'-CGCAACTCTCTACTGTGGTAAACTGATCGCTGAAGC-3' served as the forward primer and 5'-CAAAACAGCCAAGCTGAGGGCATCTTCAACGCGTGC-3' served as the reverse primer. For the amino-terminal encoding segment of the groEL gene 5'-CGCAACTCTCTACTGTAAACTGCAGGAACGCGTAGC-3' served as the forward primer and 5'-CAAAACAGCCAAGCTGAAACCCCCAGACATTTCTGC-3' served as the reverse primer. For the carboxy-terminal encoding segment of the groEL gene 5'-CAAAACAGCCAAGCTGGAAGTAAGGAGACAGGTAGC-3' served as the forward primer and 5'-CGCAACTCTCTACTGTGGAATAAAGATAATGGCAGC-3'served as the reverse primer.

Particular fragments were amplified using Pfu polymerase (Promega, Madison, WI). PCR products were sequenced using the Amersham Pharmacia Biotech Thermo Sequenase fluorescent-labeled primer cycle sequencing kit with 7-deaza-dGTP and automated sequencing with the Li-CORE system (MWG Biotech).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation of groEL44 Tr⁺ suppressor mutations:
Although groEL44 mutant bacteria were first isolated on the basis of blocking bacteriophage growth at 30°, it was subsequently shown that the bacteria themselves could not grow at temperatures above 42° (GEORGOPOULOS et al. 1973 ; see Table 1). DNA sequencing established that the groEL44 mutation resulted in the E191G amino acid substitution at codon 191 (ZEILSTRA-RYALLS et al. 1993 ) located at the hinge, next to residue G192 that controls the en bloc massive movements of the GroEL apical domain (BRAIG et al. 1994 ; XU et al. 1997 ). By using purified proteins, we showed recently that the GroEL44 mutant protein cannot interact effectively with Gp31 (RICHARDSON et al. 1999 ), the bacteriophage T4-encoded cochaperone (VAN DER VIES et al. 1994 ).

In an effort to elucidate the molecular mechanism by which the GroEL44 mutant protein limits interaction with its cochaperones, we carried out an extensive analysis of groEL44 intragenic suppressors. When groEL44 bacteria are incubated at 43° on LB-agar plates, spontaneous Tr⁺ revertants appear at a frequency of 10^-8. To ensure independent mutational events, we first plated groEL44 bacteria at 30° on LB-agar plates and streaked out 100 of the resulting colonies at 43°. Following a 24-hr incubation at 43°, 46 of the colonies gave rise to at least one Tr⁺ revertant. Only a single well-growing Tr⁺ revertant was kept from each streaked colony, both to ensure that the mutational event was an independent one and that the potential suppressor substantially restored GroEL44 protein function at 43°. The Tr⁺ revertants were purified once at 43°, and a single colony was retained. All Tr⁺ independent revertants grew well, as judged by their colony size at 30° and 43°, compared to the parental wild-type strain. We avoided using chemical or other mutagenic procedures primarily because we wanted to avoid the occurrence of multiple mutations in the groEL44 gene, which would obscure their individual contribution to the suppressed phenotype. This cautious approach paid off since none of our intragenic revertants were found to carry more than one mutation in the groEL gene besides the original groEL44 mutation (see below).

Sequencing of the groEL region of the Tr⁺ revertants:
DNA was extracted from these 46 independent spontaneous Tr⁺ revertants and the groEL DNA region was sequenced from both strands, as detailed in MATERIALS AND METHODS. Four of these Tr⁺ revertants turned out to be true revertants of the groEL44 mutations, inasmuch as the original codon 191, coding for the E191 residue, was restored (all 4 were GGA GAA transitions). Two other Tr⁺ revertants did not show any additional mutations in the groEL gene, besides groEL44. Presumably, they represent extragenic suppressors of the groEL44 mutation. Potentially, they could map either in the groES gene or in another gene, whose product can regulate GroEL levels or activity, such as rpoH (YURA et al. 1993 ). The remaining 40 Tr⁺ revertants were shown to possess both the original groEL44 mutation and an additional single mutation, located between codons 173 and 378 (Table 2). Thus, these 40 Tr⁺ revertants represent intragenic suppressors of the groEL44 mutation.

Overall, the 40 intragenic suppressors represent 23 distinct mutational events and result in 21 different amino acid substitutions in the 13 codons affected (Table 2). For example, the V174F suppressor was the consequence of a single mutational event and was isolated six independent times, thus representing the most frequent type of intragenic suppressor found in this study.

Additional highlights of the intragenic suppressors include the following. There were four different mutational events at codon 331, resulting in three different amino acid substitutions, highlighting its importance in suppressing the GroEL44 temperature-sensitive phenotype. In addition, there were three mutational events at codon 359, also resulting in three different amino acid substitutions. Overall, out of the 13 codons affected, 2 codons could mutate to code for one of three different amino acids (codons 331 and 359), 4 codons could mutate to code for one of two different amino acids (codons 174, 190,194, and 378), and the remaining 7 codons when mutated coded for a single amino acid substitution (codons 173, 176, 189, 322, 371, and 375; see Table 2). A Poisson distribution analysis of the data shown in Table 2 suggests that it is likely that we have identified most of the groEL codons that can spontaneously mutate to restore good growth at 43° to groEL44 mutant bacteria.

To verify that the groEL intragenic suppressors alone are sufficient for restoring growth at 43° to groEL44 mutant bacteria, we carried out bacteriophage P1-mediated transduction experiments. We took advantage of the presence of a Tn10 transposon (Tet^R) near the groEL region that is 60% cotransducible with the groESgroEL operon by bacteriophage P1 (FAYET et al. 1989 ). P1 lysates were grown on the putative Tr⁺ intragenic suppressors and used to transduce the isogenic B178 wild-type parent to Tet^R at 30°. Subsequently, the Tet^R transductant colonies were tested for their ability to propagate at 43°. The rationale of this experiment is as follows. If the suppressor mutations are indeed intragenic, they should be closely linked to the groEL44 mutation and thus should be cotransduced together in the vast majority of the cases. Thus, if the intragenic mutation is indeed responsible for the Tr⁺ phenotype, the vast majority of the Tet^R transductants selected at 30° should grow at 43°. If the intragenic mutation is not responsible for the Tr⁺ phenotype, only 40% of the Tet^R transductants should grow at 43°, i.e., those that do not inherit the donor groEL44 allele. We found that all Tet^R transductants tested (30 from each transduction) grew at 43°. This result is consistent with the conclusion that each of the intragenic suppressors is relatively close to the groEL44 mutation and is responsible for the Tr⁺ phenotype. The only caveat in the interpretation of this genetic experiment is that it does not exclude the formal possibility that an additional mutational event, closely linked to, but not within the groEL gene, could be responsible or coresponsible for the Tr⁺ phenotype in some of the intragenic suppressors.

Effects of intragenic suppressors on bacteriophage growth:
We tested the effect of our groEL44 intragenic suppressor mutations on the growth of bacteriophages , T4, and RB49. Previously, we had shown that the groEL44 mutation blocks the growth of all these bacteriophages, and that the T41 and RB4922 bacteriophage mutants can overcome this block (GEORGOPOULOS et al. 1972 , GEORGOPOULOS et al. 1973 ; ANG et al. 2001 ; see Table 1). The T41 mutation results in the L35I amino acid substitution in the gene 31 product (KEPPEL et al. 1990 ), and the RB4922 mutation results in the Q36R substitution in the gene cocO product (ANG et al. 2001 ). In vitro biochemical experiments with purified proteins have shown that the GroEL44 mutant protein cannot productively interact with either Gp31 or CocO, but does interact productively with the Gp311 and CocO22 mutant proteins. In addition, the mutant RB4922 bacteriophages do not form plaques on the B178 wild-type host at 37° but do so at 43° (Table 1 and Table 3). Thus, the preferential growth of either the T41 or RB4922 mutant over the corresponding wild-type bacteriophage should indicate that the suppressed protein still exhibits a GroEL44-like phenotype. For example, the V190L suppressor blocks bacteriophages T4 and RB49 wild-type growth at 43°, but permits growth of T41 and RB4922, thus demonstrating the retention of at least part of the GroEL44-like phenotype at 43° (Table 3). However, at 37° all bacteriophages tested propagate on the V190L suppressor, suggesting that the suppressed GroEL protein exhibits a phenotype intermediate between that of GroEL⁺ and GroEL44 at 37°.

Table 3 shows that all of the Tr⁺ intragenic suppressors are also capable of supporting the growth of bacteriophages , T4, and RB49, at least to some extent. For example, all of the Tr⁺ suppressors permit growth of bacteriophages and T4 at 37°, and to a more restricted extent at 43°, in the presence of the groEL44 mutation. Specifically, the V174F suppressor restores growth of , T4, and RB49 at either 37° or 43°. This result suggests that V174F could be a relatively "strong" suppressor of the groEL44 mutation. This conclusion is supported by the fact that the V174F suppressor strain is only one of two strains (the other being G375C) that block the growth of RB4922 at 37°, thus behaving like B178 wild-type bacteria in this respect (Table 1 and Table 3).

In contrast to V174F, the G173S suppressor restores growth at 37° but not at 43° for , T4, and RB49 wild type (Table 3). Thus, it appears that the G173S suppressor results in a protein with substantial GroEL⁺ phenotype at 37° but not at 43°. The general conclusion from the results reported in Table 3 is that most Tr⁺ suppressor mutations restore substantial function to the GroEL44 mutant protein, as judged by the pattern and the extent of growth of wild-type and mutant bacteriophages.

Another interesting finding revealed by the results shown in Table 3 is that, in the majority of the cases, the bacteriophage plating pattern observed with different suppressor mutations at the same codon is identical. For example, the D359Y, D359N, and D359G suppressors allow the propagation of all bacteriophages tested at 37° but block the growth of T4 and RB49 wild type at 43° (Table 3). Similarly, the T331S, T331A, and T331N suppressors behave identically with regard to bacteriophage propagation both at 37° and 43°, as do the two suppressors at codon 194 (Q194H or Q194P; Table 3). Finally, the two suppressors at codon 378 (V378G or V378A) are both permissive for all bacteriophages tested at either 37° or 43°. The only exceptions to this pattern of behavior are the two suppressor mutations V190L and V190I. As can be seen in Table 3, the V190I suppressor allows growth of bacteriophage T4 and RB49 wild type at 43°, whereas the V190L suppressor does not.

Some of the Tr⁺ suppressor mutations, by themselves, block bacteriophage growth:
Because all of our intragenic suppressors by definition carry the original groEL44 mutation, we wondered about their individual phenotype in terms of propagation of the various bacteriophages. Since many of them are located very close or even adjacent to groEL44, it would be very difficult to separate them from groEL44 using bacteriophage P1-mediated transduction. However, 3 of the 21 intragenic suppressors were already in our bacterial collection because they were encountered in a previous study and thus existed as individual mutations in the groEL gene. In the previous study, we identified the V174F, V190I, and G375C substitutions in the GroEL protein because they permit growth of bacteria carrying groES temperature-sensitive mutations (ZEILSTRA-RYALLS et al. 1994 ).

We took advantage of existing E. coli strains carrying these three single groEL mutations to study their ability to propagate bacteriophages T4, T41, and RB49 at various temperatures. The bacteriophage T431 (T31A) mutant was also included in this study because of previous evidence that the T31A substitution results in a lower affinity for GroEL (RICHARDSON and GEORGOPOULOS 1999 ; RICHARDSON et al. 1999 ). Although these bacterial strains also carry a groES42 or groES619 mutation, neither mutation should affect the pattern of T4 growth (TILLY et al. 1981 ; TILLY and GEORGOPOULOS 1982 ). Table 4 summarizes the results of this study. All three groEL mutants block RB49 wild-type growth at 30° and 37°. Bacteriophage T4 wild type is also blocked at 30° and 37°, but not as severely as RB49. In contrast, these groEL mutants block in a more pronounced way the growth of T41, whose mutant cochaperone interacts more strongly with GroEL (RICHARDSON et al. 1999 ). Taken collectively, our genetic and biochemical results suggest that the V174F, V190I, and G375C substitutions in GroEL block T4 or RB49 bacteriophage growth by somehow enhancing the interaction between the mutant GroEL and the bacteriophage-encoded cochaperones, in a fashion similar to the interaction between these mutant GroELs and the mutant GroES42 and GroES619 proteins (ZEILSTRA-RYALLS et al. 1994 ). In agreement with this conclusion, and in contrast to T4 and T41, the bacteriophage T431 (T31A) mutant, which encodes a cochaperone with decreased affinity for GroEL, grows very well on all three groEL mutants (Table 4). Thus, it appears that, at least in the case of the V174F, V190I, and G375C substitutions, the groEL44 suppressor mutations function by somehow restoring a more effective interaction with the GroES wild-type protein, thus allowing bacterial growth at 43°.

Suppressor analysis of the groEL673 mutations:
Previously, we sequenced the groEL gene of groEL673 mutant bacteria, which are unable to form colonies at temperatures above 43°. We showed that there are two mutations, one resulting in the G173D substitution, and another resulting in the G337D substitution (ZEILSTRA-RYALLS et al. 1993 ). As noted above, G173 is located in the hinge region, while G337 is located on the exterior of GroEL (BRAIG et al. 1994 ).

We carried out a limited suppressor analysis of the groEL673 mutations by simply isolating independent temperature-resistant colonies at 43° exactly as described for the groEL44 mutation above. The DNA from four such temperature-resistant colonies was extracted, the groEL region was amplified by PCR, and its sequence was determined as described in MATERIALS AND METHODS. Surprisingly, we found that all independent revertants had acquired changes uniquely at codon 173. In two of these revertants, termed type C, the original G173 wild-type codon was restored (Table 5A). In the remaining two revertants, the DNA at codon 173 was changed to code for either an A residue (type A revertant), or an N residue (type B revertant). These results highlight the primary role of the G173D substitution in restricting bacterial growth at 43°.

The importance of the G173D amino acid change in the groEL673 phenotype in blocking bacteriophages , T4, and RB49 growth is also highlighted by the following studies. Table 5B shows that all three types of reversion events at codon 173 restore growth of bacteriophages , T4, and RB49 at 37°. However, at 43° a different spectrum of bacteriophage platings is observed. Type A revertants (G173A) do not plate bacteriophage , whereas type B (G173N) and type C (G173 wild type) do plate bacteriophage (Table 5B). As far as the growth of bacteriophage RB49 goes, revertant type A or B blocks its growth at 43°, whereas type C is permissive, again arguing that the G173D substitution is the critical one in preventing bacteriophage growth on the groEL673 mutant host.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

GroEL/GroES is the only chaperone machine of E. coli that is absolutely essential for bacterial survival under all laboratory conditions tested (FAYET et al. 1989 ). GroEL/GroES homologs are found in all organisms except in some Archaea species (MACARIO et al. 1999 ) and the recently sequenced Ureaplasmum urealyticum mycobacterium (GLASS et al. 2000 ). The E. coli GroEL chaperone can function not only with its own GroES cochaperone, but also with bacteriophage-encoded cochaperones, such as the bacteriophage T4-encoded Gp31 cochaperone (VAN DER VIES et al. 1994 ) or the bacteriophage RB49-encoded CocO cochaperone (ANG et al. 2000 , ANG et al. 2001 ). Apparently, the bacteriophage-encoded cochaperones are uniquely qualified to help in the folding of the major bacteriophage-encoded capsid protein, Gp23 (LAEMMLI et al. 1970 ; GEORGOPOULOS et al. 1972 ; VAN DER VIES et al. 1994 ; ANDREADIS and BLACK 1998 ; ANG et al. 2000 ). Yet, Gp31 and CocO can also help GroEL in its generalized chaperone function, since either can substitute for GroES in E. coli growth (VAN DER VIES et al. 1994 ; ANG et al. 2001 ; F. KEPPEL and C. GEORGOPOULOS, unpublished results).

The GroEL44 mutant protein is defective in its interactions with its various cochaperones as evidenced by the facts that (a) groEL44 mutant bacteria do not propagate bacteriophage , which requires the GroES cochaperone for its proper morphogenesis (GEORGOPOULOS et al. 1973 ; TILLY et al. 1981 ); and (b) groEL44 mutant bacteria block bacteriophage T4 and RB49 growth, but mutations in their cochaperone genes, 311 and cocO22, respectively, can overcome the groEL44-imposed block (GEORGOPOULOS et al. 1972 ; ANG et al. 2001 ). These in vivo observations were reproduced in a purified in vitro protein refolding system; i.e., the GroEL44/Gp31 and GroEL44/CocO combinations do not assist in the folding of citrate synthase, whereas the GroEL44/Gp311 and GroEL44/CocO22 combinations can (RICHARDSON et al. 1999 ; ANG et al. 2001 ). Furthermore, it was shown that the GroEL44 mutant protein cannot form a stable, physical complex with Gp31, but can with the Gp311 mutant protein (RICHARDSON et al. 1999 ). All of these observations, taken together, strongly indicate that GroEL44's molecular defect is an inability to properly interact with its various cochaperones.

The X-ray crystal structures of GroES and Gp31 permitted the recognition of certain structural features that help explain the similarities between the two cochaperones (HUNT et al. 1996 , HUNT et al. 1997 ). The overall structures of GroES and Gp31 are very similar, although the two cochaperones share only 14% identity (KOONIN and VAN DER VIES 1995 ). The most important common feature, besides the sevenfold symmetric structure, is the GroEL-binding mobile loop, first identified in both GroES and Gp31 by nuclear magnetic resonance experiments and limited proteolysis (LANDRY et al. 1993 , LANDRY et al. 1996 ). Upon binding to GroEL, the mobile loop becomes immobilized, adapting a similar 3:5 ß-hairpin structure, centered around a universally conserved glycine (G) residue and followed by three universally conserved hydrophobic residues. All of the groES, 31, and cocO mutations studied thus far in our laboratory affect the mobile loop region (KEPPEL et al. 1990 ; LANDRY et al. 1993 ; RICHARDSON and GEORGOPOULOS 1999 ). For example, the T4311 mutation results in the L35I substitution and the RB49cocO22 mutation in the Q36R substitution in the corresponding mobile loop (KEPPEL et al. 1990 ; ANG et al. 2001 ).

The amino acid mutational change in GroEL44 is at codon 191 (E191G), and the adjacent residue G192 controls the rotation that leads to the massive en bloc movements of the apical domain (60° upward and 90° clockwise), which are induced by ATP binding and needed for GroEL's interaction with its cochaperones (XU et al. 1997 ; SIGLER et al. 1998 ). On the basis of this, we previously hypothesized (RICHARDSON et al. 1999 ; ANG et al. 2000 ) that GroEL44's inability to interact with its cochaperones is due to its limited ability to carry out the apical domain rotations, which are induced by ATP binding and required for cochaperone binding (reviewed by SIGLER et al. 1998 ). Of course, GroEL44 can still undergo these apical domain rotations, since it still functions in vivo with GroES, Gp311, and CocO22, at least at 37°. We hypothesize that the primary mechanism of action of most, if not all, of the groEL44 intragenic suppressors uncovered in this study is to somehow allow the GroEL44 mutant protein to visit more readily the open conformation, thus favoring interaction with all of its cochaperones. The fact that GroEL44, carrying any of the intragenic suppressor mutations identified in this work, is capable of supporting not only bacterial growth at 43° but also bacteriophage growth demonstrates a more efficient interaction of the suppressed GroEL44 protein with GroES. The fact that the suppressed GroEL44 protein can also support, at least partially, bacteriophage T4 and RB49 growth demonstrates an improved interaction with Gp31 and CocO, respectively, as well.

Such influence on the distribution of the conformational states of GroEL can be exerted by many mechanisms. For example, the presence of a suppressor amino acid may destabilize the closed GroEL state, thus indirectly favoring the open GroEL state. Alternatively, a suppressor amino acid may favor the open state by stabilizing it. We took a close look at the positions of our suppressor amino acids in the crystal structure of both the GroEL closed and open states, and came up with the following examples to explain their effects. Fig 1A shows the location of all 13 amino acids identified in our groEL44 Tr⁺ suppressor study. Without exception, all 13 amino acids are located at or in the vicinity of the hinge that controls the massive en bloc movements of the GroEL apical domain. The tracing of a single GroEL monomer in either the "closed" or the "open" conformation is shown, along with the 13 amino acid positions identified by our suppressor study (Fig 1A). Clearly, some of the suppressors alter residues that occupy positions adjacent to E191G, encoded by the groEL44 mutation, namely, V190I and V190L. Other suppressors occupy positions near E191G in the linear sequence of GroEL, such as V189E, Q194H, or Q194P. The rest of the suppressors are found in the vicinity of E191G in the three-dimensional structure of GroEL, albeit not in the linear sequence. It appears that in the closed GroEL state, there exists a very hydrophobic cluster consisting of the V174/V190/V376/I333 residues (Fig 1B, left). Interestingly, in the open GroEL state, the I333 residue is found far away from the V174F/V190/V376/I333 cluster (Fig 1B, right). Thus, we propose that our V174F, V190I, and V190L mutations act by somehow destabilizing the "cozy" arrangement of the V174/V190/V376/I333 cluster in the closed state, thus indirectly favoring the open GroEL state and hence cochaperone binding. A similar mechanism can be envisioned for the effects of the G173S and G375C mutations located directly adjacent to the V174/V190/V376 hydrophobic cluster. Interestingly, XU et al. 1997 showed that the G375 residue plays an important role in the rotation of the apical domain, as does G192, again arguing that somehow our intragenic suppressors influence the quality and/or quantity of the apical domain movements.

View larger version (34K): In this window In a new window Download PPT slide   Figure 1. Location of the GroEL44 suppressor residues on the GroEL crystal structure. (A) The positions of the 13 amino acid residues that were found to suppress the GroEL44 defects are shown in the "closed" (top; BRAIG et al. 1994 ) and "open" (bottom; XU et al. 1997 ) GroEL crystal structures. For simplicity only the backbone tracing of a single GroEL subunit is shown. The actual amino acid residues depicted on the GroEL subunit are found in wild-type GroEL (see the text for more details). Only one of the suppressing mutations at each amino acid position is indicated; i.e., G173S refers to position 173, glycine is the wild-type residue, and serine is the GroEL44 suppressing residue. The structures were modeled using the Swiss PDB viewer. (B) A possible mechanism of suppressor action. The backbone tracing is that of a single GroEL subunit in the closed (left) and open (right) states, similar to that shown above. The proximity of the I333 residue to the V174/V190/V376 hydrophobic cluster in the closed state, but not in the open state, is shown. Some of the suppressors may interfere with the stability of this hydrophobic cluster in the closed state, thus indirectly favoring the open state (see DISCUSSION for more details).

Two genetic lines of experiments suggest that the alterations in the G173 codon encountered in this work lead to a GroEL mutant with higher intrinsic affinity for its various cochaperones. First, the G173D mutation was shown to be the key alteration in the groEL673 mutant, which limits both bacterial growth at 43° and bacteriophage growth at all temperatures (Table 5). The fact that groEL673 mutant bacteria plate T4 wild type but not the T41 mutant suggests that the GroEL673 mutant protein exhibits an increased affinity toward the Gp31 cochaperone. Reversion of G173D back to wild type reverses both phenotypes. Second, the G173S suppressor mutation reverses the inability of the GroEL44 mutant protein to propagate bacteriophage T4 at 37° (Table 3). Since GroEL44 does not make a stable complex with Gp31 (RICHARDSON et al. 1999 ), the G173S suppressor must somehow increase the affinity of GroEL44 for Gp31. Because of its close spatial proximity, its molecular mechanism of action could be the destabilization of the V174/V190/V376/I333 hydrophobic cluster (Fig 1B).

The suppressor mutation that converts R322 to a G residue may be suppressing the GroEL44 temperature-sensitive phenotype by a molecular mechanism analogous to the one mentioned above. It turns out that the R322 residue forms a salt bridge with the E178 residue in the closed state (BRAIG et al. 1994 ) but not in the open state (XU et al. 1997 ; BROCCHIERI and KARLIN 2000 ). Thus, our R332G suppressor substitution should abolish the formation of this salt bridge in the closed state, thus indirectly favoring the open state and consequently better interaction between GroEL and its cochaperones. Interestingly, R322 is also in close spatial proximity in the open state to the I333 residue mentioned above.

The specific interactions that favor one GroEL state over another can be either intramolecular in nature (among amino acid residues of the same GroEL polypeptide chain), like the examples given immediately above, or intermolecular in nature (among amino acid residues in adjacent GroEL subunits). A potential example of a suppressor affecting an intermolecular interaction is at the D359 position. It turns out that in the closed state one of the neighboring residues, Y360, is in close proximity to the A383/L183/F281 residues of an adjacent subunit (BRAIG et al. 1994 ). In the open state Y360 and F281 are away from the A383/L183 residues, these latter two now being in close proximity to the K80 residue of an adjacent subunit (XU et al. 1997 ). Thus, the D359Y, D359N, or D359G suppressors may function by somehow interfering with the interactions of the neighboring residue, Y360, with the A383/L183/F281 cluster in the closed state, thus indirectly favoring the open state. In this respect, it is interesting to note that groEL515 mutant bacteria, although they do not block T4 wild-type growth, do block growth of the T41 mutant (GEORGOPOULOS et al. 1972 ; Table 1). The groEL515 mutation results in the A383T change and we previously proposed that its differential block on T4311 growth is due to a higher intrinsic affinity for its cochaperones (RICHARDSON et al. 1999 ; ANG et al. 2000 ).

Many of the amino acid residues identified by our intragenic suppressor analysis were extremely well conserved in the evolution of the GroEL/Hsp60 superfamily. Generally, the GroEL homologs are at least 40% identical among prokaryotes, eukaryotes, and Archaea (MACARIO et al. 1999 ; BROCCHIERI and KARLIN 2000 ). We carried out a comparative study with the GroEL sequences found in GenBank. It turns out that the E191 residue, altered to a G residue by the groEL44 mutation, was conserved in 100/117 GroEL-like sequences. None of the GroEL-like sequences possess a G residue at position 191. The G375 suppressor residue, identified both in our genetic selection reported here and that of ZEILSTRA-RYALLS et al. 1994 , was 100% conserved among all GroEL homologs. This may not be totally surprising, since, as stated earlier, the G375 residue is at the hinge of GroEL apical domain movements. In our study, we selected for the G375C mutational change, and in the study of ZEILSTRA-RYALLS et al. 1994 both G375C and G375S were detected. Four additional intragenic suppressor residues identified in our study were extremely well conserved as well. Residues G173 (mutated to G173S), V378 (mutated to V378G or V378A), K371 (mutated to K371N), and V174 (mutated to V174F or V174I) were conserved in 98, 97, 97, and 95% of the GroEL-like sequences, respectively. The T331 and D359 residues, which were frequent targets of mutagenesis in our selection screen, are 80 and 78% conserved, respectively. Such a high degree of conservation among most of our suppressor residues suggests that they play important regulatory roles in the efficient and timely functioning of the GroEL/GroES chaperone machine.

	FOOTNOTES

¹ Permanent address: Department of Biochemistry, University of Gdansk, Kladki 24, 80-822 Gdansk, Poland.

	ACKNOWLEDGMENTS

We thank Sam Landry, Alexandra Richardson, France Keppel, Jill Zeilstra-Ryalls, Olivier Fayet, and Debbie Ang for useful suggestions, discussions, and/or strains; Kyle Tanner for excellent help with the computer illustrations; Françoise Schwager for excellent technical assistance; Debbie Ang for a critical reading of the manuscript; Colette Rossier for excellent sequencing service; and Luli Billecchi-Mestre for expert and cheerful editorial assistance. This work was supported by a grant from the Swiss National Science Foundation (FN 31-47283-96) and the Canton of Geneva.

Manuscript received February 9, 2001; Accepted for publication March 14, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ANDREADIS, J. D. and L. BLACK, 1998  Substrate mutations that bypass a specific Cpn10 chaperone requirement for protein folding. J. Biol. Chem. 273:34075-34086[Abstract/Free Full Text].

ANG, D., F. KEPPEL, G. KLEIN, A. RICHARDSON, and C. GEORGOPOULOS, 2000  Genetic analysis of bacteriophage-encoded cochaperones. Annu. Rev. Genet. 34:439-456[Medline].

ANG, D., A. RICHARDSON, M. MAYER, F. KEPPEL, and H. KRISCH et al., 2001  Pseudo T-even bacteriophage RB49 encodes CocO, a cochaperone for GroEL, which can substitute for Escherichia coli's GroES and bacteriophage T4's Gp31. J. Biol. Chem. 276:8720-8726[Abstract/Free Full Text].

BRAIG, K., Z. OTWINOWSKI, R. HEGDE, D. C. BOISVERT, and A. JOACHIMIAK et al., 1994  The crystal structure of the bacterial chaperone GroEL at 2.8 Å. Nature 371:578-586[Medline].

BROCCHIERI, L. and S. KARLIN, 2000  Conservation among HSP60 sequences in relation to structure, function, and evolution. Protein Sci. 9:476-486[Abstract].

CHANDRASEKHAR, G. N., C. WOOLFORD, R. HENDRIX, and C. GEORGOPOULOS, 1986  Purification and properties of the groES morphogenetic protein of Escherichia coli. J. Biol. Chem. 261:12414-12419[Abstract/Free Full Text].

DOERMANN, A. H. and L. D. SIMON, 1984  Bacteriophage T4 bypass31 mutations that make gene 31 nonessential for bacteriophage T4 replication: mapping bapass31 mutations by UV rescue experiments. J. Virol. 51:315-320[Abstract/Free Full Text].

EWALT, K. L., J. P. HENDRICK, W. A. HOURY, and F. U. HARTL, 1997  In vivo observation of polypeptide flux through the bacterial chaperone system. Cell 90:491-500[Medline].

FAYET, O., T. ZIEGELHOFFER, and C. GEORGOPOULOS, 1989  The groES and groEL heat shock gene products of Escherichia coli are essential for bacterial growth at all temperatures. J. Bacteriol. 171:1379-1385[Abstract/Free Full Text].

FENTON, W. A. and A. L. HORWICH, 1997  GroEL-mediated protein folding. Protein Sci. 6:743-760[Abstract].

GLASS, J. I., E. J. LEFKOWITZ, J. S. GLASS, C. R. HEINER, and E. Y. CHEN et al., 2000  The complete sequence of the mucosal pathogen Ureaplasma urealyticum. Nature 407:757-762[Medline].

GEORGOPOULOS, C., R. W. HENDRIX, S. R. CASJENS, and A. D. KAISER, 1973  Host participation in bacteriophage lambda head assembly. J. Mol. Biol. 76:45-60[Medline].

GEORGOPOULOS, C., R. W. HENDRIX, A. D. KAISER, and W. B. WOOD, 1972  Role of the host cell in bacteriophage morphogenesis: effects of a bacterial mutation on T4 head assembly. Nat. New Biol. 239:38-41[Medline].

HARTL, F. U., 1996  Molecular chaperones in cellular protein folding. Nature 381:571-580[Medline].

HORWICH, A. L., E. U. WEBER-BAN, and D. FINLEY, 1999  Chaperone rings in protein folding and degradation. Proc. Natl. Acad. Sci. USA 96:11033-11040[Abstract/Free Full Text].

HOURY, W. A., D. FRISHMAN, C. ECKERSKORN, F. LOTTSPEICH, and F. U. HARTL, 1999  Identification of in vivo substrates of the chaperone GroEL. Nature 402:147-154[Medline].

HUNT, J. F., A. J. WEAVER, S. J. LANDRY, L. GIERASCH, and J. DIESENHOFER, 1996  The crystal structure of the GroES cochaperone at 2.8 Å resolution. Nature 379:37-45[Medline].

HUNT, J. F., S. M. VAN DER VIES, L. HENRY, and J. DEISENHOFER, 1997  Structural adaptations in the specialized bacteriophage T4 cochaperone Gp31 expand the size of the anfinsen cage. Cell 90:361-371[Medline].

KEPPEL, F., B. LIPINSKA, D. ANG, and C. GEORGOPOULOS, 1990  Mutational analysis of the phage T4 morphogenetic gene 31, whose product interacts with the E. coli GroEL protein. Gene 90:19-25.

KOONIN, E. V. and S. VAN DER VIES, 1995  Conserved sequence motifs in bacterial and bacteriophage chaperones. Trends Biochem. Sci. 20:14-15[Medline].

LAEMMLI, U. K., F. BEGUIN, and G. GUJER-KELLENBERGER, 1970  A factor preventing the major head protein of bacteriophage T4 from random aggregation. J. Mol. Biol. 47:69-85[Medline].

LANDRY, S. J., J. ZEILSTRA-RYALLS, O. FAYET, C. GEORGOPOULOS, and L. M. GIERASCH, 1993  Characterization of a functionally important mobile domain of GroES. Nature 364:255-258[Medline].

LANDRY, S. J., A. TAHER, C. GEORGOPOULOS, and S. VAN DER VIES, 1996  Interplay of structure and disorder in bacteriophage T4 and Escherichia coli chaperone—10 mobile loops. Proc. Natl. Acad. Sci. USA 93:11622-11627[Abstract/Free Full Text].

LORIMER, G., 1997  Folding with a two-stroke motor. Nature 388:720-723[Medline].

MACARIO, A. J., M. LANGE, B. K. AHRING, and E. C. DE MACARIO, 1999  Stress genes and proteins in the archaea. Microbiol. Mol. Biol. Rev. 63:923-967[Abstract/Free Full Text].

MONOD, C., F. REPOILA, M. KUTATELADZE, F. TÉTARD, and H. M. KRISCH, 1997  The genome of the pseudo T-even bacteriophages, a diverse group that resembles T4. J. Mol. Biol. 267:237-249[Medline].

NIVINSKAS, R. and L. W. BLACK, 1988  Cloning, sequence, and expression of the temperature-dependent phage T4 capsid. Gene 73:251-257[Medline].

RICHARDSON, A. and C. GEORGOPOULOS, 1999  Genetic analysis of bacteriophage T4-encoded cochaperone Gp31. Genetics 4:1449-1457.

RICHARDSON, A., S. J. LANDRY, and C. GEORGOPOULOS, 1998  The ins and outs of a molecular chaperone machine. Trends Biochem. Sci. 23:138-143[Medline].

RICHARDSON, A., S. VAN DER VIES, F. KEPPEL, A. TAHER, and S. J. LANDRY et al., 1999  Compensatory changes in GroEL/Gp31 affinity as a mechanism for allele-specific genetic interaction. J. Biol. Chem. 274:52-58[Abstract/Free Full Text].

SIGLER, P. B., Z. XU, H. S. RYE, S. G. BURSTON, and W. A. FENTON et al., 1998  Structure and function in GroEL-mediated protein folding. Annu. Rev. Biochem. 67:581-608[Medline].

SIMON, L. D. and B. RANDOLPH, 1984  Bacteriophage T4 bypass31 mutations that make gene 31 nonessential for bacteriophage T4 replication: isolation and characterization. J. Virol. 51:321-328[Abstract/Free Full Text].

STERNBERG, N., 1973  Properties of a mutant of Escherichia coli defective in bacteriophage head formation (groE) II. J. Mol. Biol. 76:25-44[Medline].

TILLY, K. and C. GEORGOPOULOS, 1982  The groEL and groES morphogenetic gene products of Escherichia coli interact in vivo. J. Bacteriol. 149:1082-1088[Abstract/Free Full Text].

TILLY, K., H. MURIALDO, and C. GEORGOPOULOS, 1981  Identification of a second Escherichia coli groE gene whose product is necessary for bacteriophage morphogenesis. Proc. Natl. Acad. Sci. USA 78:1629-1633[Abstract/Free Full Text].

VAN DER VIES, S. M., A. A. GATENBY, and C. GEORGOPOULOS, 1994  Bacteriophage T4 encodes a cochaperone that can substitute for Escherichia coli GroES in protein folding. Nature 368:654-656[Medline].

XU, Z., A. L. HORWICH, and P. B. SIGLER, 1997  The crystal structure of the asymmetric GroEL-GroES-(ADP)₇ chaperone complex. Nature 388:741-750[Medline].

YURA, T., H. NAGAI, and H. MORI, 1993  Regulation of the heat shock response in bacteria. Annu. Rev. Microbiol. 47:321-350[Medline].

ZEILSTRA-RYALLS, J., O. FAYET, L. BAIRD, and C. GEORGOPOULOS, 1993  Sequence analysis and phenotypic characterization of groE mutations that block and T4 bacteriophage growth. J. Bacteriol. 175:1134-1143[Abstract/Free Full Text].

ZEILSTRA-RYALLS, J., O. FAYET, and C. GEORGOPOULOS, 1994  Two classes of extragenic suppressor mutations identify functionally distinct regions of the GroEL chaperone of Escherichia coli. J. Bacteriol. 176:6558-6565[Abstract/Free Full Text].

This article has been cited by other articles:

				W. Zheng, B. R. Brooks, and D. Thirumalai Allosteric Transitions in the Chaperonin GroEL are Captured by a Dominant Normal Mode that is Most Robust to Sequence Variations Biophys. J., October 1, 2007; 93(7): 2289 - 2299. [Abstract] [Full Text] [PDF]

				L. Figueiredo, D. Klunker, D. Ang, D. J. Naylor, M. J. Kerner, C. Georgopoulos, F. U. Hartl, and M. Hayer-Hartl Functional Characterization of an Archaeal GroEL/GroES Chaperonin System: SIGNIFICANCE OF SUBSTRATE ENCAPSULATION J. Biol. Chem., January 9, 2004; 279(2): 1090 - 1099. [Abstract] [Full Text] [PDF]

				A. S. Kondrashov, S. Sunyaev, and F. A. Kondrashov Dobzhansky-Muller incompatibilities in protein evolution PNAS, November 12, 2002; 99(23): 14878 - 14883. [Abstract] [Full Text] [PDF]