Construction and functional analyses of a comprehensive σ⁵⁴ site-directed mutant library using alanine–cysteine mutagenesis

Abstract

The σ⁵⁴ factor associates with core RNA polymerase (RNAP) to form a holoenzyme that is unable to initiate transcription unless acted on by an activator protein. σ⁵⁴ is closely involved in many steps of activator-dependent transcription, such as core RNAP binding, promoter recognition, activator interaction and open complex formation. To systematically define σ⁵⁴ residues that contribute to each of these functions and to generate a resource for site specific protein labeling, a complete mutant library of σ⁵⁴ was constructed by alanine–cysteine scanning mutagenesis. Amino acid residues from 3 to 476 of Cys(-)σ⁵⁴ were systematically mutated to alanine and cysteine in groups of two adjacent residues at a time. The influences of each substitution pair upon the functions of σ⁵⁴ were analyzed in vivo and in vitro and the functions of many residues were revealed for the first time. Increased σ⁵⁴ isomerization activity seldom corresponded with an increased transcription activity of the holoenzyme, suggesting the steps after σ⁵⁴ isomerization, likely to be changes in core RNAP structure, are also strictly regulated or rate limiting to open complex formation. A linkage between core RNAP-binding activity and activator responsiveness indicates that the σ⁵⁴-core RNAP interface changes upon activation.

INTRODUCTION

Major regulation of gene expression mainly depends upon the control of transcription initiation. In particular, the activity of RNA polymerase is regulated to control the level of gene expression. In bacteria, RNA polymerase holoenzyme is composed of core RNAP and a σ factor, which is responsible for the specific recognition and melting of promoter sequence (1). The σ factors in bacteria can be divided into two classes: the σ⁷⁰ and σ⁵⁴ class (2). Although they bind to the same core RNAP, these two classes of σ factors have no similarity in primary sequence and the modes of their transcription regulation are very different.

Unlike σ⁷⁰-holoenzyme, the initiation rates of σ⁵⁴-holoenzyme are mainly controlled via regulation of the DNA melting step (3,4). The σ⁵⁴ -holoenzyme first binds promoter DNA in a transcriptionally silent closed complex (5–7). Conversion of the closed complex to a transcriptionally proficient open complex requires a specialized activator protein which interacts with σ⁵⁴-holoenzyme through DNA looping and brings about a conformational change in the closed promoter complex in a nucleotide hydrolysis dependent way, leading to σ⁵⁴-holoenzyme isomerization and promoter melting (7–11).

σ⁵⁴ plays an important role in the process of open complex formation and interacts extensively with activator protein, promoter DNA and core RNAP to regulate the proper formation of the open complex. Open complex formation by σ⁵⁴-holoenzyme appears to involve the activator dependent relocation of three structurally distinct σ⁵⁴ domains within the holoenzyme to (i) allow positioning of promoter DNA for entry into the active site of RNAP and (ii) removal of a σ⁵⁴ associated protein density that would otherwise prevent DNA delivery (12). Based on biochemical and genetic characterization and fragmentation analyses, σ⁵⁴ can be divided into three regions (2,13–16) (Figure 1). Region I comprises the amino terminal 56 residues and is important for activator responsiveness (10,17,18). Region II comprises residues 57–106 and is characterized by a predominance of acidic residues. Region II is not essential for the function of σ⁵⁴. In the carboxyl terminal (residues 107–477) lies Region III which contains a primary core RNAP-binding domain located between residues 120 and 215 (16), and a DNA-binding domain located between residues 329 and 477 (13,19,20). A modulation domain lies between these two domains and influences DNA-binding activity indirectly (14).

Figure 1.

Diagram of the functional regions of σ⁵⁴ and summary of the results. Grey bar below the diagram of σ⁵⁴ indicates the predicted helical region in Region I. Black bars below the diagram of σ⁵⁴ indicate the helices in the carboxyl terminal DNA binding domain based on the partial structure of σ⁵⁴ from Aquifex aeolicus (20,21). Darker color is used to represent higher activity. The function each color represents is indicated in the left of the figure. The squares at the left of the colored bars indicate the color corresponding to the activity of Cys(–)σ⁵⁴. In the bar representing the activity of σ⁵⁴-activator interaction, the dark blue color indicates the variants that form additional complexes in the presence of PspF₁₋₂₇₅ and ADP·AlFx. The data are from Table 1.

Due to the lack of a high-resolution structure of σ⁵⁴, it is still not clear which residues in σ⁵⁴ are responsible for the interactions with itself or other elements such as core RNAP and promoter DNA in the process of transcription initiation. Nor are the primary σ⁵⁴ sequences of densities evident in electron microscopy studies precisely delineated (12). To systematically define the functions of each residue in σ⁵⁴, a site-directed mutant library of σ⁵⁴ was constructed using alanine–cysteine scanning mutagenesis. Mutation of residues other than glycine to alanine can be viewed as side-chain deletion and alanine scanning has been widely used in the research of protein functions (21,22), while mutation to cysteine can facilitate site specific protein modification (23–25). Alanine–cysteine scanning mutagenesis used in this article combines the advantages of these two kinds of mutations and facilitates both structural and functional analyses of a protein.

Here, residues 3–476 of a cysteine free form of Klebsiella pneumoniae σ⁵⁴ were changed to alanine and cysteine in groups of two consecutive residues, producing a total of 237 variants of σ⁵⁴. This single cysteine mutant library is very helpful in both the structural and functional analyses of σ⁵⁴, and this paper focuses on the latter.

The results indicate that: (i) two parts of the Region I are the main targets for activator interaction; (ii) the carboxyl terminal of modulation domain can affect σ⁵⁴ isomerization, probably through it influencing the −12 promoter sequence interaction; (iii) the carboxyl terminal part of DNA-binding domain is important for maintain the in vivo integrity of σ⁵⁴, but less important in the regulation of σ⁵⁴ activity. The linkage between different functions of σ⁵⁴ and possible mechanisms are discussed. The single cysteine mutant library of σ⁵⁴ provides a powerful tool for further structural analyses using ensemble and single molecule Foster Resonant Energy Transfer (FRET) approaches (26).

MATERIALS AND METHODS

Site-directed mutagenesis

The pET28 based plasmid pSRWCys(−) (25), which directs the synthesis of the amino-terminal 6-His-tagged Cys(−)σ⁵⁴, was used as the template to create the single cysteine variants of σ⁵⁴ using the Quickchange mutagenesis kit (Stratagene). The whole rpoN gene directing the synthesis of σ⁵⁴ protein in each mutated plasmid was sequenced to ensure that only the desired substitutions were present.

β-Galactosidase assays

Plasmids carrying mutated rpoN genes were transformed to Escherichia coli rpoN knockout strain TH1/pMB221 (27). pMB221 contains both the reporter gene lacZ and the activator gene nifA. The reporter gene lacZ is under the control of K. pneumoniae nifH promoter, and the activator NifA is expressed from bla promoter. Transformants were grown at 37°C and then 100 μl of the overnight culture was used to inoculate 10 ml of LB with 50 μg/ml kanamycin and 34 μg/ml chloromycetin. Cells were grown at 37°C until A₆₀₀ reaches 0.8. σ⁵⁴-dependent gene expression in this system depend on a basal T7-RNAP independent expression of rpoN from pET28. The average β-galactosidase results from three independent colonies are presented.

Immunoblotting

Mutated plasmids were transformed to TH1/pMB221 and were cultured under the same condition used in β-galactosidase assays. Cells (1 ml) were then collected by centrifugation and resuspended in 60 μl of 10 mM Tris, 0.1 mM EDTA, pH 7.9. Ten microliters of the concentrated cell solution was lysed with 10 μl of 2× SDS sample buffer, heated at 95°C and 10 μl of each were loaded. Proteins were separated on denaturing 8% SDS–PAGE gels and blotted onto PVDF membranes. Anti-σ⁵⁴ (prepared by Institute of Genetics and Developmental Biology, Chinese Academy of Sciences) and alkaline phosphatase conjugated anti-rabbit IgG (Pierce) were used for detection.

Protein purification

Mutated σ⁵⁴ proteins were overexpressed in E. coli B834 (DE3). Two hundred microliters of overnight cultures were used to inoculate 20 ml of LB with 50 μg/ml kanamycin and cells were grown at 37°C until A₆₀₀ reached 0.6. The cultures were then shifted to 25°C and IPTG was added to 1 mM. Incubation continued for 2 h at 25°C. Cells were collected by centrifugation, and resuspended in 500 μl of Buffer A [25 mM sodium phosphate (pH 7.0), 50 mM NaCl, 5% (v/v) glycerol]. The concentrated cells were lysed by adding 55 μl of FastBreak™ Reagent (Promega) into the buffer and were shaken for 15 min at room temperature. The lysates were then incubated for 3 min with 450 μl of Ni–NTA resin (Qiagen) in the columns, which were then washed with 500 μl of Buffer B (25 mM sodium phosphate (pH 7.0), 50 mM NaCl, 30 mM imidazole, 5% (v/v) glycerol). Mutated σ⁵⁴ proteins were eluted in 400 μl of Buffer C [25 mM sodium phosphate (pH 7.0), 50 mM NaCl, 1 M imidazole, 5% (v/v) glycerol]. Purified σ⁵⁴ proteins were dialysed against 10 mM Tris (pH 7.5), 0.1 mM EDTA, 100 mM NaCl, 50% glycerol, and stored at −20°C.

Core RNAP-binding assays

E. coli core RNAP (Epicentre, 250 nM) and different amounts of σ⁵⁴ variants were mixed together in Tris–NaCl buffer (40 mM Tris–HCl (pH 8.0), 10% (v/v) glycerol, 0.1 mM EDTA, 1 mM DTT, 100 mM NaCl) and incubated for 10 min at 30°C, followed by the addition of glycerol bromophenol blue loading dye. Samples were loaded onto native 4.5% PAGE gels and run at 50 V for 2 h at room temperature in TG buffer (25 mM Tris, 200 mM glycine, pH 8.6). Proteins were visualized by Coomasie Blue staining.

Gel mobility shift assay

³²P-labeled 88-mer top strand −12/−11 mismatched heteroduplex DNA fragment consisting of the −60 to +28 Sinorhizobium meliloti nifH promoter sequence was used as a template for gel mobility shift assays. The reactions contained 20 nM DNA, 1 μM σ⁵⁴ in STA buffer [25 mM Tris–acetate (pH 8.0), 8 mM magnesium acetate, 10 mM KCl, 3.5% (w/v) PEG 8000] and were incubated for 10 min at 30°C. Where indicated, PspF₁₋₂₇₅ (5 μM), dGTP (4 mM) were added for 5 min prior to gel loading. Free DNA and σ⁵⁴ bound DNA were separated on 4.5% native PAGE gels and run at 80 V for 70 min at room temperature in TG buffer (25 mM Tris, 200 mM glycine, pH 8.6). Quantitative data were from phosphorimager analyses of the gels.

σ⁵⁴-activator interaction

Twenty micromolar PspF₁₋₂₇₅ and 1 μM of σ⁵⁴ were incubated at 30°C for 5 min in the above STA buffer with ADP (0.2 mM) and NaF (5.0 mM). After addition of AlCl₃ (0.2 mM) the samples were incubated for a further 10 min and loaded onto 4.5% native PAGE gels and run at 120 V for 50 min at room temperature in the above TG buffer. Proteins were visualized by Coomassie Blue staining.

Abortive transcription assay

Open complexes were formed on plasmid pMKC28 (28) (10 nM final concentration) in the presence of 100 nM σ⁵⁴-holoenzyme (constituted using 2.5:1 molar ratio of σ⁵⁴ over core RNAP), 20 U RNase inhibitor, 4 mM dATP and 5 μM of PspF₁₋₂₇₅. A mix containing 100 μg/ml heparin, 0.5 mM UpG, 4 μCi [α-³²P] GTP was added to the reaction for the abortive synthesis of a four nucleotide transcript (UGGG). The reactions were separated on a 20% denaturing gel and quantified as in gel mobility shift assay. Replicate assays were conducted to confirm the low activities of the variants defective in abortive transcription.

RESULTS

Mutagenesis

σ⁵⁴ from K. pneumoniae contains two cysteine residues at positions 198 and 346 (29). Substitution of these two cysteine residues with alanine does not affect the function of σ⁵⁴ significantly (25). Therefore, in order to construct variants carrying a single cysteine, the pET28b based plasmid pSRWCys(–), which directs the synthesis of the Cys(–)σ⁵⁴, was used as the template (25). Residues of Cys(–)σ⁵⁴ from 3 to 476 were systematically substituted with alanine–cysteine in groups of two adjacent residues at a time, for a total of 237 variants of σ⁵⁴. Substitution with alanine can be viewed as a side-chain deletion, whereas substitution with cysteine facilitates side-chain modification at specific positions. Each mutated rpoN gene was sequenced in its entirety to establish the presence of only the expected substitutions.

Region II is not essential for the in vivo activity of σ⁵⁴

β-Galactosidase assays were used to test the in vivo activities of the variants. The plasmid carrying the mutated rpoN gene and another plasmid, pMB221 carrying the reporter and activator genes (27), were transformed to E. coli TH1, which lacks functional σ⁵⁴. The expression of mutated σ⁵⁴ depends on the low level of basal expression of rpoN gene from the pET28 based vector. The β-galactosidase gene is under the control of K. pneumoniae nifH promoter and the activator NifA is constitutively expressed from the bla promoter. NifA acts with σ⁵⁴ to activate the expression of β-galactosidase gene from σ⁵⁴ dependent nifH promoter. Therefore, the activity of β-galactosidase can be used to indicate the in vivo activity of σ⁵⁴ variants.

The mutations that affected the in vivo activity of σ⁵⁴ are distributed extensively in Region I and III, and are relatively concentrated in the carboxyl terminal of σ⁵⁴ that includes the −24 DNA-binding bihelical structure (30) (Table 1). Consistent with the fact that Region II is not well conserved in sequence, none of the Region II variants exhibited poor activity.

Table 1.

Summary of the activities of σ⁵⁴ variants^a

σ54 mutant	Functional region	In vivo trxnb	Core binding	Activator contact	Early melted DNA binding	Isomerization	Abortive trxn
Cys(–)		100	wt-like	+	100	100	100
Region I
AC3/4	Activation	31	wt-like	+	86	34	39
AC5/6	Activation	59	weak	+	93	83	83
AC7/8	Activation	51	wt-like	+	85	22	48
AC9/10	Activation	66	wt-like	+	88	41	49
AC11/12	Activation	35	wt-like	+/−	91	0	18
AC13/14	Activation	76	wt-like	+	107	0	98
AC15/16	Activation	18	wt-like	+	89	0	44
AC17/18	Activation	23	wt-like	−	66	0	16
AC19/20	Activation	9	weak	+	46	59	42
AC21/22	Activation	27	wt-like	+	29	78	40
AC23/24	Activation	1	wt-like	+	63	52	19
AC25/26	Activation	6	wt-like	+	21	0	25
AC27/28	Activation	47	wt-like	+	77	0	37
AC29/30	Activation	16	wt-like	−	19	0	14
AC31/32	Activation	72	wt-like	+	95	0	42
AC33/34	Activation	16	wt-like	+	7	0	15
AC35/36	Activation	75	wt-like	+	90	0	34
AC37/38	Activation	8	wt-like	−	15	144	21
AC39/40	Activation	75	wt-like	+	92	52	29
AC41/42	Activation	109	wt-like	+	72	0	47
AC43/44	Activation	49	wt-like	+	58	55	31
AC45/46	Activation	30	wt-like	+	95	26	71
AC47/48	Activation	4	wt-like	−	14	0	23
AC49/50	Activation	84	wt-like	+	92	168	109
AC51/52	Activation	72	wt-like	+	102	110	59
AC53/54	Activation	81	wt-like	+	98	109	87
AC55/56	Activation	71	wt-like	+	92	105	57
Region II
AC57/58	Acidic	72	wt-like	+	95	72	46
AC59/60	Acidic	81	wt-like	+	97	101	128
AC61/62	Acidic	71	wt-like	+	78	98	78
AC63/64	Acidic	83	wt-like	+	94	89	80
AC65/66	Acidic	117	wt-like	+	90	93	45
AC67/68	Acidic	108	wt-like	+	89	94	76
AC69/70	Acidic	81	wt-like	+	93	96	87
AC71/72	Acidic	92	wt-like	+	88	84	38
AC73/74	Acidic	44	wt-like	+	88	98	75
AC75/76	Acidic	128	wt-like	+	90	99	46
AC77/78	Acidic	133	wt-like	+	86	97	51
AC79/80	Acidic	37	wt-like	+	88	91	74
AC81/82	Acidic	102	wt-like	+	89	104	76
AC83/84	Acidic	70	wt-like	+c	99	192	51
AC85/86	Acidic	97	wt-like	+	92	119	100
AC87/88	Acidic	47	wt-like	+	92	95	145
AC89/90	Acidic	76	wt-like	+	87	164	92
AC91/92	Acidic	38	wt-like	+	87	90	117
AC93/94	Acidic	39	wt-like	+	74	117	117
AC95/96	Acidic	78	wt-like	+	81	89	120
AC97/98	Acidic	28	wt-like	+	86	102	148
AC99/100	Acidic	122	wt-like	+	79	108	111
AC101/102	Acidic	104	wt-like	+	87	97	117
AC103/104	Acidic	108	wt-like	+	88	93	89
AC105/106	Acidic	109	wt-like	+	95	98	72
Region III
AC107/108	Core binding	110	wt-like	+	94	110	29
AC109/110	Core binding	66	wt-like	+	100	60	34
AC111/112	Core binding	111	wt-like	+	92	78	23
AC113/114	Core binding	72	wt-like	+	90	81	25
AC115/116	Core binding	97	wt-like	+	81	85	57
AC117/118	Core binding	79	wt-like	+	79	66	85
AC119/120	Core binding	33	wt-like	+	101	189	46
AC121/122	Core binding	88	weak	−	53	181	38
AC123/124	Core binding	8	wt-like	+/−	93	225	51
AC125/126	Core binding	14	wt-like	+/−	107	131	20
AC127/128	Core binding	8	wt-like	+/−	96	89	37
AC129/130	Core binding	60	wt-like	+	93	66	148
AC131/132	Core binding	63	wt-like	+	95	83	33
AC133/134	Core binding	16	wt-like	−	65	138	46
AC135/136	Core binding	65	wt-like	+	92	69	127
AC137/138	Core binding	19	wt-like	−	87	95	36
AC139/140	Core binding	61	weak	−	88	113	152
AC141/142	Core binding	100	wt-like	+	88	75	174
AC143/144	Core binding	107	wt-like	+	97	102	114
AC145/146	Core binding	16	wt-like	+/−	92	78	24
AC147/148	Core binding	89	wt-like	+	102	67	124
AC149/150	Core binding	96	wt-like	−	75	27	77
AC151/152	Core binding	6	weak	−	81	83	64
AC153/154	Core binding	24	wt-like	+	107	0	37
AC155/156	Core binding	83	wt-like	+	98	55	185
AC157/158	Core binding	102	wt-like	+	100	75	153
AC159/160	Core binding	32	wt-like	−	44	85	115
AC161/162	Core binding	27	wt-like	−	90	113	86
AC163/164	Core binding	100	wt-like	+	85	77	117
AC165/166	Core binding	83	wt-like	−	102	95	87
AC167/168	Core binding	87	wt-like	+	72	85	104
AC169/170	Core binding	79	wt-like	+	97	98	142
AC171/172	Core binding	106	wt-like	+	103	97	172
AC173/174	Core binding	65	wt-like	−	82	87	67
AC175/176	Core binding	65	weak	−	91	106	126
AC177/178	Core binding	74	wt-like	−	80	168	103
AC179/180	Core binding	23	weak	−	80	218	29
AC181/182	Core binding	52	wt-like	−	78	169	136
AC183/184	Core binding	47	wt-like	+	81	123	26
AC185/186	Core binding	2	wt-like	−	83	143	43
AC187/188	Core binding	9	wt-like	−	90	115	27
AC189/190	Core binding	115	wt-like	−	69	71	154
AC191/192	Core binding	92	wt-like	+	87	90	156
AC193/194	Core binding	21	wt-like	+	90	147	190
AC195/196	Core binding	30	wt-like	+/−	81	153	113
AC197/198	Core binding	99	wt-like	+	88	181	104
AC199/200	Core binding	12	wt-like	−	57	201	26
AC201/202	Core binding	27	wt-like	−	64	131	27
AC203/204	Core binding	55	wt-like	−	64	169	64
AC205/206	Core binding	98	wt-like	−	95	134	138
AC207/208	Core binding	85	wt-like	+	28	188	77
AC209/210	Core binding	69	wt-like	−	21	148	56
AC211/212	Core binding	87	wt-like	+	93	107	69
AC213/214	Core binding	96	wt-like	−	86	122	47
AC215/216	Core binding	71	wt-like	−	44	136	38
AC217/218	Modulation	97	wt-like	+	69	180	61
AC219/220	Modulation	23	wt-like	−	38	269	26
AC221/222	Modulation	89	wt-like	+	77	126	74
AC223/224	Modulation	25	wt-like	−	43	138	16
AC225/226	Modulation	108	wt-like	+	79	144	35
AC227/228	Modulation	5	wt-like	+	91	140	46
AC229/230	Modulation	59	wt-like	+	72	144	61
AC231/232	Modulation	61	weak	−	57	122	42
AC233/234	Modulation	62	wt-like	+	33	230	56
AC235/236	Modulation	30	wt-like	−	46	175	33
AC237/238	Modulation	69	wt-like	+	47	131	40
AC239/240	Modulation	90	wt-like	+/−	56	63	43
AC241/242	Modulation	74	weak	−	20	143	32
AC243/244	Modulation	77	wt-like	−	78	109	57
AC245/246	Modulation	70	wt-like	+	75	171	56
AC247/248	Modulation	72	wt-like	−	66	85	49
AC249/250	Modulation	89	wt-like	+	92	115	76
AC251/252	Modulation	52	wt-like	+/−	47	124	56
AC253/254	Modulation	23	weak	−	87	139	62
AC255/256	Modulation	74	wt-like	−	53	135	60
AC257/258	Modulation	89	wt-like	+	71	145	75
AC259/260	Modulation	24	wt-like	+	84	187	20
AC261/262	Modulation	86	wt-like	+	80	145	44
AC263/264	Modulation	102	wt-like	+	85	125	60
AC265/266	Modulation	107	wt-like	+	98	114	76
AC267/268	Modulation	98	wt-like	+	98	146	82
AC269/270	Modulation	47	wt-like	+	96	118	50
AC271/272	Modulation	86	wt-like	+	97	92	74
AC273/274	Modulation	71	wt-like	+	82	102	50
AC275/276	Modulation	102	weak	−	69	153	18
AC277/278	Modulation	78	weak	−	59	128	31
AC279/280	Modulation	91	weak	+/−	51	153	33
AC281/282	Modulation	106	wt-like	+	67	136	51
AC283/284	Modulation	102	wt-like	+/−	73	103	43
AC285/286	Modulation	7	weak	−	88	95	27
AC287/288	Modulation	61	wt-like	+	106	100	63
AC289/290	Modulation	49	wt-like	+/−	81	68	43
AC291/292	Modulation	78	wt-like	+	79	89	35
AC293/294	Modulation	92	wt-like	+	88	103	50
AC295/296	Modulation	103	wt-like	+	17	0	36
AC297/298	Modulation	28	weak	+/−	8	153	30
AC299/300	Modulation	8	wt-like	+c	18	0	27
AC301/302	Modulation	96	wt-like	+c	50	142	38
AC303/304	Modulation	20	wt-like	+c	62	96	32
AC305/306	Modulation	55	wt-like	+c	51	133	60
AC307/308		92	wt-like	+c	58	185	51
AC309/310		119	wt-like	+c	37	163	42
AC311/312		68	wt-like	+c	39	173	31
AC313/314		103	wt-like	+/−	50	122	51
AC315/316		95	wt-like	+/−	61	237	34
AC317/318		17	weak	+c	67	141	42
AC319/320		6	wt-like	+	64	54	20
AC321/322		124	wt-like	+	90	181	36
AC323/324		52	wt-like	+	83	0	22
AC325/326		48	wt-like	+	48	92	21
AC327/328		12	wt-like	+	39	59	15
AC329/330	Xlink	7	wt-like	+	4	451	16
AC331/332	Xlink	39	wt-like	+	81	72	32
AC333/334	Xlink	5	wt-like	+/−	13	294	21
AC335/336	Xlink	9	weak	+	7	0	45
AC337/338	Xlink	89	wt-like	+/−	80	71	26
AC339/340	Xlink	34	wt-like	+/−	13	265	30
AC341/342	Xlink	73	weak	−	46	152	36
AC343/344	Xlink	93	weak	−	85	73	22
AC345/346	Xlink	86	wt-like	+	93	107	40
AC347/348		12d	weak	−	10	290	13
AC349/350		99	wt-like	+	101	158	58
AC351/352		33	weak	+	11	77	41
AC353/354		41	weak	−	61	144	39
AC355/356		4	wt-like	−	3	193	15
AC357/358		16	weak	+/−c	20	504	27
AC359/360		93	weak	–	58	148	52
AC361/362		37	weak	+/−c	50	175	54
AC363/364		49	wt-like	+/−c	7	306	42
AC365/366	HTH	95	weak	−	41	235	49
AC367/368	HTH	75	weak	−	17	343	43
AC369/370	HTH	29	weak	−	12	246	16
AC371/372	HTH	69	wt-like	+	95	109	29
AC373/374	HTH	79	wt-like	+	97	94	43
AC375/376	HTH	91	weak	+	92	78	20
AC377/378	HTH	68	weak	+/−	88	124	33
AC379/380	HTH	2	wt-like	+	10	1654	19
AC381/382	HTH	45	weak	−	10	600	32
AC383/384	HTH	0	wt-like	+	1	637	16
AC385/386	HTH	10	wt-like	+	78	66	66
AC387/388		23	wt-like	+	56	27	30
AC389/390		16	weak	−	67	111	31
AC391/392		98	wt-like	+	104	120	39
AC393/394		50	weak	−	71	146	28
AC395/396		42	weak	−	50	146	60
AC397/398		85	weak	−	84	118	81
AC399/400		23	weak	−	14	142	30
AC401/402		25	weak	+/−	30	122	69
AC403/404		7	weak	+/−	30	131	25
AC405/406		45	wt-like	+	95	123	41
AC407/408		20	wt-like	+/−	74	179	31
AC409/410		61	wt-like	+c	71	153	55
AC411/412		39	wt-like	+c	59	205	70
AC413/414		87	wt-like	+c	94	123	52
AC415/416		53	wt-like	+c	9	929	54
AC417/418		27	wt-like	+	81	133	34
AC419/420		72	wt-like	+	90	113	67
AC421/422		122	wt-like	+	98	128	44
AC423/424		10	wt-like	−	28	369	115
AC425/426		12	weak	+c	18	574	68
AC427/428		16	wt-like	+/−	39	296	107
AC429/430		90	wt-like	+/−	91	119	85
AC431/432		0d	wt-like	+/−	3	202	41
AC433/434		46	wt-like	+/−	59	177	79
AC435/436		18	wt-like	−	62	190	96
AC437/438		8d	wt-like	+	78	124	77
AC439/440		0d	wt-like	+/−	10	339	34
AC441/442		15	wt-like	+	17	356	135
AC443/444		96	wt-like	+c	89	138	82
AC445/446		13	wt-like	+	84	103	99
AC447/448		96	wt-like	+c	89	116	92
AC449/450		8d	wt-like	+	69	171	59
AC451/452		21	wt-like	+c	58	150	112
AC453/454	RpoN box	34	wt-like	+c	60	210	67
AC455/456	RpoN box	0	wt-like	+	67	0	23
AC457/458	RpoN box	0d	wt-like	+	2	251	65
AC459/460	RpoN box	23	wt-like	+c	35	210	121
AC461/462	RpoN box	0d	wt-like	+	42	156	28
AC463/464	RpoN box	86	wt-like	+c	104	141	71
AC465/466		52	wt-like	+c	74	109	99
AC467/468		1d	wt-like	+	24	303	62
AC469/470		99	wt-like	+	93	146	61
AC471/472		66	wt-like	+c	10	132	58
AC473/474		0d	wt-like	+	19	174	23
AC475/476		32	wt-like	+c	32	134	70

+: σ⁵⁴ variants bind PspF_1–275 normally. +/−: σ⁵⁴ variants bind PspF_1–275 weakly. −: σ⁵⁴ variants cannot bind PspF_1–275.

^aActivities in all the assays are measured relative to Cys(–)σ⁵⁴. Isomerization ability of σ⁵⁴ is calculated as a percentage of initially bound DNA converted to ssσ–DNA in an activator and dGTP dependent way.

^bTested using β-galactosidase assay.

^cNew complexes were formed when PspF_1–275 was trapped with σ⁵⁴ in the presence of ADP·AlFx.

^dLow level of in vivo expression.

Low activities of the variants could be caused by protein instability. Immunoblotting was therefore carried out to measure the expression levels of the variants which retained <20% activity compared to Cys(–)σ⁵⁴. σ⁵⁴ variants were expressed under the same condition used in the in vivo activity assay. Variants found at lower than wild-type expression levels have substitutions mainly located in the DNA-binding domain [AAs 329–477, defined previously by proteolysis experiments and including the −24 promoter recognition structure (30,31)], especially in the carboxyl terminal, between residues 431 and 476 (Figure 2), indicating that many residues in this region are important for maintaining the structural integrity of σ⁵⁴ in vivo, and that the instabilities of these variants may be one of the reasons for their low in vivo activities. However, the series of purified σ⁵⁴ variants changed in residues 457–476 migrated as discrete bands in native gels, each with a mobility similar to the Cys(–)σ⁵⁴ (when loaded at higher concentrations for detection), suggesting that their gross structural integrity is not seriously affected in vitro (Figure S1A). These variants were detected in native gels at concentrations where the Cys(–)σ⁵⁴ was not easily detected, indicating a less diffuse behavior, and by inference some differences in conformation compared to the control protein Cys(–)σ⁵⁴.

Figure 2.

In vivo stability of σ⁵⁴ variants which retained <20% in vivo activity compared to Cys(-)σ⁵⁴. The functional regions in which these variants are located are also shown. The asterisks indicate the variants with very low expression levels. Immunoblots were of lysates of TH1 cells used in in vivo activity assays, containing pMB221 and mutant rpoN genes. Equal amounts of each cell extract were loaded. Immunoblots of purified σ⁵⁴ samples are also shown presented at 4, 8 and 12 ng.

Residues contributing to core RNAP binding are extensively distributed throughout Region I and III

Next, all the variants as well as Cys(–)σ⁵⁴ were purified by Ni affinity chromatography for further functional analyses in vitro. An essential and first step en route to open complex formation is holoenzyme forming between σ and core RNAP. Native gel mobility shift assay was used to measure the binding of σ⁵⁴ variants to core RNAP. Holoenzyme can be detected in Coomassie blue stained gels based on its different mobility versus core RNAP.

Consistent with the view that the core RNAP interface of σ⁵⁴ comprises at least two functionally important but distinct sequences: 1–56 and 120–215 (16), many mutations located in these two sequences altered the interaction of σ⁵⁴ with core RNAP: Variants AC 5/6, AC 19/20, AC 121/123, AC 139/140, AC 151/152, AC 175/176 and AC 179/180 shifted all core RNAP at a molar ratio of 4:1 (σ⁵⁴: core RNAP), whereas Cys(–)σ⁵⁴ can shift all core RNAP at a molar ratio of 1:1 (Figure 3). In Region I, mutations that affected binding of σ⁵⁴ to core RNAP mainly locate in its amino terminus, while mutations between residues 39–56 had no effect upon holoenzyme formation (Figure S1A). Furthermore, in this assay, many residues in modulation domain or DNA-binding domain were found to be important for core RNAP binding (Figure 3), indicating residues affecting interaction of σ⁵⁴ with core RNAP are widely distributed across Regions I and III. Although Region II is reported to play a role in the binding of σ⁵⁴ to core RNAP (32), none of the mutations in Region II affected the core RNAP-binding activities in our assays (Figure S1B).

Figure 3.

Formation of holoenzyme by σ⁵⁴ variants which bind core RNAP with low affinity. Holoenzyme were formed at 1:1, 1:2 and 1:4 ratios of core RNAP (250 nM) to σ. Migration positions of core RNAP, holoenzyme and free σ are indicated.

Regions I and III cooperate for activator interaction

σ⁵⁴-Holoenzyme initiates transcription in response to an interaction with an activator protein, and σ⁵⁴ is the major and direct target for this interaction (7,33,34). To identify the residues in σ⁵⁴ responsible for the stable interaction with activator protein, an assay to measure the interaction between the σ⁵⁴ and the activator protein by native-PAGE of the reaction was used. A DNA-binding domain deletion form of the E. coli phage shock protein F (PspF₁₋₂₇₅) served as a model activator in these assays (35). Since the interaction between σ⁵⁴ and its activators is normally very transient, stable complex formation between σ⁵⁴ and PspF₁₋₂₇₅ depends on ADP·AlFx—an ATP analog that mimics the state of ATP at the point of hydrolysis, which ‘traps’ the transient complex between PspF₁₋₂₇₅ and σ⁵⁴ so that it can be resolved and detected on native gels (34).

Results show that four Region I variants (AC17/18, AC29/30, AC37/38 and AC47/48.) failed to form the ADP·AlFx dependent stable complex with PspF₁₋₂₇₅, indicating the substituted residues contribute to part of the interface σ⁵⁴ makes with activator protein (Figure 4). Mutations of these residues may alter this interface and prevent the variants from making stable interaction with activator protein. Low in vivo activities of these four variants may also be attributed to their altered interaction with activator protein.

Figure 4.

σ⁵⁴ variants which cannot bind PspF_1–275 normally in the presence of ADP·AlFx. ADP·AlFx dependent complexes formed between σ⁵⁴ and PspF_1–275 were detected by Coomassie staining. σ⁵⁴ and PspF_1–275 were presented at 1 and 20 μM separately. Arrow (a) indicates the complex formed between σ⁵⁴ and PspF_1–275 in the presence of ADP·AlFx. Arrow (b) indicates the changed position of PspF_1–275 complex in the presence of ADP·AlFx. Arrow (c) and (d) indicates the new bands formed in this assay.

Many Region III variants also cannot form a detectable stable complex with PspF₁₋₂₇₅, implying that determinants contributing to the interface between σ⁵⁴ and activator protein are widely distributed. Considering that Region I constitute the primary target for activator interaction (34), it is probable that substitution of these Region III residues indirectly affect the stable σ⁵⁴-activator complex formation through the interaction with Region I.

Interestingly, some σ⁵⁴ variants in the trapping reactions showed formation of a new complex which has not been observed before. This new complex has a slower mobility compared to the normal σ⁵⁴-activator complex (Figure 4). We speculate that this new complex might be a σ⁵⁴-activator complex with a new conformation, caused by the altered interaction of σ⁵⁴ with PspF₁₋₂₇₅. The new complexes were detected under conditions supporting PspF–ADP·AlFx formation, but we did not determine if all the reaction components (AlCl, NaF, ADP) were required. However, the presence of ADP alone was not sufficient for formation of new complexes (data not shown). Among the variants with this phenotype, most have substitutions located in Region III and only one variant was in Region II, residues 83–84.

Promoter DNA binding

σ⁵⁴ is a DNA-binding protein that can in vitro occupy some of its cognate promoters in the absence of core RNAP (36). Interaction of σ⁵⁴ with repressive fork junction promoter sequences around −12 plays an important role in the regulation of open complex formation (12,37–39). To determine the σ⁵⁴ sequences that contribute to promoter DNA binding, we conducted DNA-binding assay using heteroduplex promoter DNA (opened at −12 and −11, termed early melted DNA) to mimic the state of DNA in the closed complex (Figure 5A). Mutations that affected early melted DNA binding mainly located in two regions: the carboxyl terminal of Region I (residues from 25 to 48) and the DNA-binding domain (residues from 329 to 477). In Region I (residues from 25 to 38), it is interesting to note the periodicity of the DNA-binding defective phenotype (Figure 5B). Some other mutations located in modulation domain (residues 207–210, 241–242 and 295–300) also affect early melted DNA binding, consistent with the view that modulation domain functions to assist DNA binding (14).

Figure 5.

(A) S. meliloti nifH heteroduplex promoter probe used for DNA binding and σ⁵⁴ isomerization assay. The consensus GG and GC elements are underlined. The mismatched region and transcription start sequence are boxed in black. (B) The amounts of σ–DNA and ssσ–DNA complex resulting from the action of PspF_1–275 and dGTP are plotted. Only the examples of variants that fail to bind DNA, form ssσ–DNA complex, or variants that more readily form the ssσ–DNA complex are shown. I. Sigma binding to early melted DNA. II. ssσ–DNA complex formation in the presence of PspF_1–275 and dGTP.

Variants defective in early melted DNA-binding function abnormally in isomerization assay

σ⁵⁴ bound to early melted DNA responds to activator protein in the presence of an hydrolysable NTP source and isomerizes independently of core RNAP to form a new slower migrating ‘supershifted (ss)’ DNA complex (33). To characterize the sequences in σ⁵⁴ that contribute to activator responsiveness, we attempted to form the isomerized supershifted σ–DNA complex (ssσ–DNA) using activator PspF₁₋₂₇₅ and hydrolysable nucleotide dGTP. For each variant, a comparison of the fraction of initially bound DNA converted, in an activator and hydrolysable nucleotide dependent way, to the new isomerized ssσ–DNA species was made. Reactions without hydrolysable nucleotide were also conducted to potentially identify any variant that can response to activator protein in a hydrolysable nucleotide independent way. However, we failed to detect any variant with this phenotype, indicating the importance of hydrolysable nucleotide to the normal function of activator protein (data not shown).

The σ⁵⁴ isomerization assay revealed two classes of variants that behave differently from Cys(–)σ⁵⁴. The first class of variants is represented by some Region I variants which failed to form the isomerized ssσ–DNA species. This phenotype is also shared with some other variants that have substitutions located in core RNAP-binding domain (153–154), carboxyl terminal of modulation domain (295–296, 299–300), and DNA-binding domain (323–324, 335–336 and 455–456) (Figure 5B). However, residues that contribute to σ⁵⁴ isomerization mainly located in two parts of Region I (11–18 and 25–36), indicating that Region I comprises the major elements influencing activator responsiveness, but likely interacts with other parts of σ⁵⁴. The second class of variants is represented by some Region III variants. In striking contrast to the first class, this class of variants can form isomerized ssσ–DNA species much more efficiently than Cys(–)σ⁵⁴ (Figure 5B). As expected, early melted DNA binding seems to be important to the normal formation of ssσ–DNA, since most variants defective in early melted DNA binding either failed to form ssσ–DNA (represented by Region I variants), or formed ssσ–DNA much more efficiently (represented by Region III variants) (Figure 5B and Table 1). The latter group of variants seem to have a reduced barrier to being isomerized, implying Region III residues keep isomerization in check and that σ⁵⁴ could function through concerted changes in the organization of its domains.

Transcription activities of the variants in carboxyl terminal of DNA-binding domain are restored in vitro

To further define the elements in σ⁵⁴ that contribute to open complex formation and transcription initiation, abortive transcription assays were conducted from a supercoiled S. meliloti nifH promoter template (pMKC28) (28), whose sequence from −1 to +3 is TGGG (Figure 5A). If the activator PspF₁₋₂₇₅ and dATP (needed for the ATPase activity of the activator protein) were provided, the tetranucleotide UGGG can be synthesized by σ⁵⁴-holoenzyme in the presence of the priming dinucleotide UpG, and transcription substrate GTP. Heparin was also added together with GTP and after dATP plus PspF to measure transcription from a single round of activation.

The in vitro transcription activities of σ⁵⁴ variants were largely in accordance with their in vivo activity, but in the carboxyl terminal of DNA-binding domain, many variants with low in vivo activity functioned normally in the abortive transcription assay (Table 1 and Figure S3). Considering the low in vivo expression levels of these variants detected by immunoblotting, the low in vivo activities of these variants could be attributed to their in vivo instability or reduced abilities to compete with other σ factors in vivo, and these defects are not operational in vitro. Notably for many variants that had improved activities in forming the isomerized ssσ–DNA species, the abilities of the holoenzyme they formed to initiate transcription were seldom improved, sometimes even impaired (Figure 6 and Table 1), demonstrating the influence of steps after σ⁵⁴ isomerization to open complex formation and suggesting coordinated changes in σ⁵⁴ and core RNAP structure are required for making activator dependent open complexes.

Figure 6.

Mutations in σ⁵⁴ that affect transcription initiation. Autoradiograph of the denaturing gel showing the synthesis of abortive transcripts from holoenzyme formed with σ⁵⁴ variants which affect transcription initiation.

DISCUSSION

Cryo-EM reconstruction of σ⁵⁴–holoenzyme reveals four density regions (D1, Db, D2 and D3) attributable to σ⁵⁴. Region I sequence, including the −12 recognition region, is within Db density which is located at the position of DNA loading into the RNAP active centre, and is suggested to prevent the active centre from accessing the template DNA strand, therefore inhibiting the spontaneously formation of the open complex from double stranded DNA. Interaction of activator protein and σ⁵⁴ likely causes a conformational change of Region I which removes the steric obstruction and modifies the interaction with −12 promoter to allow the loading of the template strand into the active centre, leading to open complex formation (Figure 7) (12). Our results fully support this model, indicating that Region I is the major element interacting with activator protein and plays an important role in coordinating activator responsiveness. Although several distinct activities reside in discrete parts of σ⁵⁴, residues of σ⁵⁴ contributing to core RNAP binding or activator contacting are widely distributed across Regions I and III of σ⁵⁴, consistent with the observation that multiple interactions with core RNAP occur with several domains of σ⁵⁴ and that the main connecting density between PspF₁₋₂₇₅ and σ⁵⁴–holoenzyme occurs not only at the interface of Db, but also at D1 and D2 which contain major core RNAP-binding element of σ⁵⁴ and part of Region III of σ⁵⁴ (12). These residues probably constitute the surfaces of the density regions through which σ⁵⁴ exerts its influence upon the reorganization of RNAP.

Figure 7.

The left panel shows the cryo-EM reconstruction of σ⁵⁴-holoenzyme bound to the activator protein in the absence of promoter DNA (12). Shown in green and yellow are the positions of the σ⁵⁴ densities before and after binding to the activator protein, respectively. The σ⁵⁴-holoenzyme is shown alone (light blue) and in complex with the activator PspF_1–275 (red). The middle and right panels show cartoons depicting the structural changes inferred from cryo-EM reconstruction of the σ⁵⁴-holoenzyme and σ⁵⁴-holoenzyme bound by the activator protein. The promoter DNA is shown as an orange line: In the ‘CC conformation’ σ⁵⁴ domains D2 and D3 contact the −24 and −12 consensus promoter sequences upstream of the +1 site, respectively, thereby distorting DNA next to −12. The interaction with the activator protein (red/pink) in a mixed nucleotide state is thought to enable σ⁵⁴-holoenzyme to adopt a conformation similar to that in the intermediate complex (stabilized experimentally with ADP·AlFx).

Roles of Region I sequences

The amino terminal Region I plays a central role in activator responsiveness and may act as an organizing centre that brings the key σ⁵⁴, core RNAP and DNA component together (10,40,41). Mutations affecting σ⁵⁴ function located between residues 15 and 47 have been described previously (18,21), and we identified one additional sequence affecting activator responsiveness of σ⁵⁴, residues 11–14 (Figure 1). Most mutations affecting σ⁵⁴ isomerization were located between residues 11–18 and 25–36, suggesting these two sequences are the primary sequences for activator responsiveness. Since the −12 promoter sequence was shown to have a role in contributing to supershifted ssσ–DNA complex formation (33), for some variants defective in early melted DNA binding (AC25/26, 29/30, 33/34, 47/48), the failure to form isomerized ssσ–DNA species may be attributed to the altered interaction of σ⁵⁴ with −12 promoter sequence. Although some variants in DNA-binding domain also bound DNA poorly and less ssσ–DNA complexes were formed, the percentage of the initial bound DNA converted to ssσ–DNA is much higher compared to Cys(–)σ⁵⁴, suggesting that for these Region I variants, poor DNA binding per se does not limit the isomerization of sigma–DNA complex. Indeed tight DNA binding at −12 may limit isomerization. It has been shown before that interaction of σ⁵⁴ and −12 promoter sequence influences the regulation of open complex formation (37,39,42). Consistent with this, our results show a very clear correlation between Region I variants that have a defect in early melted DNA binding and their poor activities in activated transcription, contrasting the result that many other variants function normally in activated transcription, even if they have the same overall defect in early melted DNA binding (Figure 1). It has been suggested that an α-helix in Region I could be required for normal σ⁵⁴ function (21), and in our assay, it is interesting to note that mutations impairing DNA binding distribute periodically between residues 25–38 (Figures 1 and 5B).

Region I sequence binds core RNAP weakly and can be footprinted by core RNAP (16,31), and our results identified the sequences in Region I contributing to this function, associated with inhibiting the RNA polymerase from making open complexes (12). It has been demonstrated previously that residue 36 is proximal to core RNAP (25). Consistent with this, our results show that mutations around residue 36 (residues 33–34 and 37–38) affect core RNAP binding slightly (Figure S2). The variants with reduced core RNAP-binding activity are also impaired in σ⁵⁴ isomerization and/or activator responsiveness, suggesting that the substituted residues contribute to the interface σ⁵⁴ makes with core RNAP through which Region I exerts its influence upon holoenzyme to regulate function.

Roles of Region II sequences

Region II is very variable in sequence. Groups of acidic residues in Region II are involved in promoter melting and deletion of Region II affects holoenzyme formation and activator responsiveness of σ⁵⁴ (32,43). In our activator interaction assays, one Region II variant AC83/84, behaves differently from Cys(–)σ⁵⁴ (Figure 4), suggesting that an altered interaction with activator protein may occur. The activity of this variant in isomerized ss σ–DNA formation is also improved compared to Cys(–)σ⁵⁴, implying the substituted residues may influence activator responsiveness (Table 1). However, we failed to identify any Region II variants significantly affecting the transcription activity of the assembled σ⁵⁴-holoenzyme both in vivo and in vitro.

Roles of Region III sequences

Region III includes a major determinant for core RNAP binding at its amino terminal and sequences for DNA binding located at the carboxyl terminal (13,15,16). In the cryo-EM reconstruction of σ⁵⁴–holoenzyme, the primary core RNAP-binding domain and carboxyl terminal DNA-binding domain are assigned to D1 and D3, respectively, while some other Region III residues located in D2 with Region I contribute to maintaining the stable closed complex (Figure 7) (28,41,44,45).

Our data show that although the primary core RNAP-binding sequence is located between residues 120 and 215, many other residues outside this region also affect core RNAP binding (Figure 3). This is consistent with the fact that sequences footprinted by core RNAP lie outside the primary core RNAP-binding domain, and are near or within the DNA-binding domain (31). On the basis of the model of the complex formed between the carboxyl terminal DNA-binding domain of σ⁵⁴ (AAs 396–465) and promoter −24 sequence, it is proposed that the first loop of this domain may interact with the β-flap region of core RNAP (20,30). Interestingly, we find that mutations between residues 423–426, which locate in the first loop of the carboxyl terminal DNA-binding domain based on the structure of σ⁵⁴ from A. aeolicus (Figure 1), indeed affect the core RNAP-binding activity (Figure 3).

Residues affecting activator interaction are also distributed widely in Region III (Figure 4). Since Region I is the major determinant for activator interaction (33), these Region III variants may influence activator interaction through the interaction between Regions III and I (28,41). Region III variants affecting activator interaction can be divided into two classes. Class one variants are somewhat defective in activator interaction, but unlike Region I variants with this same phenotype, many of these Region III variants can still respond to activator protein. Class two variants can interact with activator protein, but an additional complex with slower mobility was formed (Figure 4). This new complex has not been observed before and we suggest that this new complex might reflect a conformational change in the σ⁵⁴-activator complex, implying an altered interaction of σ⁵⁴ with activator protein. Residues around 307 have been proposed to be a site for activator interaction (46). Interestingly, our data show that mutations between residues 297 and 318 either lead to the failure of σ⁵⁴ variants to form stable complex with activator protein, or the formation of an additional complex (Table 1), both cases reflecting the altered interaction with activator protein. Interactions that activator protein mediates between Regions III and I can be used to achieve the apparently concerted domain movements of σ⁵⁴ seen in structural reconstructions of the PspF₁₋₂₇₅ bound σ⁵⁴-holoenzyme (Figure 7) (12).

In contrast to the results of core RNAP-binding activity and σ⁵⁴-activator interaction, residues affecting DNA binding are mainly located in the previously delineated DNA-binding domain, residues 329–477, fully consistent with the reconstruction model of σ⁵⁴-holoenzyme–DNA complex where D2 and D3 densities cover −12 and −24 promoter sequences, with D1 containing the primary core RNAP-binding domain (residues 120–215). Based on the structure of the carboxyl terminal DNA-binding domain of σ⁵⁴ from A. aeolicus, the residues D439, T457 and N471 of σ⁵⁴ from K. pneumoniae can interact with −24 promoter sequence through hydrogen bonding (30). In accordance with this, our results show that substitutions of these three residues all affect the DNA-binding ability significantly (Table 1). Specific residues in modulation domain affecting DNA binding were also identified, confirming the view that modulation domain functions to assist DNA binding (14).

Unlike Region I variants, there is no close linkage between the defect in early melted DNA binding and ssσ–DNA formation. Among the DNA-binding defective variants in Region III, only three of them (AC295/296, 299/300 and 355/356) are defective for early melted DNA binding (Table 1). It has been suggested that residue 336 is proximal to −12 promoter sequence and interacts with sequences downstream to −12 GC element to regulate RNAP function together with Region I (28,41). Thus, like the Region I variants, the defect of variant AC335/336 in ssσ–DNA formation could also result from the altered interaction of σ⁵⁴ and the −12 promoter sequence. In contrast to the properties of these variants, in the carboxyl terminal DNA-binding domain variants defective in DNA binding (AC439/440, AC441/442, AC457/458 etc.) are not impaired in ssσ–DNA formation, indicating that it is an interaction with −12 promoter sequence that is crucial for σ⁵⁴ isomerization, with the rpoN box providing the −24 promoter sequence contact that is not directly involved in the isomerization process. Variants AC295/296 and AC299/300 share the phenotype of AC335/336 and some Region I variants, implying that residues 295–300 located in the carboxyl terminal of modulation domain may also have a role in interactions with −12 promoter sequence, which may have a basis in communication with Region I.

Although most Region III variants are not impaired in ssσ–DNA formation, a distinctive class of variants is revealed on the basis of their greatly improved activities in ssσ–DNA formation. This class of variants can be roughly divided into two groups based on the linkage with other functional defects. Group I isomerization changed variants are represented by the variants in core RNAP-binding domain whose core RNAP-binding activities are impaired. Mutations leading to this phenotype are mainly located in two patches around residues 120 and 180 (Figure 1). It has been suggested that some interactions between σ⁵⁴ and core RNAP are changed upon isomerization of the σ⁵⁴–DNA complex and recent structural analyses of the activator bound σ⁵⁴-holoenzyme are in agreement with the activator causing a reorganization of σ⁵⁴ domains within the holoenzyme (12,47). The impaired core RNAP-binding activities of these variants described here may reflect an altered core RNAP-binding surface which is closer to the conformation of σ⁵⁴ in the isomerized state, hence leading to the improved activity in isomerized ss σ–DNA formation. The activities of these σ⁵⁴ variants in transcription assay are somehow impaired, suggesting that the substituted residues may contribute to the communication with core RNAP and that when they are substituted, σ⁵⁴ isomerization is uncoupled from the changes that must occur in the core RNAP, required for the later stages of the DNA opening process and stable open complex formation (48,49).

Group II isomerization changed variants are in the DNA-binding domain. For these variants there is a clear linkage between the defect in DNA binding and an improved activity in ssσ–DNA formation (Figure 1). The weak contact these variants make with DNA may facilitate the changes in σ–DNA interactions which are important for ssσ–DNA complex formation (33,47). However, increased formation of the ssσ–DNA complex has little overall effect on the normal function of holoenzyme in the context of in vitro transcription assays, arguing for the existence of other steps limiting open complex formation. For most of the variants in carboxyl terminal of σ⁵⁴ (residues 431–476), the normal functions of the holoenzyme in in vitro transcription assay, together with their low expression levels in vivo, strongly indicates the carboxyl terminal of σ⁵⁴ plays an important role in maintaining the structural integrity perhaps related to turnover of σ⁵⁴ in vivo.

In conclusion, by constructing the single cysteine mutant library of σ⁵⁴ and analyzing the functions of σ⁵⁴ variants both in vivo and in vitro, σ⁵⁴ sequences interacting with core RNAP, DNA and activator protein are systematically identified. Linkages between different functions are revealed and the possible mechanisms for those linkages are also discussed. Furthermore, the characterized σ⁵⁴ library provides a powerful resource for functional analyses using site specific probes, and for interpretations of molecular structures.

SUPPLEMETARY DATA

Supplementary Data are available at NAR Online.

FUNDING

Human Frontier Science Program [RGP22/2007 to Y.P.W.]; Natural Science Foundation of China [No. 30830005 to Y.P.W.]; the Program of Introducing Talents of Discipline to Universities [No. B06001 to Y.P.W.]; 973 program on nitrogen fixation to Y.P.W.; BBSRC project grant to M.B. and BBSRC David Phillips Fellowship [BB/E023703 to S.R.W.].

Conflict of interest statement. None declared.