Transcriptional Analysis of Genes Encoding Shiga Toxin 2 and Its Variants in Escherichia coli

Abstract

Six of 37 non-O157 Escherichia coli strains possessing Shiga toxin (Stx) 2 gene variant stx_2d or stx_2e secreted no detectable Stx. These isolates produced significantly less stx mRNA than Stx2d, Stx2e, Stx2c, or Stx2 secretors did. Standard screening procedures miss a significant subset of E. coli harboring stx₂ variants.

Shiga toxin (Stx)-producing Escherichia coli strains (STEC) cause diarrhea, hemolytic uremic syndrome (HUS), or asymptomatic infections. Stx1 and Stx2 are two major Stx types, and each has variants (5, 8, 9, 12, 16, 17, 22). Whereas STEC containing stx₂ or stx_2c cause HUS (3), strains possessing stx_2d (12) or stx_2e are less virulent (3). Also, STEC from HUS patients produce significantly higher levels of Stx1 or Stx2 than bovine STEC do (14). In this study, we applied Vero cell cytotoxicity assays (6), enzyme immunoassays (EIA), latex agglutination (LA), and colony immunoblot assays to characterize Stx production in non-O157 E. coli clinical isolates that possess an stx₂ variant as their sole stx gene.

Broth cultures of 37 non-O157 E. coli strains were investigated (Table 1) using Vero cell cytotoxicity assays, EIA (Ridascreen Verotoxin; R-Biopharm GmbH, Darmstadt, Germany), verotoxin-producing E. coli reverse passive LA (VTEC-RPLA; Denka Seiken Co., Tokyo, Japan), and colony immunoblot assays (Shigatoxin [Verotoxin] Colony Immunoblot; Sifin GmbH, Berlin, Germany) (3, 6, 19). Selected polymyxin B (5,000 U/ml; Sigma, Taufkirchen, Germany) bacterial extracts were also tested for Stx. Complete stx_2d and stx_2e genes were amplified with primer pairs SD-a (5′-TTCTAAGCAATCGGTCACT-3′)-SD-b (5′-GTAACTACATTGCTGCACAC-3′) and SE-a (5′-GAGCAGACGACACGATAACA-3′)-SE-b (5′-AACAGCATCCACAACACTA-3′), respectively, purified (PCR purification kit; QIAGEN, Hilden, Germany), sequenced, analyzed (DNASIS program; Hitachi Software), and compared with sequences from the National Center for Biotechnology Information database. A one-step quantitative reverse transcription (RT)-PCR was performed with the LightCycler system (Roche, Mannheim, Germany) and the QuantiTect SYBR green RT-PCR kit (QIAGEN) to compare stx_2c, stx_2d, and stx_2e mRNA levels; two stx₂⁺ E. coli strains served as positive controls. Total RNA was isolated (RNeasy Mini kit; QIAGEN) without mitomycin C and after mitomycin C (0.5 μg/ml; Sigma) induction. PCR was performed in microcapillary tubes in 10-μl volumes containing 100 ng of total RNA, 2× QuantiTect SYBR green RT-PCR master mix (5 μl), QuantiTect RT mix (0.1 μl), MgCl₂ (4 mM), and a 0.5 μM concentration of each primer. The primers RT-stx2F (5′-CGACCCCTCTTGAACATA-3′) and RT-stx2R (5′-TAGACATCAAGCCCTCGTAT-3′), GK5 and GK6 (15), stx2dB-1 (5′-AAGAAGATATTTGTAGCGG-3′) and stx2dB-2 (5′-CGTCATTCCTGTTAACTGTGCG-3′), FK1 and FK2 (2), and icdA-F and icdA-R (21) were used to amplify stxA₂, stxA_2c, stxA_2d, and stxA_2e; stxB₂ and stxB_2c; stxB_2d; stxB_2e; and the housekeeping gene icdA, respectively. RT-PCR included reverse transcription (50°C, 30 min) and preliminary denaturation (95°C, 15 min) steps and 40 cycles of denaturation (94°C, 10 s), annealing (50°C, 20 s), and extension (72°C, 20 s). Melting curve analysis was performed with continuous fluorescence reading from 60 to 95°C. For each mRNA assayed, relative quantification was performed using an external standard (purified icd and the respective stx₂ variant PCR products). Data were analyzed by the fit-point method with LightCycler version 3.5 software. stx mRNAs were normalized to icdA mRNA. Control reactions were performed without reverse transcriptase to confirm that the target detected was RNA.

TABLE 1.

Comparison of Stx detection in supernatants of 37 E. coli strains^a using the Vero cell cytotoxicity assay, enzyme immunoassay, and latex agglutination assay

stx geno- typeb	No. of strains	Disease associationc			No. of strains with indicated Stx detection resultsd
		HUS	D	A	Vero+, EIA+, VTEC- RPLA+	Vero+, EIA−, VTEC- RPLA+	Vero+, EIA−, VTEC- RPLA−	Vero−, EIA−, VTEC- RPLA−
stx2c	9e	5	2	2	7	2	0	0
stx2d	16f	0	5	11	11	0	0	5
stx2e	12g	0	8	4	7	1	3	1
Total	37	5	15	17	25	3	3	6

^aIsolated at the Institute of Hygiene and Microbiology, University of Würzburg, Würzburg, Germany, and at the Institute of Hygiene, University Hospital Münster, Münster, Germany, using a protocol described previously (3).

^bstx_2c, stx_2d, and stx_2e genotypes were determined as described previously (3); all 16 stx_2d-harboring strains contained an stx_2d allele which was amplified with primers VT2cm-VT2f (12) and encodes Stx2d not activatable by intestinal mucus (12); stx_{2d-activatable} (9) was absent from all 16 of these strains.

^cD, diarrhea; A, asymptomatic infection; HUS, hemolytic-uremic syndrome.

^dVero, Vero cell cytotoxicity assay; EIA, Stx enzyme immunoassay; VTEC-RPLA, latex agglutination assay; +, positive result; −, negative result.

^eSerotypes (number of strains, if more than one, in parentheses): O77:H⁻ (nonmotile), O23:H⁻, O30:HNT (H nontypeable), O60:H19, O92:HNT, O120:HNT, O145:H⁻ (2 strains), Orough (autoagglutinable):H⁻.

^fSerotypes (number of strains, if more than one, in parentheses): O40:H8 (6 strains), O86:H⁻, O91:H⁻ (2 strains), O146:H28 (2 strains), Orough:H45, Orough:HNT, ONT (O nontypeable):H12, ONT:H32, ONT:HNT.

^gSerotypes (number of strains, if more than one, in parentheses): O8:HNT, O60:H2, O60:H⁻ (2 strains), ONT:H⁻ (8 strains).

Detection of Stx.

Five of 16 stx_2d- and 1 of 12 stx_2e-containing isolates, respectively, did not release Stx that was detectable by any assay (Tables 1 and 2). Polymyxin extracted minute amounts of Stx from the periplasmic space of each of the five stx_2d-harboring strains but not from the stx_2e-containing strain (Table 2).

TABLE 2.

Detection of Stx in polymyxin B extracts of the E. coli strains which secreted no detectable Stx

Strain no./yr isolated	Serotype	Diseasea (reference)	stx genotype	Detection of Stx in culture supernatants by:			Detection of Stx in polymyxin extracts by:			Detection of Stx by colony immunoblotingg
				Vero cell assayb	Stx EIAc	VTEC-RPLAd	Vero cell assaye	Stx EIAf	VTEC-RPLAd
24196/97	O86:H−	A	stx2d	−	−	<1:2	−	−	1:4	−
6451/98	Or:HNT	D	stx2d	−	−	<1:2	+	−	1:2	−
5293/98	O40:H8	A	stx2d	−	−	<1:2	+	−	1:4	−
5299/98	O40:H8	A	stx2d	−	−	<1:2	+	+	1:2	−
5300/98	O40:H8	A	stx2d	−	−	<1:2	+	+	1:2	−
26725/97	ONT:H−	A	stx2e	−	−	<1:2	−	−	<1:2	−
EH 250	ONT:H12	D (12)	stx2d	+	+	1:16	+b	+c	1:32	+
4797/97	O91:H−	A	stx2d	+	+	1:4	+b	+c	1:16	+
2771/97	ONT:H−	D (10)	stx2e	+	+	1:4	+b	+c	1:16	+

^aD, diarrhea; A, asymptomatic infection.

^b−, Stx titer of <1:5; +, Stx titer of >1:5 (usually >1:50).

^cStx enzyme immunoassay (Ridascreen Verotoxin); −, negative (optical density at 620 nm [OD₆₂₀] values below the cutoff value); +, strongly positive (OD₆₂₀ values comparable to that of the positive control).

^dThe titers are reciprocals of the highest dilution that caused agglutination of latex particles sensitized with a polyclonal antibody against Stx2 (Stx2 reagent; VTEC-RPLA).

^e−, titer of <1:5; +, titer equal to 1:5.

^f−, negative (OD₆₂₀ values below the cutoff value); +, weakly positive (OD₆₂₀ values slightly above the cutoff value).

^gAll the other strains investigated gave positive signals in the colony immunoblot analysis.

Sequence analysis of stx genes.

The stxB_2d and stxB_2e genes from the six Stx nonsecretors were 100% identical to stxB_2d and stxB_2e of prototype strains EH250 and 2771/97 (GenBank accession numbers AF043627 and AJ249351, respectively), which secrete the corresponding Stx (Table 2). stxA_2d and stxA_2e from these strains were 97.1 to 99.9% homologous to the respective subunits of strains EH250 and 2771/97. The deduced amino acid sequences of StxA subunits of these strains are compared in Fig. 1. There was no frame shift, deletion, insertional inactivation, or truncation of stx in any nonproducing strain.

FIG. 1.

Comparison of deduced amino acid (aa) sequences of StxA subunits from stx_2d- or stx_2e-harboring strains which secrete no detectable Stx with those from Stx2d-producing strain EH250 (GenBank accession number AF043627) and Stx2e-producing strain 2771/97 (GenBank accession number AJ249351). The amino acid changes in StxA subunits are located outside positions containing conserved amino acids (4) and do not affect glutamic acid at position 167, which is an active-site StxA residue (4); glutamic acid at position 167 is conserved in all strains. The sequences of the StxB subunits were 100% identical to those of the reference strains. The stx_2d sequence from strain 5293/98 was identical to those of the other two stx_2d-harboring Stx nonsecretors, 5299/98 and 5300/98.

Quantitative RT-PCR analysis and mitomycin C effect.

The five Stx2d nonsecretors (Fig. 2, lanes 1, 3, 5, 7, and 9) and one Stx2e nonsecretor (lanes 15) had significantly less cognate stx mRNA (both subunits) than the Stx2d (lanes 11 and 13) and Stx2e (lanes 17) secretors. Mitomycin C did not significantly augment stxA (Fig. 2A, lanes 2, 4, 6, 8, 10, and 16) or stxB (Fig. 2B, lanes 2, 4, 6, 8, 10, and 16) transcription. In contrast, in Stx2d or Stx2e secretors, mitomycin C increased both stxA and stxB transcription approximately 100-fold (Fig. 2, lanes 12, 14, and 18).

FIG. 2.

Relative stxA_2d, stxA_2e, stxA_2c, and stxA₂ (A) and stxB_2d, stxB_2e, stxB_2c, and stxB₂ (B) mRNA expression levels in Stx nonsecretors and Stx secretors analyzed by quantitative RT-PCR. In lanes 1 to 26, the following strains are depicted (the stx₂ allele, Stx secretion [−, mRNA levels measured without mitomycin C induction; +, mRNA levels measured after mitomycin C induction], and diagnosis [A, asymptomatic infection; D, diarrhea; HUS, hemolytic uremic syndrome] are shown in parentheses): lanes 1 and 2, 24196/97 (stx_2d, Stx−, A); 3 and 4, 6451/98 (stx_2d, Stx−, D); 5 and 6, 5293/98 (stx_2d, Stx−, A); 7 and 8, 5299/98 (stx_2d, Stx−, A); 9 and 10, 5300/98 (stx_2d, Stx−, A); 11 and 12, 4797/97 (stx_2d, Stx+, A); 13 and 14, EH250 (stx_2d, Stx+, D); 15 and 16, 26725/97 (stx_2e, Stx−, A); 17 and 18, 2771/97 (stx_2e, Stx+, D); 19 and 20, 4911/99 (stx_2c, Stx+, HUS); 21 and 22, 3032/03 (stx_2c, Stx+, D); 23 and 24, 1309/99 (stx₂, Stx+, HUS); 25 and 26, 3733/97 (stx₂, Stx+, D). The levels of stx mRNAs were normalized to those of icdA mRNA. Data are means ± standard deviations of results from three independent experiments. Comparison of the means and comparison of groups of the relative levels of stx mRNA were performed using the Student's t test and χ² test, respectively. P values of <0.05 were considered significant. The means of results for stxA_2d (lanes 1, 3, 5, 7, and 9) and stxB_2d (lanes 1, 3, 5, 7, and 9) nonsecretors were compared to those for stxA_2d (lanes 11 and 13) and stxB_2d (lanes 11 and 13) secretors, in the absence of mitomycin C (t = 12.4; P = 0.0001; 95% confidence interval [CI₉₅], 9.55 to 14.56 for stxA_2d and t = 9.1; P = 0.0003; CI₉₅, 5.59 to 10.03 for stxB_2d) and in the presence of mitomycin C (t = 30; P < 0.0001; CI₉₅, 965 to 1,142 for stxA_2d and t = 13.73; P < 0.00001; CI₉₅, 1,388 to 2,030 for stxB_2d). The means of results for stxB_2d and stxB_2e nonsecretors (lanes 1, 3, 5, 7, 9, and 15) were compared to those for stxB_2d and stxB_2e secretors (lanes 11, 13, and 17) (t = 28.6; P < 0.0001; CI₉₅, 8.8 to 10.2). Differences in the relative level of stx mRNA among secretors and nonsecretors between groups in the presence and absence of mitomycin C were compared for stxA_2d (χ² = 49.7; P < 0.00001; CI₉₅, 5.6 to 150) and stxA_2e (χ² = 68.9; P < 0.00001; CI₉₅, 6.87 to 347). Differences in means in the presence of mitomycin C were compared among nonsecretors (t = 3; P = 0.17; CI₉₅, 0.84 to 6.40 for stxA_2d and t = 1.9; P = 0.1; CI₉₅, 0.5 to 4.55 for stxB_2d) and among secretors (t = 16.1; P = 0.0038; CI₉₅, 755 to 1,336 for stxA_2d and t = 7.242; P = 0.0185; CI₉₅, 651 to 2,757 for stxB_2d). Differences in the means of relative levels of stx mRNA in the absence of mitomycin C of secretors (lanes 11, 13, and 17) of stxA_2d were compared for stxA_2e (t = 16.4; P = 0.0230; CI₉₅, 1 to 156), for stxA_2c (t = 16.4; P = 0.0230; CI₉₅, 1 to 156), and for stxA₂ (t = 24.3; P = 0.0001; CI₉₅, 97 to 126). Differences in the means of relative levels of stx mRNA in the presence and absence of mitomycin C were compared for stxA_2c (t = 86.2; P < 0.00001; CI₉₅, 11,848 to 12,591), stxB_2c (t = 24.9; P < 0.00001; CI₉₅, 9,738 to 11,866), stxA₂ (t = 12.0; P < 0.00001; CI₉₅, 2,789 to 2,961), and stxB₂ (t = 27.7; P < 0.00001; CI₉₅, 2,357 to 2,815).

Comparison of stx₂ and stx₂ variant transcription in Stx secretors.

stxA (Fig. 2A) and stxB (Fig. 2B) mRNA levels were higher in strains containing stx_2e (Fig. 2, lanes 17), stx_2c (lanes 19 and 21), or stx₂ (lanes 23 and 25) than in those containing stx_2d (lanes 11 and 13), but the differences reached statistical significance only for stxA. No significant differences in stx_2d mRNA levels were detected between Stx2d secretors isolated from diarrhea (strain EH250 [Fig. 2, lanes 13]) and asymptomatic infection (strain 4797/97 [lanes 11]). Isolates from HUS (strains 4911/99 and 1309/99 [lanes 19 and 23]) and from diarrhea (strains 3032/03 and 3733/97 [lanes 21 and 25]) had similar stx₂ or stx_2c mRNA levels. However, mitomycin C significantly increased stx_2c (Fig. 2, lanes 20 and 22) and stx₂ (lanes 24 and 26) transcription in all secretors of the respective Stx.

These data suggest that Stx detection can be compromised by Stx nonsecretors and that molecular methods detecting stx might offer advantages (13, 18). The reason for Stx nonexpression is not the insertion sequence IS1203v within stxB (7), because stx genes in the Stx nonproducers are structurally intact. Stx production appears to be controlled transcriptionally, and its failure is not remedied by mitomycin C induction of the lytic cycle (11, 20), suggesting that stx genes of these strains are not phage borne or are located on defective phages. In contrast, mitomycin augments stx transcription in strain 2771/97 and the Stx2d secretors EH250 and 4797/97, in agreement with our previous report that stx_2e in strain 2771/97 is located on an intact phage (10), which suggests that also stx_2d genes in both Stx2d secretors are phage encoded. Notably, we found no association between the stx transcription level and clinical outcome of the infection within producers of Stx2c, Stx2d, or classical Stx2, in concordance with a recent North American study of E. coli O157:H7 infections (1). Interestingly, although Stxs are A₁B₅ toxins, the mRNA levels were similar for the A- and B-subunit genes (Fig. 2), suggesting that the stoichiometry of the subunits is not transcriptionally regulated. In summary, we have demonstrated discordance between stx genotype and Stx expression, for stx_2d or stx_2e. Expression appears to be controlled at the level of transcription. Phenotypic detection of Stx will miss STEC that poorly secrete Stx.

Nucleotide sequence accession numbers.

Nucleotide sequences for stx_2d from E. coli strains 24196/97, 6451/98, and 5293/98 and stx_2e from E. coli strain 26725/97 have been entered into the EMBL database under accession numbers AJ567995 to AJ567998, respectively.