The study was conducted at Aligarh hospital, India (from April 2007 to November 2008). Seventy-seven diabetic patients having clinically infected foot ulcers admitted to the endocrinology ward were studied in a consecutive series. Wagner classification was employed to grade the ulcers 9. Thirty-two study factors were recorded for each patient. BMI i.e. Body Mass Index (< 18.5 kg/m² = underweight; 18.5-22.9 kg/m² = normal weight; 23.0- 24.9 kg/m² = overweight), presence of nephropathy (creatinine ≥ 150 μmol/l or presence of micro- or macroalbuminuria), neuropathy (absence of perception of the Semmes-Weinstein monofilament at 2 of 10 standardized plantar sites on either foot) and peripheral vascular disease (ischemic symptoms and intermittent claudication or rest pain, with or without absence of pedal pulses), were some of them. Osteomyelitis was diagnosed on suggestive changes in the radiographs and bone scans. Regarding a fasting blood glucose level of < 110 mg/dl and/or postprandial level of < 160 mg/dl as glycemic control, the number of patients achieving glycemic control over the hospital stay was compared with respect to sex. In case a patient had multiple ulcers, specimens from all the ulcers were taken. If any of the ulcers was found to be infected with MDRGNB, the patient was grouped in the MDRGNB-infected category. Fasting HDL and VLDL were measured and LDL was calculated by subtracting their sum from total cholesterol. Clinical assessment for signs of infection (swelling, exudate, surrounding cellulitis, odor, tissue necrosis, crepitation, and pyrexia) was made. Ulcer size was determined by multiplying the longest and widest diameters and expressed in centimeters squared. Demographic and other relevant details were retrieved from medical records.

Culture specimens were obtained at the time of admission, after the surface of the wound had been washed vigorously by saline, and followed by debridement of superficial exudates. Specimens were obtained by scraping the ulcer base or the deep portion of the wound edge with a sterile curette. The soft tissue specimens were quickly sent to the laboratory and processed for Gram-negative bacteria. Standard methods for isolation and identification of bacteria were used 10. Antimicrobial susceptibility testing of the isolates was performed by the standard disc diffusion method as recommended by the Clinical and Laboratory Standards Institute (CLSI) 11. The test-antibiotics included gentamycin, tobramycin, amikacin, kanamycin, streptomycin, cefalothin, cefazolin, cefuroxime, cefoxitin, ceftazidime, cefotaxime, ceftriaxone, cefepime, ciprofloxacin, norfloxacin, tetracycline, piperacillin, imipenem as well as combinations such as amoxyclav and piperacillin/tazobactam. Minimum Inhibitory Concentrations (MICs) for the same antimicrobials were determined by the CLSI microbroth dilution method 11. The ESBL synergy test for the study isolates was performed as described by Jarlier et al 12. For this study, bacteria resistant to at least 4 of the test antibiotics were designated as MDRGNB. All ESBL-producers were considered as MDRGNB.

Enterobacterial repetitive intergenic consensus (ERIC)-PCR was used for the analysis of genetic relatedness of the ESBL-producers as described elsewhere 13. A search for bla_CTX-M, bla_TEM-1, armA, rmtA and rmtB genes in the genomic and plasmidic DNA of isolates which were positive for ESBL synergy test was performed by PCR amplification, as described previously 141516. Plasmids were obtained by the method of Kado and Liu 17. Screening of integrons was also performed as described elsewhere 18.

PCR amplified bla_CTX-M genes, were ligated into 'pDrive-easy-PCR-cloning-vector' and transformed into competent E. coli C600 cells using Qiagen Cloningplus Kit (Qiagen, USA). Transformed colonies were selected on Luria-Bertani agar plates, containing isopropyl-β-D-thiogalactopyranoside (1 mM), 5-bromo-4-chloro-3-indolyl-β-D-thiogalactopyranoside (40 mg/L) and kanamycin (30 mg/L). Plasmid extracted from the culture of a well-isolated white colony was used as template for PCR amplifications of bla_CTX-M. The same plasmids were transformed 19 into azide-resistant E. coli J53 cells. The transformants were selected on Luria-Bertani agar plates containing azide (250 mg/L), cefotaxime (2 mg/L) and kanamycin (30 mg/L). The transformed cells were re-checked for ESBL-production. DNA sequences were analyzed with ABI 3130 Genetic Analyzer (Applied Biosystems). The Basic Local Alignment Search Tool (BLAST) program of the National Center for Biotechnology Information http://www.ncbi.nlm.nih.gov was used to search databases for similar nucleotide sequences.

E. coli strains that were PCR-positive for any of the tested genes were used as donors in transconjugation experiments as described elsewhere 20. Azide-resistant E. coli J53 was used as the recipient. Transconjugants were selected on Luria-Bertani agar plates containing azide (250 mg/L) and either cefotaxime (2 mg/L) or gentamicin (50 mg/L).

Qualitative variables were expressed as percentages and quantitative variables were expressed as means ± SD (Standard Deviation). Significance of the study variables for "sex-associated MDRGNB infection status" of the foot ulcers was tested by using Student's t test or Fisher's exact test as appropriate. All risk factors that proved to be statistically significant on univariate analysis were entered by blockwise entry in a logistic regression model after exclusion of multicollinearity and interaction between variables. A two-tailed p value of < 0.05 was taken as significant. Statistical analysis was performed using SPSS (version 11.5, Chicago). The results were again confirmed using the SISA (Simple Interactive Statistical Analysis) program http://www.quantitativeskills.com/sisa/distributions/binomial.htm

BLAST-P was performed to retrieve suitable templates for homology-modeling using the bla_CTX-M sequences obtained in this study. Protein Data Bank (PDB) IDs of these templates are shown in table 1. Models were prepared using Swiss-Model-Server 21 and verified by Procheck 22 and verify3D programmes 23. Discovery Studio2.0 (Accelrys) was used to prepare Ramachandran plots. Z-scores were calculated using 'WHAT IF'-package. Cefotaxime was docked into each of the modeled structures employing Hex 5.1.

A total of 185 Gram-negative bacteria were isolated from 95 foot ulcer samples of the 77 diabetic patients admitted during the study-period. Out of these bacteria, 149 (80.5%) were found to be MDRGNB. Sixty-six samples which belonged to 55 patients were found to harbor these MDRGNB. Fig 1 shows the distribution of MDRGNB obtained from infected diabetic foot ulcers divided into 2-month periods from April 2007 to November 2008. Characteristics of male and female patients having MDRGNB-infected foot ulcers were recorded. Among the 55 MDRGNB-infected patients, 46 (83.6%) were males while 9 (16.4%) were females. Grade-5 ulcer was found in 1 male patient only. The majority of subjects (89.1%) had type 2 diabetes. Fifty-one (92.7%) were hypertensive, 22 (40%) had retinopathy, 37 (67.3%) had nephropathy, 34 (61.8%) had neuropathy and 33 (60%) had peripheral vascular disease. Nine (16.4%) patients had osteomyelitis. Three male patients died during the hospital stay. However, there were only 2 subjects who possessed foot ulcer infected with more than 1 ESBL-producing strain.

Forty-three (28.9%) ESBL-producers were identified among the 149 MDRGNB. The highest ESBL production was noted in Pseudomonas aeruginosa (20/61; 32.8%), followed by E. coli (14/51; 27.5%) and K. pneumoniae (9/38; 23.7%). All the MDRGNB showed 100% susceptibility to imipenem (data not shown). All the ESBL-producers, with the exception of P. aeruginosa strains were found to be susceptible to cefoxitin, imipenem and piperacillin/tazobactam while the same were resistant to ceftazidime, cefotaxime and piperacillin. However, the ESBL producing strains of P. aeruginosa were invariably resistant to cefoxitin. Amikacin was the most effective of all the aminoglycosides against ESBL-producers. Imipenem alone and the piperacillin/tazobactam combination were found to be the therapies of choice. Hence, the 43 ESBL-producers could be broadly divided into 3 different resistance patterns. Strains resistant to aminoglycosides and quinolones but susceptible to cefepime constituted pattern 1. Pattern 2 comprised of strains resistant to cefepime but susceptible to either aminoglycosides or quinolones or both. Strains resistant to all the three mentioned classes of antibiotics constituted pattern 3. Resistance pattern 1 was found to be the most common among ESBL-producers. The other aerobic Gram-negative organisms isolated from foot ulcers included Acinetobacter baumannii (5 strains), Proteus species (20 strains), Enterobacter species (1 strain) and Morganella morganii (1 strain). None of these were ESBL-producers. Only 13 anaerobes were isolated of which 8 were Gram-negative. The Gram-negative anaerobes consisted of Veillonella species (3 strains) and Bacteroides species (5 strains) none of which happened to be MDRGNB.

Only 2 strains were confirmed to harbor bla_CTX-M by PCR. Strain D253 was positive for both bla_CTX-M and bla_TEM-1 while strain D281 possessed only bla_CTX-M. None of the strains were positive for any of the 16S rRNA methyl transferase genes (i.e. armA, rmtA or rmtB). Seven integrons were detected whose sizes ranged from 600 bp to 1.5 Kb. All these integrons were found to be Class1 type and were positive for int1 and sul1 genes of the expected sizes 845 bp and 840 bp respectively. None of the genes were located in the integrons. E. coli C600 cells harboring cloned bla_CTX-M (C600 ^blaCTX-M) were again found to be PCR-positive for the said gene in all instances. Moreover, cultures derived from C600 ^blaCTX-M cells gave a positive ESBL synergy test. Azide resistant E. coli J53 cells transformed with the plasmidic DNA extracted from C600 ^blaCTX-M cells became cefotaxime resistant. Also, cultures derived from these transformed E. coli J53 cells were found to be positive for the ESBL test. Sequencing results confirmed all the bla_CTX-M genes obtained in this study as bla_CTX-M-15.

The ceftazidime and cefotaxime resistance markers were located on the plasmidic DNA of both the bla_CTX-M-15 positive strains and were transmissible through conjugation. Quinolone (ciprofloxacin) resistance was co-transferred with cefotaxime resistance, in only one of the bla_CTX-M-15 positive isolates. Transfer frequencies (Total no. of transconjugants per plate/Total no. of recepients per plate) observed were of the order 10^-5 (Table 2). Randomly taking a single representative strain in case of identical susceptibility patterns, ERIC-PCR was performed for 40 ESBL-producers. P. aeruginosa, K. pneumoniae and E. coli displayed 5, 5 and 8 major ERIC-profiles respectively (Fig 2). Both the bla_CTX-M-15 positive strains were found to be clonally related and displayed ERIC-profile M1 (Table 1).

Over 90% of the amino acid residues in the protein structures modeled from bla_CTX-M-15 genes obtained in this study were found to be present in the most favorable regions as revealed by their respective Ramachandran plots. The Ramachandran Z-score for modeled enzyme from strain D253 was found to be 0.027 (Fig 3a), which further confirms the accuracy of the modeled structure. Cefotaxime was docked into the modeled enzyme-structures. It was found that the strain (D253) having a higher cefotaxime MIC displayed a higher negative binding energy as well (Table 1). Analysis of the docked structures by Discovery Studio2.0 revealed that the amino acid residues Asn132, Glu166, Pro167, Val172, Lys234 and Thr235 make important contacts with cefotaxime (Fig 3b).

In the univariate analysis, many study variables including mean age, mean ulcer duration, smoking, Wagner grade and 'glycemic control at discharge' (all p = 0.00) were found to be significant for "sex-associated MDRGNB infection status" of the foot ulcers. Hence, we went for multivariate analysis. The multiple logistic regression model confirmed 'glycemic control at discharge' to be the single independent risk factor for the same. (p = 0.001). ORs and 95% CIs can be seen in Additional file 1. Hence our main finding is that, male diabetic patients having MDRGNB-infected foot ulcers have poor glycemic control during the hospital stay.

The accession numbers of the E. coli genes sequenced in this study are [GenBank: FJ997866], [GenBank: FJ997867] and [GenBank: GQ145220].