Inactivation of genes in oxidative respiration and iron acquisition pathways in pediatric clinical isolates of small-colony variant Enterobactericeae

Isolation of bacterial small colony variants (SCVs) from clinical specimens is not uncommon and can fundamentally change the outcome of the associated infections. Bacterial SCVs often emerge with their normal colony phenotype (NCV) co-isolates in the same sample. The genetic and biochemical basis of SCV emergence in vivo is not well understood in Gram-negative bacteria. In this study, we interrogated the causal genetic lesions of SCV growth in three pairs of NCV and SCV co-isolates of Escherichia coli, Citrobacter freundii, and Enterobacter hormaechei. We confirmed the isogenic basis of SCV emergence, as there were only 4 single nucleotide variants in SCV for E. coli, 5 in C. freundii, and 8 in E. hormaechei, with respect to their NCV co-isolate. In addition, a 10.2kb chromosomal segment containing 11 genes was deleted in the E. hormaechei SCV isolate. Intriguingly, each SCV had at least one coding change in a gene associated with bacterial oxidative respiration and another involved iron capture. Chemical rescue confirmed the causal role of heme biosynthesis in E. coli and C. freundii and lipoic acid in E. hormaechei SCV isolates. Genetic rescue restored normal growth under aerobic conditions for fes and hemL in C. freundii; hemL in E. coli; and lipA in E. hormaechei SCV isolates. Prototrophic growth in all 3 SCV Enterobacteriaceae species was unaffected under anaerobic culture conditions in vitro, illustrating how SCVs may persist in vivo by abandoning the highly energetic lifestyle in an iron-limiting environment. We propose that the selective loss of functions in oxidative respiration and iron acquisition is indicative of bacterial virulence attenuation for niche specialization and persistence in vivo. Importance Small colony variant (SCV) bacteria are routinely isolated in the clinical microbiology laboratory and can be notoriously difficult to treat. Most studies of the genetic underpinnings of SCV clinical isolates have examined Staphylococcus aureus and few have looked at how SCV emerge in Gram-negative bacteria. Here, we undertook detailed characterization of three clinical isolates of SCV in Escherichia coli, Citrobacter freundii, and Enterobacter hormaechei along with their NCV co-isolates. Genomic sequencing revealed that each SCV had at least one coding change in genes involved in both bacterial oxidative respiration and iron capture. Chemical and genetic rescue revealed that both pathways could be responsible for the small colony variant. Each of the SCV showed no growth defect compared to NCV when incubated under anaerobic conditions, indicating a potential mechanism for SCV survival in vivo. We hypothesize that by retreating to anaerobic environments and avoiding escalating iron competition with the host, SCV have adapted to live to see another day.


60
To survive in the hostile host environment, bacteria may take two separate paths. The 61 first and most commonly discussed is an arms race of iron competition and acquisition of 62 antimicrobial resistance genes and pathogenicity factors (1-3). The alternative path is to adapt 63 to persist through reductions in metabolic needs and an attenuated growth rate. The isolates 64 that display this alternative phenotype are known as small colony variants (SCVs). Bacterial 65 SCVs were first described in Salmonella typhi over a hundred years ago, prior to the antibiotic 66 era (4). Isolation of SCVs is especially common in recurrent or persistent infections involving 67 the respiratory tract, urinary tract, mid-ear, foreign body-related implants, and bone and joint (5-

88
However, the basis of SCV formation and persistence in clinical isolates of Gram-negative 89 bacteria has been less well characterized (10, 24).

90
Laboratory recognition, isolation, characterization and appropriate report of bacterial 91 SCVs have suffered from a lack of established standards or guidelines. We previously reported 92 a method practical in clinical laboratories for recognition and phenotypic characterization of SCV 93 S. maltophilia from airway secretions of CF patients (8). Our lab has since implemented a 94 systematic, culture-based approach that checks not only for colony variation in size, texture, 95 color, or hemolysis, but also inability to grow on the standard Mueller-Hinton (MH) medium for 96 susceptibility testing. Using this systematic approach, we identified 3 pairs of clinical NCV and 97 SCV Enterobacteriaceae co-isolates from blood and urine cultures. We then used whole 98 genome sequencing to screen for the molecular mechanisms distinguishing the SCVs from their 99 NCV co-isolates. Confirmatory chemical and genetic rescues were performed on the SCVs to 100 determine which mutations were causal for the altered phenotype.

103
Isolation and characterization of SCV

108
Bruker Daltonics, Inc.). Of note, the SCVs isolates described here did not grow on Mueller-

112
Case 1 -A previously healthy 6-week-old male with no history of hospitalization or receipt of 113 antibiotics, presented to the emergency department with fever. His white blood cell count was 114 elevated at 21,600/ml and a 2+ urine leukocyte esterase at the time of emergency visit. Given 115 the patient's age, the patient was admitted for a rule out sepsis workup and the patient was 116 started on empiric ceftriaxone. Urine cultures grew 10 3 -10 4 cfu/mL Escherichia coli and 10 3 -117 10 4 cfu/mL SCV Escherichia coli. It was felt the urine culture did not support the diagnosis of 118 UTI, and no additional antibiotics were indicated. The patient recovered fully.

120
Case 2 -A 2-month-old female with right duplicated collection system presented to the 121 emergency department with fever and foul smelling urine. The patient had experienced two 122 urinary tract infections (UTIs) in the previous month (Escherichia coli and Citrobacter spp.) that 123 were treated with amoxicillin and cephalexin, respectively. Given that the patient was currently 124 receiving antibiotics for the previous Gram-negative UTI, the decision was made to admit the 125 patient for likely IV antibiotic treatment, and the patient was started on empiric piperacillin-126 tazobactam. Her urine leukocyte esterase was 3+ with elevated red and white blood cells in 127 urine at the time of culture. Urine cultures grew >10 5 cfu/mL Citrobacter spp. and 5•10 4 -10 5 128 cfu/mL SCV Citrobacter spp. and therapy was switched to ciprofloxacin. The patient completed patient received ceftazidime line-lock therapy. The patient completed treatment and recovered 138 fully. Though the isolates were resulted out as E. cloacae, they were later identified as E.       All assemblies are available from NCBI BioProject PRJNA523376

215
Clinical cases and isolates 216 Case histories are depicted in Figure 1 and described in the Materials and Methods.

217
Briefly, the paired E. coli isolates were from a urine culture on a 6-week old otherwise healthy 218 term male infant with fever. His white blood cell count was elevated at 21,600/ml and a 2+ urine

228
Antimicrobial resistance pattern is explained by ampC 229 Co-isolation of both NCV and SCV of the Enterobacteriaceae strains were common to all 230 3 cultures during routine culture workups. The antibiotic susceptibility pattern for the three NCV 231 isolates is shown in Table 1 and followed expected patterns of resistance given the case 232 histories. All three corresponding SCV isolates failed to grow on MH medium for susceptibility

251
Sequencing the Citrobacter freundii NCV strain yielded an assembly of 220 contigs >200 252 bp with an N50 of 61,653 bp. Mapping of the Citrobacter freundii SCV reads to the NCV 253 assembly yielded a total of 6 variants ( Table 2). Two of these variants were related to heme-254 producing genes. Most notable among these were a 1 bp deletion in the glutamate-1-

307
Over the course of the rescue experiments, the C. freundii SCV also reverted to normal 308 growth. This clone had NCV-like fes and hemL sequence, while the same SCV coding 309 mutations were seen in araC, fdnG, and pta, as well as the intergenic mutation upstream of   while PQQ, heme and its biosynthetic intermediates L-glutamate and ALA failed to increase 320 SCV growth (Figures S3d-g). These results conclusively demonstrate that disruption of the 321 lipoylation pathway was responsible for the small growth phenotype in our E. hormaechei SCV 322 isolate. We did not observe reversion to normal growth for the E. hormaechei SCV.                        Figure S1 -Chemical rescue of Escherichia coli SCV growth was successful with δ-aminolevulinic acid (a), but not with L-glutamate (b), lipoic acid (c), or pyrroloquinoline quinone (d).  Figure S2 -Chemical rescue of Citrobacter freundii SCV growth was successful with δ-aminolevulinic acid (a), but not with L-glutamate (b), lipoic acid (c), or pyrroloquinoline quinone (d). h) i) Figure S3 -Failure to rescue Enterobacter hormaechei SCV growth with deleted and mutated genes crcB, cspE, cusA, cusB, cusF, cysD, hyfB, rnr, tatE, tolC, and udk under aerobic conditions (a-c). L-glutamate (d), δ-aminolevulinic acid (e), heme (f), and pyrroloquinoline quinone (g) failed to rescue SCV growth under aerobic conditions. SCV grew similar to NCV under all anaerobic growth conditions for each of the genetic rescues attempted (h-j).