Strains used in this study are described in Table 1. NTHi strains were grown at 37°C, 5% CO₂ on chocolate agar (Biomérieux), MH-F agar (Biomérieux), or brain-heart infusion (BHI) agar supplemented with 10 μg/ml hemin and 10 μg/ml nicotinamide adenine dinucleotide (NAD), referred to as sBHI. NTHi liquid cultures were grown in sBHI (37°C, 5% CO₂). Thymidine (Sigma-Aldrich) was dissolved in sterile distilled water (stock solution, 10 mg/ml). When necessary, media were supplemented with thymidine by (i) using sterile paper discs soaked on thymidine 300 μg/ml or 10 mg/ml, (ii) thymidine 300 μg/ml spreading on 20 ml chocolate agar or MH-F agar plates, (iii) thymidine 300 μg/ml addition into sBHI. Erythromycin 11 μg/ml (Erm₁₁) or spectinomycin 50 μg/ml (Spec₅₀) were used when required. Escherichia coli was grown on Luria Bertani (LB) agar at 37°C, supplemented with ampicillin 100 μg/ml (Amp₁₀₀), erythromycin 150 μg/ml (Erm₁₅₀) or Spec₅₀, when necessary.

For thyA disruption, a DNA fragment containing the thyA gene and its respective adjacent regions (2,852 bp), was PCR amplified with Phusion polymerase (Thermofisher) using NTHi375 genomic DNA as template and primers thyA-F1 (5′-TGCCTGAATATTCGCTCGGTTACATTTA) and thyA-R1 (5′-CGATGATACTTAAAAGTAATCGCGACCAAAAATTCGG). The gene-containing fragment was cloned into pJET1.2 (Thermofisher), generating pJET1.2-thyA. This cloned PCR product was disrupted by inverse PCR with Phusion polymerase, using primers thyA-F2 (5′-CGTTCCTGTGATGTTCCGCTTGGA) and thyA-R2 (5′-ATATCATCAACGCCTCTACGATGC). An internal 226-bp fragment (nucleotides 312–537 in the thyA coding sequence) was replaced by a blunt-ended erythromycin resistance cassette excised by SmaI digestion from pBSLerm (Allen et al., 2005), generating pJET1.2-thyA::ermC. This plasmid was used as a template to amplify the thyA::ermC disruption cassette with primers thyA-F1 and thyA-R1, which was used to transform NTHi375 using the MIV method (Herriott et al., 1970). Transformants were screened by plating bacteria on sBHI agar with Erm₁₁ to obtain NTHi375ΔthyA. Same approach and disruption cassette were used to generate RdKW20ΔthyA. For mutant confirmation, NTHi375ΔthyA and RdKW20ΔthyA genomic DNA were used as template to be PCR amplified with four primer pairs: (i) thyA-F1 and thyA-R1, rendering a 3,814 bp product; (ii) thyA-F1 and pBSLerm-down (5′-GGTACACGAAAAACAAGTTAAGGG), rendering a 2,421 bp product; (iii) pBSLerm-up (5′-ATAAAGAGGGTTATAATGAACGAG) and thyA-R1, rendering a 2,190 bp product; (iv) pBSLerm-up and pBSLerm-down, rendering a 797 bp product (data not shown). Recombination events resulting from integration of the thyA disruption cassette in the NTHi375 and RdKW20 genomes were further verified by NTHi375ΔthyA and RdKW20ΔthyA genomic DNA PCR amplification with primers thyA-F1 and thyA-R1, and PCR product sequencing with primers thyA-F1, pBSLerm-up, pBSLerm-down and thyA-R1. When necessary, the thyA gene was PCR amplified with primers ThyA_Pro_F (5′-TGCGCCTTTGATTCCGTTTG) and ThyA_R2 (5′-TCACCTAACGCTTCCGCTTT) for DNA sequencing.

The nupC gene and its respective adjacent regions (2,664 bp) was amplified by PCR with Phusion polymerase using RdKW20 genomic DNA as template and primers nupC-F1 (5′-ATGAACAGGTTATGGAGGCAGTTCCAT) and nupC-R1 (5′-GTGAGTACGAATATGGTCAGACACGGT). The gene-containing fragment was cloned into pJET1.2, generating pJET1.2-nupC. A Spec^r cassette was PCR amplified from pRSM2832 using gene-specific mutagenic primers nupC-F2 (5′-GTAGAATAAGCCGAATTTTATTAACTTAACTAATCTAGGGGAATCAAATGATTCCGGGGATCCGTCGACC) and nupC-R2 (5′-TACCGCACTTTTTAATTGATTAAGATAAATTAAAGTGCTGCAGCACCTAAGCCTGTAGGCTGGAGCTGCTTCG), as described previously (Tracy et al., 2008). Primers were designed to delete sequences between the start codon and the last seven codons of nupC. E. coli SW102 cells were prepared for recombineering, co-electroporated with pJET1.2-nupC (Amp^r; 50 ng) and the nupC-specific mutagenic cassette (Spec^r; 200 ng) as previously described (Sinha et al., 2012), and mutagenized clones containing pJET1.2-nupC::spec were selected on LB agar with Amp₁₀₀, Spec₅₀. This plasmid was used as a template to amplify the nupC::spec disruption cassette with primers nupC-F1 and nupC-R1, which was used to transform NTHi375 and RdKW20 using the MIV method. Transformants were selected on sBHI agar with Spec₅₀, to obtain NTHi375ΔnupC and RdKW20ΔnupC mutant strains.

Laboratory records of Microbiology Department at the Bellvitge University Hospital (Spain) regarding the difficulties of some NTHi isolates to growth on conventional antimicrobial susceptibility testing medium (Mueller-Hinton agar+5% defibrinated horse blood and 20 mg/l β-NAD, MH-F) were reviewed from 2010 to 2014. Screening for thymidine auxotrophs among NTHi clinical isolates was based on the interpretation of their growth characteristics. Isolates identified as having growth problems were further screened for thymidine auxotrophy by comparative growth on chocolate agar plates and low-thymidine MH-F agar plates. Thymidine auxotrophy was confirmed by testing strain growth on MH-F agar in the absence or presence of discs soaked with thymidine.

TxS susceptibility testing of NTHi thymidine auxotrophs failed to produce results by disc diffusion (Becton Dickinson), E-test (Biomérieux) or broth microdilution when inoculated on MH-F and incubated for 24 h, as specified by the European Committee on Antimicrobial Susceptibility Testing (EUCAST; http://www.eucast.org/clinical_breakpoints). Given that chocolate agar supported growth for all strains after 24 h of incubation, this medium was used for E-test- or TxS discs-based determination of minimal inhibitory concentrations (MIC). Strains were grown on chocolate agar, to generate bacterial suspensions normalized in phosphate-buffered saline (PBS) to OD₆₀₀ = 1. Normalized suspensions were spread on chocolate agar or, when necessary, on MH-F agar, in the presence of E-test or TxS discs (23.75 mg/1.25 mg SMX:TMP) and incubated for 24 h before assessing the diameter of the growth inhibition zones.

To monitor growth, NTHi strains grown on chocolate agar for 16 h were inoculated (2–5 colonies) in 20 ml sBHI, with or without thymidine, and incubated for 11 h with shaking. Cultures were diluted in 40 ml sBHI, with or without thymidine, to OD₆₀₀ = 0.01 (RdKW20) or OD₆₀₀ = 0.05 (NTHi375), incubated with agitation, and OD₆₀₀ was recorded every hour for 8 h. Every 2 h, culture samples were serially diluted and plated on sBHI agar. Data are shown both as OD₆₀₀ and c.f.u./ml. At the final time point, 10 μl of the bacterial cultures were placed on glass coverslips and fixed with 3.7% paraformaldehyde (PFA) in PBS pH 7.4 for 15 min at room temperature. Bacteria were labeled with a polyclonal rabbit anti-NTHi primary antibody diluted 1:600 and a donkey anti-rabbit conjugated to Cy2 (Jackson Immunological) secondary antibody diluted 1:100. Samples were analyzed with a Carl Zeiss Axioskop 2 plus fluorescence microscope and a Carl Zeiss Axio Cam MRm monochrome camera.

NTHi strains were grown on chocolate agar for 16 h, in the absence or presence of thymidine, and a colony was aseptically spread with an inoculation loop over a drop of distilled water on a microscopy slide. Samples were air-dried and stained for 15 min in the darkness with the cell-permeable fluorescent nucleic acid stain SYTO 9 (Life Technologies), following the manufacturer's instructions. Samples were washed twice with distilled water and fluorescence was observed by confocal laser microscopy. Images were acquired using a Leica TCS-SL filter-free spectral confocal laser-scanning microscope (Leica Microsystems) equipped with a 488 nm argon laser, 543 nm and 633 nm He/Ne lasers (Centres Científics i Tecnològics-Campus de Bellvitge, Universitat de Barcelona, Spain) using a 63× magnification oil immersion objective (1.4 numerical aperture), and an image resolution of 1024 × 1024 pixels. Images were acquired randomly and analyzed using the Leica Confocal Software 2.5 (Leica Microsystems).

H. influenzae strains were examined by TEM after growth on chocolate agar following established procedures (Remuzgo-Martinez et al., 2015). Briefly, bacteria were applied to Formvar-coated grids, air dried, negatively stained with 1% phosphotungstic acid in distilled water for 10 s, and examined with a JEM-1011 transmission electron microscope (JEOL) operating at 80 kV and equipped with an Orius SC1000 charge-coupled device (CCD) camera (Gatan).

NTHi strains were grown for 16 h on chocolate agar. Bacteria (2–5 colonies) were inoculated into 20 ml sBHI, grown for 11 h, with or without thymidine, diluted into 40 ml fresh sBHI to OD₆₀₀ = 0.05, in the absence or presence of thymidine, and grown to OD₆₀₀ = 0.6. Bacterial total RNA was isolated using TRIzol reagent (Invitrogen). Total RNA quality was evaluated using RNA 6000 Nano LabChips (Agilent 2100 Bioanalyzer). All samples had intact 16S and 23S ribosomal RNA. Complementary DNA (cDNA) was synthesized from total RNA (1 μg) using SuperScript II Reverse Transcriptase reagents (Invitrogen). Real-time quantitative PCR was performed using Thermo Scientific Luminaris HiGreen qPCR Master Mix (Thermo Scientific) and fluorescence data were analyzed with BioRad CFX96 qPCR System (Bio-Rad). Relative quantities of mRNAs were calculated using the comparative threshold cycle (Ct) method and normalized using 16S ribosomal RNA (16SrRNA) as an endogenous control. Primer pairs were designed with Primer3 software: for nupC, nupC-RT-Fw (5′-ATGTTAATCGCATTCGTTGGTT) and nupC-RT-Rv (5′-ATTTGTCCTGCAATACCTGCTT); for tdk, tdk-RT-Fw (5′-TGTGCTTTGTTATGGCTTGC) and tdk-RT-Rv (5′-CTTTAATGACCTCGCCCTGA); for 16SrRNA, 16S-Fw (5′-GGCGTTGATGACCGTGAAAC) and 16S-Rv (5′-GCCAGTAATAATCGCCCTCTTCTAG). Data are expressed as relative expression on mutant strains compared to their parental wild-type strains, considered to be 1. All measures were carried out in duplicate and at least three times (n ≥ 6).

Three to four colonies of NTHi grown on chocolate agar for 16 h were inoculated into 20 ml sBHI, grown for 11 h, diluted in sBHI to OD₆₀₀ = 1 and left standing at room temperature for 5 h (starting volume ~25 ml). The viability of each culture was tested by serial dilution and plating on sBHI agar at the beginning of each experiment (t = 0; data not shown). OD₆₀₀ readings were performed at regular time intervals on 700 μl aliquots collected from the top of each bacterial suspension. At least four independent experiments (n ≥ 4) were performed for each strain.

Three to four colonies of NTHi grown on chocolate agar for 16 h were inoculated into 20 ml sBHI, grown for 11 h, diluted into 40 ml sBHI to OD₆₀₀ = 0.05, grown to OD₆₀₀ = 0.6 in the absence or presence of thymidine, serially diluted, plated on sBHI agar for c.f.u. determination, and used to generate stocks stored at −80°C in sBHI with 20% glycerol as single use aliquots for further experiments. For PCho determination, ~1 × 10⁷ c.f.u. were incubated for 1 h at 37°C with TEPC-15, a mouse monoclonal antibody specific for PCho (Sigma-Aldrich) diluted 1:25 in PBS-0.05% Tween 20. Samples were washed twice with PBS-0.05% Tween 20, and incubated with a fluorescein isothiocyanate (FITC)-conjugated rabbit anti-mouse (Serotec) diluted 1:300 in PBS-0.05% Tween 20 for 30 min at 4°C under dark conditions. Bacteria were washed with PBS-0.05% Tween 20, fixed in 3% paraformaldehyde (PFA) for 2–3 min at room temperature, and analyzed on a FACSCalibur flow cytometer (BD Biosciences) using forward and side scatter parameters to gate on at least 25,000 bacteria. Results are expressed as a relative percent fluorescence index (RFI), to measure both the proportion of fluorescent bacteria positive for PCho and the intensity of fluorescence (Ramos-Sevillano et al., 2015). Assays were performed in quadruplicate in at least three independent occassions (n ≥ 12).

Bacterial strains were grown as indicated for PCho determination. C3b deposition was analyzed as explained previously (Ramos-Sevillano et al., 2015). Briefly, a bacterial suspension containing ~1 × 10⁷ c.f.u. was opsonized with human serum diluted 1:4 in PBS-0.05% Tween 20, and detected with a FITC-conjugated polyclonal goat anti-human C3b antibody (ICN-Cappel) diluted 1:300 in PBS-0.05% Tween 20 for 30 min at 4°C under dark conditions. CRP binding was measured as previously described (Ramos-Sevillano et al., 2015), by incubating ~1 × 10⁷ c.f.u. with human serum diluted 1:4 in PBS-0.05% Tween 20 and detected with a polyclonal rabbit anti-human CRP antibody (Calbiochem) for 1 h at 37°C, followed by two washes with PBS-0.05% Tween 20. Bacterial suspensions were then incubated for 30 min with a FITC-conjugated polyclonal goat anti-rabbit antibody in PBS-0.05% Tween 20 for 1 h at 37°C. For both C3b and CRP binding, bacteria were finally washed with PBS-0.05% Tween 20, fixed in 3% PFA, and analyzed on a FACSCalibur flow cytometer as described above. Results are expressed as a RFI. Assays were performed in quadruplicate in at least three independent occassions (n ≥ 12).

A549 human alveolar basal epithelial cells (ATCC CCL-185) were maintained as described (Morey et al., 2011), seeded to 6 × 10⁴ cells/well for 32 h, and serum-starved 16 h before infection. NCI H-292 mucoepidermoid pulmonary human carcinoma epithelial cells (ATCC CRL-1848) were maintained as described (Euba et al., 2015c), and seeded to 4 × 10⁵ cells/well 16 h before infection. Adhesion and invasion assays were performed and processed as described (Morey et al., 2011; Lopez-Gomez et al., 2012; Euba et al., 2015a,c). For infection, PBS-normalized bacterial suspensions (OD₆₀₀ = 1) were prepared by using NTHi strains grown on chocolate agar for 16 h, or in 20 ml sBHI for 11 h in the absence or presence of thymidine. A multiplicity of infection (MOI) of ~100:1 was used. To monitor adhesion, cells were infected for 30 min. Although this assay does not completely exclude a possible internalization of some bacteria, experimental conditions were previously set to mainly monitor adhesion (Morey et al., 2011; data not shown). For invasion assays, cells were incubated with bacteria for 2 h, washed three times with PBS, incubated for 1 h with RPMI 1640 medium containing 10% FCS, Hepes 10 mM, and gentamicin 200 μg/ml to kill extracellular bacteria.

When necessary, A549 cells were infected for 30 min with bacterial suspensions (OD₆₀₀ = 1) generated with bacteria grown on chocolate agar, in the absence or presence of thymidine; infections were performed in RPMI 1640, and in the absence or presence of purified C3 (16 μg/ml), 2% C3-deficient human serum (HS), or C3-depleted HS reconstituted with purified C3. Human C3-deficient serum and human complement C3 were purchased from Sigma-Aldrich. Cells were then washed three times with PBS, lysed with 300 μl of PBS-saponin 0.025% for 10 min at room temperature, and serial dilutions were plated on sBHI agar. All infections were performed in triplicate at least three independent times (n ≥ 9). Results are expressed as c.f.u./ml.

For A549 cell stimulation by NTHi, bacteria grown on chocolate agar, in the absence or presence of thymidine, were collected with PBS, suspensions were normalized to OD₆₀₀ = 1, and used for a 2 h infection with a MOI of ~100:1. Cells were washed three times with PBS, and incubated for 6 h in RPMI 1640 medium containing 10% FCS, Hepes 10 mM and gentamicin 100 μg/ml. Supernatants were collected from the wells, cell debris removed by centrifugation and samples frozen at –80°C. IL-8 levels in the supernatants were measured by ELISA (Abnova KA0115) with sensitivity <2 pg/ml. Infections were performed in duplicate and at least twice (n ≥ 4). Results are expressed as IL-8 pg/ml.

A549 cells were seeded on glass coverslips and infected as described above, by using bacteria previously grown on chocolate agar. Coverslips were fixed in ice-cold 3% glutaraldehyde for 20 min at 4°C. Samples were dehydrated in a series of graded acetone, dried by the critical point method, coated with gold in a Fine coat ion sputter JFC-1100 226 (JEOL, Ltd) and observed with an Inspect S microscope (FEI Company) working at 15 or 20 kV (Lazaro-Diez et al., 2016).

A CD1 mouse model of NTHi lung infection was used, as described previously (Morey et al., 2013; Euba et al., 2015a,b,c). CD1 female mice (18–20 g) aged 4–5 weeks were purchased from Charles River Laboratories (France), housed under pathogen-free conditions at the Institute of Agrobiotechnology facilities (registration number ES/31-2016-000002-CR-SU-US), and used at 22–25 g. Animal handling and procedures were in accordance with the current European (Directive 86/609/EEC) and National (Real Decreto 53/2013) legislations, following the FELASA and ARRIVE guidelines, and with the approval of the Universidad Pública de Navarra (UPNa) Animal Experimentation Committee (Comité de Ética, Experimentación Animal y Bioseguridad) and the local Government authorization. NTHi375 and NTHi375ΔthyA were used for lung infection, and mice were randomly divided into two groups (n = 5): (i) control, vehicle solution (0.1 ml PBS) administered by oroesophageal gavage (Popper & Sons Inc.); (ii) one dose of TxS each 6 h administered by oroesophageal gavage, starting at 6 h post-infection (hpi). TxS (Septrin, 8 mg/40 mg/ml TMP:SMX) treatment was performed at a dose of 960 mg/kg of body weight in 0.1 ml PBS. Infecting bacteria were previously grown on chocolate agar, in the absence or presence of thymidine. For NTHi intranasal infection, 40 μl of a NTHi suspension containing ~5 × 10⁸ c.f.u./ml (~2 × 10⁷ c.f.u./mouse) were placed at the entrance of the nostrils until complete inhalation, in mice previously anesthetized with ketamine-xylazine (3:1). At 12, 24, or 48 hpi, mice were euthanized using cervical dislocation. BALF samples were obtained by perfusion and collection of 0.7 ml of PBS, with help of a sterile 20G (1.1 mm diameter) Vialon™ intravenous catheter (Becton-Dickinson) inserted into the trachea. An aliquot of each recovered BALF was serially 10-fold diluted in PBS, and plated on sBHI agar to determine the number of viable bacteria. Results are expressed as mean ± SD of individual log₁₀ c.f.u./BALF. In parallel, lungs were removed; the left one was processed for viable bacterial counts (as detailed above), and the right lung was fixed in 10% neutral buffered formalin for histological purposes. Heads and necks containing upper airways, larynx, and tracheas were fixed in the same buffered formalin for histology. Uninfected mice receiving PBS or TxS were used as controls when necessary.

For cell counting, the remaining volume of each BALF sample was centrifuged at 5,000 r.p.m. for 3 min at 4°C. Each pellet was resuspended in 1 ml RPMI 1640 with 10% FCS and Hepes 10 mM, and total cell count determined using a hemocytometer. ~5x10⁴ cells in 200 μl RPMI 1640 with 10% FCS and Hepes 10 mM were used for cytospin preparation (1,500 r.p.m. for 10 min at room temperature, Thermo Shandon Cytospin). Giemsa stains were performed with an automated hematology slide preparation unit (SP-10, Sysmex Cosporation) according to the manufacturer's instructions. Preparations were examined in a double-blinded manner with an optical microscope (BX, Olympus).

Heads and necks were rinsed in running tap water for 1 h, immersed in 5% nitric acid for 24–36 h until complete decalcification, and 7–8 transaxial slices were made every 3–4 mm beginning at the nostrils and finishing in the caudal tracheas. Transaxial slices and lungs were embedded in paraffin, and 4–6 μm sections were stained with hematoxylin and eosin (H&E) by standard procedures, and examined by microscopy to determine the presence and extent of inflammatory lesions. Sections were examined blind as sets by a trained veterinary pathologist (Dr. M. Barberán). Parameters characterizing an acute inflammatory reaction in upper airways, larynx, trachea, and lung, including hemorrhages, hyperemia, polymorphonuclear cell infiltrates (PMNs), and alveolar macrophages, were subjectively scored on a scale of 0–3 (0: absent, 1: mild, 2: moderate, 3: severe). For tissue control, similar organs obtained from non-infected control and TxS treated mice were processed in an identical manner to the infected tissues.

For self-aggregation, gene expression, diameter of growth or inhibition zones, PCho level, complement deposition, cell infection, bacterial loads in lungs and BALF samples, and inflammatory cells in BALF samples, mean ± SD were calculated and statistical comparison of means performed using the two-tail t-test. For histopathology scoring, means ± SD were also calculated and statistical comparisons performed using one-way analysis of variance (ANOVA) followed by Fisher's protected least significant difference (PLSD) multiple-comparison test. In all cases, a p < 0.05 value was considered statistically significant. Analyzes were performed using Prism software, version 6 for Mac (GraphPad Software) statistical package.

Among 2,542 clinical NTHi isolated between 2010 and 2014 at Bellvitge University Hospital (Spain), 119 (4.7%) isolates formed slow-growing colonies on MH-F agar. Thymidine auxotrophy was screened on the 119 isolates by comparative growth on chocolate agar and on low-thymidine MH-F agar. Thymidine auxotrophy was confirmed by absence of bacterial growth on MH-F agar, and further growth on MH-F agar in the presence of discs soaked in thymidine. One thymidine auxotroph NTHi isolate, named strain 8233, was identified in a sputum sample recovered from a COPD patient who had received TxS for 36 days to treat an acute exacerbation by Stenotrophomonas maltophilia. NTHi8233 produced small colonies on chocolate agar (data not shown) and did not grow on MH-F agar, except when discs soaked with thymidine 300 μg/ml or 10 mg/ml were added to the plate. Bacterial growth showed to be dependent on thymidine concentration (Figure 1A). The thymidylate synthase encoding gene thyA_NTHi8233 was sequenced, and a 6 nt duplication 267GAAAAT rendering a two-amino acid insertion (E90N91) was found, when compared to that of the RdKW20 genome sequenced reference strain (Fleischmann et al., 1995). A prediction of ThyA_{H. influenzae} tertiary structure, based on Burkholderia thailandensis ThyA protein (PDB: 3V8H), was generated by the automated modeling tool of the Swiss Model web service (http://swissmodel.expasy.org). ThyA_NTHi8233 displayed 11 α-helices, 6 β-sheet strands and several coil connecting segments (Figure S1A). Homology between ThyA_NTHi8233 and ThyA_RdKW20 predicted structures is shown by structural alignment (Figure S1D).

NTHi8233 morphology was analyzed by confocal microscopy, and showed non-septated bacterial filaments ~5–10 times longer than the average NTHi bacteria (Figure 1A, right). The absence of NTHi8233 growth on MH-F agar did not allow assessing its antibiotic susceptibility following standard procedures (http://www.eucast.org/clinical_breakpoints), and TxS resistance was tested by E-test on bacteria grown on chocolate agar. The MIC of TxS for NTHi8233 was >32 μg/ml. NTHi8233 thymidine auxotrophy was found to be easily reversible, as described for other pathogens (Kahl, 2014). Thymidine prototroph reverted NTHi8233 strain did not present a filamented morphology (data not shown) and the sequence of the thyA gene did not present the 6 nt duplication described above, even though its MIC of TxS remained unchanged.

Identification of NTHi8233 supports the notion that thymidine auxotrophy may be a bacterial response to TxS treatment that could be underestimated due to the lack of growth on MH-F agar and given that, as previously stated for other NTHi isolates (Platt et al., 1983), was shown to be reversible. This observation prompted us to generate mutant strains lacking the thyA gene on previously characterized NTHi genetic backgrounds for a detailed study of the impact of thymidine auxotrophy for this pathogen.

H. influenzae genome sequenced strains NTHi375 and RdKW20 were employed to generate thymidine dependent mutants by disruption of the thyA gene (thyA accession numbers NF38_0045 and HI0905, respectively). At the protein level, ThyA_NTHi375 and ThyA_RdKW20 displayed 98.2% identity (Figure S1E). NTHi mutants lacking the thyA gene were selected on sBHI agar and rendered normal size colonies (data not shown), did not grow on MH-F agar, and grew around discs soaked with thymidine on MH-F agar. Growth was dependent on the thymidine concentration present on the discs (Figure 1B and Table 2). Thus, the thyA mutants displayed a larger growth zone around discs soaked with thymidine 10 mg/ml than around discs soaked with thymidine 300 μg/ml (NTHi375ΔthyA, p < 0.005; RdKW20ΔthyA, p < 0.05). In contrast, wild-type (WT) bacteria presented a normal growth on MH-F agar. For NTHi375 WT and ΔthyA strains, the MICs of TxS were 0.25 and 0.38–0.5 μg/ml, respectively. For RdKW20 WT and thyA mutant strains, the MICs of TxS were 0.12 and 0.5–0.75 μg/ml, respectively. For purpose of illustration, larger growth inhibition zones were observed for WT than for thyA mutant strains on chocolate agar around TxS discs (Figure 1B). The morphology of NTHi375ΔthyA and RdKW20ΔthyA strains was assessed by confocal microscopy, showing non-septated filaments longer than their respective WT bacteria. Bacterial median length, measured by TEM in ~70 bacteria per strain, was (i) 1.42 ± 0.56 and 3.78 ± 2.86 μm for NTHi375 WT and ΔthyA strains, respectively (p < 0.0001); (ii) 1.43 ± 0.24 and 4.6 ± 4.98 μm for RdKW20 WT and ΔthyA strains, respectively (p < 0.0001). In agreement, SEM showed enlarged thymidine dependent bacteria on the surface of infected A549 human airway epithelial cells, which may modify the infectious process, compared to that shown by the WT strains (Figure 1B).

NTHi self-aggregates, which may promote microcolony formation on host cell surfaces (Meng et al., 2011; Mell et al., 2016). We asked whether the observed thymidine auxotrophy-driven increased bacterial size differs in its self-aggregation, by using tube-settling assays and monitoring the optical density of bacterial suspensions over time. Thymidine dependent mutants self-aggregated faster than their respective WT strains (Figure 1C).

Together, inactivation of the thyA gene in NTHi (i) impaired bacterial growth on MH-F agar, compensated by addition of external thymidine, (ii) triggered enlarged bacilli which self-aggregate faster than their isogenic WT strains, (iii) increased resistance to TxS.

Growth in sBHI was analyzed for WT and thyA mutant strains. NTHi375ΔthyA had an extended lag phase and lower final OD₆₀₀ than the WT strain, which correlated with a reduced viability, measured as c.f.u./ml at the indicated time points (Figure 2A). Although not as pronounced, RdKW20ΔthyA also had a slightly extended lag phase and lower numbers of viable bacteria than those of the WT strain (Figure 2B). In both thyA mutants, the observed growth defects were partially restored by sBHI supplementation with thymidine 300 μg/ml, independently of its addition in the pre-culture used for further dilution in sBHI and OD₆₀₀ recording, or in the actual sBHI culture used for growth monitoring over time (Figures 2A,B). Immunofluorescence microscopy at the final time point of the growth curve showed non-septated long bacterial filaments for NTHi375ΔthyA grown in sBHI, but a mixture of non-septated filaments and bacteria with the average NTHi WT size for NTHi375ΔthyA pre-grown in sBHI supplemented with thymidine 300 μg/ml (Figure 2A, right). This could relate to the observed restoration of final OD₆₀₀ but not of final bacterial counts for NTHi375ΔthyA when pre-cultured in sBHI supplemented with thymidine.

The thyA mutants, previously grown on chocolate agar with thymidine (chocolate+Thy), grew on MH-F agar around discs in a thymidine dependent manner, displaying a larger growth zone around discs soaked with thymidine 10 mg/ml than with thymidine 300 μg/ml (NTHi375ΔthyA, p < 0.005; RdKW20ΔthyA, p < 0.005). Of note, thyA mutants previously grown on chocolate+Thy rendered a better growth on MH-F agar around discs soaked in thymidine than that of the same strains previously grown on chocolate agar (for thymidine 300 μg/ml, NTHi375ΔthyA, p < 0.05; for thymidine 10 mg/ml, NTHi375ΔthyA, p < 0.05, RdKW20ΔthyA, p < 0.005; Figures 1B, 2 bottom and Table 2). Similarly, external thymidine (chocolate+Thy) rendered thyA bacterial length and morphology similar to those shown by their respective isogenic WT strains (Figures 1B, 2 bottom).

Both thyA mutants growth on MH-F agar around discs soaked in thymidine, and morphology/growth restoration in the presence of external thymidine, prompted us to speculate that these mutants may use external thymidine to overcome their nucleoside dependency. It has been previously shown that the nucleoside transporter encoding gene nupC is overexpressed in S. aureus thymidine dependent-SCVs, and that a ΔnupC mutant fails to use external thymidine for growth under TxS challenge (Chatterjee et al., 2008; Kriegeskorte et al., 2014). HI0519 and NF38_02480 are annotated as nupC in the RdKW20 and NTHi375 genomes, respectively (Fleischmann et al., 1995; Mell et al., 2014). Expression of the nupC gene in H. influenzae was assessed by RT-qPCR, showing a trend to be higher in the ΔthyA than in their isogenic WT strains grown in sBHI, with stronger evidence in NTHi375 (Figure S2A). As shown in Figure 2, sBHI supplementation with thymidine partially restored the growth defects shown by the thyA mutants. Following the notion that thymidine supplementation may restore mutant-related phenotypes, expression of the nupC gene was analyzed in ΔthyA bacteria when pre-cultured in sBHI with thymidine and then grown in sBHI, or when grown in sBHI supplemented with thymidine. Unexpectedly, when mutant bacteria were pre-cultured in sBHI supplemented with thymidine 300 μg/ml and then grown in sBHI, expression of the nupC gene was even higher than when pre-cultured and grown in sBHI (Figure S2A). Of note, a total absence of growth inhibition was observed for thyA mutant strains on chocolate agar around TxS discs, when pre-grown on chocolate+Thy (data not shown).

Next, NTHi375ΔnupC and RdKW20ΔnupC strains were generated, showing growth rates comparable to those of their respective WT strains in sBHI (Figure S2B), chocolate agar and MH-F agar (data not shown). To assess the biological function of nupC as a pyrimidine transporter responsible for the uptake of extracellular thymidine in H. influenzae, we tested the nupC mutants for their ability to use external thymidine for growth under TxS challenge in a TxS disc diffusion assay. Inhibition zones on chocolate agar and MH-F agar were comparable for WT and ΔnupC strains. Growth inhibition around TxS discs was also tested on chocolate agar and MH-F agar plates supplemented with thymidine. Unfortunately, no conclusive differences were observed because, as already described, bacterial growth on plates with high concentration of thymidine may produce haze or fine growth areas (Lorian, 1996). Following this notion, we performed a search for pyrimidine transporters in all available genome sequenced H. influenzae strains, which did not reveal the presence of nucleoside transporters additional to NupC in this bacterial species. We also unsuccessfully attempted to generate a NTHi double mutant strain lacking both the thyA and nupC genes (data not shown), further supporting a role for NupC in the uptake of external thymidine.

Moreover, thymidylate is synthesized either by the thymidylate synthase ThyA or by the thymidine kinase TK, and inactivation of the former has been shown to result in upregulated expression of the later one in Mycoplasma pneumoniae (Wang et al., 2010). HI0529 and NF38_02430 are annotated as a tdk thymidine kinase in the RdKW20 and NTHi375 genomes, respectively (Fleischmann et al., 1995; Mell et al., 2014). Different to nupC, the tdk gene expression was shown to be similar in WT and ΔthyA strains grown in sBHI, independently of thymidine addition (Figure S2C).

Together, these results suggest that nupC may be a transporter for external thymidine in H. influenzae, whose increased expression could contribute to bypass the effects of TxS upon de novo thymidylate biosynthesis.

We have previously reported that bacteria lacking PCho self-aggregate slightly faster (Morey et al., 2013). Thymidine auxotroph mutants showed faster self-aggregation compared to that of their WT strains (Figure 1C), which could relate to changes in cellular morphology (i.e., elongation) or, alternatively, to a different amount of PCho residues on the bacterial surface. The level of PCho was measured on NTHi375 and RdKW20 WT and thyA mutant strains by flow cytometry using bacteria grown in sBHI. Thymidine auxotrophy was associated with a decreased detection of PCho in both NTHi375ΔthyA and RdKW20ΔthyA mutants, which was restored in NTHi375ΔthyA by bacterial growth in sBHI in the presence of thymidine. Indeed, PCho detection was higher in the thymidine dependent mutants grown in sBHI supplemented with thymidine than in their isogenic WT strains (Figure 3A and Figure S3A). Decreased levels of PCho in the thyA mutants may modify the bacterial ability to bind C-reactive protein (CRP; Weiser et al., 1998). However, CRP deposition on the bacterial surface was similar in WT and thyA mutant strains (Figure 3B and Figure S3B). Finally, to assess the possibility that cellular morphology may affect complement interaction, C3b deposition was analyzed. Thymidine auxotrophy was associated with decreased detection of C3b in NTHi375ΔthyA, which was restored by bacterial growth in sBHI supplemented with thymidine (Figure 3C). No differences were observed for C3b deposition between RdKW20 strains (Figure S3C).

Together, inactivation of the thyA gene caused a reduction in the amount of PCho residues on H. influenzae surface, leading to impaired C3b deposition in at least NTHi375, which suggests that thymidine dependency may trigger changes affecting bacterial recognition by this key complement component by a CRP-independent mechanism.

The interplay of NTHi with the human respiratory epithelium plays a determinant role in the progression of infection (Clementi and Murphy, 2011). Next, we assessed if auxotrophy-related bacterial morphology changes could alter such interplay, by infecting A549 human type II pneumocytes with WT and thyA mutant strains (Morey et al., 2011; Lopez-Gomez et al., 2012; Euba et al., 2015c). Invasion by RdKW20 WT and thyA mutant strains was not assayed given the poor invasiveness of this genetic background (Mell et al., 2016). A549 cell adhesion and invasion by NTHi375ΔthyA was lower than that shown by the WT strain (Figure 4A). We also asked if external thymidine could restore such deficiency by using mutant bacteria grown in sBHI, in the absence or presence of thymidine (sBHI±Thy). A549 cell adhesion of NTHi375ΔthyA grown in sBHI was lower than that shown by the WT strain, which was restored by thymidine addition (Figure 4A, right). Similar results were obtained for RdKW20ΔthyA grown in sBHI, when compared to the WT strain, and to growth in sBHI with thymidine (Figure S4A). Given the impaired host cell interaction due to thymidine auxotrophy, we asked if inactivation of the thyA gene modifies the epithelial cell inflammatory response upon NTHi infection. Of note, the amount of secreted IL-8 was higher in A549 cells infected by NTHi375ΔthyA than by the WT strain, and such increase was reduced to WT levels upon infection by NTHi375ΔthyA previously grown on chocolate+Thy (Figure 4C). NTHi375ΔthyA lower epithelial infection rate was also observed during infection of NCI H-292 human bronchial epithelial cells (Euba et al., 2015c), for both adhesion and invasion (Figure 4B). NTHi375ΔthyA grown in sBHI showed lower adhesion to NCI H-292 cells than the WT strain, which was restored by thymidine supplementation, and adhesion by NTHi375ΔthyA grown in sBHI+Thy was higher than that shown by the WT strain (Figure 4B, right).

Complement opsonization, specifically C3, has been shown to enhance the bacterium-epithelial cell interaction for poorly encapsulated strains (de Astorza et al., 2004). Based on the observed differential C3b deposition by NTHi375 WT and ΔthyA strains, we next tested the effect of C3 in NTHi infection of A549 cells. NTHi375 and NTHi375ΔthyA strains were grown on chocolate agar (ΔthyA) or chocolate+Thy (ΔthyA+Thy), and A549 cells were infected in the presence of human purified C3, C3-deficient serum, or C3-deficient serum reconstituted with human purified C3. The level of bacterial attachment did not change in the presence of C3, compared to that observed in control untreated cells (CON) or cells co-incubated with C3-deficient serum (Figure S4B).

In summary, thymidine auxotrophy modified NTHi ability to infect airway epithelial cells in terms of bacterial location and triggered inflammatory response. Changes in adhesion and IL-8 secretion could be restored by addition of external thymidine in the bacterial growth medium. The presence of C3 at the onset of infection did not modulate bacterial adhesion.

Finally, we sought to determine the impact of thymidine auxotrophy in vivo, by using a mouse NTHi respiratory infection model system previously used for NTHi375 (Morey et al., 2013; Euba et al., 2015a,b). Mice were infected with NTHi375 WT and thyA mutant strains grown on chocolate agar, and bacterial loads were quantified in lungs and BALF samples at 12, 24, and 48 hpi. In lungs, NTHi375ΔthyA bacterial numbers were lower than those recovered for the WT strain at the three infection time points tested. Thymidine dependency was also associated with significantly reduced bacterial counts in BALF at 12 and 24 hpi (Figures 5A,B).

To further investigate if thymidine auxotrophy, despite attenuation, confers an advantage to H. influenzae under TxS treatment, we used a therapeutic regimen consisting of one TxS dose (960 mg/kg, 1:5 TMP:SMX) every 6 h, starting at 6 hpi. Results indicate that TxS treatment was more efficient in reducing bacterial counts of the WT than of the thyA mutant strain. Thus, a significant reduction of the WT population in lungs at 12, 24, and 48 hpi, and BALF samples at 24 hpi, was observed under treated—compared to that of untreated conditions. In contrast, bacterial counts of NTHi375ΔthyA were unchanged at 12 and 24 hpi in both lungs and BALFs, despite TxS administration. The protective effect of thyA mutation against TxS was not observed at 48 hpi, maybe due to NTHi375ΔthyA severe clearance at this post-infection time point.

Based on the differences observed at 24 hpi, we attempted to restore the thyA mutant phenotypes by infecting with mutant bacteria previously grown in the presence of thymidine. However, intranasal inoculation of NTHi375ΔthyA grown on chocolate+Thy did not restore bacterial counts, and lung and BALF counts for NTHi375 WT were higher than those obtained for ΔthyA grown on chocolate+Thy. Unexpectedly, NTHi375ΔthyA previous growth on chocolate+Thy increased TxS efficiency, i.e., in TxS treated animals, lung, and BALF counts were higher for NTHi375ΔthyA grown on chocolate agar than for NTHi375ΔthyA grown on chocolate+Thy. Also, lung and BALF counts for NTHi375ΔthyA grown on chocolate+Thy were higher in untreated—than in TxS treated animals (Figures 5A,B).

Microscopy score of the average histopathological lesion in samples of mice infected with WT or ΔthyA strains was also determined along the respiratory tract, and compared for untreated and TxS-treated mice (Table 3). Histopathological analysis of upper airways, larynxes, tracheas, and lungs from mice intranasally infected with WT or with ΔthyA rendered inflammatory lesions. Lungs and airways from control mice instilled with PBS did not show significant inflammation or pathological changes, independently of TxS administration. In both WT and ΔthyA-infected mice, the upper airways showed PMN infiltration of the lamina propria and lumens. PMN infiltration was found in larger numbers in the lumens of ΔthyA infected mice than in those infected with WT bacteria, at both 12 and 24 hpi. Conversely, PMN infiltration of the upper airway lamina propria was found to be larger in WT—than in ΔthyA infected mice at 24 hpi. Mild PMN infiltration of the lamina propria and lumens containing red blood cells and PMNs were the main findings in larynxes and tracheas. A tendency towards increased red blood cells containing lumens in WT-infected compared to ΔthyA infected mice was apparent at 24 hpi, although the differences did not reach statistical significance. As previously described (Euba et al., 2015b), lungs in infected mice showed areas of acute bronchopneumonia where alveolar septa were thickened with edema and hyperemic septal capillaries. Neutrophils, alveolar macrophages and scattered small hemorrhages were observed in alveolar spaces. Comparison of scored lesions at 12 hpi showed significantly more hemorrhages at the lower airway of mice infected with WT than at those infected with ΔthyA bacteria in control untreated mice. Hemorrhages were also found in larger proportions at the alveoli of mice infected with NTHi375 at 12 than at 24 hpi, but a lower proportion of alveolar macrophages was found at the lower airway of mice infected with NTHi375 at 12 than at 24 hpi. Significant differences were not found at the alveoli of ΔthyA infected mice between post-infection time points. Analysis of scored lesions in the airways of WT infected mice showed differences between control untreated- and TxS treated-mice. We observed a higher proportion of PMN infiltration in (i) the upper airway lumens at 24 hpi in TxS treated—than in control untreated mice, and (ii) the alveoli in control untreated—than in TxS treated mice at 12 hpi. Conversely, ΔthyA infection caused comparable lesions in both untreated- and TxS treated-mice (Table 3). Intranasal inoculation of NTHi375ΔthyA previously grown on chocolate+Thy rendered similar inflammatory lesions in both untreated- and TxS treated-animals; as a sole difference, a higher proportion of PMN infiltration was observed in the alveoli of TxS treated—than of control untreated mice (p < 0.05).

Last, we quantified accumulation of immune cells in the collected BALF samples at 12 hpi. NTHi375 WT pulmonary infection increased the accumulation of PMNs and alveolar macrophages, compared to that observed in the BALF of ΔthyA infected mice. TxS treatment showed a trend to reduce immune cell accumulation in BALF samples of WT infected mice. Unexpectedly, TxS treatment increased immune cell accumulation in BALF samples of ΔthyA infected mice (for both PMNs and alveolar macrophages; Figure 5C).

In summary, thymidine auxotrophy reduces NTHi virulence but, concomitantly, it confers an advantage under TxS treatment in vivo, suggesting that uptake of external thymidine from the infected tissue may contribute to bypass the bactericidal effect of TxS. Moreover, overall higher inflammatory traits were observed in WT—than in ΔthyA infected mice.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.