In Vitro Anti-NTHi Activity of Haemophilin-Producing Strains of Haemophilus haemolyticus.

Nontypeable Haemophilus influenzae (NTHi) is a leading causative organism of opportunistic respiratory tract infections. However, there are currently no effective vaccination strategies, and existing treatments are compromised by antibiotic resistance. We previously characterized Haemophilus haemolyticus (Hh) strains capable of producing haemophilin (HPL), a heme-binding protein that restricts NTHi growth by limiting its access to an essential growth factor, heme. Thus, these strains may have utility as a probiotic therapy against NTHi infection by limiting colonization, migration and subsequent infection in susceptible individuals. Here, we assess the preliminary feasibility of this approach by direct in vitro competition assays between NTHi and Hh strains with varying capacity to produce HPL. Subsequent changes in NTHi growth rate and fitness, in conjunction with HPL expression analysis, were employed to assess the NTHi-inhibitory capacity of Hh strains. HPL-producing strains of Hh not only outcompeted NTHi during short-term and extended co-culture, but also demonstrated a growth advantage compared with Hh strains unable to produce the protein. Additionally, HPL expression levels during competition correlated with the NTHi-inhibitory phenotype. HPL-producing strains of Hh demonstrate significant probiotic potential against NTHi colonization in the upper respiratory tract, however, further investigations are warranted to demonstrate a range of other characteristics that would support the eventual development of a probiotic.


S1.1. Triplex Real-Time PCR Assay Design, Optimisation and Validation
A triplex PCR assay was designed to detect and quantify NTHi, Hh and the HPL ORF. The targets used for discrimination of Hh (HypD) and NTHi (SiaT) have previously been described and validated [1]. For detection of the HPL ORF, primers were designed based on the HPL ORF of Hh-BW1 (GenBank MN720274) [2]. Primer specificity was confirmed using discontiguous megaBLAST analysis, performed across 115 complete and 862 draft genome assemblies for Haemophilus.spp, available from Genbank. A nonredundant nucleotide (nr/nt) collection BLAST search was also conducted to determine amplicon specificity in non-Haemophilus genomes.
Following optimisation in singleplex format, the three assays were merged into triplex format. Annealing temperature was optimised using an 8-step temperature gradient ranging from 53-63 °C. Specificity of amplicons was determined by gel electrophoresis, and the optimal temperature was selected based on highest yield of amplicons of the correct size in the absence of nonspecific amplification. PCR specificity for NTHi, Hh and HPL was determined using 13 Hh strains with varying HPL sequence similarity to BW1, as previously defined [2], 13 Hi strains and 9 other genera representing common upper respiratory tract flora (see Supplementary Table 2).
Reaction efficiency of triplex reaction was determined using 10-fold dilutions of control strains over the range of 2 to 2 × 10 −8 ng (Hi ATCC49247, Hh ATCC 33390 and Hh BW1). Limits of quantification (LoQ) values were determined for hypD and siaT targets in triplex format based on criteria where replicates at a given dilution with a cycles to threshold (Ct) standard deviation (σ) of ≥ 0.8 were considered to exceed the LoQ, with one or more amplification failures also deemed an LoQ failure. The upper LoQ value was not determined due to the unlikelihood of encountering such highpathogen DNA concentrations in clinical specimens. The lower limit of detection was also determined for HPL target and defined from linearity data ( Figure S1).

S1.2. Boiled DNA Extraction Protocol
Bacterial suspensions were heated to 99 °C for 10 min in a heat block with shaking (1000 rpm). The tube was vented and vortexed for 10 s, followed by a further 10 min incubation at 99 °C. Boiled suspensions were spun for 10,000 g for 10 min in a benchtop microcentrifuge. The supernatant was diluted 1/5 in molecular grade water for use in PCR reactions. Given the instability of gDNA extracted by this method, PCR was conducted immediately or stored at 4 °C for no more than 2 h before use.

S1.3. Validation of Boiled DNA Extraction Method
A series of 10-fold dilutions (10 0 -10 −10 ) were made made from a heavy suspension (OD600 = 3.0) of either Hh-BW1 or NTHi ATCC 49427 in TSB. To detect the presence of PCR inhibititors, each dilution was tested in duplicate and a nonlinear regression was generated. A slope between −3.1 and −4.1 (equivalent to amplification efficiency of 75%-110%) and a correlation coefficient > 0.98 was considered acceptable [3].
To test the methods accuracy in quantifying both targets simultaneously, each combination of NTHi and Hh dilutions in the range of 10 0 -10 −6 was mixed (50:50) prior to extraction and subsequent PCR. This produced varying proportions of NTHi and Hh to mimic possible scenarios of co-culture.
To compare OD600 readings and GE, a growth curve experiment was conducted where 110 uL aliquots were taken from a 5 mL culture in TSB every hour for 8 h and at 18 and 28 h. The OD600 of each aliquot was measured in a 96-well microtiter plate (Infinite 200 PRO, Tecan Life Sciences). Of this aliquot, 10 uL was used to produce dilutions for quantitation by colony count on chocolate agar and 50 uL was taken for gDNA extraction by boiling.

S2.1. Validation of Triplex Real-Time PCR
In silico specificity for the HPL amplicon revealed 97%-100% primer and probe nucleotide sequence identity for 26 of 61 complete or draft Hh genomes available in Genbank. Sequence similarity to the HPL amplicon was detected in 20 H. influenzae genomes out of 757 complete and draft genome assemblies. However, all alignments contained a minimum of four SNPs in the reverse primer and Taqman probe and did not contain any sequence homology with the forward primer. HPL was also detected in three genome assemblies available for Haemopilus quentini, which was expected based on previous analysis of the HPL ORF [2]. Despite close relatedness to Hh, isolation of this strain has only been described in the genitourinary tract so is unlikely to be co-isolated from respiratory specimens [4]. Comparison of 11 previously sequenced HPL ORFs [2] to these databases yielded the same results, indicating high sensitivity to known HPL sequence variants (ranging from 85%-100% homology to BW1-HPL). PCR of these isolates, and additional Hh, Hi and common upper respiratory tract colonisers confirmed specificity for each target.

S2.2. Quantification by PCR of Boiled DNA is Comparable to Optical Density
Serial dilutions did not reveal the presence of PCR inhibitors as regression parameters produced amplification efficiency of 99.37% and 91.57% for the SiaT and HypD targets, respectively ( Figure S2). Similarly, amplification efficiencies did not differ between species, and their measurement was not affected when present in dramatically different, or similar, ratios ( Figure S3). Growth patterns generated by OD600 could be replicated by PCR using DNA extracted by boiling. This was observed visually and by regression statistics directly comparing OD600 to GE ( Figure S4).