Early oseltamivir reduces risk for influenza-associated aspergillosis in a double-hit murine model.

Invasive pulmonary aspergillosis (IPA) is a life-threatening fungal infection occurring mainly in immunocompromised patients. We recently identified IPA as an emerging co-infection with high mortality in critically ill, but otherwise immunocompetent influenza patients. The neuraminidase inhibitor oseltamivir is the current standard-of-care treatment in hospitalized influenza patients; however, its efficacy in influenza-associated pulmonary aspergillosis (IAPA) is not known. Therefore, we have established an imaging-supported double-hit mouse model to investigate the therapeutic effect of oseltamivir on the development of IAPA. Immunocompetent mice received intranasal instillation influenza A or PBS followed by orotracheal inoculation with Aspergillus fumigatus 4 days later. Oseltamivir treatment or placebo was started at day 0, day 2, or day 4. Daily monitoring included micro-computed tomography and bioluminescence imaging of pneumonia and fungal burden. Non-invasive biomarkers were complemented with imaging, molecular, immunological, and pathological analysis. Influenza virus-infected immunocompetent mice developed proven airway IPA upon co-infection with Aspergillus fumigatus, whereas non-influenza-infected mice fully cleared Aspergillus, confirming influenza as a risk factor for developing IPA. Longitudinal micro-CT showed pulmonary lesions after influenza infection worsening after Aspergillus co-infection, congruent with bioluminescence imaging and histology confirming Aspergillus pneumonia. Early oseltamivir treatment prevented severe influenza pneumonia and mitigated the development of IPA and associated mortality. A time-dependent treatment effect was consistently observed with imaging, molecular, and pathological analyses. Hence, our findings underscore the importance of initiating oseltamivir as soon as possible, to suppress influenza infection and mitigate the risk of potentially lethal IAPA disease.

Introduction

Invasive pulmonary aspergillosis (IPA) is a life-threatening fungal infection characterized by acute infection of the lungs occurring mainly in immunocompromised patients [1]. In the last decade, severe influenza has gained a foothold worldwide as an emerging risk factor for IPA [1,2]. Recently, we published two retrospective case–control studies on influenza-associated pulmonary aspergillosis (IAPA) [1,3]. IPA was found as a co-infection in 19% of patients with severe influenza at a median of 3 days after intensive care unit (ICU) admission [1]. Remarkably, the cohort of co-infected patients had a high mortality risk of 51%, whereas that of severe influenza without IAPA was 28% [1]. By June 2018, 128 IAPA cases were published, with 28% of the patients having no underlying medical condition [4]. These incidence rates emphasize the importance of recognizing influenza as an independent risk factor for IPA and the high morbidity and mortality associated with it. Hence, an international group of experts has published a case definition of IAPA to facilitate research in epidemiology, diagnosis, and management of this emerging disease [5].

Globally, influenza is estimated to affect 5–10% of the population each year. On average, 3–5 million cases of severe influenza occur annually worldwide, leading to 290,000–650,000 deaths [6,7]. Virus-induced lung epithelial damage and host immune response aberrations are hallmark manifestations of severe influenza infection [8-11]. These pathological processes are the key to success for secondary pathogenic microorganisms to enter the lung parenchyma and colonize patients [12]. Bacterial superinfections with Staphylococcus aureus and Streptococcus pneumoniae are well-defined complications of a severe influenza infection and different mouse models of influenza-bacterial-co-infection have been developed [13-16]. However, until today, influenza-associated fungal co-infection models are not yet widely experimentally addressed. To preclinically assess potential treatment strategies, a robust animal model recapitulating IAPA disease in humans is needed.

Currently, the neuraminidase inhibitor (NAI) oseltamivir, is the standard-of-care to suppress influenza. This drug prevents successful virus replication by inhibiting the viral neuraminidase. Several clinical studies have corroborated that early oseltamivir treatment improves the outcome in critically ill patients suffering from influenza [17-19]. Also in mice, oseltamivir is known to improve survival, body weight loss, and to reduce viral lung titers, thereby limiting the severity of lung damage [20,21]. In addition, in a mouse model of influenza-bacterial-co-infection both prophylactic and early oseltamivir treatments improved the survival and reduced the extent of the secondary pneumococcal pneumonia [22-24]. Likewise, we hypothesize that oseltamivir may reduce the susceptibility to IAPA in mice and men through attenuation of influenza infection.

Here, we establish a double-hit mouse model for IAPA and report on its imaging capabilities to longitudinally follow-up influenza pneumonia and exacerbation after challenge with A. fumigatus conidia. In this model, we assessed whether IAPA is prevented by oseltamivir treatment initiated at different time points during influenza infection.

Materials and methods

Influenza virus strain

A mouse-adapted influenza virus A/H3N2/Ishikawa/7/82 was kindly provided as a mouse lung homogenate by M. Hosoya (Fukushima, Japan) [25]. The virus was expanded in Madin–Darby canine kidney (MDCK) cells, which were also used to determine the virus titer by standard plaque assay. Original virus stocks were thawed and diluted to a final concentration of 500 PFU/mL.

Aspergillus strain

In all experiments, an A. fumigatus derivative of strain CBS144.89 expressing a red-shifted thermostable firefly luciferase from the akuB locus was used. In brief, a codon-optimized red-shifted and thermostable firefly luciferase (lucOPT_red_TS, GenBank accession number: MT554554) was synthesized (GenScript) and cloned under control of the 820 bp promoter of the glyceraldehyde-3-phosphate dehydrogenase (gpdA) from Aspergillus nidulans and fused with a 322 bp gpdA terminator region. This luciferase expression construct was cloned into the ptrA-pJET1 plasmid [26] containing the pyrithiamine resistance gene. The entire 4785 reporter_ptrA cassette was amplified with oligonucleotides PgpdAIFKu80up_f (5ʹ – CCC TGA GGC GGC CGC GGG CTG AGT AAT AAG – 3ʹ) and ptrAIFKU80do_r (5ʹ – AAC TTT GGG CGG CCG CGT ATT ATA CTG TC – 3ʹ) to allow subsequent in vitro recombination with the upstream and downstream flanking regions of the akuB gene from A. fumigatus. Genomic DNA of the A. fumigatus wild-type strain CBS144.89 served as template for amplification of the 819 bp upstream and 666 bp downstream region of the akuB gene. The akuB upstream region was amplified with oligonucleotides KU80upIFpUC_f (5ʹ – CGA GCT CGG TAC CCG GGA CTC AAT TTC TAT TCT AGA GCA TC- 3ʹ) and KU80upIFPgpdA_r (5ʹ – GCGGCCGCCTCAGGGTCGTCAAAGTCAGTAC – 3ʹ). The downstream region was amplified with oligonucleotides KU80doIFptrA_f (5ʹ – GCG GCC GCC CAA AGT TAA AGG GCG CAA GC) and KU80doIFpUC_r (5ʹ – CTC TAG AGG ATC CCC GGG CTA TCA CTT TGC CCA GTC – 3ʹ). A pUC19 plasmid was linearized with SmaI and the three PCR products were assembled with the plasmid by in vitro recombination using the InFusion HD cloning kit (Takara/Clonetech). Assembled plasmids were amplified in Escherichia coli DH5α. After isolation of the plasmid (NucleoSpin plasmid, Machery-Nagel) the entire cassette was excised from the plasmid backbone by SmaI restriction and used for protoplast transformation of the A. fumigatus wild-type strain CBS144.89 using 0.1 μg pyrithiamine/ml (Sigma) as selection marker. Transformants were pre-screened by cultivation on Aspergillus minimal medium [27] containing 0.2 mM D-luciferin (Promega) and imaging of plates in a Chemi Doc XRS system (BioRad) under chemiluminescent setting. Selected transformants were analyzed for single copy integration of the reporter construct into the akuB locus and strain Af_ΔakuB::lucOPT_red_TS_ptrA No. 5 was selected for further use. For preparation of conidia suspensions, the strain was cultured on Sabouraud agar for 3 days at 42°C. A. fumigatus conidia were harvested by detaching with NaCl-0.1% Tween 80 (Sigma-Aldrich, Diegem, Belgium) solution, followed by vigorous shaking and filtering of the suspension. Finally, spores were counted with a Neubauer hemocytometer and diluted to a final concentration of 5 × 109 conidia/mL.

Mouse model

All animal experiments were approved by the KU Leuven Ethical Committee for animal research (license p074/2018). Eight to ten weeks old immunocompetent male C57BL/6NTac mice (internal stock of the Laboratory Animal Center, KU Leuven) were kept in a conventional animal facility in individually ventilated cages with free access to food and water. To prevent bacterial infection, an antibiotic (50 mg/kg/day, Baytril®, Bayer) was added to the drinking water. Animals were randomly assigned to experimental groups. Mice received intranasal instillation of 10 plaque forming units (PFU) influenza virus or PBS (10 µL per nostril) under inhalation anesthesia with isoflurane (1.5–2% in oxygen, Primal Critical Care). Mice were orotracheally inoculated with 20 µL of A. fumigatus spore solution under injection anesthesia (Ketamine (75 µg/g) and Medetomidine (1 µg/g)) (Figure 1(a)) or isoflurane (Figure 2a, Figure 4a). Oseltamivir (10 mg/kg) or PBS treatment was administered twice daily by oral gavage. During the experiments, mice were monitored daily for body weight, general condition, and respiratory parameters, non-invasively scanned with micro-computed tomography and bioluminescence imaging and sacrificed when the humane or predefined experimental endpoint was reached, by an overdose of pentobarbital via intraperitoneal injection [28]. Subsequently, blood was drawn by cardiac puncture, bronchoalveolar lavage (BAL) fluid was obtained by three lung lavages with 700 µL 0.9% NaCl and right lung lobes were aseptically collected, weighted, and homogenized in PBS. The left lung was inflated with 400 µL 10%-formalin, 24 h post fixated and then stored in PBS-0.1% sodium-azide at 4°C. Subsequently, histopathology analysis was performed. BAL and blood were centrifuged at respectively 1000 x g for 10 min at 4°C and 13,817 x g for 4 min at 4°C. BAL supernatant, serum and lung homogenates were used for further quantification of virus titers, fungal burden, and cytokine analysis.

Figure 1.

Second-hit with A. fumigatus during active influenza infection drives risk for clinical deterioration with IPA. (a) experimental setup for the development of double-hit mouse model: immunocompetent mice received influenza A virus on day 0, followed by orotracheal inoculation with A. fumigatus conidia on day 3, 4, 5, 6 or day 7 post-influenza infection. control groups of mock-infected (sham), influenza only and aspergillus only were included. body weight was measured daily. (b) body weight evolution of sham-infected and infected mice. (c) maximum body weight loss per animal for different experimental arms. red box represents mice which were sacrificed because humane endpoints were reached (d) grocott’s methenamine silver staining of left lungs from mice with A. fumigatus, influenza-A. fumigatus day 3 co-infected, influenza-A. fumigatus day 4 co-infected mice sacrificed at respectively day 7 and day 8 post influenza infection (magnification ×400). Aspergillus conidia (red box) and hyphae (yellow box)

Figure 2.

Immunocompetent mice develop IPA in a background of a sublethal influenza-pneumonia. (a) experimental setup: immunocompetent mice received influenza A virus on day 0 followed by orotracheal inoculation with A. fumigatus conidia on day 3 and 4 post-influenza infection. sacrifice was performed 2 days after co-infection. Single control groups of either mock infected (sham) and influenza alone with sacrifice on day 6 were included. (b) body weight evolution of immunocompetent mice infected with influenza on baseline (day 0) and co-infected with a. fumigatus 3 or 4 days later. Parallel sham-infected and influenza-infected mice were included as controls. (c) graph representing the log10 CCID50 per gram lung tissue from sham-infected (sacrificed on d6), influenza-infected (sacrificed on d6), influenza – A. fumigatus day 3 (sacrificed on d5) and influenza-A. fumigatus day 4 co-infected mice (sacrificed on d6). (d) Log10 colony forming unit (CFU) count per gram lung tissue from influenza-A. fumigatus day 3 (sacrifice d5) and influenza-A. fumigatus day 4 co-infected mice (sacrifice d6). (e) Scatter plot of cytokine response (IL-1β and IL-6) in serum and bronchoalveolar lavage fluid (BALF) of mock infected (sham), influenza infected and influenza-Aspergillus day 4 co-infected mice sacrifice on day 6. dashed line represents lower detection limit of the standards (LDL). In influenza-A. fumigatus day 4 co-infected mice. all graphs are represented as mean or as mean ± SD, **** P < 0.0001

Figure 4.

Oseltamivir treatment improves survival and body weight reduction. (a) experimental setup of the establishment of the relationship between influenza and risk to IPA through modulation of influenza by antiviral treatment oseltamivir: immunocompetent mice received influenza A on day 0, followed by inoculation with A. fumigatus on day 4. oseltamivir (10 mg/kg) or PBS treatment twice daily by oral gavage was started at day 0 (n = 10), day 2 (n = 10) or day 4 (n = 5) up to predefined experimental endpoint (day 7 or day 12) or when humane endpoints were reached. during the experiment, mice were monitored daily (body weight, general condition, respiratory parameters) while performing micro-computed tomography (µCT) and bioluminescence imaging (BLI). (b) Body weight evolution of immunocompetent influenza- Aspergillus day 4 co-infected mice without oseltamivir (blue, n = 10), oseltamivir treatment from day 4 (orange, n = 5), oseltamivir treatment from day 2 (red, n = 10), oseltamivir treatment from day 0 (green, n = 10). Longitudinal Statistical analysis: two-way ANOVA with Tukey’s multiple comparisons test repeated measurements performed from day 0 until day 4 (*** p = 0,0001; * (orange, green) p = 0,0416; * p (red, green) = 0,0220) and day 4 until day 7 (**** P < 0.0001, ** p = 0,0092, * p = 0,0453) (c) Kaplan-Meier survival curve of influenza-aspergillus day 4 co-infected mice with no oseltamivir (blue, n = 5), oseltamivir treated from day 2 (red, n = 5) and oseltamivir treated from day 0 (green, n = 5). all graphs are represented as mean ± SD

Micro-computed tomography (µCT)

A dedicated whole-body small animal µCT scanner (SkyScan 1278, Bruker micro-CT, Kontich, Belgium) was used to acquire lung µCT data of free-breathing anesthetized (isoflurane, 1.5–2% in oxygen, Piramal Critical Care) animals in supine position. Following scan parameters were used: 50 kVp X-ray source, 1 mm aluminum X-ray filter, 350 µA current and 150 ms exposure time per projection acquiring projections with 0.9° increments over a total angle of 220°. Acquisition resulted in respiratory weighted reconstructed 3D dataset with isotropic voxel size of 50 µm in a total scanning time of 3 min, associated with a radiation dose exposure of 69–89 mGy [29]. Temperature, respiration rate, and visual information were used to monitor mice while scanning. µCT data was reconstructed, visualized, and processed using NRecon, DataViewer, and CTan software, provided by the manufacturer. Reconstruction parameters were: smoothing 2, beam hardening correction of 10%; post alignment and ring artifact reduction were optimized for each individual scan. µCT data were processed and calibrated to Hounsfield units (HU) as described before [30]. In brief, regions of interests (ROI), resulting in a volume of interest (VOI) covering the entire lung, were manually delineated on transverse images, thereby avoiding the heart and main blood vessels [30,31]. Subsequently, imaging-derived biomarkers of lung pathology (total lung volume (TLV), aerated lung volume (ALV), non-aerated lung volume (NALV) and mean lung densities) were quantified. To distinguish ALV from NALV, a fixed manually set threshold of 94 HU was used based on histogram in CTan (TLV ranges from −1000 to 1160, ALV from −1000 to (−94), NALV −85 to1160). Visual observations of transverse µCT images were semi-quantitatively scored by two blinded researchers (LS and BL) according to [31] with minor modifications. In brief, four transverse images at different, defined positions in the lung were selected for each animal at each time point on the reconstructed µCT. Per animal, a score of zero was assigned to each image at baseline. Subsequently, scores of +0.25, 0 or −0.25 depending on observed worsening, stabilization, or improvement of pulmonary lesions (pulmonary infiltrate), respectively, were assigned and added to the score of the previous day to obtain a cumulative score.

Bioluminescence imaging (BLI)

For in vivo BLI, the mice were intraperitoneally injected with D-luciferin (in PBS, 50 mg/mL, 500 mg/kg), then anesthetized (isoflurane inhalation, 1.5–2% in oxygen, Piramal Critical Care) and placed in supine position in an IVIS Spectrum System (Perkin-Elmer, Hopkinton, MA, USA). Five consecutive images were acquired starting 10 min after injection. Following settings were used: exposure time of 1 min, medium binning, F/stop of 1 and subject height of 1.5 cm. Quantification of BLI data reflecting fungal lung burden was performed with software provided by the manufacturer (Living Image Software version 4.7.3). Total photon flux (p/s) was measured through a ROI of 6.9 cm2, covering the lungs, and the maximum from consecutive acquisitions used for further analysis.

For ex vivo BLI, 1:5 serial dilutions of homogenized right lung lobes in black 96-well plates (Nunc® Microwell, Thermo Fisher Scientific, Merelbeke, Belgium) were placed in the IVIS Spectrum after addition of 10% D-luciferin (in PBS, 50 mg/mL). Images were acquired until maximum signal intensity was reached with following settings: exposure time of 1 min per image, medium binning, F/stop of 1 and subject height of 1 cm. ROIs, covering one well, were created to measure total photon flux. The highest total photon flux from consecutive acquisitions was used for further analysis.

Quantification of virus and fungal titers

Serial dilutions of right lung homogenates were also used to assess the fungal load by colony-forming unit (CFU) counting. Dilutions were plated on Sabouraud agar plates and incubated at 37°C for 48 h. Reported CFU counts represent average values of duplicate plating of each dilution. DNA was extracted from lung homogenates and BAL fluid, for subsequent AsperGenius® multiplex real-time PCR assay, as previously described [32]. For influenza virus titration, serial dilutions of right lung homogenate supernatants were added to MDCK cells in quadruplicate. After 72 h incubation at 35°C, viral cytopathic effect was assessed by microscopic scoring. Virus titers were calculated by the method of Reed and Muench as log10 CCID50/g of lung tissue [33].

Cytokine analysis

Cytokine levels were quantified in BAL fluid and serum using an electrochemiluminescence based V-PLEX Proinflammatory Panel 1 Mouse kit from Meso Scale Diagnostics (MSD) [34].

Histopathology

Left lung 5-µm paraffin sections were cut and stained with hematoxylin-eosin and Grocott’s methenamine silver staining. Staining were scored by a blinded pathologist (EV).

Statistical analysis

Statistics were performed using GraphPad Prism (version 8.1.2, GraphPad Software, San Diego, CA, USA). Longitudinal and end point data were analyzed with, respectively, two-way ANOVA or one-way ANOVA with Tukey’s multiple comparisons test (repeated measurements performed from day 0 to day 4 and day 4 to day 7). All graphs represent mean values ± SD. P values are indicated in figures and are described in full in figure or figure legends and differences were considered significant if p was smaller or equal to 0.05. *P < 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001. n values represent the number of animals.

Results

Immunocompetent mice develop IPA in a background of a sublethal influenza-pneumonia

We first set out to investigate the risk of developing IPA in the setting of an influenza virus infection. To determine the effect of timing of secondary Aspergillus infection toward developing IAPA, mice were co-infected with conidia of a bioluminescent A. fumigatus strain at day 3, 4, 5, 6, or 7 post-influenza virus infection (Figure 1(a), Suppl Fig. S1). Mice infected with influenza virus alone showed moderate body weight loss and clinical worsening up to day 8 followed by recovery toward day 12 in all mice. Likewise, mice infected with only A. fumigatus showed limited transient body weight loss without overt clinical illness, and all these animals fully recovered. In contrast, infection with influenza virus followed by A. fumigatus resulted in faster and more substantial body weight loss as well as more severe clinical deterioration in some but not all mice (Figure 1(b-c)). In day 3 or day 4 co-infected fatally ill mice, histopathological examination confirmed the presence of hyphae invading the airway wall indicating airway invasive pulmonary aspergillosis which is in sharp contrast with Aspergillus only infected mice where only Aspergillus conidia were present (Figure 1(d)).

Given the observed disease severity in these mice, similarity with early IAPA development in critically ill influenza patients in ICU and therapeutic window for immune modulation in future research, in a next experiment, we zoomed in on the pathology of Aspergillus co-infection 3 and 4 days after influenza infection (Figure 2(a)). Consistently, these co-infected animals showed markedly faster reduced body weight as well as clinical deterioration immediately after Aspergillus inoculation in comparison with influenza-only infected mice (Figure 2(b)). At day 2 post A. fumigatus infection, we measured extensive fungal load and high viral load (Figure 2(c-d)). In influenza- A. fumigatus day 4 co-infected mice, sacrificed at day 6, the cytokine profile showed a strong inflammatory burden with high levels of IL-1β and IL-6 in BAL fluid and serum which could point toward a possible myeloid activation (Figure 2(e), Suppl Fig. S2). Furthermore, histopathology of day 4 co-infected mice exhibited severe diffuse acute purulent mainly neutrophilic bronchitis and pneumonia with epithelial erosion at day 6 (2 days post A. fumigatus infection) (Figure 3(a)). In contrast, single infection with influenza virus resulted in diffuse bronchiolitis with epithelial damage and multifocal lymphocytic alveolitis at day 6 (Figure 3(a)). Histopathology further confirmed the presence of A. fumigatus conidia with hyphae invading the airway wall in all day 4 (Figure 3(a)) and some of the day 3 co-infected mice (data not shown). Together, these data demonstrate that sublethal influenza pneumonia predisposes to airway invasive pulmonary aspergillosis in immunocompetent mice.

Figure 3.

Histopathological confirmation of airway invasive pulmonary aspergillosis in influenza Aspergillus day 4 co-infected mice. (a) Hematoxylin and eosin (H&E) of left lung of (on the top) single infection with influenza virus (magnification ×50, ×200) showing a diffuse bronchiolitis with epithelial damage and multifocal lymphocytic alveolitis at day 6 and (on the bottom) influenza-Aspergillus day 4 co-infected mice sacrificed at day 6 (magnification ×100, ×400) showing a severe diffuse acute purulent mainly neutrophilic bronchitis and pneumonia with epithelial erosion at day 6. grocott’s methenamine silver staining (GMS) of influenza-Aspergillus day 4 co-infected mice sacrificed at day 6 (magnification x400). Black arrows represent hyphae invading the airway wall

Oseltamivir modulates severity of influenza infection and mitigates the susceptibility to IPA in a time-dependent manner

Having established a double-hit mouse model of IAPA, we next explored the efficacy of oseltamivir in modulating the susceptibility to develop IPA. The antiviral drug was started at day 0, 2, 4 after influenza virus infection (Figure 4(a)). The non-treated day 4 co-infected mice experienced severe body weight loss by day 4, significantly worsening toward day 7, and all these animals had to be euthanized by day 8 or 9 (Figure 4(b-c)). In sharp contrast, oseltamivir initiated at day 0 or 2 generated a survival rate of 100%, in line with significantly less body weight loss and body weight recovery from day 7 onwards (Figure 4(b-c)). Also, mice treated from day 4, lost significantly less body weight loss in comparison with the non-treated group was seen toward day 7 (Figure 4(b)). Hence, oseltamivir proved highly effective the earlier the treatment was started.

We further analyzed the precise impact of oseltamivir on the severity of influenza-pneumonia through longitudinal non-invasive µCT. Toward day 4 (during influenza infection and before Aspergillus co-infection), semi-quantitative scores derived from µCT-thorax scans revealed multifocal hyperdense pulmonary lesions in the lungs of non-treated mice, while oseltamivir treated mice remained largely spared from lesions development the earlier oseltamivir was started (Figure 5(a-b) Suppl Fig. S3). In support of these visual observations, non-aerated lung volume, a quantitative biomarker of pulmonary infiltrates and consolidation, was markedly lower in oseltamivir versus non-treated mice (Figure 5(c)). Accordingly, the aerated lung volume was consistently higher in drug-treated animals (Figure 5(d)). Together, these results showed substantially reduced severity of influenza pneumonia the earlier the start of oseltamivir treatment.

Figure 5.

µCT revealed longitudinal visualization and quantification of the disease process of IAPA and the alleviated or abolished effect from oseltamivir treatment. (a) Longitudinal transversal lung µCT images of sham-treated (no oselt), oseltamivir treated from day 4 (oselt d4), oseltamivir treated from day 2 (oselt d2) and oseltamivir treated from day 0 (oselt d0) double-infected mice and coronal images on day 7. Blue arrow: pulmonary infiltrates, red arrow: bronchial dilation (b) Semi-quantification of image score of transverse lungs µCT images (no oseltamivir, blue, n = 10; oseltamivir d4, orange, n = 5; oseltamivir d2, red, n = 10; oseltamivir d0, green, n = 10). Statistical analysis: two-way ANOVA with Tukey’s multiple comparisons test repeated measurements performed from day 0 until day 4 (* p = 0,0257) and day 4 until day 7 (* (red, green) p = 0,0158; * (blue, orange) p = 0,0349; **** P < 0.0001) (c-d) graph represents quantification of biomarkers non-aerated lung volume (c) and aerated lung volume (d) from transverse lung µCT images (no oseltamivir, blue, n = 10; oseltamivir d4, orange, n = 5; oseltamivir d2, red, n = 10; oseltamivir d0, green, n = 10). Statistical analysis for all longitudinal graphs: two-way ANOVA with Tukey’s multiple comparisons test repeated measurements performed from day 0 until day 4 (C = non-aerated lung volume: * p = 0,0388; D = aerated lung volume: * (red, green) p = 0,0359, * (orange, green) p = 0,0484, * (blue, green) p = 0,0137) and day 4 until day 7 (C = non-aerated lung volume: * (orange, green) p = 0,0106, * (red, green) p = 0,0185, ** (blue, orange) p = 0,0080, **** P < 0.0001; D = aerated lung volume: * p = 0,0444, ** p = 0,0068, **** P < 0.0001). Graphs are represented as mean ± SD. (e) Spearman correlation between % body weight change on day 4 and non-aerated lung volume on day 7. The spearman’s correlation coefficient and corresponding P-value are given on the graph

Toward day 7 (i.e. 3 days after co-infection with Aspergillus), we observed that oseltamivir suppressed or even prevented disease exacerbation resulting from IAPA. Obvious consolidation of entire lung lobes and bronchial dilation was seen in non-treated mice (Figure 5(a-b) Suppl Fig. S3). On the contrary, less to no pulmonary infiltrates were present in mice treated with oseltamivir, and this was more prominent the earlier treatment was started (Figure 5(a-b) Suppl Fig. S3). During Aspergillus co-infection (between day 4 and day 7), non-treated mice showed significant gradual increase in non-aerated and decreased aerated lung volume in comparison with treated mice (day 0 – day 2) (Figure 5(c-d)). Overall, the earlier treatment was initiated, the more significant these differences became. Although oseltamivir treated mice (day 4) do have a significantly lower non-aerated lung volume, there was no difference in aerated lung volume in comparison with non-treated mice (Figure 5(c-d)). Finally, by pooling the data from all mice, we observed that the non-aerated lung volume on day 7, strongly correlated with the body weight reduction on day 4 (r = −0,8255; P < 0.0001) (Figure 5(e)). Collectively, our analysis of lung abnormalities using imaging and µCT-derived biomarkers demonstrated that oseltamivir modulates the severity of influenza and subsequent Aspergillus co-infection in a time-dependent manner during IAPA infection.

Next, we explored the impact of oseltamivir on IAPA by longitudinal follow-up of fungal burden through bioluminescence-based imaging (BLI) and microbiological gold-standard techniques. Toward day 7 (3 days post Aspergillus co-infection), in vivo bioluminescence signal intensity was lower in oseltamivir-treated conditions compared to non-treated mice; however, almost all mice elicited comparable ex vivo bioluminescent signal and fungal burden (CFU count) (Figure 6(a-d)). Moreover, histopathology detected clear morphological differences between treated and non-treated mice: all mice treated with oseltamivir presented with lymphocytic inflammation combined with macrophages with intracellular A. fumigatus conidia, whereas non-treated mice showed a diffuse acute purulent neutrophilic inflammation with A. fumigatus hyphae invading the airway wall (Figure 7(a), Suppl. Table. S1).

Figure 6.

Oseltamivir treatment mitigates the risk to develop IPA. (a) Bioluminescence images (BLI) of double-infected mice acquired on day 7, 8 and day 12 post influenza. (b) in vivo BLI: quantification of total photon flux from BLI images based on region of interest (ROI) covering the complete lung region of influenza-Aspergillus day 4 co-infected mice without oseltamivir (blue, n = 10), oseltamivir treatment from day 4 (orange, n = 5), oseltamivir treatment from day 2 (red, n = 10), and oseltamivir treatment from day 0 (green, n = 10). Longitudinal Statistical analysis: two-way ANOVA with Tukey’s multiple comparisons test repeated measurements performed from day 4 until day 8 (*** (green, black) p = 0,0002; ** (blue, red) p = 0,0024, *** (blue, green) p = 0,0006) and day 4 until day 12 (*** p (red, black) = 0,0006). (c) ex vivo BLI: scatter plot represents the quantification of total photon flux from BLI images based on ROI covering the well containing undiluted lung homogenates on day 7 and endpoint (no oseltamivir day 8, oseltamivir treated groups day 12) from influenza-A. fumigatus day 4 co-infected mice without oseltamivir, oseltamivir treatment from day 4, oseltamivir treatment from day 2, and oseltamivir treatment from day 0, **** P < 0.0001. (d) Graph representing the log10 CFU count per gram lung tissue from influenza-Aspergillus day 4 co-infected mice without oseltamivir, oseltamivir treatment from day 4, oseltamivir treatment from day 2, and oseltamivir treatment from day 0, on day 7 and endpoint (no oseltamivir d8, oseltamivir groups d12), **** P < 0.0001. (e) Representative A. fumigatus quantification with AsperGenius qPCR from influenza-A. fumigatus day 4 co-infected without oseltamivir, oseltamivir treatment from day 2, and oseltamivir treatment from day 0 at endpoint (no oseltamivir d8, oseltamivir groups d12), **** P < 0.0001. (f) Spearman correlation between the log10 total flux from ex vivo BLI on experimental endpoint or humane endpoint and mean clinical score on day 4 of influenza-A. fumigatus day 4 co-infected without oseltamivir, oseltamivir treatment from day 4, oseltamivir treatment from day 2, and oseltamivir treatment from day 0. The spearman’s correlation coefficient and corresponding P-value are given on each graph. All graphs are represented as mean ± SD

Figure 7.

Oseltamivir treated mice are able to clear A. fumigatus in lungs. (a) Histological hematoxylin and eosin (H&E) (left, magnification ×50, ×200) and grocott’s methenamine silver staining (GMS) (right, magnification ×400) staining of left lung of different groups (no oseltamivir, oseltamivir treatment from day 4, oseltamivir treatment from day 2, oseltamivir treatment from day 0) sacrificed on day 7. Hyphae (black arrow), intracellular conidia (arrowhead). (b) Histological H&E (left, magnification ×50, ×200) and GMS (right, magnification ×200) staining of left lung of different groups (oseltamivir treatment from day 2 and oseltamivir treatment from day 0) sacrificed on day 12

At later time points, clearance of fungal infection became increasingly evident in oseltamivir treated animals and at endpoint, CFU counts and ex vivo BLI signals were significantly lower (compared to non-treated), in agreement with higher Ct-values when detection was done by PCR (Figure 6(c-e)). Again, these findings nicely correlated with the histopathology since mice treated with oseltamivir from day 0 or day 2 manifested, respectively, no or only slight lymphocytic inflammation with notable signs of epithelial recovery (Figure 7(b), Suppl. Table. S1). In addition, very local or no A. fumigatus conidia were perceived (Figure 7(b)). On the contrary, in non-treated mice, we observed mainly neutrophilic inflammation in bronchovesicular bundles and alveolar parenchyma with clear diffuse epithelial damage and bronchitis accompanied by A. fumigatus hyphae invading the airway wall (Suppl. Table. S1). Finally, mean clinical score on day 4 correlated with ex vivo bioluminescence on endpoint (r = 0.6455, P < 0.0001) (figure 6(f)). To conclude, early oseltamivir prevented severe lung pathology from influenza virus-Aspergillus co-infection and facilitated the pathogen clearing process.

Discussion

Here, we report the establishment of the first imaging-supported mouse model for the sequential interaction between influenza and Aspergillus fumigatus, experimentally confirming influenza as an independent risk factor for the development of IAPA in an immunocompetent host. Furthermore, we demonstrated that early oseltamivir treatment modulates the severity of influenza and mitigates the susceptibility to IAPA in a time-dependent manner. We herewith identify an early treatment to reduce the incidence of IAPA.

Until now, mouse models for influenza-associated co-infections have been limited to bacterial infections and more recently to Cryptococcus [35,36]. Although several clinical reports revealed influenza as an independent risk factor for IPA and a case definition of IAPA was published [1,4,5], animal models for superinfection by Aspergillus spp. were lacking. We provide a robust mouse model that corroborates our earlier retrospective data on IAPA in critically ill patients with severe influenza and confirm that IAPA develops in immunocompetent mice even without corticosteroids [1]. In line with the findings from Tobin et al. (2020) [38]. In our model, immunocompetent mice showed severe body weight loss and clinical deterioration when first infected with influenza virus and superinfected 4 days later with A. fumigatus conidia, whereas single infection with influenza or Aspergillus alone generated only transient disease. High titers of virus and fungus were present at 2 days after co-infection, pointing to an important link between high viral replication and increased susceptibility to IPA, similarly observed in other influenza models with fungal or bacterial co-infection [35,37].

Immunocompetent mice can spontaneously clear In line with the findings from Tobinet al. (2020) [38]A. fumigatus conidia, with complete restoration of lung tissue architecture, although proinflammatory cytokines, neutrophil infiltrates, Aspergillus conidia and CFUs were reported to be present during the first three days of infection [39-41]. By using longitudinal imaging, we further uncovered the disease progression in co-infected mice. Notable multifocal lesions were present over the course of influenza infection, which worsened into consolidation of total lung lobes and high fungal burden after co-infection with A. fumigatus. Corresponding to the lesions on µCT, severe diffuse acute purulent mainly neutrophilic bronchitis and pneumonia with epithelial erosion were discovered with histopathology. Despite the immunocompetent status, all influenza virus pre-infected mice developed proven airway invasive aspergillosis when co-infected with A. fumigatus on day 4, in sharp contrast to the animals infected with A. fumigatus only, which proved fully capable of clearing the fungus. Hence, we experimentally corroborate influenza as a strong risk factor for developing IPA, which is independent of other immunosuppressive regimens or conditions.

There is ample evidence from clinical studies that early oseltamivir treatment improves the outcome in patients with severe influenza, by reducing the duration of hospitalization, complications, and hospital mortality [17-19,42,43]. The importance of early influenza treatment was also seen in a ferret model [44]. Likewise, oseltamivir is capable of reducing the viral load and the severity of lung infection/lung damage [20,21]. On top of this evidence, we observed that oseltamivir not only reduces the severity of influenza, but also mitigates the susceptibility to IAPA. As far as we know this has never been demonstrated so far. However, a pre-requisite is that oseltamivir started early enough. This conclusion is supported by our comprehensive clinical, radiological, pathophysiological and microbial analysis. First, early oseltamivir fully prevented the fatal outcome of IAPA. Secondly, antiviral therapy resulted in less pulmonary infiltrates, less severe influenza pneumonia and smaller lesions on µCT after A. fumigatus co-infection. Thirdly, the increase in fungal burden was strongly (or even completely) inhibited in mice that received early oseltamivir. Finally, histopathological examination of lung tissue damage confirmed that the drug-treated mice were able to completely recover from this serious co-infection. To conclude, early oseltamivir administration in mice with sublethal influenza successfully lowers the susceptibility to develop IAPA. Our pre-clinical findings warrant future clinical studies focusing on reducing IAPA incidence, by impacting on influenza severity via oseltamivir.

Regarding Aspergillus clearing, we observed that oseltamivir-treated mice carried macrophages with intracellular conidia of Aspergillus at 3 days after its inoculation. This concurs with an active antifungal immune response where macrophages phagocytose Aspergillus conidia [45]. Remarkably, we did not observe this in non-treated mice suggesting that influenza might cause some immune dysregulation, which is prevented by early treatment with oseltamivir. Influenza is known to induce a strong IFN-γ/Th1 response decreasing the macrophage phagocytic activity in an influenza/Streptococcus pneumoniae model secondary to a downregulation of a scavenger molecule (MARCO) at the cell surface [15]. Considering the histology findings in our model, the latter hypothesis could be appropriate with a severe reduction in phagocytic properties in mice co-infected with influenza and A. fumigatus while this phagocytic function could be restored once the viral burden decreases after oseltamivir treatment. However, our observations argue for a more comprehensive study. The transient presence of Aspergillus conidia inside macrophages also provides an explanation for why bioluminescence signals and CFU counts were positive yet consistently low on day 7, but negative on day 12. Thus, oseltamivir treatment reduces the severity of the infectious pathology and drives/facilitates an active antifungal clearing process.

Importantly, the timing of oseltamivir treatment remains essential. Early treatment reduces influenza disease symptoms accompanied by a decreasing risk of acquiring IAPA. However, once IAPA is present, it is not clear whether neuraminidase inhibition is beneficial or detrimental. There is evidence that blocking neuraminidase activity in a murine model of aspergillosis is detrimental [46,47]. Blocking neuraminidase might have a negative impact on anti-Aspergillus host defense itself [48-50]. However, our results show that this possibly negative effect on the fungal host response will probably be balanced out by the positive effect on influenza.

In conclusion, we established the first successful preclinical imaging-supported double-hit mouse model of IAPA. This model recapitulates and confirms influenza as a risk factor for the development of invasive and potentially fatal pulmonary aspergillosis. We demonstrated that early oseltamivir treatment mitigates the risk to develop IAPA. Besides reinforcing the recommendation to start oseltamivir as soon as possible (within 48–72 h) in case of severe influenza our findings indicate that early onset of oseltamivir might also reduce the high mortality and morbidity associated with IAPA [1]. In a broader context, this double-hit mouse model is the key for future research aimed at unraveling the underlying immune mechanism of IAPA.

Click here for additional data file.