Pneumonia initiates a tauopathy

Abstract

Pneumonia causes short- and long-term cognitive dysfunction in a high proportion of patients, although the mechanism(s) responsible for this effect are unknown. Here, we tested the hypothesis that pneumonia-elicited cytotoxic amyloid and tau variants: (1) are present in the circulation during infection; (2) lead to impairment of long-term potentiation; and, (3) inhibit long-term potentiation dependent upon tau. Cytotoxic amyloid and tau species were recovered from the blood and the hippocampus following pneumonia, and they were present in the extracorporeal membrane oxygenation oxygenators of patients with pneumonia, especially in those who died. Introduction of immunopurified blood-borne amyloid and tau into either the airways or the blood of uninfected animals acutely and chronically impaired hippocampal information processing. In contrast, the infection did not impair long-term potentiation in tau knockout mice and the amyloid- and tau-dependent disruption in hippocampal signaling was less severe in tau knockout mice. Moreover, the infection did not elicit cytotoxic amyloid and tau variants in tau knockout mice. Therefore, pneumonia initiates a tauopathy that contributes to cognitive dysfunction.

Keywords: dementia; extracorporeal membrane oxygenation; long-term potentiation; lung; prion disease.

FIGURE 1

Pneumonia elicits a hyperdynamic circulatory state coincident with tau oligomer distribution to peripheral organs and suppressed long‐term potentiation. A, P aeruginosa was introduced directly into the airway of animals and peripheral organ function was tested 24 and 48 hours later. This primary infection is designated by a “1”. B, P aeruginosa increased left ventricular ejection fraction at 24 hours (closed squares) and 48 h (open squares) post‐infection, characteristic of a hyperdynamic circulatory state (P = .002 using Student's t‐test; left‐hand panel). Data represent mean ± SD. Representative ultrasound images of diastole and systole in the long axis parasternal view are shown in control and infected animals (right‐hand panel). Ao = aorta; LA = left atrium; LV = left ventricle; RV = right ventricle. C, Representative ultrasound assessment of the lung shows the pleural line (red arrows) and hyperechoic regularly spaced horizontal repetitive lines (A lines) in control lungs (top panel). Irregular focal hyperechoic opacities distorting the pleural line are seen in infected subjects, characteristic of lung consolidation (white arrowheads). Edema and alveolar filling is evident from the hyperechoic vertical lines that obscure the normal A lines (highlighted with the red arrowheads). The B lines were consistent with the heterogeneous consolidation and lung injury seen on gross inspection (bottom panel; arrows). D, Sarkosyl extraction of plasma, heart and brain was performed and the resulting detergent insoluble fraction was probed for oligomeric tau using the T22 antibody. A representative western blot illustrates high T22‐immunoreactive tau variants 48 hours following infection, with little immunoreactivity in an uninfected control. E, The normalized field excitatory post synaptic potential (fEPSP) slope was measured at the Schaffer collateral synapses in control and infected animals over a 60‐minute time course (left‐hand panel). A theta burst stimulation (TBS) was used to initiate long term potentiation (LTP). The dashed line represents the baseline values at time 0. Each group was comprised of 3 separate animals and LTP was recorded from 4‐5 hippocampal slices per animal. The distribution of hippocampal slice recordings from different animals is plotted in the right‐hand panel. Whereas LTP was not reduced 24 hours following infection (p = ns vs control), it was decreased by ~50% 48 hours following infection (P = .0003 vs control and P = .017 vs 24 hours). Summary data were quantified from the last 5 minutes of the LTP response. Statistical differences were determined using one‐way ANOVA with a Tukey's post hoc test

FIGURE 2

Amyloid and oligomeric tau variants present in the circulation post‐infection cause end organ injury when introduced into the airways of naïve animals, that is, uninfected animals. A, Amyloid and tau species were captured from the plasma 48 hours after infection; the control and infected subjects are designated as “1”. The amyloid and tau present in 1% of the plasma volume from these animals was introduced into the airway of a naïve subject, designated as “2”, and peripheral organ function was tested at the indicated time points. B, Left ventricular ejection fraction was similar in control and experimental groups. Two groups comprised control subjects, including uninfected animals from which the A11 and T22 antibodies were used to capture circulating amyloid and tau (●) and infected animals from which an IgG antibody was used to non‐selectively capture molecular species (○). Control animals were studied 48 hours after amyloid and tau introduction into the airway. Left ventricular ejection fractions were not different between these antibody controls and the uninfected control experiments that were reported in Figure 1 and are shown as a dashed line (p = ns using Student's t‐test). Whereas amyloid and tau variants isolated from infected animals [24 hours (■), 48 hours (□), and 1 mo (■)] did not negatively impact left ventricular ejection fraction relative to antibody controls (p = ns), these values were significantly lower than the uninfected control animals (P = .007). Statistical differences were examined using one‐way ANOVA with a Tukey's post hoc test and a Student's t‐test. C, Representative ultrasound images of lungs from animals receiving amyloid and tau retrieved by A11 and T22 antibodies in both control and infected animals. Control and experimental animals demonstrate the usual pleural line (red arrows) and A lines (top panel). B lines were not commonly observed in these experiments. However, intermittent evidence for lung involvement was seen on gross morphology (arrows). D, The normalized fEPSP slope was measured at the Schaffer collateral synapses in animals receiving amyloid and tau isolated from control and infected subjects over a 60‐minute time course (left‐hand panel). A TBS was used to initiate LTP. The dashed line represents the baseline values at time 0. The control group included three animals inoculated with amyloid and tau collected by A11 and T22 antibodies in uninfected animals and three animals inoculated with molecular species collected using a non‐specific IgG antibody in infected animals, each at the 48 hours time point. Controls were compared to an experimental group of animals that were inoculated with amyloid and tau collected by A11 and T22 antibodies from infected animals (n = 3). LTP was recorded from 4‐5 hippocampal slices per animal. The distribution of hippocampal slice recordings from different animals is plotted in the right‐hand panel. LTP was reduced at 24 hours (P =.001), 48 hours (P = .0001) and 1 mo (P = .0001) following amyloid and tau inoculation. Summary data were quantified from the last 5 minutes of the LTP response. Statistical differences were determined using one‐way ANOVA with a Tukey's post hoc test

FIGURE 3

Amyloid and oligomeric tau variants present in the circulation after their intratracheal inoculation elicit a replicating injury in the absence of bacteria. A, Amyloid and oligomeric tau were captured from the plasma fraction of blood in the abdominal aorta 1 mo after amyloid and tau species were inoculated in the airways; this series of inoculations is designated as “2”. The amyloid and tau present in 1% of the plasma volume were introduced into the airway of a naïve animal, designated as “3”, and peripheral organ function was tested at the indicated time points. Note that control animals were not initially infected (“1—uninfected control” → “2—amyloid and tau inoculation”) whereas the experimental group was initially infected with P aeruginosa (“1—P aeruginosa infection” → “2—amyloid and tau inoculation” → “3—amyloid and tau inoculation”). B, Because amyloid and tau did not impair organ function in control subjects, they were not collected from the 2nd control animals and inoculated into the 3rd animals in series; control values reported in Figure 2B are shown by a dashed line. Infection‐derived amyloid and tau that were collected from the 2nd animals and introduced through the airway into the 3rd animals in series elicited a non‐significant (P = .10) reduction in left ventricular ejection fraction when considering all time points together [24 hours (■), 48 hours (□), and 1 mo (■)], although the 48 hours (P = .02) and 1 mo (P = .03) time points were both reduced when compared to controls. Statistical differences were determined using one‐way ANOVA with a Tukey's post hoc test and a Student's t‐test. C, Representative lung ultrasound images are shown of subjects receiving intratracheal delivery of amyloid and tau collected from the plasma by A11 and T22 antibodies in control‐ and amyloid‐ and tau‐treated animals. The control animal demonstrates the usual pleural line (red arrows) and A lines (top panel), whereas B lines were seen in the animal receiving amyloid and tau post‐infection (highlighted with the red arrowheads). The B lines were consistent with the heterogeneous and minor lung injury (arrows) seen on gross inspection (bottom panel). D, The normalized fEPSP slope was measured at the Schaffer collateral synapses in animals receiving amyloid and tau from an infection series (“1—P aeruginosa infection” → “2—amyloid and tau inoculation” → “3—amyloid and tau inoculation”) over a 60‐minute time course (left‐hand panel). A TBS was used to initiate LTP. The dashed line represents the baseline values at time 0. The control groups are representative of data plotted in Figure 1E (ｰｰｰｰ) and Figure 2E (……). Controls were compared to an experimental group of animals that were inoculated with amyloid and tau collected by A11 and T22 antibodies from animals (n = 3) that received intratracheal delivery of amyloid and tau (“1—P aeruginosa infection” → “2—amyloid and tau inoculation” → “3—amyloid and tau inoculation”). LTP was recorded from 4‐5 hippocampal slices per animal. The distribution of hippocampal slice recordings from different animals is plotted in the right‐hand panel. LTP was reduced 24 hours (P = .023), 48 hours (P = .034) and 1 mo (P = .017) following amyloid and tau inoculation. Summary data were quantified from the last 5 minutes of the LTP response. Statistical differences were determined using one‐way ANOVA with a Tukey's post hoc test

FIGURE 4

Amyloid and tau in ECMO oxygenator effluent of pneumonia patients are cytotoxic. A, Amyloid and tau captured from ~0.7% of the membrane oxygenator effluents (designated “1”) using A11 and T22 antibodies were introduced into the airway of previously uninfected subjects (designated “2”). Controls included amyloid and tau collected from the membrane oxygenator effluent of patients 2, 7, and 10, none of whom harbored infection during extracorporeal support. Experimental samples were obtained from patients 1, 4, 9, and 11, all of whom had ongoing pneumonia at the time of ECMO oxygenator decannulation (see also Table S1). B, Amyloid and tau collected from the membrane oxygenator effluent of uninfected patients had no effect, whereas those isolated from the membrane oxygenator effluent of infected patients led to a time‐dependent decrease in left ventricular ejection fraction, where 1 mo values (■) were lower than those at both 24 hours (■; P = .02) and 48 hours (□; P = .01). Statistical differences were determined using Student's t‐test and one‐way ANOVA with a Tukey's post hoc test. C, Representative lung ultrasound images are shown. The control animal demonstrates the usual pleural line (red arrows) and A lines (top panel). Lung consolidation (white arrowheads) with B lines characteristic of edema (highlighted with the red arrowheads) is shown. The B lines were consistent with the lung injury (arrows) seen on gross inspection (bottom panel). D, The normalized fEPSP slope was measured at the Schaffer collateral synapses in animals receiving amyloid and tau from the membrane oxygenator effluent of uninfected and infected subjects over a 60‐minute time course (left‐hand panel). A TBS was used to initiate LTP. The dashed line represents the baseline values at time 0. Whereas LTP was normal in control subjects, it was significantly reduced at all time points in the experimental groups (24 hours, P = .0001; 48 hours, P = .0001; and, 1 mo, P = .0001). LTP was recorded from 4‐5 hippocampal slices per animal. The distribution of hippocampal slice recordings from different animals is plotted in the right‐hand panel. Summary data were quantified from the last 5 minutes of the LTP response. Statistical differences were determined using one‐way ANOVA with a Tukey's post hoc test

FIGURE 5

Infection‐induced amyloid and tau cytotoxicity requires tau. A, Lung capillaries express tau. Fluorescence of wild type (left panel) and tau knockout (KO; right panel) mice were tested. Cell nuclei were stained with NucBlue and the endothelium was labeled with tomato lectin (Lycopersicon esculentum) through the circulation. The circulation and airways of wild type and knockout mice were then filled with gelatin and agarose, respectively. The B6.129S4(Cg)‐Mapt ^{tm1(EGFP)/Klt}/J mouse harbors a tau knockout with an EGFP insertion in exon 1 of the tau locus so that tau‐expressing cells can be visualized. Three hundred micron‐thick lung sections were cut, and the alveolus was imaged by high resolution confocal microscopy. Lung capillaries exhibited green, that is, EGFP, and red, that is, tomato lectin, fluorescence, indicating that they express endothelial tau under basal conditions. Yellow arrows show areas of overlap between the tau reporter and tomato lectin labeling of capillary endothelium. B, P aeruginosa was introduced directly into the airway of wild type and tau knockout mice and long‐term potentiation was measured 48 hours later. C, The normalized fEPSP slope was measured at the Schaffer collateral synapses in control and infected animals over a 60‐minute time course (left‐hand panel). A TBS was used to initiate LTP. The dashed line represents the baseline values at time 0. Each group was comprised of 3 separate animals and LTP was recorded from 4‐5 hippocampal slices per animal. The distribution of hippocampal slice recordings from different animals is plotted in the right‐hand panel. LTP was similar in uninfected wild type and tau knockout mice (p = ns). Whereas LTP was reduced following infection of wild type mice (P = .0001), it was not decreased following infection of tau knockout mice (p = ns). Summary data were quantified from the last 5 minutes of the LTP response. Statistical differences were determined using one‐way ANOVA with a Tukey's post hoc test. D, Amyloid and tau were isolated from the plasma of wild type and tau knockout mice following infection, using the A11 and T22 antibodies. Amyloid and tau were then introduced into wild‐type animals and long‐term potentiation was measured 48 hours later. E, The normalized fEPSP slope was measured at the Schaffer collateral synapses in control and infected animals over a 60‐minute time course (left‐hand panel). A TBS was used to initiate LTP. The dashed line represents the baseline values at time 0. Each group was comprised of 3 separate animals and LTP was recorded from 4‐5 hippocampal slices per animal. The distribution of hippocampal slice recordings from different animals is plotted in the right‐hand panel. LTP was reduced in animals receiving amyloid and tau from wild type mice (P = .002), but it was not decreased following introduction of amyloid and tau from tau knockout mice (p = ns). Summary data were quantified from the last 5 minutes of the LTP response and statistical significance determined using a Student's t‐test and a Mann–Whitney test. F, Amyloid and tau collected from ECMO patient 11 (M morganii pneumonia) were introduced into the tail vein of wild type and tau knockout mice. Forty‐eight h later LTP was measured. Whereas the amyloid and tau inhibited LTP in wild type animals, LTP was preserved in tau knockout animals (n = 2 wild type and tau knockout animals analyzing 14 hippocampal brain slice recordings; P = .003 by Mann‐Whitney test)

FIGURE 6

Pneumonia elicits the production of cytotoxic amyloid and tau within the lung. Lung infection (red and green rods represent bacteria and the green sphere represents a virus) promotes endothelial production of oligomeric tau. This oligomeric tau can be detected in the airways, circulation, cerebrospinal fluid, and the brain. Pneumonia leads to impairment of long‐term potentiation in the hippocampus, and this effect requires tau

FIGURE 7

Pneumonia‐induced oligomeric tau (red) accesses the brain where it impairs LTP in the hippocampus. Oligomeric tau is generated by endothelium within the lung and disseminates through the circulation. Important questions remain unanswered regarding whether oligomeric tau: (1) is transported across the blood–brain barrier, increases blood–brain barrier permeability, or both (astrocyte is green, pericyte is orange, endothelium is light red); (2) induces amplification by proteopathic seeding in the circulation or the brain (microtubule is gray); and, (3) interacts with other potentially injurious biomolecules, like some forms of heparan sulfates/proteoglycans (bottom, blue) and beta amyloid (top, green), to promote neuroinflammation and cytotoxicity