Alveolar regeneration following viral infection is independent of tuft cells

Severe injuries following viral infection cause lung epithelial destruction with the presence of ectopic basal progenitor cells (EBCs), although the exact function of EBCs remains controversial. We and others previously showed the presence of ectopic tuft cells in the disrupted alveolar region following severe influenza infection. Here, we further revealed that the ectopic tuft cells are derived from EBCs. This process is amplified by Wnt signaling inhibition but suppressed by Notch inhibition. Further analysis revealed that p63-CreER labeled population de novo arising during regeneration includes alveolar epithelial cells when Tamoxifen was administrated after viral infection. The generation of the p63-CreER labeled alveolar cells is independent of tuft cells, demonstrating segregated differentiation paths of EBCs in lung repair. EBCs and ectopic tuft cells can also be found in the lung parenchyma post SARS-CoV-2 infection, suggesting a similar response to severe injuries in humans.


Introduction 60
While the deadly "Spanish" influenza pandemic killed approximately 50 million people worldwide during 1918-1920, seasonal influenza (flu) continued to be deadly claiming 291,000 to 646,000 lives each year (CDC, 2017). Moreover, the COVID-19 pandemic caused by SARS-CoV-2 virus has already claimed over 5.9 million lives and infected over 432 million people thus far (JHU, 65 2022). Post-mortem examination of H1N1 influenza-infected lungs revealed extensive tissue remodeling accompanied by the presence of ectopic cytokeratin-5 positive (KRT5 + ) basal cells in the lung parenchyma (Xi et al., 2017). These ectopic basal cells (EBCs) are also present in the mouse parenchyma following infection with modified H1N1 influenza PR8 virus or treatment with a high dose of bleomycin (Kanegai et al., 2016;Vaughan et al., 2015;Xi et al., 2017;Yuan et al., 70 2019; Zacharias et al., 2018;Zuo et al., 2015). Initial studies suggest that these EBCs contribute to alveolar regeneration (Zuo et al., 2015). However, studies from other groups suggest that EBCs do not meaningfully become alveolar epithelial cells, but rather provide structural supports to prevent the lung from collapsing (Basil et al., 2020;Kanegai et al., 2016;Vaughan et al., 2015).
These seemingly contradictory results necessitate a better understanding of EBCs. 75 Tuft cells are a minor cell population critical for chemosensory and relaying immune signals in multiple organs including the intestine and trachea (Howitt et al., 2016;Montoro et al., 2018;von Moltke et al., 2016). In the intestine tuft cells serve as immune sentinels during parasitic infection (Gerbe et al., 2016;Howitt et al., 2016;McGinty et al., 2020b;von Moltke et al., 2016).

Responding to helminth infection, tuft cells release the alarmin interleukin (IL)-25 which activates 80
type 2 innate lymphoid cells (ILC2s) and their secretion of IL-13, initiating type 2 immune response to eliminate parasitic infection (Gerbe et al., 2016;Howitt et al., 2016;von Moltke et al., 2016). Tuft cells were initially identified in the rat trachea over six decades ago (Rhodin and Dalhamn, 1956). However, we just have begun to appreciate their functions in the respiratory system (Bankova et al., 2018;Ualiyeva et al., 2020). Recent single cell RNA sequencing confirmed 85 the presence of tuft cells in the mouse trachea where they express several canonical genes, such as Alox5ap, Pou2f3, and Gfi1b (Montoro et al., 2018). Following repeated allergen challenges, tuft cells expand and amplify the immune reactions (Bankova et al., 2018). Intriguingly, tuft cells were found ectopically present in the lung parenchyma, co-localized with EBCs following PR8 viral infection (Rane et al., 2019). More recently, we reported that ectopic tuft cells were also present 90 in the parenchyma of COVID-19 lungs, and ablation of tuft cells dampens macrophage infiltration at the acute phase following viral infection in a mouse model (Melms et al., 2021). However, the role of tuft cells in lung regeneration following viral infection remains undetermined. It is also unknown what cells generate these tuft cells and what signaling pathways control the derivation.
In this study, we show that EBCs serve as the cell of origin for the ectopic tuft cells present 95 in the parenchyma during viral infection. Upon screening multiple signaling pathways we identified that Notch inhibition blocks tuft cell derivation, while Wnt inhibition significantly enhances tuft cell differentiation from EBCs. We then used multiple mouse models to demonstrate that p63-CreER labeled subpopulations of alveolar type I and 2 (AT1/AT2) cells during regeneration regardless of the presence of tuft cells in the parenchyma. 100

Tuft cells expanded in response to challenges with influenza, bleomycin or naphthalene.
Tuft cells are rarely present in the proximal airways of normal adult mice (Montoro et al., 2018). However, they were present in the distal airways at birth when tuft cells were first detected ( Figure  105 1A and 1B). Tuft cells continued to be present throughout the airways when examined at postnatal (P)10 and P20 but were no longer detected in the terminal bronchiole at P56 ( Figure 1A and 1B).
Notably, tuft cells reappeared in the terminal airways approximately 7 days post infection (dpi) with PR8 virus ( Figure 1C). The number of tuft cells also significantly increased in the large airways at 30 dpi (4.9 ± 1.1 Vs 25.7 ± 3.3 per 1000 epithelial cells) ( Figure 1C). As previously 110 described, tuft cells were present in the EBC area (Melms et al., 2021;Rane et al., 2019). Further analysis revealed that tuft cells were first detected in the EBC area at approximately 15dpi and peaked at 21dpi ( Figure E1A). The numbers of tuft cells also expanded in the large airways following naphthalene (13.4 ± 1.1 per 1000 epithelial cells) or bleomycin challenge (8.6 ± 0.8 per 1000 epithelial cells) ( Figure 1D). In addition, tuft cells were ectopically present within EBCs 28 115 days following challenge with a high dose of bleomycin ( Figure 1D). These findings suggest that tuft cell expansion in the airways is a general response to lung injuries, and that severe injuries induce ectopic tuft cells in the parenchyma.

The ectopic tuft cells in the parenchyma did not proliferate or give rise to other cell lineages 120
Previous studies suggest that Dclk1 + cells serve as progenitor cells in the small intestine and pancreas (May et al., 2008;May et al., 2009;Westphalen et al., 2016). We asked whether the ectopic tuft cells in the lung parenchyma also possess progenitor potential. We therefore performed lineage tracing with a knock-in Pou2f3-CreER mouse line (McGinty et al., 2020b). Tuft cells (Dclk1 + ) in the large airways were specifically labeled with the lineage tracing maker (tdT + ) upon 125 three Tamoxifen (Tmx) injections into Pou2f3-CreER;R26tdT mice ( Figure 2A). Notably, ~92% tdT + cells expressed Dclk1 in the trachea, whereas all tdT + cells expressed Dclk1 in the large intrapulmonary airways (Figures 2A and Supplement file S1B). We then analyzed the mice challenged with PR8 virus. Tmx was continuously given from 14 to 27 dpi and the lungs were analyzed at 60 dpi ( Figure 2B). The majority (~93%) of the ectopic tuft cells were lineage labeled 130 in the lung parenchyma of Pou2f3-CreER;R26tdT mice ( Figure 2B). These labeled tuft cells solitarily distributed within EBCs ( Figure 2B), indicating that they had not undergone expansion.
Consistently, none of the tuft cells were positive for the proliferation marker Ki67 ( Figure 2C).
To further test whether tuft cells give rise to other cell lineages, immunostaining with various cell type markers was performed, including Scgb1a1 (club cells), FoxJ1 (ciliated cells) and Clca3 135 (goblet cells). The results showed that tuft cells did not contribute to other cell types following a chasing period up to 60 days ( Figure 2D). These data suggest that tuft cells in the lung unlikely serve as progenitor cells.

The ectopic tuft cells in the parenchyma originated from EBCs 140
Basal cells serve as the origin of tuft cells in the mouse trachea (Montoro et al., 2018). By contrast, lineage-negative epithelial stem/progenitor cells (LNEPs) were considered as the origin of tuft cells in the lung parenchyma following viral infection (Rane et al., 2019). While characterizing ectopic tuft cells in the parenchyma, we noticed approximately 5% tuft cells co-expressed Dclk1 and the basal cell marker Krt5 ( Figure 2E), suggesting that these tuft cells are in a transitioning 145 state from EBCs towards tuft cells. To test this hypothesis, p63-CreERT2;R26tdT mice were infected with PR8 virus. Given that tuft cells were initially detected at around 15dpi, we injected Tmx daily from 14 dpi to 18 dpi ( Figure 2F). The majority of tuft cells within EBCs were labeled with tdT ( Figure 2F), suggesting that EBCs serve as the cell origin for the ectopic tuft cells. EBCs can be derived from multiple cell sources including a subpopulation of Scgb1a1 + club cells (Yang 150 et al., 2018). To test whether EBCs derived from the club cell subpopulation give rise to tuft cells, we lineage labeled club cells before exposing Scgb1a1-CreER;R26tdT mice to PR8 virus. 21.0% ± 2.0 % tuft cells within EBCs are labeled with tdT when examined at 30 dpi ( Figure 2G), suggesting that club cell-derived EBCs generate a portion of tuft cells in the lung parenchyma upon viral infection. 155

Inhibition of Wnt and Notch signaling has opposite effects on the differentiation of EBCs into tuft cells
We next sought to identify the signaling pathways that can influence tuft cell differentiation from EBCs. Lineage-labeled pod cells were isolated and purified from the lungs of p63-CreER; R26tdT 160 mice following viral infection and were expanded using the protocol for culturing basal cells  (Supplement file S2A). The expanded EBCs maintained the lineage tag (tdT + ) and expressed the basal cell markers p63 and Krt5 (Supplement file S2B). In addition, they also expressed the respiratory protein markers Nkx2.1 and Foxa2 (Supplement file S2B). We then used the expanded EBCs to assess the impact of major signaling pathways which have been implicated 165 in the determination of cell fate during lung development. The tested pathways include Tgfß/Bmp, Yap, Shh, Wnt, Notch, IL-6 and IL-4/IL-13 (Ahdieh et al., 2001;Barkauskas et al., 2013a;Ikonomou et al., 2020;Lee et al., 2014b;Li et al., 2017;Nabhan et al., 2018;Pardo-Saganta et al., 2015;Tadokoro et al., 2014;Vaughan et al., 2015;Yuan et al., 2019;Zacharias et al., 2018). We observed that treatment with IL-4 or IL-13 apparently promoted tuft cell derivation from EBCs in 170 air-liquid interface (ALI) culture (data not shown), in agreement with the previous reports of IL-4Rα-dependent tuft cell expansion from the intestinal crypts (Gerbe et al., 2016;von Moltke et al., 2016). Notably, treatment of the WNT signaling activator CHIR9902 blocked the derivation of tuft cells by 35.3% ± 3.1%. Conversely, treatment with the Wnt inhibitor IWR-1 promoted tuft cell differentiation by 32.4% ± 6.4 % (Supplement file S2C). Inhibition of WNT signaling with 175 the Porcupine inhibitor Wnt-C59 also led to increased tuft cell differentiation of airway basal cells isolated from Dclk1-GFP mice ( Figure 3A). Consistently, daily injection of Wnt-C59 induced abundant tuft cells in both airways and lung parenchyma following viral infection ( Figure 3B and Supplement file S2D). By contrast, Notch inhibition with the γ-secretase inhibitor Dibenzazepine (DBZ) reduced tuft cell differentiation from EBCs in ALI culture ( Figure 3C). Daily DBZ injection 180 also decreased the number of tuft cells in the parenchyma by 63.6% ± 1.5% following viral infection ( Figure 3D). Together these findings suggest that inhibition of the WNT signaling pathway promotes while Notch inhibition blocks the differentiation of EBCs into tuft cells.

Enhanced generation of p63-CreER lineage labeled alveolar epithelium in Trpm5 -/but not 185
Pou2f3 -/mutants Controversy remains regarding the contribution of EBCs to alveolar regeneration (Vaughan et al., 2015;Zuo et al., 2015). Given tuft cell ablation reduces macrophage infiltration in PR8-infected mouse lungs (Melms et al., 2021), we asked whether loss of tuft cells improved alveolar regeneration. The first mouse model we examined was Trpm5 null mutant which demonstrates 190 ~80% reduction of tuft cells in the intestine (Howitt et al., 2016). However, the number of tuft cells were not significantly reduced following viral infection in the lung parenchyma of p63-CreER;Trpm5 -/-;R26tdT mice as compared to control mice (p63-CreER;Trpm5 +/-;R26tdT) (p>0.05, Supplement file S3A). Furthermore, we did not observe an apparent reduction in the size of Krt5 + EBC clones in the lung parenchyma (Supplement file S3B). Notably, 19.4 % ± 4.2% p63-CreER 195 lineage-labeled cells expressed the AT2 cell marker SftpC but not Krt5 as compared to the control (7.9% ± 1.4%) when Tmx was injected after viral infection (Figures 4A and Supplement file S3C).
Lineage labeled AT2 cells were further confirmed by Abca3, another mature AT2 cell marker (Supplement file S3D). In addition, lineage-labeled cells also include AT1 epithelium (Hopx + , T1a + ) which was increased in the parenchyma of p63-CreER;Trpm5 -/-;R26tdT mutants (6.5% ± 200 0.9% Vs 2.5% ± 0.8% at 60 dpi) ( Figure 4B and Supplement file S3E). To assess whether Trpm5 -/mice exhibited improved pulmonary mechanics after viral infection, we assessed airway resistance when challenged with increasing concentrations of methacholine. Total respiratory system resistance (Rrs) and central airway resistance (Rn) were significantly attenuated in Trpm5 -/mice ( Figure 4C). 205 We next examined the contribution of p63-CreER lineage-labeled cells to lung regeneration in another mutant (Pou2f3 -/-) which has no tuft cells . We subjected p63-CreER;Pou2f3 -/-;R26tdT mutants and controls (p63-CreER;Pou2f3 +/-;R26tdT) to viral infection followed by continuous Tmx injection from 14 to 18 dpi. While prominent bronchiolization 210 occurred in the parenchyma with the extensive presence of Krt5 + EBCs ( Figure 5A), approximately 8% of the lineage-labeled cells became Sftpc + Krt5 -AT2 cells in the controls ( Figure 5B). However, we did not detect improved contribution of lineage-labeled alveolar epithelium in the mutants as compared to the controls (p>0.05) ( Figure 5B), suggesting that loss of ectopic tuft cells has no impact on alveolar regeneration. 215 increased the derivation of tuft cells from EBCs, which is in contrast to the findings from the cosubmitted manuscript where deletion of IL-4 had no impact on tuft cell differentiation. This could be due to the extra IL-4/IL-13 we supplied to the cell culture, which may not be present in an in vivo setting during viral infection. Re-expression of the transcription factor Sox9 has been observed during the repair of the airway epithelium following naphthalene challenge (Jiang et al., 255 2021). Interestingly, Sox9 is expressed by tuft cells as shown by the accompanied manuscript. It will be interesting in future experiments to determine whether Sox9 is required to promote the generation of tuft cells.
However, other groups demonstrated that EBCs provide structural support to prevent the collapse of severely damaged lung without meaningful contribution to regeneration (Kanegai et al., 2016;Vaughan et al., 2015). Here, we used lineage tracing to demonstrate that p63-CreER labeled cells include AT1 and AT2 cells regardless of the mouse lines with/without tuft cells (p63-265 CreER;R26dtT;Trpm5 +/-, p63-CreER;R26dtT;Trpm5 -/-, p63-CreER;R26dtT;Pou2f3 +/-, p63-CreER;R26dtT;Pou2f3 -/-). EBCs have been reported to derive from multiple sources including Krt5 + basal cells (Vaughan et al., 2015;Zuo et al., 2015). In those studies, and the study accompanying this manuscript, Tmx was administrated prior to viral infection, presumably labeling preexisting basal cells with the Krt5-CreER mouse line. Lineage labeled cells did not 270 contribute meaningfully to the regenerated alveolar cells. By contrast, we performed Tmx injection after viral infection and observed significant contribution of p63-CreER lineage labeled cells to the alveolar epithelium. It is possible that we labeled bona fide newly generated cell populations that expressed p63, although the cell origin for these cells remains to be determined. In addition to the preexisting basal cells, subpopulations of EBCs can be derived from Scgb1a1 + cells following 275 viral infection (Yang et al., 2018). More recently, AT2 cells were also found to generate basal cells in vitro (Kathiriya et al., 2022). It is also possible that our strategy may label these club cells or AT2 cell-derived EBC subpopulations during the transition. In addition, at the early stage of mouse lung development (embryonic day 10.5), lineage-labeled p63 + progenitor cells give rise to AT1 and AT2 cells (Yang et al., 2018). Therefore it is also possible that subpopulations of AT1/AT2 280 cells regain the transcription program of fetal lung progenitors and transiently express p63 prior to becoming alveolar cells (Yang et al., 2018).
We noticed an increased contribution of p63-CreER labeled cells to alveolar cells in Trpm5 -/but not Pou2f3 -/mutants which lack tuft cells, suggesting that lung regeneration is independent of the presence of ectopic tuft cells in the parenchyma. Trpm5 is a calcium-activated channel 285 protein that induces depolarization in response to increased intracellular calcium (Prawitt et al., 2003). It is also expressed in B lymphocytes in addition to tuft cells (Sakaguchi et al., 2020). We

Administration of tamoxifen
Tamoxifen was dissolved in sunflower seed oil to 20 mg/mL as stock solution. For lineage analysis 310 and genetic targeting, Pou2f3-CreER mouse strain were administered 2 mg tamoxifen by oral gavage at days 0, 2, and 4 for control or 1 mg tamoxifen by oral gavage at days 14, 17, 20, 23, 25, 27 post viral infection. p63-CreER mice were administered with tamoxifen at a dose of 0.25 mg/g bodyweight by daily oral gavage for a total five doses as indicated. For Scgb1a1-CreER mice, a period of ten or twenty-one days as indicated was used to wash out the residual tamoxifen before 315 any further treatments.

For influenza virus infection, 260 plaque forming units (pfu) of influenza A/Puerto Rico/8/1934
H1N1 (PR8) virus were dissolved in 40 µl of RPMI medium and then pipetted onto the nostrils of 320 anesthetized mice, whereupon mice aspirated the fluid directly into their lungs. Post procedure, mice were weighed weekly and monitored for mortality rate. For bleomycin injury, mice were anesthetized and intratracheally instilled with 4 unit/kg body weight of bleomycin hydrochloride.
For naphthalene treatment, naphthalene solution (25mg/ml) was freshly prepared before the procedure by dissolving naphthalene in sunflower seed oil. A single dose of naphthalene was 325 delivered by intraperitoneal injection at 275 mg/kg body weight. For all procedures listed above, the administration of the same volumes of vehicle (PRMI medium or saline or sunflower seed oil) was used as control.

Mouse EBC isolation, culture and differentiation 330
p63-CreER; R26tdT mice were infected with PR8 influenza virus and were administered with tamoxifen intraperitoneally as indicated. At 18dpi mouse peripheral lungs were dissected and dissociated according to the protocol as previously described (Barkauskas et al., 2013b;Rock et al., 2011). tdT + cells were sorted by FACS and cultured using the protocol as previously reported , and the protocol for inducing the differentiation of basal-like cells was 335 previously described (Feldman et al., 2019;. Air-liquid interface (ALI) culture was used to test the effects of the major signaling pathway inhibitors on the differentiation of EBCs towards tuft and mucous cell lineages. Moreover, tracheal basal cells isolated from Trpm5-GFP mice were cultured and tested for drug effects in ALI. The tested inhibitors include the BMP signaling inhibitor Dorsomorphin (5 µM), YAP signaling inhibitor verteporfin (100 nM), NOTCH 340 signaling inhibitor DBZ (2 µM), SHH signaling inhibitor GDC-0449 (1 µM), WNT signaling inhibitor IWR-1 (5 µM), WNT signaling activator CHIR99021 (2 µM) which inhibits glycogen synthase kinase (GSK) 3, IL-6 (10 ng/ml), IL-4 (10 ng/ml) and IL-13 (10 ng/ml) were also used to treat ALI culture of EBCs.

Treatment of mice with the Porcupine inhibitor Wnt-C59 and the γ secretase inhibitor DBZ
Wnt-C59 was resuspended by sonication for 20 minutes in a mixture of 0.5% methylcellulose and 0.1% tween-80. Wnt-C59 (10 mg/kg body weight) or vehicle was administrated via oral gavage from day 14 to 29 post viral infection (n=4 for each group). For DBZ administration, either vehicle or DBZ was administered intranasally at 30 mmol/kg body weight (n=4 per group) from day 14 to 350 29 post viral infection. DBZ was suspended in sterile PBS mixed with 2.5 µg/g body weight dexamethasone.

Tissue and ALI culture processing and immunostaining
Human and mouse lung tissues were fixed in 4% paraformaldehyde (PFA) at 4℃ overnight and 355 then dehydrated and embedded in paraffin for sections (5-7 µm). ALI culture membranes were fixed with 4% PFA for direct wholemount staining or were embedded in OCT for frozen sections.
All slides and membranes were stained following the protocol reported previously (Feldman et al., 2019;. The primary antibodies used for immunostaining are listed in the Resources Table. 360

Pulmonary function assessment
Trpm5 +/and Trpm5 -/mice (21 days after viral injury) were anesthetized with pentobarbital (50 mg/kg, i.p.). After surgical anesthesia was achieved, a tracheotomy was performed for the insertion of an 18G cannula, and mice were immediately connected to a flexiVent (SciReq,365 Montreal, QC, Canada) with an FX1 module and an in-line nebulizer. Body temperature was monitored and maintained with a thermo-coupled warming blanket and rectal temperature probe.
Heart rate was monitored by EKG (electrocardiography). Mice were mechanically ventilated with a tidal volume of 10 mg/kg, frequency of 150 breaths/min, and a positive end expiratory pressure of 3 mmHg. Muscle paralysis was achieved with succinylcholine (10 mg/kg, i.p.) to prevent 370 respiratory effort and to eliminate any contribution of chest wall muscle tone to respiratory measurements. By using the forced oscillation technique, baseline measurements of lung resistance (Rrs and Rn, representing total and central airway resistance) were performed. Resistance was then measured during nebulization of increasing concentrations of methacholine (10 second nebulization, 50% duty cycle). Methacholine dissolved in PBS was nebulized at 0, 6.25, 12.5, 25 375 and 50 mg/ml and resistance (Rrs and Rn) was measured after each concentration. Values for all measurements represent an average of triplicate measurements. Statistical significance was established by comparing the area under the curve for each mouse.

COVID-19 lung specimens 380
The lung specimens from deceased COVID-19 patients with short post-mortem interval (PMI) (2.5-9hrs) were obtained from the Biobank at Columbia University Irving Medical Center. All experiments involving human samples were performed in accordance with the protocols approved by the Institutional Review Boards at Columbia University Irving Medical Center.

Microscopic imaging and quantification
Slides were visualized using a ZeissLSM T-PMT confocal laser-scanning microscope (Carl Zeiss).
The staining of cells on culture dishes and the staining on transwell membranes were visualized with the Olympus IX81 inverted fluorescence microscope. For quantification of lineage tracing, the lung sections were tiled scanned with 20X images from at least three mice for each genotype. 390 Cells were counted from at least five sections per mouse including at least three individual lung lobes. The production of various airway epithelial cell types was counted and quantified on at least 5 random fields of view with a 10X or a 20X objective, and the average and standard deviation was calculated.

Quantification and statistical analysis
Data are presented as means with standard deviations of measurements unless stated otherwise (n≥3). Statistical differences between samples are assessed with Student two-tailed T-test. P-values below 0.05 are considered significant (*p<0.05, **p<0.01, ***p<0.001).