Freshly isolated normoxic AEC2s are essentially nonproliferative, with greater than 90% of the population in G1 phase of the cell cycle, and remain quiescent in culture (20). In contrast, AEC2s isolated from hyperoxic rats are more proliferative, suggesting that the hyperoxic milieu within the lung is conducive to AEC2 growth. We obtained BAL and lung extracts from hyperoxic rats, taken to be representative of the AEC2’s in vivo damage environment, and compared their effects with BAL and lung extract from normoxic animals when added to 24-hour cultured, attached, nonconfluent normoxic AEC2s. The hyperoxic BAL and lung extracts increased the cell numbers by 2-fold in 24 hours (Figures 1A and 1B). In contrast, normoxic BAL and lung extract did not significantly affect AEC2 proliferation. Thus, the AEC2 cell in vivo hyperoxic milieu is sufficient to recapitulate the proliferation induced by in vivo hyperoxia when added to normoxic AEC2s ex vivo.

Figure 1. Alveolar epithelial cell type 2 (AEC2) damage milieu promotes AEC2 proliferation, wound healing, and migration. (A) Freshly isolated AEC2s from control rats were allowed to attach overnight, incubated a further 24 hours in DMEM with 10% FBS ± normoxic and hyperoxic bronchoalveolar lavage (BAL) or lung extracts, and examined under phase contrast light microscopy at 20× magnification (bar = 250 μm). Hyperoxic BAL or lung extracts induced pronounced proliferation of AEC2s, as confirmed by cell counts in panel B. (B) AEC2 numbers from the conditions described in panel A are expressed as fold increase over untreated. Hyperoxic BAL and lung extract increased cell numbers of AEC2s (P < 0.004 and P < 0.003, respectively; n = 5). The hyperoxic BAL and lung extract were significantly different from normoxic BAL and lung extract (P < 0.001 and P < 0.02, respectively). (C) A confluent AEC2 monolayer was damaged by micropipet tip, washed to remove damaged cells and serum, and incubated in DMEM ± BAL or lung extract from hyperoxic rats. After 24-hour incubation, the cells were stained with crystal violet, and the damage area was photographed at 20× (bar = 250 μm). This composite, which shows that hyperoxic BAL and damaged lung extract promote AEC2 wound healing, is representative of several experiments. Wound width at 24 hours: untreated, 552 ± 109 μm; hyperoxic lung extract–treated, 0 (100% healed); hyperoxic BAL-treated, 212 ± 32 μm (P < 0.04 versus untreated; n = 3). (D) The chemoattraction of AEC2s toward milieu from undamaged and damaged AEC2s was measured by plating AEC2s on porous (8 μm) filters atop wells containing AEC2 in vitro milieu (conditioned medium [CM] from undamaged and scratch damaged AEC2s) or in vivo milieu (normoxic and hyperoxic BAL or lung extract). After 24 hours, migrating cells on the underside of the filter were stained and quantitated by spectrophotometric analysis of eluted cell stain (x axis, A600 nm). Migration toward the AEC2 milieu versus basal migration toward DMEM alone: hyperoxic BAL (n = 7), P < 0.0001; hyperoxic lung (n = 7), P < 0.02; control BAL (n = 6), P < 0.01; and CM from in vitro damaged AEC2 (n = 3), P < 0.01; CM from undamaged AEC2 (n = 3), NS. Hyperoxia significantly increased the chemoattraction of the BAL and lung extracts: hyperoxic versus control BAL, P < 0.002; hyperoxic versus control lung, P < 0.03.

Download Figure | Download Powerpoint

[More] [Minimize]

We used the scratch damage model to determine whether the damage cues present in the microenvironent of damaged AEC2s could potentiate alveolar epithelial repair. Hyperoxic BAL and lung extract promoted significantly faster healing of an AEC2 monolayer scratch wound (Figure 1C). The AEC2 is normally unmigratory (18) but can be induced to migrate by in vivo hyperoxia (19). To determine whether AEC2 cell damage milieu promoted chemotaxis of AEC2s, we used an in vitro Boyden chamber approach (18, 19) to measure migration of AEC2s toward AEC2 damage milieu. In vivo and in vitro AEC2 damage milieu stimulated migration in the normally unmigratory AEC2s (Figure 1D). These migration experiments were also repeated using primary cultures of AEC1s with similar results (data not shown), suggesting that the whole alveolar epithelium is responsive to signals resulting from AEC2 damage.

Because AEC2 in vitro damage milieu stimulated AEC2 migration, we tracked exogenous fluorescent-labeled AEC2s added to a fresh monolayer scratch wound. We added 10⁵ exogenous cells to a damaged monolayer in a 5-cm² tissue culture well, replacing the estimated 5% loss caused by wounding and washing. A wound generated by growing the monolayer around a removable template served as a passive (minimally damaged) denudation control, and exogenous cells were added to nonwounded confluent monolayers in parallel to assess the pattern of engraftment to nondamaged cells. The size of the wounds generated by scratch and template was equivalent (952 ± 208 μm versus 988 ± 33 μm, respectively; n = 7). After 24 hours, exogenous fluorescent AEC2s had efficiently attached to a scratch wound (Figure 2A, left panel) and to a lesser extent to a template wound (Figure 2A, middle panel), where they were lightly attached and more likely to be removed by washing. Exogenous cells added to undamaged monolayers were diffusely scattered over the monolayer with much less efficient attachment (Figure 2A, right panel). The scratch wound had healed by 24 hours in the presence of exogenous AEC2 cells, in contrast to the template wound, which remained unhealed (Figure 2B, left panels). Endogenous wound healing was also more efficient in the scratch wound than in the template wound (Figure 2B, right panels). To determine whether exposed matrix was responsible for the increased wound healing seen when cells were damaged versus denuded, we compared scratch and template wounds immediately after damaging for protein, fibronectin, laminin, and collagen deposition. There were similar small remnants of protein detected in template and scratch wounds, mainly adjacent to the wound edge. Laminin was detected similarly (Figure 2C), but no fibronectin or collagen staining was detected in the wounds or at the wound edges of either damage model (data not shown). Although it is possible that endogenous and exogenous AEC2s are attracted to traces of laminin or other proteins left in the wound after damage, this is unlikely to account for the difference in healing properties between scratch-damage milieu and template-damage milieu because both wound types showed similar trace amounts of protein staining. Taken together, the data suggest that soluble autocrine factors released by damaged AEC2s are sufficient to expedite endogenous repair of a wounded alveolar epithelial membrane but can also attract exogenous cells to augment endogenous wound healing. In contrast, denudation without damage is a less effective stimulus for repair.

Figure 2. Exogenous AEC2s expedite healing at the site of damage in an AEC2 monolayer. (A) Exogenous CM DiI-labeled AEC2s (2 × 10⁴ exogenous cells/cm²) were added to AEC2 monolayers freshly wounded by scratch (active denudation) or template (passive denudation) and to a nonwounded monolayer in parallel and cultured for 48 hours. The initial sizes of wounds were equivalent (952 ± 208 μm scratch versus 988 ± 33 μm template; n = 7). Photographs of live cultures at 48 hours show that exogenous AEC2s attached within the damage area (arrows) of both wounds but with increased deposition in the scratch. In contrast, exogenous cells added to an undamaged monolayer show limited and diffuse attachment (original magnification, ×10; bar = 1,000 μm). (B) Exogenous and endogenous AEC2-mediated wound healing at 24 hours is more efficient in a scratch-wounded versus template wounded AEC2 monolayer. Wound width, scratch versus template at 24 hours: exogenous, 0 (100% healed) versus 535 ± 106 μm; endogenous, 313 ± 53 μm versus 742 ± 130 μm (P < 0.02; n = 4) (original magnification, ×4; bar = 1,000 μm). (C) Small remnants of protein remaining in the wound were seen predominantly near the wound edges of scratch and template wounds that were fixed in 4% paraformaldehyde immediately after damaging and washing. Similar levels of protein (Coomassie Blue staining) and laminin (immunostaining) were detected in both wound types (original magnification, ×10; bar = 400 μm). No collagen (Sirius red) or fibronectin immunostaining was detected (data not shown).

Download Figure | Download Powerpoint

[More] [Minimize]

To examine cytokine/chemokine release into the AEC2 milieu after a breach of alveolar epithelial integrity, we screened conditioned medium from undamaged versus scratch-damaged versus template-damaged AEC2s using a proteomic cytokine array of 30 rat-specific cytokines. CINC-2 and ICAM were the only cytokines significantly elevated in the medium collected from scratch-damaged cells at levels greater than 1.5-fold higher than in control medium (Figure 3A), and this difference had resolved by 48 hours (data not shown). A template wound with minimal cell damage generated 25% less ICAM and 66% less CINC-2 than scratch-damaged wounds. The damage-mediated CINC-2 and ICAM release was confirmed using rat-specific ELISAs (Figures 3A and 3B). CINC-2 was increased 1.5-fold during the first 24 hours of healing. Basal levels of ICAM were relatively high in control AEC2 conditioned medium (CM) (4,942 ± 50 pg/ml), with a 1.6-fold increase in ICAM secretion in the first 3 hours after damage, although we were not able to detect a significant difference by 24 hours after damage. Up-regulation of these cytokines has been reported previously in models of acute lung injury in vivo (21, 22). Indeed, we detected a 3.4-fold increase in CINC-2 and a 1.8-fold increase in ICAM-1 in hyperoxic BAL versus normoxic BAL (Figures 3B and 3C). To demonstrate that CINC-2 and ICAM-1 secretion was a response to damage and not due to proliferation per se, conditioned medium from undamaged proliferating and nonproliferating AEC2s was screened by protein microarray (see Figure E2 in the online supplement). Proliferation was induced by 25 ng/ml KGF treatment for 24 hours and confirmed by PCNA Western blot. There was no difference in CINC-2 expression between proliferating and nonproliferating AEC2s, and ICAM was barely detectable in CM from either population. This is in contrast to the elevated CINC-2 and ICAM levels secreted by damaged AEC2s shown in Figure 3A and confirms that release of CINC-2 and ICAM reflects an autocrine response to cell damage.

Figure 3. CINC-2 and ICAM are released by AEC2 damage and promote cell migration. (A) Cytokines induced by AEC2 damage were detected by screening conditioned medium from AEC2 wounded by scratch or template using a rat specific protein cytokine array of 30 cytokines. The CM was collected 24 hours after damage and compared with 24-hour CM from undamaged cells. Of the seven cytokines detected with high expression in AEC2 CM, only CINC-2 and ICAM were increased by AEC2 damage (scratch wounding) by greater than 1.5-fold. Template wounds, which involve minimal cell damage, did not result in increased CINC and ICAM secretion. (B and C) CINC-2 and ICAM levels were quantitated in CM from undamaged and scratch damaged AEC2s (in vitro damage) and BAL from normoxic and hyperoxic rats (in vivo damage) by rat-specific ELISA. CM was collected 24 hours after damage for CINC-2 (B) and 0 to 3 hours and 3 to 24 hours after damage for ICAM (C). CINC-2 levels were elevated in CM from damaged versus undamaged AEC2s (P < 0.05; n = 3) and in hyperoxic BAL versus control BAL (P < 0.02; n = 3 control BAL, n = 6 hyperoxic BAL). ICAM levels were elevated in damaged AEC2 CM in the first 3 hours after damage (P < 0.02) and in hyperoxic versus normoxic BAL (P < 0.02; n = 3 control BAL, n = 6 hyperoxic BAL). (D) AEC2 migration toward serum-free medium ± ICAM or CINC-2 at levels covering the range seen in damaged AEC2s was measured using 8-μm-pore filters. Migrating cells were stained and quantitated by spectrophotometric analysis of eluted cell stain. Data are expressed as fold increase over medium control (DMEM with 0.1% BSA). ICAM and CINC-2 induced migration at doses of 5 ng/ml (each P < 0.02) and 50 ng/ml (P < 0.01 and P < 0.02, respectively; n = 3).

Download Figure | Download Powerpoint

[More] [Minimize]

We measured in vitro migration of AEC2s through porous 8-μm filters toward DMEM ± CINC-2 and ICAM at concentrations at and above the level seen in damage milieu and found significantly increased migration at 5 ng/ml or greater (Figure 3D). Both cytokines also expedited in vitro wound healing, whereas endogenous wound healing was delayed by their respective neutralizing antibodies (Figures 4A and 4B), confirming a functional role for both cytokines in AEC2 autocrine repair in vitro. CINC-2 and ICAM neutralizing antibodies also delayed wound healing stimulated by AEC2 in vivo damage milieu since delayed wound healing was also seen in the presence of hyperoxic BAL (Figure E3). CINC-2 or ICAM did not promote proliferation of AEC2s when added at the time of plating or to already attached cells for a period of 24–hours, as measured by cell counts. However, when added to freshly damaged AEC2s, both cytokines stimulated proliferation significantly by 24 hours (CINC-2: 1.4 ± 0.1-fold versus untreated [P < 0.03]; ICAM: 1.6 ± 0.3-fold versus untreated [P < 0.04; n = 4]).

Figure 4. CINC-2 and ICAM contribute to endogenous AEC2 wound healing. (A) Confluent cultures of AEC2s were scratch-damaged and incubated for 24 hours with basal medium (DMEM with 0.1% BSA) ± 5 ng/ml CINC-2 or 5 ng/ml ICAM. Optimal doses of cytokines had been determined in previous dose–response experiments. The cultures were then stained with crystal violet and photographed under low power (original magnification, ×10; bar = 500 μm). CINC-2 and ICAM promoted AEC2 wound healing within 24 hours, whereas untreated wounds remained unhealed. (B) Confluent cultures of AEC2s were scratch damaged and incubated with basal medium (DMEM with 0.1% BSA) ± an excess of CINC or ICAM neutralizing antibodies. The control well received antibody vehicle (PBS with 5% trehalose). Incubation time was 48 hours, the time required for endogenous healing. CINC and ICAM neutralization delayed endogenous healing, as shown by noticeably less compacted cells within the wound compared with the untreated wound. (C) Undamaged AEC2s were cultured for 24 hours ± CINC-2 or ICAM (5 ng/ml). The cells were lysed, and the cellular proteins were analyzed by Western blot. Both cytokines resulted in ERK activation at 24 hours, as measured by the appearance of the phospho (active) form, which was not readily detected in untreated cells. Increased expression of the active, phosphorylated form of Smad2 was also apparent after 24-hour treatment with CINC or ICAM (1.8 ± 0.4- and 1.7 ± 0.4-fold basal, respectively, by densitometry; n = 3 blots). Short-term exposure of AEC2s to CINC-2 and ICAM for 5 minutes or 1 hour did not result in significant activation of ERK or Smad-2 (data not shown). (D) Freshly damaged AEC2 monolayers were cultured ± MAP kinase kinase (MAPKK) inhibitor PD 98059 (50 μM) or TGF-β1 receptor kinase inhibitor SB505124 (1 μM) for 2 hours before the addition of CINC-2 and ICAM (5 ng/ml). All wells contained carrier DMSO (0.1% final volume). After incubation for 24 hours, the cultures were stained with crystal violet and photographed under low power (original magnification, ×4; bar = 1,000 μm). Endogenous healing was delayed in the presence of either inhibitor, with less defined wound margins (wound widths: untreated, 250 ± 11 μm; SB 505124–treated, 467 ± 37 μm [P < 0.003 versus untreated]; PD 98059–treated, 413 + 37 μm [P < 0.006 versus untreated]). CINC-2 and ICAM expedited wound closure (CINC-2 wound width, 66.7 ± 63 m [P < 0.05] versus untreated; ICAM wound 88.6 + 44 m [P < 0.02] versus untreated; n = 3), while inhibition of MAP kinase activation with PD 98059, or TGF-β1 receptor kinase activity by SB 505124 decreased CINC-2 and ICAM-mediated wound healing to endogenous levels.

Download Figure | Download Powerpoint

[More] [Minimize]

To determine mechanisms of CINC and ICAM-mediated AEC2 wound healing, we examined the effect of CINC and ICAM on the activation of common signaling pathways in AEC2s. CINC-2 and ICAM were added to fresh isolates of AEC2 for 5 minutes or 1 hour or added at the time of plating for 24-hour culture. The dosage used was 5 ng/ml, which was previously shown to promote migration and expedite wound healing. The cells were lysed after 5 minutes, 1 hour, or 24 hours of exposure to the cytokines. Although short-term exposure to CINC-2 and ICAM did not induce ERK or Smad2 activation (data not shown), both cytokines induced sustained ERK and Smad2 phosphorylation after 24 hours of treatment (Figure 4C). The delayed activation of both pathways may be secondary to cytokine-mediated attachment per se or to induction and release of growth factors or cytokines. In contrast, we could not detect CINC-2– or ICAM-mediated changes in the phosphorylation of the other key signaling members of the MAP kinase family, p38 and JUNK (data not shown). Inhibition of MAP kinase kinase (MAPKK) activation by PD98059 or TGF-β1 receptor kinase activity by SB505124 slowed endogenous wound healing and decreased CINC-2 and ICAM-mediated wound healing to endogenous levels (Figure 4D). These data suggest that CINC- and ICAM-mediated epithelial repair is mediated through indirect stimulation of MAPK and TGF-β signaling.

We have previously shown that in systemically delivered uncommitted human amniotic fluid stem cells, c-kit + hAFSCs can be detected in murine lung epithelium after hyperoxic injury in vivo, albeit in low numbers (7). To see if the milieu of damaged AEC2 per se could provide an attractive environment for uncommitted stem cells, we added CM-DiI–labeled hAFSCs to AEC2 monolayers freshly damaged by scratch or by template. The AEC2 milieu generated by the damage model was more attractive to engraftment of hAFSCs than the passive denudation model since hAFSCs attached more efficiently to a scratch wound than to a template wound (Figure 5A, upper panels). Wound healing at 24 hours in the presence of hAFSCs was also more efficient in the scratch wound versus the template wound (Figure 5A, lower panels).

Figure 5. Human amniotic fluid stem cells (hAFSCs) are attracted to damaged AEC2s and damaged AEC2 milieu. (A) AEC2 monolayers, undamaged or damaged by scratch (active denudation) or template (passive denudation), were cultured for 24 hours with Dil-labeled hAFSCs (upper panels; original magnification, ×10; bar = 500 μm). hAFSCs attached sparsely and randomly to an undamaged monolayer but attached within the scratch damage site and to a lesser extent to the template damage site. The wound healing at 24 hours in the presence of exogenous hASFCs was more efficient in the scratch wound than in the template wound (175 ± 35 μm versus 545 ± 20 μm, respectively; P < 0.001; n = 3), as shown by crystal violet staining (lower panels; original magnification, ×4; bar = 1,000 μm). (B) Migration of hAFSCs toward in vitro (left panel) and in vivo (right panel) AEC2 damage milieu was measured and compared with migration toward medium alone. Migration was expressed as absorbance of crystal violet stain eluted from migrated cells (A_600nm). Photographs show the phenotype of stained cells on filters at ×10 magnification (bar = 500 μm). AEC2 CM versus medium (DMEM + 0.1% BSA), P < 0.002; damaged AEC2 CM versus medium, P < 0.001; damaged versus undamaged CM, P < 0.05; control BAL versus medium, P < 0.02; hyperoxic BAL versus medium, P < 0.0001; hyperoxic BAL versus normoxic BAL, P < 0.0001. Control med, n = 4; AEC2 CM, n = 3; AEC2 dam CM, n = 3; control BAL, n = 3; hyperoxic BAL, n = 4.

Download Figure | Download Powerpoint

[More] [Minimize]

To confirm that the microenvironment of damaged AEC2s was attractive to hAFSCs, we compared the migration of hAFSCs toward AEC2-conditioned medium from undamaged and in vitro scratch-damaged AEC2s as well as BAL from control and hyperoxic animals (Figure 5B, upper panels). hAFSCs migrated toward conditioned medium from in vitro damaged AEC2s with an approximately 5-fold increase over basal medium. The migratory response of hAFSCs toward hyperoxic BAL had a greater than 20-fold increase in migration compared with basal medium and was approximately 10-fold higher than the response to BAL from normoxic animals, which was also stimulatory. The cells migrating toward hyperoxic BAL had a predominantly fibroblast-like phenotype, in contrast to the cells migrating toward damaged AEC2 CM, which had an epithelial phenotype, or control BAL, where tight spheres of small cells were seen, only a few of which had recognizably epithelial and fibroblast morphology (Figure 5B, lower panels). Unlike rat AEC2s, hAFSCs did not migrate significantly toward CINC-2 or ICAM, the major cytokines released by AEC2 damage (data not shown). Thus, the chemotactic signals originating from damaged AEC2 milieu that are attractive to hAFSCs remain to be identified.

hAFSCs can ameliorate tissue damage by modulating cytokines in the damage milieu, as seen in a rodent model of acute tubular necrosis (23). Therefore, we screened for human cytokines released when hAFSCs were incubated in the presence of damaged AEC2 milieu using a human-specific cytokine proteomic screen of 36 cytokines (Figure 6A). Although no human cytokines were detected in damaged AEC2 milieu in the absence of hAFSCs, five human cytokines were seen at readily detectable levels when hAFSCs were cultured with AEC2 in vivo damage milieu: growth regulated protein-α, IL-6 and IL-8, macrophage migration inhibitory factor (MIF), and plasminogen activator inhibitor 1 (PAI-1), whereas MIF and PAI-1, and to a lesser extent IL-6, were detected after culture of hAFSCs with AEC2 in vitro scratch damage medium. PAI-1 and MIF were endogenously secreted by hAFSCs that were cultured in the absence of damage milieu but were strongly induced by AEC2 damage. MIF, a lymphokine involved in cell-mediated immunity, immunoregulation, and inflammation, plays a role in the regulation of macrophage function in host defense and may be involved in integrin signaling pathways. AEC2s proliferate in response to MIF and express high surface levels of CD74, which binds MIF (24). PAI-1 is a major negative regulator of pericelllular plasmin and accompanies wound repair in vitro and in vivo (25), including alveolar epithelial wound healing (26). Because MIF and PAI-1 were likely candidates to promote wound healing and were strongly induced by in vivo and in vitro damage milieu, we focused on these cytokines.

Figure 6. hAFSCs contribute healing cytokines in response to AEC2 damage. (A) A human-specific array of 36 cytokines and chemokines was used to compare the medium from hAFSCs incubated 24 hours in the presence of AEC2 in vivo or in vitro (scratch) damage milieu with AEC2 damage milieu alone. Only five human cytokines from a panel of 36 were detected at high levels. MIF and PAI-1 were highly expressed in the presence of in vitro AEC2 damage. No spots were detected in AEC2 damage milieu in the absence of hAFSCs using the human array. (B) Confluent cultures of AEC2 s were scratch damaged and incubated for 24 hours with medium (DMEM with 0.1% BSA) ± MIF or PAI-1 at 5 ng/ml. The cultures were then stained with crystal violet and photographed under low power (original magnification, ×4; bar = 1,000 μm). MIF and PAI-1 healed the wound in 24 hours, whereas untreated wounds remained unhealed. (C) Confluent cultures of AEC2s were scratch-damaged and preincubated for 2 hours ± MIF antagonist ISO-1 (50 μM) and PAI-1 neutralizing antibody before the addition of exogenous hAFSCs. As ISO-1 was dissolved in DMSO, an equivalent volume was added to control and PAI-1 neutralizing antibody–treated cultures. After 24-hour culture, the cells were stained with crystal violet and photographed (original magnification ×4; bar = 1,000 μm). hAFSCs significantly promoted AEC2 repair: gap width 133 ± 66 μm (+hAFSCs) versus 342 ± 47 μm (−hAFSCs) (n = 3; P < 0.01) but could not promote repair when MIF or PAI-1 were inhibited.

Download Figure | Download Powerpoint

[More] [Minimize]

To confirm a functional role for MIF and PAI-1 in alveolar epithelial wound healing, we added human recombinant MIF and PAI-1 (5 ng/ml) to freshly scratch-damaged monolayers of AEC2s in the absence of hAFSCs. After 24 hours, the scratch wounds treated with MIF and PAI-1 were healed (Figure 6B), in contrast to the control scratch. Proliferation of damaged AEC2s, as measured by cell numbers, was increased 2.1 ± 0.1-fold by MIF and 1.6 ± 0.1-fold by PAI-1, suggesting that expedited wound closure in the presence of the cytokines was due to increased cell proliferation. In contrast, AEC2s did not migrate toward MIF or PAI-1 at doses that promoted wound healing (data not shown). Inhibition of PAI-1 or MIF, using MIF antagonist ISO-1 (50 μM) or an excess of PAI-1 neutralizing antibody, delayed hAFSC-mediated AEC2 wound healing (Figure 6C), confirming that the major cytokines secreted by hAFSCs in the presence of damaged AEC2s contribute to a prohealing milieu.

Cytokeratin staining of cells within an AEC2 wound, 24 hours after wounding and the addition of DiI-labeled hAFSC, showed islands of cytokeratin-positive rat AEC2s in intimate proximity to cytokeratin-negative, DiI-labeled hAFSCs (Figure 7A, upper panels). The hAFSC-modified milieu also attracted cytokeratin-negative cells from the monolayer into the wound. These cells are likely fibroblasts, which are present as contaminants in the AEC2 cultures (usually ∼ 2%). Although SPC protein was not detected in hAFSC (DiI+) cells at this time (24 h), it was detected after 72 hours of culture of hAFSCs with damaged AEC2s by RT-PCR using human-specific primers (Figure 7A, bottom panel).

Figure 7. AEC2 damage milieu can induce SPC expression in a population of hAFSCs. (A) Cultures of damaged AEC2s, incubated for 24 hours with Dil-labeled hAFSCs, were fixed with 4% paraformaldehyde, immunostained for cytokeratin using an FITC-labeled secondary antibody, and counterstained with DAPI. Photographs show cells within the healing wound. Groups of cytokeratin positive (green), rat (Dil-negative) cells with the classical AEC2 phenotype are seen in intimate contact with DiI-positive (red), cytokeratin-negative hAFSCs (original magnification, ×20; bar = 100 μm). Although SPC protein costaining with DiI was not detected at this time (data not shown), human SPC, as detected by RT-PCR using human-specific primers, was induced in hAFSCs that had been cultured with damaged AEC2s for 72 hours (bottom panel). Human lung (lane 2) served as a positive control; rat AEC2s (lane 3) served as a negative control. Human SPC transcript was not detected in hAFSCs growing with damaged AEC2s for less than 72 hours (data not shown). (B) Upper three panels: hAFSCs were cultured with AEC2s in vitro damage milieu (CM from damaged AEC2) on chamber slides coated with collagen 1, fixed in methanol acetone 1:1 after various times in culture, immunostained with pro-SPC antibody and Alexa Fluor 488 secondary antibody, and mounted in DAPI mounting medium. Photograph shows that SPC protein expression appeared after 96 hours; right panels of each pair are primary antibody controls (rabbit IgG). No positive pro-SPC protein staining was seen before 72 hours (data not shown). Lower panel: RNA was extracted from hAFSCs cultured for 24, 48, and 72 hours with AEC2 milieu (normoxic and hyperoxic lung extract) on chamber slides coated with collagen 1. Human SPC was detected in hASFCs grown for 72 hours in AEC2 damage milieu by RT-PCR using human-specific primers. Coexpression of human TTF-1 reaffirmed an epithelial phenotype in SPC + hAFSCs. Human lung served as positive control, and rat AEC2s served as a negative control. Human SPC transcript was not detected before 72 hours (data not shown).

Download Figure | Download Powerpoint

[More] [Minimize]

To assess phenotypic changes mediated by AEC2 damage milieu in uncommitted hAFSCs, we cultured hAFSCs on collagen 1 with AEC2 control and damage milieu for various times and analyzed for SPC, a protein manufactured by AEC2s and a marker of distal alveolar epithelium, by immunostaining. Cells that stained for pro-SPC were not seen until after 4 days of culture with damage milieu (Figure 7B, upper panels). No SPC-staining cells were seen in hAFSCs cultured in basal medium (data not shown) or in cytospins of c-kit–positive hAFSCs assayed before each experiment (Figure E1). Human SPC was detected using RT-PCR with human primers in hAFSCs grown with AEC2 damage milieu, confirming that pro-SPC–positive immunostaining was not due to endocytosis of rat surfactant from the damage milieu, whereas coexpression of human TTF-1 reaffirmed an epithelial phenotype in SPC+ hAFSCs (Figure 7B, lower panel). The differentiation rate, as measured by SPC immunostaining, was 13 ± 1.4% (range, 8–27%; n = 11) and was similar between the various types of AEC2 damage milieu tested (damaged AEC2 CM, hyperoxic BAL, and hyperoxic lung extract). SPC-staining hAFSCs were seen in 1 of 11 undamaged AEC2 milieu tested (a sample of normoxic BAL, where approximately 2% SPC + hAFSCs were detected). ABCA3, a transporter protein that plays an essential role in surfactant lipid metabolism and lamellar body biogenesis, was also induced in hAFSCs treated with AEC2 damage milieu (Figure E4 in the online supplement). SPC expression, together with TTF-1 and ABCA3 expression, confirms induction of a phenotype consistent with distal lung alveolar epithelium in hAFSCs treated with AEC2 cell damage milieu.