Impact of payload shielding on Enterobacter cloacae viability and proteomic profile: Insights from a stratospheric weather balloon flight experiment

The TNF-α assay measures the release of tumor necrosis factor-alpha (TNF-α), a key inflammatory cytokine, from immune cells, indicating the cellular response to infection or stress. This assay is of interest because TNF-α plays a crucial role in immune regulation and its levels can provide insights into the potential inflammatory impact of E. cloacae under the tested conditions. Cellular supernatant was collected from RAW 264.7 cells infected with E. cloacae at 2- and 24-hour time points. Samples were diluted 1:10 with PBS, and TNF-α concentration was determined using a DuoSet ELISA Mouse TNF-α (R&D Systems) according to the manufacturer’s protocol for four biological replicates.

An equal amount of 25 µg protein per sample was utilized, and three replicates per sample type were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The entire gel lane corresponding to each sample was excised and diced into 1 mm² cubes, followed by ingel trypsin digestion as we did previously ^18,19. The resulting peptide samples were analyzed using a 250-mm Ultrahigh-Performance Liquid Chromatography (UHPLC) system coupled to an Orbitrap Fusion mass spectrometer (Thermo Scientific) similar to what was described previously ^18,19. After the raw data were collected, tandem mass spectra were extracted, the charge state was deconvoluted, and the charge state was deisotoped using Proteome Discoverer (Thermo Scientific) version 2.1. Mascot (Matrix Science, London, UK; version 2.7.0) which was employed to analyze all MS/MS samples. The Mascot search was performed against the UniProt E. cloacae database (5,412 entries), assuming trypsin as the digestion enzyme. Fragment ion mass tolerance was set to 1.00 Da, and parent ion tolerance was set to 10.0 PPM. Pyrrolysine with O+18 and cysteine with carbamidomethyl were specified as fixed modifications, while n-terminal Glu->pyro-Glu, deamidated asparagine and glutamine, and methionine oxidation were set as variable modifications in Mascot. Scaffold (Proteome Software Inc., version 4.11.0) was used to validate peptide and protein identifications based on MS/MS. Peptide identifications were accepted with a greater than 95.0% probability using the Scaffold Local FDR algorithm. Protein identifications were accepted if they could be established with a greater than 95.0% probability and contained at least 2 identified peptides. Protein probabilities were assigned using the Protein Prophet algorithm. Proteins that shared similar peptides and could not be distinguished based on MS/MS analysis alone were grouped, and proteins sharing significant peptide evidence were clustered. The reported protein false discovery rate (FDR) was 0.9%, and the peptide FDR was 0.06%. Protein quantification was performed using weighted spectral counts. The fold change was calculated based on the spectral counts of proteins from Faraday fabric-shielded bacteria compared to unshielded bacteria. A fudge factor of 0.5 was applied to proteins that were not identified to enable calculations. Moreover, only proteins present in 2 out of 3 replicates were reported as identified proteins. Statistical significance was determined using a Fisher’s exact test, and proteins with a p-value <0.05 were considered to exhibit statistically significant changes in abundance.

To conduct the principal component analysis (PCA) of the differentially abundant proteins, we utilized Clustvis (version 2.00). Row-wise unit variance scaling was applied, and SVD with imputation was used to calculate the principal components. Prediction ellipses were generated to encompass new observations from the same group with a probability of 0.95, ensuring that they fall within the ellipse. Additionally, we performed a hierarchical clustering analysis. Both rows and columns were clustered using correlation distance, and average linkage was employed. Before clustering, rows were centered and unit variance scaling was applied to ensure accurate data representation.

For functional analysis of differentially abundant proteins, we utilized ShinyGO (version 0.80), using E. cloacae ATCC 13047 STRINGdb as the selected species. A false discovery rate (FDR) cutoff of 0.05 and a minimum size of 2 proteins per pathway were applied. Redundancies in the enrichment results were removed. For the enrichment of Uniprot terms and InterPro domains, as well as the generation of Local Network Clusters using STRING, we employed the respective tools and databases within ShinyGo.

This study was part of the 2022 Florida Space Trek Academy, organized by Atlantis Educational Services and sponsored by NASA’s Florida Space Grant Consortium. Undergraduate and graduate university student team members underwent preparatory meetings and submitted written proposals seven weeks before the stratospheric balloon flight. Daily debriefings and support sessions ensured progress and cohesion. The program involved weather balloon deployment discussions, role assignments, and weather analysis for suitable launch sites. Activities included parachute creation, sensor component setup, and students’ payload preparation, launch, and retrieval efforts followed by payload release.

A payload of frozen mid-exponential phase E. cloacae cells was flown on a small high-altitude weather balloon (Figure 1). E. cloacae samples were prepared in a frozen state for our stratospheric flight experiment to closely simulate potential microbial conditions within spacecraft environments and to preserve the initial bacterial state, facilitating direct assessment of stratospheric impact on viability and behavior. This approach not only mirrors the states in which microbes may exist onboard, including the cold zones and biological sample storage areas, but also allows for comparative analysis with existing studies employing desiccated samples ^16,20. In brief, two boxes, each containing 40 replicate aliquots of mid-exponential phase E. cloacae cells, were loaded into a styrofoam cooler containing both external and internal sensors that captured a variety of environmental measurements (see Figures 1–2). One box of E. cloacae aliquots was wrapped in a triple-layered Faraday fabric. This fabric, composed of conductive metals copper and nickel, was designed to shield against a wide range of electromagnetic frequencies, encompassing microwaves, infrared, and UV light ²¹. The efficacy of shielding is predicated on the fabric’s material properties, which are particularly effective at attenuating electromagnetic radiation across these mentioned spectra. The fabric’s effectiveness in attenuating electromagnetic radiation across these spectra is contingent upon its material properties, such as weave density and metal conductivity ^22,23.

Figure 1.

Figure 2.

On the flight day, the weather balloon containing the E. cloacae shielded and unshielded payloads (Figure 2) was launched into the stratosphere. The time from launch to landing spanned 1 hour and 29 minutes, with the payload spending approximately 1 hour and 2 minutes in the stratosphere. The balloon and payload were exposed to various external environmental conditions during the flight (summarized in Table 1, Figure 3, Table S1). The balloon’s ascent reached a maximum altitude of 27 kilometers and was exposed to a range of temperatures (−21°C minimum to −5°C in the stratosphere, Figure 3A), atmospheric pressures (21.87 kPa to 1.83 kPa, Figure 3C), and humidity levels (26% to 7.4%, Table S1) during the flight. Internal temperature was measured as well (Figure 3B). Although other internal sensor data was also measured for our payload (POD_1 in Table S1), water contamination of the POD_1 sensor by an unrelated payload contained in this same stratospheric balloon flight (monitored on POD_3 in Table S1) created technical issues and malfunctions with both these sensors. Hence, we did not expand upon these data here. Upon landing and retrieval, the samples were immediately frozen on dry ice and stored at −80°C.

Figure 3.

First, we evaluated the effect of the flight in the presence or absence of shielding on E. cloacae viability. We determined the colony-forming units per milliliter (CFU/mL) to assess bacterial survival after the stratosphere flight. The postflight samples that were shielded or not shielded were compared to each other and to the pre-flight samples (i.e., −80°C stored E. cloacae aliquots prepared as described in Materials and Methods, in a manner identical to the balloon flight samples). Interestingly, shielded and unshielded postflight samples had a significant but modest decrease in viability (~0.2 log) compared to pre-flight samples, but the shielding itself did not affect the bacterial viability (Figure S1). It is also important to note that long-term storage at −80°C did not affect bacterial viability (Figure S2).

Next, we performed a gentamicin protection assay in RAW 264.7 murine macrophages to evaluate whether Faraday fabric-based payload shielding during the weather balloon flight affects the ability of E. cloacae to survive within host cells and to assess the potential virulence of this bacterium. We observed that shielding the samples significantly increased the survivability (measured by intracellular CFU/mL) within infected macrophages at 2 hours post-infection (hpi) compared to cells subjected to the flight without shielding (Figures 4A, C). A similar trend was also observed at 24 hpi (Figures 4B, D).

Figure 4.

To complement these findings, we executed a ground control (GC) experiment designed to replicate the ambient temperature conditions encountered during the flight. Here, frozen bacterial samples were subjected to a 2-hour and 57-minute incubation at room temperature within a styrofoam cooler, mirroring the complete duration of the balloon flight experiment. Post incubation, these samples were swiftly returned to −80°C storage pending further analysis. Analysis revealed that the ground control samples exhibited lower overall viability than the flight samples (Figure S1) despite experiencing consistent temperature exposure to the balloon flight. Additionally, in the intracellular survival assay conducted with ground control bacteria, a decrease in bacterial count was observed from 2 to 24 hours post-infection (Figure S3, MOI of 43:1). It is worth noting that direct comparison of ground samples with the flight samples might be somewhat skewed as the MOI for the flight-exposed samples was nearly double that of the ground control bacteria. Despite efforts to standardize, such as maintaining ground controls at room temperature within styrofoam coolers to closely simulate flight conditions, discrepancies between the hypothetical and actual MOIs were evident. These differences were particularly evident in ground controls, where retests confirmed a consistently lower actual MOI. The objective to achieve an infection MOI of 80:1 was specifically challenged by these findings, highlighting the biological variability.

We subsequently assessed the release of tumor necrosis factor-alpha (TNF-α) from murine macrophages after exposure to E. cloacae. A comparative analysis of macrophages infected with shielded and unshielded E. cloacae samples exhibited substantial TNF-α release (Figure 5). While no statistically significant differences were observed in the levels of TNF-α releases between macrophages infected with shielded and unshielded E. cloacae, a pronounced increase in the pro-inflammatory cytokine was observable when ground control E. cloacae was used for infection (Figure 5). This observation is noteworthy, particularly considering the lower Multiplicity of Infection (MOI: ~43:1) applied for ground control E. cloacae, in contrast to the higher MOI (~80:1) used for the flight-exposed E. cloacae.

Figure 5.

Next, the impact of payload shielding on the proteome of E. cloacae during the weather balloon flight was investigated. To achieve this, proteome profiling was performed on E. cloacae shielded and unshielded postflight samples. Bacteria from each postflight sample were lysed, and cellular proteins were extracted, followed by SDS-PAGE analysis and tryptic digestion of entire gel lanes. The resulting peptides were then analyzed using Orbitrap Fusion Mass Spectrometry.

In our study, 1355 E. cloacae proteins were identified, with 1217 proteins detected in both the shielded and unshielded samples during the weather balloon flight (Figure 6A). Additionally, we found 24 unique proteins in the shielded samples and 114 unique proteins in the unshielded samples (Figure 6A). Label-free quantitative mass spectrometry analysis revealed that 97 proteins exhibited altered abundance between the shielded and unshielded conditions (Fisher’s test p<0.05, fold change greater than −1.5/+1.5). Among these, 22 proteins displayed increased abundance in the shielded conditions, while 75 proteins were increased in the unshielded conditions (Figure 6A).

Figure 6.

To explore the relationship between the proteins with altered abundance in shielded and unshielded samples, we performed principal component analysis (PCA), which revealed a distinct grouping of the samples based on their proteomic profiles (Figure 6B). Furthermore, unsupervised hierarchical clustering analysis of the proteins in shielded versus unshielded bacteria identified clusters of proteins that specifically showed an increase in abundance in each condition, providing a reliable means of distinguishing between shielded and unshielded samples (Figure 6C).

To gain insights into the biological functions of the differentially abundant proteins in E. cloacae under shielded or unshielded conditions during the weather balloon flight, we conducted further bioinformatic analyses (Figure 7, Table S2–S7). We initially investigated the enrichment of specific Uniprot-assigned functions among the differentially abundant proteins. Notable functional categories included transferases, ATP-binding proteins, nucleotide-binding proteins, coiled-coil proteins, kinases, ribosomal proteins, ribonucleoproteins, helicases, metalloproteases, proteins involved in glycolysis, and other metabolic proteins (Figure 7A). Additionally, InterPro domain mapping revealed that a significant number of the differentially abundant proteins belonged to specific domains, including O-loop containing nucleoside triphosphate hydrolases, pyridoxal phosphate-dependent transferases, tetratricopeptide-like helical domains, CoA-binding proteins, and histone-like DNA-binding proteins (Figure 7B). Moreover, STRING analysis identified various functional associations of the differentially abundant proteins, such as alcohol dehydrogenases, ribosomal proteins, lysine tRNA ligases, proteins involved in glycogen metabolism and glycolysis, DNA-directed DNA polymerases, and others (Figure 7C).

Figure 7.

Furthermore, through manual inspection of the differentially abundant proteins in shielded versus unshielded conditions, we observed a decrease in chemotaxis proteins in the shielded samples compared to the unshielded samples (Table S2), such as FlgK (A0A0H3CJP9, −10.00) or FliL (A0A0H3CQG8, −2.00). This finding suggests that the mobility of E. cloacae may be affected by shielding. Moreover, a notable decrease was observed in the levels of proteins involved in DNA repair, replication, and transcription in the shielded samples compared to the unshielded samples (Table S3), such as RapA (A0A0H3CEW2, −10.00), RnK (A0A0H3CMR3. −10.00), RuvA (A0A0H3CIN6, −5.00), MalT (A0A0H3CQR0, −3.33), PolA (A0A0H3CRP3, −2.50), or IhfA (A0A0H3CLA6, −2.00).

In addition, decreased abundance of proteins related to peptidoglycan synthesis, turnover, cell division, and other cell wall processes were observed in E. cloacae under shielded conditions (Table S4), such as in AnmK (A0A0H3CJQ3, −5.00), ParC (A0A0H3CRH6, −3.33), or Ddl (A0A0H3CIQ4, −3.33). Similarly, proteins involved in proteolysis, such as the chaperone BepA (A0A0H3CNZ5. −2.50), showed primarily a decreased abundance in shielded conditions (Table S5). In terms of proteins involved in protein translation and transport, these molecules exhibited increased abundance in the shielded samples compared to unshielded conditions (Table S6), such as SecB (A0A0H3CD34, +2.22), a protein involved in protein transport, or translation proteins RplV (A0A0H3CUA0, +1.90), RplR (A0A0H3CSK2, +1.80), RplQ (A0A0H3CSJ2, +1.70), RplL (A0A0H3CF84, +1.60), or RplB (A0A0H3CRH9, +1.50). However, other proteins involved in the translation and protein transport showed a decreased abundance in the shielded samples, including LysS (A0A0H3CP92, −1.67), SrmB (A0A0H3CNG0, −1.67), Ffh (A0A0H3CSE5, −1.67), and ArgS (A0A0H3CGG9, −1.67) (Table S6).

Finally, the largest group of differentially abundant proteins was associated with various metabolic processes. Notably, tricarboxylic acid cycle proteins SucD (A0A0H3CPV5, +1.60) and Mdh (A0A0H3CQC4, +1.60) exhibited increased abundance in shielded conditions, as did several proteins involved in fatty acid biosynthesis and amino acid metabolism (Table S7). Conversely, proteins involved in the metabolism with a decrease in abundance in the shielded versus unshielded conditions were phosphofructokinase (PfkA, A0A0H3CTK4, −1.67), enolase (Eno, A0A0H3CEM8, −1.67), amino acid biosynthetic proteins AroB (A0A0H3CSQ2, −2.00), Asd (A0A0H3CRS7, −2.00), ProB (A0A0H3CG68, −2.00), adenosine deaminase (Add, A0A0H3CKK4, −2.00), Coenzyme A biosynthesis bifunctional protein CoaBC (A0A0H3CEK0, −2.50), PurE (A0A0H3CJT3, −2.50), glycogen synthase (GlgA, A0A0H3CST7, −3.33), GpmB (A0A0H3CII5, −3.33) that is involved in the glycolytic process, and Lgt (A0A0H3CPZ5, −3.33) that regulates the lipoprotein biosynthetic processes.