Bacterial strains of A. johnsonii 2P08AA and A. radioresistens 50v1 were obtained through the Planetary Protection Culture Collection at the Jet Propulsion Laboratory. Stocks of ethanol (ENG Scientific) and Kleenol 30 (Mission Laboratories, Los Angeles, CA; Clovis Janitorial) were sterile filtered and stored as 200 μL aliquots at 4°C. Stock solutions of 25 mM Fe²⁺ were prepared using ferrous ammonium sulfate (Fe(NH₄)₂(SO₄)₂·6H₂O; EM Science) in ultrapure water (18 MΩ⁻¹), followed by sterile filtration (0.2 μm syringe filter, VWR), and storage as 200 μL aliquots at 4°C (with no visible precipitation over long term storage).

Concentrated minimal media (5x MM) were prepared using 30.0 g Na₂HPO₄·7H₂O (Sigma), 15.0 g KH₂PO₄ (EM Science), 2.50 g NaCl (Fisher Scientific), 5.00 g NH₄Cl (EM Science) per liter of ultrapure water. Low-osmolarity media (0.2x M9) were prepared by adding 0.4929 g MgSO₄·7H₂O (EM Science) and 0.0147 g CaCl₂·2H₂O (EM Science) to 200.0 mL 5x MM and dilution to 1.00 L using ultrapure water (to yield 1x M9), followed by an additional 5-fold dilution (200 mL 1x M9 in 1.00 L) to yield 0.2x M9. Final concentrations for ions in 0.2x M9 were 20 μM Ca²⁺, 400 μM Mg²⁺, 4.4 mM K⁺, 10.7 mM Na⁺, and 3.7 mM NH₄⁺, along with 5.4 mM Cl⁻, 4.5 mM H₂PO₄⁻, 4.4 mM HPO₄²⁻, and 0.4 mM SO₄²⁻.

Lysogeny broth media (LB) were prepared using 10.0 g tryptone (VWR), 5.0 g yeast extract (Amresco), 5.0 g NaCl (Fisher Scientific) and 1 mL of 1 M NaOH (Sigma-Aldrich) per liter of ultrapure water. Agar plates were prepared using 1.0 L LB and 15 g agar (Amresco).

Cultivations were performed in 15 mL screw cap test tubes (Pyrex, borosilicate glass; 13 × 100 mm), which were washed with tap water, rinsed with distilled water, and autoclaved between experiments. All screw cap test tubes were capped tightly and wrapped with parafilm during cultivation to prevent loss of ethanol, which served as a sole supplied carbon source. Cultivations were performed with mild agitation (200 rpm) using a New Brunswick Scientific Innova 4,200 incubator. Cell densities were monitored during cultivation by following agar plate counts and changes in optical density (OD) at 600 nm (Spectronic 20 Genesys). All media were autoclaved at 121°C and 15 psi for 45 min.

Glycerol stocks of A. radioresistens 50v1 and A. johnsonii 2P08AA were separately streaked onto LB agar plates and incubated at 32°C for ~24 h. Isolated colonies were inoculated into 2.0 mL 0.2x M9 containing 25 μM Fe²⁺ and 0–2.0% v/v Kleenol 30 (or 0–40.0 μL). Pre-cultures were initiated by addition of 20.0 μL 90% v/v ethanol to yield final concentrations of 150 mM or 1.0% v/v ethanol. Pre-cultures were grown to late log phase (OD ~0.6 at ~12 h). Target cultures were prepared using fresh media (2.00 mL) which were inoculated with 20.0 μL (1:100 dilution) of the respective pre-cultures and initiated by addition of 1.0% v/v ethanol.

Temporal changes in OD were followed for 0–15 h. Growth kinetics were characterized by regression analysis (Microsoft Excel) using a modified version of the Gompertz equation (Begot et al., 1996), which describes a non-linear bacterial growth model, and yields the parameters of growth rate (k), lag time (L), and an estimate of the maximum change in relative biomass ( log ( N / N 0 ) ). All regressions were minimized by least squares analysis. Control cultures containing no inoculate showed no growth (OD ≤ 0.002), as did cultures containing inoculate but no ethanol (OD ≤ 0.002), which indicated negligible biological contamination and accumulation of abiotic particles during cultivation.

Survivals of late-log phase cultures (OD ~0.4) were assessed by plating onto LB agar plates. In control experiments, plate counts using 0.2x M9 agar plates supplemented with ethanol (just prior to use) yielded irreproducible results, when compared to LB agar plate, likely due to variances in adsorption of ethanol (under our conditions). For the plate count assays, therefore, aliquots (100 μL) of the cultures (in 0.2x M9/Fe) were transferred to 2.5 mL microcentrifuge tubes, decimally diluted by 10⁶-fold using 0.2x M9, and spread (20 μL) onto LB agar plates using sterile plastic cell spreaders. All plates were sealed with parafilm and incubated at 32°C for 24 h. Plates with ≤300 colonies were enumerated and expressed as colony forming using per mL of the parent culture (cfu mL⁻¹).

Cultures of A. radioresistens 50v1 and A. johnsonii 2P08AA were prepared as described and harvested during early stationary phase (OD ~0.6–0.7, 7–8 h) by centrifugation (3,500 rpm) at 4°C for 15 min (Beckman Coulter Allegra 21R). After centrifugation, the cell pellets and supernatant fractions were separated and, respectively, treated. The supernatants were saved as 500 μL aliquots, dried using a DNA 110 Savant DNA SpeedVac, and stored at −80°C. The pelleted cells were washed twice with 0.2x M9 and stored at −80°C.

Samples were characterized by the West Coast Metabolomics Center using gas chromatography and time-of-flight mass spectrometry (GC-TOF/MS). In brief, dried cells and culture broth were separately resuspended in 2 mL of pre-chilled (−20°C) and degassed extraction solvent (acetonitrile:isopropanol:water, 3:3:2), vortexed for 30 s, shaken for 5 min at 4°C, clarified by centrifugation (~12,000 g), and the resulting supernatant evaporated to dryness. Samples were derivatized by resuspension in 10 μL of 40 mg/mL methoxyamine hydrochloride in pyridine (30°C, 1.5 h) followed by addition of 41 μL N-methyl-N-(trimethylsilyl) trifluoroacetamide (80°C, 30 min); fatty acid methyl esters (e.g., C8-C30) were additionally added to serve as retention index markers (Barupal et al., 2019).

Samples were transferred to crimp top vials and separated on an Agilent 6,890 Gas Chromatograph equipped with a Gerstel automatic liner exchange system (ALEX), multipurpose sample (MPS2) dual rail, Gerstel CIS cold injection system (Gerstel, Muehlheim, Germany), and built-in gas purifier (Airgas, Radnor PA). Chromatographic separation was afforded using a 30 m (0.25 mm i.d.) Rtx-5Sil MS column (0.25 μm 95% dimethyl 5% diphenyl polysiloxane film) with an additional 10 m integrated guard column (Restek, Bellefonte PA). Carrier gas was 99.9999% pure Helium with a constant flow of 1 mL/min.

Sample volumes of 0.5 μL were injected with a 10 μL s⁻¹ injection speed on a spitless injector with a purge time of 25 s. Temperature profile included 1 min at 50°C for the oven temperature, an increase of 20°C min⁻¹ to 330°C across 14 min, with a final hold for 5 min at 330°C. A temperature of 280°C was used for the transfer line between the gas chromatograph and Leco Pegasus IV time of flight mass spectrometer (single mass analyzer). The measured mass range was 85–500 Da, scan rate was at 17 spectra s⁻¹, electron energy was 70 eV, ion source temperature was 250°C, and detector voltage was 1850 V. Spectra were acquired using the Leco ChromaTOF software vs. 2.32 (St. Joseph, MI) and processed using the BinBase database system (Fiehn et al., 2005, 2010; Trigg et al., 2019), which quantifies (signal-to-noise ratio of 5:1) and matches mass peaks to the Fiehn mass spectral library using retention index (RI) and reference mass spectral information (which are internally compiled by the West Coast Metabolomic Center). Metabolomic data (Study ID ST002380, DatatrackID: 3552) were stored at the NIH Common Fund’s National Metabolomics Data Repository (Sud et al., 2016).

Comparisons of the cultivation kinetic parameters were conducted using unpaired Student’s t-tests and one- and two-way ANOVA analyses (Microsoft Excel), where statistical relevance was accepted at p < 0.05, normal distribution of the data was supported by Shapiro–Wilk tests (Past 4.10; Hammer et al., 2001), and variances were assumed as equal. Metabolomic data were compared using univariate, multivariate, and visual approaches.

Discrete changes in the metabolomes (e.g., trends for a single metabolite) were identified using unpaired Student’s t-tests (Microsoft Excel) with corrections for multiple testing using a false discovery rate (FDR) of FDR ≤ 0.20 (Benjamini and Hochberg, 1995). For the 700 metabolites in the total study (whole cell and extracellular metabolomes from 2 bacterial strains treated with and without Kleenol 30), normal distributions (Shapiro–Wilk Tests) were indicated for ~92% (646) of the metabolites. To account for potential underestimations of normality given the sample size (n = 3), metabolites exhibiting thresholds for normality of p > 0.03 (Shapiro–Wilk Tests) and significance of p < 0.05 (Student’s t-tests) were carried forward in the univariate and visual assessments of the data.

Broad structural and metabolic trends were obtained by visualizing the changes in the metabolomes (p < 0.05) using MetaMapp ¹ and Cytoscape 3.9.1 ² (Barupal et al., 2012), which constitute a statistical organizational approach to yield visual maps of metabolites arranged by known structural patterns and metabolic pathways. Confirmation of broad changes in the metabolomes were obtained using ChemRich ³ (Barupal and Fiehn, 2017), a statistical enrichment tool that compares groups of metabolites based on structural and biochemical classes. For this study, the standard and user-defined classifications in ChemRich included amino acids, monosaccharides, sugar acids, organic acids, fatty acids, lipid-related metabolites, nucleotide-related metabolites, and compatible solutes. Multivariate tests were conducted using canonical correspondence analyses (Past 4.10; Hammer et al., 2001) to correlate the metabolomic trends to differing cultivation growth parameters and conditions.

In Figure 1A we show that the cleanroom-associated Acinetobacter tested in this study, A. johnsonii 2P08AA and A. radioresistens 50v1, readily grow in the presence of ≤1% v/v Kleenol 30 (K30) under low-osmolarity aqueous conditions with ethanol serving as the sole supplied carbon source. Cultivations on 2% v/v K30 did not yield measurable cell densities.

In 0.1% v/v K30, cultures of the floor-associated A. johnsonii 2P08AA show no apparent loss in survival, as plate counts (~late log phase) effectively show no difference (p > 0.05) in the presence (2.1 ± 0.5 × 10⁷ cfu mL⁻¹) and absence (2.6 ± 0.1 × 10⁷ cfu mL⁻¹) of the detergent. These trends indicate quantitative survival of the 2P08AA strain in the presence of 0.1% v/v K30.

For the spacecraft surface-associated A. radioresistens 50v1, in contrast, survivability in 0.1% v/v K30 readily decreases (p = 0.006) to 46 ± 6% (propagated error) – as calculated by comparison of cultures cultivated in the presence (4.0 ± 0.4 × 10⁷ cfu mL⁻¹) and absence (8.8 ± 0.7 × 10⁷ cfu mL⁻¹) of 0.1% v/v K30. These trends are indicative of ~50% survival in the presence of 0.1% v/v K30.

In 1.0% v/v K30, survivabilities for the 2P08AA strain (26 ± 12%) and 50v1 strain (22 ± 8%) decrease to similar values – as indicated by comparison of cultures cultivated in the presence (2P08AA, 0.7 ± 0.3 × 10⁷ cfu mL⁻¹; 50v1, 1.9 ± 1.1 × 10⁷ cfu mL⁻¹) and absence (2P08AA, 2.6 ± 0.1 × 10⁷ cfu mL⁻¹; 50v1, 8.8 ± 0.7 × 10⁷ cfu mL⁻¹) of 1.0% v/v K30. These trends are indicative of ~70–80% inhibition in the presence of 1.0% v/v K30 across both strains.

Displayed in Figures 1B–D and Figure 2 are the cultivation kinetic parameters and associated growth curves for A. johnsonii 2P08AA and A. radioresistens 50v1. Optical measurements (Figure 2, empty blue circles) for cultures were converted to the ratiometric changes in relative biomass ( N / N 0 ) by assuming a direct relationship between optical transmittance and cell density for the initial culture media ( N 0 ) and at the time of measurement ( N ), as outlined in Begot et al. (1996). Changes in relative biomass (Figure 2, filled red circles, fitted line) were expressed as a log function ( log ( N / N 0 ) ), plotted over time, and fit to a modified version of the Gompertz equation (Eq. 1), as detailed in Begot et al. (1996) and Mogul et al. (2018). Minimized regressions yielded the parameters of growth rate (k; cell divisions h⁻¹), estimated time in lag phase (L; h), and log of the maximum increase in relative biomass ( log ( N / N 0 ) max ; unitless dimension).

To account for optical scattering, the maximum increases in relative biomass ( log ( N / N 0 ) max ) were broadly interpreted as maximum increases in intact cells, cellular aggregation, and/or intra- and extracellular polymeric substances at stationary phase. Reported in this study are the averaged values from the minimized regressions along with standard errors (n = 3–5 biological replicates).

Comparisons of the growth rates (k) show ~18% decreases for cultivations of the 2P08AA strain in 0.1% v/v K30 (p = 0.029) – as calculated by comparison of the difference in growth rates (Figure 1B) between the absence (0.50 ± 0.01 h⁻¹) and presence (0.41 ± 0.02 h⁻¹) of 0.1% v/v K30. The 50v1 strain, in contrast, shows no change in growth rate (p = 0.05) in 0.1% v/v K30 – as calculated by comparison of rates (Figure 1B) in the absence (0.39 ± 0.01 h⁻¹) and presence (0.40 ± 0.02 h⁻¹) of 0.1% v/v K30. When cultivated in 1.0% v/v K30, growth rates for the 2P08AA strain (0.22 ± 0.03 h⁻¹) and 50v1 strain (0.35 ± 0.01 h⁻¹) decrease by ~56% (p = 0.001) and ~ 13% (p = 0.043), respectively, as calculated by comparison rates (Figure 1B) in the absence (2P08AA, 0.50 ± 0.01 h⁻¹; and 50v1, 0.39 ± 0.01 h⁻¹) of the detergent.

Comparisons by one factor ANOVA analyses confirm that cultivations in the presence of 0.1–1.0% v/v K30 significantly impact the respective growth rates for the 2P08AA (p = 0.038, f = 4.256) and 50v1 (p = 0.0002, f = 4.459) strains. Additionally, two factor ANOVA analyses confirm that cultivations in the presence of K30 differentially impact the growth rates of the 2P08AA and 50v1 strains (p = 0.015, f = 5.318). These combined trends for exponential phase behavior are indicative of higher tolerances toward K30 by the 50v1 strain at 0.1 and 1.0% K30. These trends are opposite to those from plate counts at late-log phase, which indicate a higher tolerance for the 2P08AA strain.

Comparisons of the lag times (L) in 0.1 and 1.0% v/v K30 show decreases for both strains during cultivation (Figure 1C), which is unexpected since the addition of detergents was presumed to inhibit growth and increase the time in lag phase. For the 2P08AA strain, the native lag time of 0.77 ± 0.23 h⁻¹ (absence of K30) reduces to negligible values (~0 h) in 0.1 and 1.0% v/v K30. For the 50v1 strain, the native lag time of 2.61 ± 0.24 h⁻¹ is ~3-fold longer than 2P08AA strain and decreases by ~70% in 0.1% v/v K30 (0.76 ± 0.15 h⁻¹) and to negligible values in 1.0% v/v K30.

Comparisons by one factor ANOVA analyses confirm that cultivations in the presence of 0.1–1.0% v/v K30 significantly impact the respective lag times for the 2P08AA (p = 0.00008, f = 4.256) and 50v1 (p = 0.015, f = 4.459) strains. Additionally, two factor ANOVA analyses confirm that cultivations in the presence of K30 differentially impact the lag times for the 2P08AA and 50v1 strains (p = 0.006, f = 5.318). These combined trends indicate that K30 induces accelerated entry into the exponential phase in a concentration dependent manner, where the lag phase for the 2P08AA strain is effectively eliminated at ~10-fold lower K30 abundances when compared to the 50v1 strain – under these conditions.

Comparisons of the maximum change in relative biomass ( log ( N / N 0 ) max ) for cultivations in 0.1% v/v K30 show unexpected increases for both strains (Figure 1D). For the 2P08AA strain, cultivations in 0.1% v/v K30 yield ~40% increases (p = 0.026) in maximum apparent biomass when comparing values obtained in absence (2.30 ± 0.02) and presence of (2.44 ± 0.04) of 0.1% v/v K30 and accounting for the log transformation. Similarly, for the 50v1 strain, cultivations in 0.1% v/v K30 yield ~55% increases (p = 0.048) in maximum apparent biomass when comparing values obtained in absence (2.28 ± 0.01) and presence of (2.47 ± 0.04) of 0.1% v/v K30 and accounting for the log transformation. When cultivated in 1.0% v/v K30, the 2P08AA strain (1.53 ± 0.18) and 50v1 strain (2.07 ± 0.03) exhibit decreases of ~80% and ~ 40% in maximum apparent biomass, respectively, which is consistent with the decreased growth rates under these conditions (~56 and ~ 13%, respectively).

Comparisons by one factor ANOVA analyses confirm that cultivations in the presence of 0.1–1.0% v/v K30 significantly impact the respective maximum changes in relative biomass for the 2P08AA (p = 0.000003, f = 4.256) and 50v1 strains (p = 0.000004, f = 4.459). Additionally, two factor ANOVA analyses confirm that cultivations in the presence of K30 differentially impact the maximum changes in relative biomass for the 2P08AA and 50v1 strains (p = 0.007, f = 5.318). Combined, these trends suggest that 0.1% v/v K30 induces changes in apparent biomass in stationary phase.

Displayed in Figure 3 are biochemical maps that compare the extracellular metabolomes from A. johnsonii 2P08AA and A. radioresistens 50v1 (early stationary phase). Per strain, dried cultivation broth (cell free) yielded a total of 56 metabolites, which were processed and visualized as described above. Listed in Table 2 are the extracellular metabolomic changes in the 2P08AA and 50v1 strains after cultivation in the absence and presence 0.1% v/v K30. Pathway analyses were not conducted for extracellular metabolites.

When grouped by biochemical classes, cultivations of the 2P08AA strain in K30 yield differential impacts to extracellular sugar acids, monosaccharides, and organic acids (di- and monocarboxylic acids). Yet, univariate tests show no changes after correction for multiple testing. Impacts to sugar acids (p = 1.4×10⁻⁵; Q = 6.9×10⁻⁵; ChemRich) are supported by substantial increases in extracellular gluconolactone (22-fold) and glycerate (34-fold). Impacts to di- and monocarboxylic acids, which were grouped as organic acids (p = 3.0×10⁻⁴; Q = 7.6×10⁻⁴; ChemRich), are supported by increases in oxalate and decreases in adipic acid (hexanedioic acid) and 3,4-dihydroxybenzoate (protocatechuate). Impacts to extracellular monosaccharides (p = 7.6×10⁻³; Q = 1.3×10⁻²; ChemRich) are supported by substantial increases in extracellular fructose (4.5-fold), galactose (22-fold), and mannose (57-fold) and decrease (2.5-fold) in extracellular ribose (pentose).

In contrast, cultivations of the 50v1 strain in K30 – after correction for multiple testing (n = 56) – yield significant changes in the extracellular galactonate, gluconate, gluconolactone, glycerate, mannose, ribose, tartrate, 4-hydroxybutyrate, and α-ketoglutarate. Likewise, grouping by biochemical classes are reveals significant changes in the extracellular metabolome.

Impacts to sugar acids (p = 1.9×10⁻⁸; Q = 9.6×10⁻⁸; ChemRich) are supported by substantial increases in extracellular galactonate (36-fold), gluconate (46-fold), gluconolactone (8.8-fold), glycerate (9.3-fold), and 2-methylglycerate (1.9-fold). Impacts to extracellular organic acids (p = 7.8×10⁻²; Q = 0.16; ChemRich) are suggested by the significant increases in tartrate and significant decreases in α-ketoglutatate, along with the potential decreases in citramalate and 4-hydroxybutyrate. Impacts to extracellular monosaccharides (p = 9.6×10⁻²; Q = 0.16; ChemRich) are suggested by significant increases in extracellular mannose (20-fold) and significant decreases in ribose (2.5-fold).

Molecular insights into the tolerance of spacecraft-associated Acinetobacter toward Kleenol 30, a NASA cleanroom floor detergent, were obtained through cultivation, kinetic, and metabolomic approaches. Comparisons were conducted using A. johnsonii 2P08AA and A. radioresistens 50v1, which were isolated at, respectively, unique locations within the spacecraft assembly facilities, which were subjected to differing cleaning regimes.

The strain A. johnsonii 2P08AA was isolated from the floor of the Mars Phoenix lander assembly facility, which was routinely cleansed with Kleenol 30. The strain A. radioresistens 50v1 was isolated from the surface of the pre-flight Mars Odyssey orbiter, which was likely cleaned with isopropyl alcohol (2-propanol, isopropanol, IPA) and never exposed to Kleenol 30.

Hence, comparisons across these differing sub-environments provide initial insights into the impacts of repeated cleansing, under the conditions of spacecraft assembly, toward biochemical and cultivation selection.

Cultivations containing 0–2.0% v/v Kleenol 30 (K30) were conducted in a low-osmolarity minimal media (0.2x M9, 25 μM Fe) containing 150 mM ethanol, which served as the sole supplied carbon source. Cultivation media containing 2.0% v/v K30 were bactericidal for the tested Acinetobacter, which suggests a biocide-type of activity for K30.

In contrast, in 1.0% v/v K30, appreciable cultivation tolerances were observed for the 2P08AA and 50v1 strains. Cultures containing 1.0% v/v K30 exhibited (1) ~20–30% survivals, (2) ~66 and 12% decreases in growth rates, and (3) eliminated lag phase times for the 2P08AA and 50v1 strains, respectively. These cultivation tolerances of <2% v/v K30 are significant since current NASA planetary protection practices utilize ~0.8-1.6% v/v K30 for floor cleansing. Our results, therefore, suggest that the spacecraft-associated Acinetobacter harbor a tolerance toward the current floor cleansing practices for NASA cleanrooms.

In 0.1% v/v K30, differential impacts were observed across the cultivation parameters. For the 2P08AA strain, cultures exhibited (1) no measurable changes in cell density at late-log phase (plate counts), (2) modest decreases in the growth rate (~18%), and (3) negligible times in the lag phase. These results indicate that K30 impacts the 2P08AA strain by accelerating growth out of the lag phase, slightly inhibiting growth during the exponential phase, and by yielding no changes cell density after cultivation to late-log phase.

For the 50v1 strain, cultures exhibited (1) ~50% survivals, (2) no changes in growth rate, and (3) ~70% decreases in the lag phase time. This is indicative of K30 impacting the 50v1 strain by mildly accelerating growth out of the lag phase, yielding no change in the growth rates during the exponential phase, and inhibiting cell density by ~50% when cultivated to late-log phase.

The differential cultivation trends observed in 0.1% v/v K30 are indicative of strain and growth-phase dependent responses to K30. In turn, these trends suggest that K30 reduces nutrient bioavailability in culture (e.g., alkaline earth and transition metals) and/or elicits specific biochemical responses (e.g., metabolism of K30 and/or enzymatic inhibition by K30) – which together manifest as acceleration out of the lag phase for both strains, slight inhibition in rates of cell division during the exponential phase for the 2P08AA strain, and moderate reductions in total cell density at late-log phase for the 50v1 strain.

To obtain metabolic insights into the tolerances toward Kleenol 30 (K30), we used cultures grown in 0.1% v/v K30 to obtain reliable biomasses at late-log phase for both strains (see comparison of 0.1 and 1.0% w/w K30 in Figures 1D). For A. johnsonii 2P08AA, cultivations with 0.1% v/v K30 exhibited (1) no significant and discreet changes in the abundances of intracellular metabolites, (2) several significant extracellular changes when considering broader classifications such as sugar acids, monosaccharides, organic acids (di- and monocarboxylic acids), and fatty acids, and (3) potential support for biodegradation and metabolism of both detergents in the K30 formulation.

The combined results for the 2P08AA strain suggest that the native and intracellular metabolic state of the 2P08AA strain readily accommodates cultivation in the presence of 0.1% v/v K30. The trends for the 2P08AA strain also suggest that survival is associated with adjustments in the extracellular metabolome and a potential for biodegradation of the Kleenol 30 detergents.

For A. radioresistens 50v1, cultivations in 0.1% v/v K30 exhibited (1) differential changes in the abundances of intracellular amino acids, compatible solutes, nucleotide-related metabolites, dicarboxylic acids, and saturated fatty acids, (2) substantial and differential changes in the abundances of extracellular sugar acids, monosaccharides, and organic acids, and (3) no clear statistical support for biodegradation and metabolism of the detergents from the K30 formulation (when excluding the potential for degradation via benzene metabolism).

These results for the 50v1 strain indicate that the intracellular and extracellular metabolomes of the 50v1 strain significantly adjust to accommodate cultivation in the presence of 0.1% v/v K30. These trends also suggest that survival is associated with intracellular changes to amino acid and peptide metabolism, nucleotide metabolism, central metabolism, fatty acid metabolism, and pyrimidine metabolism.

For the 50v1 strain, the changes in amino acid metabolism include decreases in lysine, proline, glycine, ornithine, homoserine, valine and others listed in Table 1. This is relevant since decreases in lysine and valine are observed during biofilm formation in A. baumanni 1,656–2 (Yeom et al., 2013), while decreases in amino acid metabolism are observed after treatments of A. baumanni AB5075 with mixed antibiotics (polymyxin B and rifampicin) (Zhao et al., 2021). In E.coli UTI189, decreases in intracellular lysine and valine are associated with biofilm formation (Lu et al., 2019). In Bifodobacterium bifudum, decreases in intracellular lysine, arginine, and proline are associated with biofilm formation (Sadiq et al., 2020). These observations suggest that the Acinetobacter undergo stress-like responses toward K30 that inhibit total cell mass (e.g., survival), which could include changes in early biofilm development.

Trends across nucleotide abundances and the cultivation terms are similarly suggestive of early biofilm-type development (Sarkar and Chakraborty, 2008; Castillo-Juarez et al., 2017) and/or quorum sensing related changes such as intracellular PHA synthesis (Kessler and Witholt, 2001; Xu et al., 2011). These trends include increases in the maximum relative biomass terms for the 2P08AA and 50v1 strains, decreases in 5′-MTA and 5’-AMP for the 50v1 strain, and substantial increases in extracellular monosaccharides (e.g., mannose, galactose, and fructose for the 2P08AA strain).

In comparison, biofilms of E. coli UTI189 show decreases in 5′-MTA and 5’-CMP, as well increases in metal-binding siderophores, when Fe³⁺ concentrations in culture decrease from 10 to 1 μM (Guo et al., 2021). Lowered 5′-MTA abundances are associated with biofilms of Pseudomonas aeruginosa PAO1 (Tielen et al., 2013). In Pseudomonas fluorescens PF08, exogenously added 5’-AMP inhibits biofilm formation (Wang et al., 2021). In Streptococcus pyogenes, the import of mannose induces lysozyme resistance via a quorum sensing pathway (Chang et al., 2015); while exogenously added galactose differentially inhibits and activates biofilm development in differing species of Streptococcus (Ryu et al., 2016).

To obtain biochemical insights into the mechanisms of tolerance, the metabolomic trends were interpreted in context of the cultivation conditions. In this study, cultivations were conducted in low-osmolarity media (0.2x M9) supplemented with 25 μM Fe²⁺ (Fe(NH₄)₂(SO₄)₂), which slowly oxidized to Fe³⁺ under the aerobic cultivations in the capped glass tubes (per absorbance spectroscopy). However, in control experiments, cultures supplemented with [Fe(EDTA)]¹⁻ grew slower, when compared to Fe(NH₄)₂(SO₄)₂, while cultivations in acid-washed glassware yielded no growth. Therefore, under the cultivation conditions, trace elements and species such as Zn²⁺, Co²⁺, Cu²⁺, Mn²⁺, BO₃³⁻ (borate; a B source), and MoO₄²⁻ (molybdate; a Mo source) were likely obtained from trace contaminants in the media and/or the glassware.

These details are relevant since K30 contains EDTA and sodium metasilicate, which are chelators that sequester transition and alkaline earth metals. Hence, in culture, chelation by EDTA would likely restrict the bioavailability of transition metals and inhibit growth – as observed in control studies. Limited bioavailabilities for Ca²⁺ and Mg²⁺ are also expected since sodium metasilicate (Na₂SiO₃) is a water softening agent that functions through ion exchange via chelation of Ca²⁺ or Mg²⁺ (and release of Na⁺).

Under these considerations, therefore, our review of the extracellular metabolomes reveals changes in several biochemicals possessing metal-binding properties. Appraisal of the literature shows that sugar acids such as galactonate, gluconate, and gluconolactone form appreciably stable complexes with transition metals such as Fe³⁺, Zn²⁺, Co²⁺, Cu²⁺, and Mn²⁺ (Sawyer, 1964; Carper et al., 1989; Frutos et al., 1998; Gyurcsik and Nagy, 2000). For the alkaline earth metals of Ca²⁺ and Mg²⁺, stable complexes are formed with sugar acids such as gluconolactone and glycerate (Taga et al., 1978; Meehan et al., 1979; Tajmir-Riahi, 1990), monosaccharides such as mannose (Angyal, 1973), and organic acids such as oxalate (Chutipongtanate et al., 2013; Izatulina et al., 2018) and tartrate (Hawthorne et al., 1982). In addition, borate and Ca²⁺ form stable complexes with gluconate, glycerate, and tartrate (van Duin et al., 1987); while molybdate yields stable complexes with mannose and fructose (Spence and Kiang, 1963; Sauvage et al., 1996).

Biochemically, therefore, these observations suggest that the 2P08AA strain targets trace Ca²⁺ and Mg²⁺ acquisition through potential increases in extracellular glycerate, gluconolactone, and oxalate, but not via gluconate and galactonate, which bind transition and alkaline earth metals but were not detected. In contrast, the 50v1 strain potentially targets both transition and alkaline earth metals given the increases (p < 0.05, FDR ≤ 0.20) in extracellular gluconate, galactonate, glycerate, gluconolactone, and tartrate. In fact, increases in glycerate export for the 50v1 strain are supported by the increases in extracellular glycerate and decreases in intracellular glycerate. Control of intracellular Ca²⁺ retention is also suggested by increases in extra- and intracellular tartrate. Multivariate tests support these total trends and suggest that chelator concentrations (from the K30 formulation) influence the abundances of extracellular metal-binding biochemicals.

Thus, our total results suggest that up-regulations (per se) in extracellular trace element acquisition strategies are common survival features to the spacecraft-associated Acinetobacter. Future studies using concentrated and clarified cultivation media may yield spectral support for the trace metal chelation. Additional common survival features may include regulation of the bulk osmolarity, as suggested by decreases in intracellular compatible solutes for the 50v1 strain (e.g., glycerate, 3-phosphoglycerate, and mannitol), statistical associations (CCA plot) between survival and changes in intracellular compatible solutes (e.g., trehalose) for the 2P08AA strain, and increase in an extracellular compatible solute for the 2P08AA strain (e.g., sorbitol).

When considering intracellular changes, differing responses to 0.1% v/v K30 are observed by the 2P08AA and 50v1 strains. While limited intracellular responses are observed for the 2P08AA strain, substantial intracellular changes are observed for the 50v1 strain. Multivariate tests for the 50v1 strain support a potential for the synthesis of light scattering PHAs during stationary phase in response to 0.1% v/v K30. Relatedly, univariate tests suggest decreases in the export of α-ketoglutarate and 4-hydroxybutanoate (e.g., decreases in extracellular abundances), while multivariate tests show correlations between intracellular PHA synthesis (e.g., α-ketoglutarate, 4-aminobutanoate, and 4-hydroxybutanoate) and increases in the apparent maximum relative biomass.

For the 2P08AA strain, multivariate analyses highlight a potential for partial metabolism of the detergents (from the Kleenol 30 formulation) – which would serve as an additional biochemical strategy toward survival. While these exact trends were not observed for the 50v1 strain, multivariate tests do support correlations between survival and changes in the degradation of aromatic compounds (benzene metabolism). In support, GC–MS studies in Mogul et al. (2018) provided evidence for scission of polyethylene glycol mono-nonylphenyl ether by the 50v1 strain. Future targeted cultivation studies are required to confirm the extent of biodegradation of the detergent by these strains.

When considering impacts of K30 on the lag phase, we observe a differential concentration dependence, where the time in lag phase is eliminated in the presence of 0.1 v/v K30 for the 2P08AA strain, and in 1.0% v/v K30 for the 50v1 strain. We venture that metabolism of the alkyl sides chains on the detergents (via ω-oxidation of the terminal carbon and subsequent cycles of β-oxidation) may serve as activators for acceleration out of the lag phase, and that degradation of the aromatic detergent cores (via benzene metabolism) may sustain the 50v1 strain during exponential phase, where growth rates show no inhibition in the presence of 0.1% v/v K30. If so, carbon and energy acquisition from the detergents could serve as an added biochemical strategy toward survival. Alternatively, the K30 detergents could potentially disperse any low abundance cellular aggregates during the lag phase, thereby yielding higher OD values and the apparent increases in biomass. Under this assumption, the observed concentration dependence across the strains would be suggestive of the cellular aggregates from both strains having differing stabilities.