| 10 Mar 2026
Ice-nucleating Bacteria Enriched in the Sea-Surface Microlayer as Potential Sources of Atmospheric Ice-Nucleating Particles
Abstract. The sea-surface microlayer (SML) forms the ocean’s thin biological and chemical film that mediates air-sea exchange. This interface is enriched in surface-active organic matter and biogenic particles that can act as ice-nucleating particles (INP), yet the microbial sources of marine INPs remain poorly resolved. Here, we isolated and characterized ice-nucleating bacteria from the SML and underlying water (UW) of a semi-enclosed coastal inlet in Japan. Among 92 bacterial isolates, six strains induced freezing above –15 °C, indicating active ice nucleation. These isolates, affiliated with Flavobacteriia and Gammaproteobacteria, were heat-labile, and lost activity after 0.22 μm filtration, consistent with large (>100 kDa) outer-membrane ice-nucleation proteins. Flow-cytometric assays confirmed that heating up to 100 °C did not cause a decline in bacterial cell numbers, indicating that INA loss resulted from denaturation of heat-labile proteins rather than cell lysis. In contrast, Proteinase K treatment caused marked membrane disruption, suggesting that proteolytic inactivation and cell damage jointly contributed to INA reduction. One strain, 4U17, retained high INA after Proteinase K treatment despite extensive cell lysis, suggesting a protease-resistant nucleator that may be non-proteinaceous or shielded within extracellular or cell-derived particles. Amplicon sequencing of environmental samples revealed that taxa related to these ice-nucleating bacteria were consistently enriched in the SML relative to UW. Together, our results identify the marine SML as a reservoir for biogenic ice nucleators that may contribute to the oceanic source of atmospheric INPs.
General comments
This manuscript presents exciting data on ice nucleation-active bacterial strains isolated from the sea surface microlayer in the Sagami Bay of Japan. The authors identify potentially novel ice nucleation-active strains affiliated with Flavobacteriia and Gammaproteobacteria and characterize their ice nucleation activity through heat treatment, Proteinase K treatments, filtration, and size fractionation of the ice-nucleating material. By demonstrating enrichment of taxa related to the isolates in the sea surface microlayer compared to the underlying water using 16S rRNA amplicon sequencing data, the authors highlight the sea surface microlayer as an important reservoir of biogenic ice nucleators that may contribute to atmospheric ice-nucleating particles.
The manuscript is generally well-written with a logical flow, and the motivation, workflow, and reasoning are clearly presented. However, several sections would benefit from additional clarification, as detailed below. In addition, I recommend a thorough read-through to further improve clarity and streamline the manuscript.
The authors generally do a good job describing the sampling methods. However, it lacks a clear description of when the samples were collected, how the samples were stored prior to analyses, and whether the isolated strains and amplicon sequencing data originated from the same samples or not. Please ensure that this information is clearly presented and easy for the reader to follow.
The authors perform flow cytometry using SYBR Green I Nucleic Acid Gel Stain to assess whether heat treatments (50°C and 100°C) and Proteinase K treatments affect cell integrity. However, to my knowledge, SYBR Green I is generally considered membrane permeable and can enter intact bacterial cells. Consequently, the use of this dye alone does not allow discrimination between intact and ruptured cells. For live/dead discrimination, a combination of dyes such as SYBR Green I and Propidium Iodide (PI) could be used. PI is largely excluded from intact cells and enters primarily cells with compromised membranes.
In this context, the authors report similar cell counts before and after heat treatment at 100°C and conclude that cell membranes largely remained intact. I find this interpretation difficult to support. A more plausible explanation is that, without double-staining, intact and disrupted cells cannot be distinguished, and that this limitation is reflected in the values presented in Table 2. Furthermore, the reduced number of detected events following Proteinase K treatment could result from the flow cytometry analysis itself and because of applied threshold values. Since Proteinase K degrades membrane-associated proteins and may compromise cell envelope integrity, DNA could be released as diffuse fragments that fall below the fluorescence threshold or outside the gating parameters of the flow cytometer, rather than being counted as cellular events.
I therefore suggest that the authors reconsider the purpose of the flow cytometry analysis and the suitability of using SYBR Green I alone for assessing cell integrity in their isolates. If the isolates are still available in stock, it may be possible to repeat the flow cytometry analysis using a dual-staining approach (SYBR Green / PI) for a live/dead assay and distinguish between intact and membrane-compromised cells? Regardless of whether additional experiments are performed, the Methods, Results (including Table 2), and Discussion sections should be revised to ensure that the interpretation of the flow cytometry data is fully supported by the methodology employed.
The authors state that 92 strains were tested for ice nucleation activity and refer to Supplementary Table 1 (page 11, line 249). However, I was unable to find any supplementary information.
Statistical tests were applied to evaluate differences between the control, heat-treated, and Proteinase K-treated samples. The authors use the pairwise Student’s t-test. However, I encourage them to justify the use of this test by assessing whether the data satisfy the assumption of normality (e.g., using the Shapiro-Wilk test). I am concerned that the ice nucleation data may not be normally distributed, in which case a non-parametric approach (Wilcoxon signed rank test) would be more suitable for the data. If three or more dependent groups/treatments (control, 50°C, and 100°C) are compared, a Friedman ANOVA test followed by an appropriate post hoc test may be more suitable. I recommend that the authors evaluate this and include a brief description in the Method section explaining the rationale for their choice of statistical analyses.
In addition, please clarify which metric (e.g., T50) was used for the statistical comparisons.
Specific comments
Throughout the manuscript, please consistently use either “underlying water” or “UW”, and ensure the abbreviation is defined upon first use.
Page 5, lines 103-106: Please clarify the number of water samples collected and the exact dates of collection.
Page 6, lines 122-126: Please specify whether the amplicon sequencing data and INA isolates originate from the same samples. At present, it is unclear whether the UW and SML samples used for these analyses were collected at the same time or on different dates.
Page 7, lines 159-160: The first letter of each sampler abbreviation used in the isolate codes appears to be missing.
Page 8, line 173: Clarify that the minimum of 106 bacterial cells per mL required to detect freezing events at high subzero temperatures is based on the strain of P. syringae studied by Maki et al. (1974). It is unclear whether this assumption applies to other INA bacterial strains, and this should be stated explicitly if generalized.
Page 8, lines 175-176: Could you clarify whether 8 or 16 droplets were used in the droplet freezing assay? The Method section states: “For each strain, eight replicates of 100 μL cell suspension were dispensed into an 8-strip PCR tube (two sets per strain) for analysis”, whereas the Results (page 11, line 254-255) state: “In addition, a threshold of ≥50% frozen droplets (4 of 8 replicates)”.
Page 8, line 178-179: Please provide a reference supporting the statement: “The PCR tubes were incubated at 4℃ for 2 h to promote the expression of ice nucleation proteins”.
Page 9, line 195: The manuscript sometimes uses “IN factor”, other times “INA factor”. Please ensure consistent terminology throughout.
Page 9, line195-198: Please specify the incubation temperature and duration used for Proteinase K treatment.
Page 10, line 228: Does the “4°C control” refer to the “unheated control” described in the previous Method section (page 9, line 204)?
Page 10, line 243: Please clarify how the retentate (>100 kDa) fraction was tested for ice-nucleating activity. Was the material recovered from the filter membrane using sterile 1 x PBS?
Table 1, caption: “six-potential ice-nucleating bacteria isolates” should be “six potentially ice-nucleating bacterial isolates”.
Page 12, lines 272- 274: The statement: Although the onset freezing temperature decreased in all strains after both treatments, the final freezing temperature (the point at which all droplets froze) remained largely unchanged for strains 4U19, 5M3, and 5G21” appears to be supported only for strain 4U19, based on the results shown in Fig. 2. Please clarify.
Page 12, line 291: Please specify what statistical test was used?
Table 2: Why is the initial cell concentration for replicate 2 not normalized to a relative value of 1, as done for replicate 1? While this approach may facilitate within-replicate comparison, it complicates comparison across treatments.
Page 14, lines 337-340: A supporting reference is needed.
Page 15, line 344-345: The statement “with droplets freezing completely at -20°C” should be revised, as droplet either freezes or do not. Please rephrase to indicate that all droplets froze at the respective temperatures.
Page 15, line 363: Please ensure consistent use of the abbreviation “UW”.
Page 16, line 384-387: It would be more accurate to associate Type I, II, and III classifications with ice-nucleating proteins rather than bacteria, as these categories refer to protein types rather than bacterial taxa.
Page 16, lines 387- 389: The statement “Although these bacteria nucleate ice at colder temperatures, they remain relevant to atmospheric processes due to their potential release from marine environments into the lower troposphere” requires a supporting reference.
Page 16: lines 390- 392: “Previous studies have shown that most culturable INA-positive strains from precipitation are Gammaproteobacteria (Failor et al., 2017), though INA activity has also been reported among other bacterial lineages (Ponder et al., 2005; Mortazavi et al., 2008)”.
Please specify which other bacterial lineages are referred to here?
Page 17, line 411 and 414: “INA” is defined here as ice-nucleation activity, but this abbreviation has not been previously introduced. Earlier in the manuscript “INA” has been referred to as “ice nucleation-active”. Please ensure consistent terminology throughout.
Page 19, lines 454- 456: A reference is needed to support the following sentence: “Once airborne, such cells or fragments could act as ice nuclei, initiating freezing in mixed-phase clouds at relatively warm subzero temperatures”.
Page 19, lines 458-460: The statement regarding halotolerant ice-nucleating microbes in coastal precipitation would benefit from inclusion of the taxonomic identities of the organisms discussed, to improve clarity for the reader: “The detection of halotolerant, ice-nucleating microbes in coastal precipitation (Beall et al., 2021) supports the idea that marine-derived bacteria, including those reported here, can survive atmospheric transport and influence cloud glaciation”.
Technical corrections
Page 2, line 32: “INA” has not been defined previously.
Page 3, line 54: “films” should be “film”.
Page 3, line 62. “underlying water” is introduced for the first time here. Please define the abbreviation “UW” and ensure consistency.
Page 3, line 67: “a major contributors” should be “a major contributor” or “major contributors”, depending on intended meaning.
Page 4: line 81: “atmopsheric” -> “atmospheric”.
Page 4: line 87: “intristic” -> “intrinsic”.
Page 4: line 90: “a relatively warm subzero temperatures” -> “relatively warm subzero temperatures”
Page 4, line 93: “underlying water” -> “UW”.
Page 5, line 96: “underlying water” -> “UW”.
Page 5, line 106: “clam” -> “calm”.
Page 7, line 145: “length16S” -> “length 16S”.
Page 7, line 165: “subsequently” -> “Subsequently”.
Page 9, line 212: “ml-1” -> “mL-1”.
Page 10, line 240: “the” should be “then”.
Page 11, line 255: “are” -> “is”
Page 14, line 332: remove “(IN)”.
Page 15, line 358: “INA-active strains” -> “INA strains”.
Page 15, line 364: “INA bacteria isolates” -> “INA bacterial isolates”.
Page 16, line 383: Remove the abbreviation “(INB)” as it is not used elsewhere in the manuscript.
Page 16, line 392: “INA activity” should be “ice nucleation activity” or “IN activity”.
Page 18, line 417: “INA activity” -> “ice nucleation activity” or “IN activity”.
Page 18, line 437: “INA activity” -> “ice nucleation activity” or “IN activity”.
Page 18, line 439: “INA-active” should just be “INA”.
Page 19, line 444: “INA-active” should just be “INA”.
Page 19, line 454: “sea-spray aerosols (SSA)” -> “SSA” as it has already been defined earlier in the manuscript.