Introduction

An environment is often best understood by the physiology of the organisms that inhabit it. In microbial ecology, the description of how microorganisms contribute to the functioning of their ecosystem is a daunting task, owing to their phylogenetic and physiological diversity, and our limited capacity to study them in their habitat, e.g., the deep sea. Categorization of microbes into functional groups that reflect either specific processes or habitat constraints is one approach to address this. However, this requires robust definitions of such groups. Where specialized structures or metabolic pathways exist, definition of such groups based on diagnostic genes or proteins is often possible. In contrast, where groups are to be defined based on habitat constraints for cell growth multidimensional stimuli must be considered, which is often challenging. The variation of hydrostatic pressure (HP), temperature (T), salinity, pH, oxygen availability, water activity, and radiation over evolutionary times contributed to shape the present microbial diversity, and still define the countless ecological niches on Earth. The biochemical adaptation to these different physicochemical stimuli may be similar (e.g., increase in unsaturated fatty acids [in Bacteria] or glycerol ether lipids [in Archaea] at high HP or low T; increase in ω-alicyclic fatty acids at high T or low pH; accumulation of polar organic solutes at low water activity or high T, HP, salinity or radiation; Mn2+ accumulation at high radiation or low water activity [1]). Nonetheless, some trade-offs exist which limit the capacity of microbes to grow under different combinations of conditions [2]. A clear definition of the operative boundaries imposed by such factors is key to the conceptual and practical use of functional groups. HP in the deep sea is one example.

HP is the force exerted on an area by a fluid at rest. The most pervasive effect of HP is its influence on intermolecular distances, where in addition to gas volumes it affects conformation of polynucleotides (DNA and RNA), lipid bilayers, and multimeric proteins [3]. Increased HP is experienced by planktonic crustaceans, and marine invertebrates, fish, and deep-diving mammals, and is characteristic of the largest microbial habitats on Earth i.e. the deep sea and subseafloor [4] where Bacteria and Archaea thrive at tens of MPa [5]. In food and medicine production, elevated HPs are managed to inactivate HP-sensitive pathogens and viruses [6]. However, a systematic categorization of microbial fitness constraints in response to HP is missing. Such preference is generally referred to as piezophily, descriptive of microorganisms growing better at increased HP, with obligate piezophiles (or hyperpiezophiles) unable to grow at ambient pressure, and piezosensitive strains growing best at ambient pressure. Although such simple concepts recognize the effect of HP on microbes, they do not yet provide a useful basis to truly define how HP gradients influence biogeochemical processes. At what depth are piezosensitive microorganisms outcompeted by piezophiles? Do such transitions shape the rates of key processes for ecosystem modeling? Are transition points similar in all water bodies or do they vary with other physicochemical properties such as T or salinity?

HP linearly increases by 1 MPa every 100 m below seawater level (bswl), and about three times as much below seafloor level (bsfl) [3]. Enhanced growth at elevated HP has been used to locate the beginning of the piezosphere at corresponding depths (e.g. 1000 m bswl if HP was equal to the corresponding ~10 MPa). However, different HP thresholds were proposed for both piezophiles and hyperpiezophiles: ≥10 MPa [7]; ≥10 MPa for piezophiles and ≥50 MPa for hyperpiezophiles [8]; >0.1 MPa for piezophiles and ≥60 MPa for hyperpiezophiles [9], but indicating moderate piezophily if the HP optimum [HPopt] of an isolate under investigation was between 10 and 30 MPa [10]; ≥40 MPa (as the average oceans depth is ~3750 m bswl [11]). Following deep-sea sampling, isolation, and lab-scale testing, a seminal work by Yayanos in 1986 proposed that ‘true’ piezophiles populated waters at least 2000 m bswl (≥20 MPa). This work introduced the significant concept of PTk diagrams, where growth rates are plotted in three-dimensional graphs versus HP and T. Such multidimensional space describing growth limits aligns well with the principle of ecological niches formalized by Hutchinson [12]. Defining piezophily by relationship between HP and growth rate is limited by the fact that other parameters (e.g. T) also impact growth rate. PTk diagrams provide a means to visualize the relationship between HP, T and growth rate, with the maximum growth rate [μmax] observed when concomitantly adjusting HP and T to an optimum. Yayanos work was almost exclusively based on psychrophiles (microorganisms with a T optimum [Topt] ≤15 °C). The isolation of several new piezophiles in recent years, particularly with Topt >15 °C, is an opportunity to revise the role of T on the growth of piezophiles.

Deep-sea environments exposed to elevated HPs commonly experience low T (<5 °C). At polar regions, T decreases <5 °C already at ~100 m bswl, while at warmer low and middle latitudes T decreases rapidly along a permanent thermocline to ~5 °C at ~1000 m bswl [13]. Exceptions are the deep, warm Sulu, Mediterranean, and Red Sea (~10, 13.5, and 20 °C at seafloor, respectively, at depths of ~4400, 5270, and 3040 m bswl). At hydrothermal vents as deep as 5800 m bswl (http://vents-data.interridge.org) microorganisms may grow >110 °C. In deep sub-seafloors, T increases 25 °C every km underground [9, 14]. Notwithstanding the many HP–T combinations in the environment, this correlation has not been systematically addressed. This lack of information can lead to a poor understanding of the true operative boundaries of a microorganism. For instance, thermophilic isolates collected from surface waters and theoretically belonging in that environment have shown to grow best at higher T when concomitantly increasing HP, in both Archaea (Methanococcus thermolithotrophicus [15]) and Bacteria (Clostridium paradoxum [16]).

Piezophiles separate in three functional groups based on T

To date, the most commonly accepted definition for piezophily states that piezophiles show a μmax above the atmospheric pressure of 0.1 MPa. At the time of writing, there are only 86 documented microbial isolates with a HPopt > 0.1 MPa, an incredibly small number for a condition featuring the largest reservoir of prokaryotes on the planet [4]. To determine if T exerts an ecologically-relevant effect on the relationship between HP and optimal growth rate, all μmax values in described piezophiles were plotted versus either HPopt (Fig. S1A) or Topt (Fig. S1B). This would explain growth based on HP or T independently of one another. High HPopt is consistent with low μmax while the opposite is true for Topt. However, no strict correlation appears evident. Piezophiles were then divided for T preference neglecting HPopt. Microbial preference for T does not result in subgroups with precise boundaries [17]. However, microorganisms with Topt ≤15 °C are generally referred to as psychrophiles [18] and those with Topt ≥50 °C as thermophiles [19], the resulting mesophiles having 16 < Topt < 49 °C. The average μmax in piezopsychro-, piezomeso- and piezothermophiles increases with T namely 0.22 ± 0.13, 0.42 ± 0.38 and 1.25 ± 0.84 h−1 (Table 1, all data in Table S1). Only the average μmax in piezothermophiles is significantly different from the other groups (t-test, p < 0.0011). The opposite approach (i.e. neglecting T classification) was attempted using the HP thresholds suggested by Yayanos [9]: piezophiles >0.1 MPa up to 60 MPa; and hyperpiezophiles >60 MPa, however, no significant difference was found (p > 0.05, average μmax equal to 0.55 ± 0.61 and 0.41 ± 0.69 h−1, respectively).

Table 1 Described piezophilic isolates and their main features. In (A), piezopsychrophiles; in (B) piezomesophiles; in (C) piezothermophiles.

When accounting for the combined effects of HP and T on growth, a distinction of functional piezophilic groups emerges. Since both HP and T influence growth rate, the HPopt and Topt of every piezophile was divided by their μmax to normalize the relationship across the available strains (Fig. 1). This revealed a correlation between HP and growth rate only when separating piezophiles in the three T-defined subgroups piezopsychro-, piezomeso- and piezothermophiles, as it can be inferred by the large variations in the axes of Fig. 1. Within each subgroup, the normalized HPopt and Topt of all strains have an exponential correlation (R2 equal to 0.62 (n = 37), 0.78 (n = 12) and 0.58 (n = 17), respectively). The data points for Profundimonas piezophila YC-1 and Rhodobacterales bacterium PRT1 were removed from piezopsychrophiles, as they are the slowest piezophiles isolated so far (μmax < 0.02 h−1) and plot out of scale. Their inclusion increases R2 from 0.62 to 0.87. For the same reason Archaeoglobus fulgidus VC-16T was removed from piezothermophiles, its inclusion slightly reducing R2 from 0.58 to 0.54.

Fig. 1: Correlation between optimal HP and temperture with respect to maximum growth in piezophiles.
figure 1

The rate increase between HPopt (HPopt/μmax) and Topt (Topt/μmax) is described for piezopsychro- (A), piezomeso- (B), and piezothermophiles (C). Statistical correlation was obtained with GraphPad Prism 5, nonlinear regression, exponential growth equation, least square (ordinary) fit.

Variations in T and HP can affect membrane fluidity (phospholipid fatty acids packing and conformation, affecting proton influx/efflux) and protein folding (structural disruption affecting activity). The subdivision of piezophiles into three functional T subgroups suggests that the evolutionary cellular adaptation to the constraints imposed by HP occurred within the boundaries of three subsets of T. However, the environment is not a two-dimensional space described by HP and T alone. Aside T [20], growth rates at increased HP may also depend on diversity and concentration of available nutrients [21], main carbon and energy substrate [7, 22], pH [23] and salinity [2]. For instance, variations in salinity in cold (e.g. Orca Basin, 5 °C), warm (e.g. Nereus Basin, 30 °C) and hot (e.g. Atlantis II Deep, up to 68 °C) deep brine pools (~2400, 2500, and 2200 m bswl, respectively) may have additionally imposed a considerable water activity stress, affecting protein folding and turgor pressure. Cells maintain iso-osmosis with the environment by intracellular accumulation of salts (requiring the molecular adaptation of all intracellular enzymes) or organic compatible solutes (an energy-demanding strategy) [24]. Understanding how these halophiles’ evolution has been constrained by concurrent HP and T boundaries requires a sizeable collection of isolates.

Estimating habitat preferences

The relationship between piezophiles’ capture depth and HPopt is reported (Fig. 2A). The almost linear correlation found with piezopsychrophiles (R2 = 0.69, n = 48) may reflect their habitat, the contiguous cold seawaters where HP increases linearly to the seafloor (Fig. 2B, as derived from Figs. S2–S4). The small discrepancy between the linear increase in HP with capture depth (HPcapt) and HPopt (dotted vs. straight line, Fig. 2A) indicates that in piezopsychrophiles HPopt < HPcapt, consolidating previous observations on few isolates [25, 26]. As cold T and high HP impose similar constraints on cells [27], piezopsychrophiles may also inhabit permanently cold surface waters at polar regions (−1.8 to 5 °C [13]; yellow lines, Fig. 2B). Piezomesophiles inhabit warm and deep anaerobic sediments up to ~2500 m bsfl. However, in 4/6 isolates collected underground HPopt«HPcapt (at least 2.4 times), suggesting that definition of piezomesophiles true maximum HPopt in sub-seafloors may require improved HP retainers. The warm, geographically limited seawaters of the Mediterranean, Sulu, and Red Sea (in orange, Fig. 2B) are another habitat for piezomesophiles, although rarely collected there (1/11, Table S1B). The irrelevant correlation between their capture depth and HPopt (R2 = 0.03, n = 11) highlights the high resilience of sinking piezomesophiles in colder, deeper seawaters (HPcapt ≥ HPopt in 9/11) where they may compete with piezopsychrophiles. Similarly, high T tolerant piezomesophiles (as those collected close to hydrothermal vents; 4/11, Table S1B) may compete with piezothermophiles. Piezothermophiles are mostly Archaea (15/21). No correlation between capture depth and HPopt is evident (R2 = 0.07, n = 21), even when removing the six Bacteria (R2 = 0.11, n = 15). While piezothermophiles’ most obvious habitat are deep hydrothermal vents (18/21, Table S1C; in red, Fig. 2B), those collected at hot vents have HPopt > HPcapt (16/18), possibly because the cellular constraints imposed by high T can be compensated with increased HP [15, 16, 28,29,30,31].

Fig. 2: Correlation between optimal HP and capture depth in piezophiles, and estimation of their habitat.
figure 2

In (A) the straight line indicates the linear regression for piezopsychrophiles (R2 = 0.69, n = 48) as obtained with GraphPad Prism 5, linear regression; the dotted line indicates the linear increase of HP with increasing depth in seawater; all data in Table S1. Keys reported in the graph. In (B) the global topography reporting oceans depth, surface temperature, and plate boundaries was obtained by over imposing three maps from the NOAA (National Center for Environmental Information; all data available in Supplementary Information). Keys: color scale for oceans seawater depth reported on the map as meters below seawater level (m bswl). Yellow lines indicate where surface temperature at polar seas is permanently <5 °C (data averaged from 1985 to 2010); thus, at low and mid latitude in between yellow lines surface temperature can be higher than 5 °C. Red lines indicate plate boundaries, where hydrothermal vents most commonly are found. Red dots indicate hydrothermal vents not on plate boundaries. The warm and deep Mediterranean, Sulu and Red Sea are highlighted in orange.

Competitive advantage of piezophiles

The piezosphere is also inhabited by piezosensitive microorganisms [32]. The minimum HP setting a competitive advantage for piezophiles remains fairly unclear. The lowest HPopt observed so far differs according to T: in piezopsychrophiles, eight isolates have HPopt between 10 and 20 MPa (except for Shewanella sp. SC2A at 7 MPa, whose growth rate is however almost identical until 14 MPa, 0.076 vs. 0.072 h−1, respectively [25]); in piezomesophiles, the lowest HPopt is 10 MPa; in piezothermophiles it is 20 MPa (except for Thioprofundum lithotrophica 106 at 15 MPa). The present data thus suggests that at ≥20 MPa piezophiles consistently possess a competitive advantage irrespective of T. This aligns with Yayanos’ proposal on piezopsychrophiles [20]. Nonetheless, piezopsychro- and piezomesophiles may have an increasingly competitive advantage already ≥10 MPa. This may indicate that piezophiles could be collected from deep freshwater lakes as Lake Baikal and Tanganyika (1642 and 1470 m depth, respectively, at 16 and 14 MPa).

Hyperpiezophiles

Hyperpiezophiles are microorganisms that cannot grow at ambient pressure. There are several hyper-piezopsychrophiles (11/48), no isolated hyper-piezomesophiles, and just one isolated hyper-piezothermophile (Pyrococcus yayanosii CH1) (Table 1). The lowest HPopt of hyper-piezopsychrophiles is 50 MPa (3/11 strains) and their shallowest capture depth is 6000 m bswl (Table S1), which aligns with the depth of abyssal plains. The fact that HP at the onset of abyssal plains is slightly higher than hyper-piezopsychrophiles’ lowest HPopt (~60 vs. 50 MPa) mirrors the small discrepancy noted for all piezopsychrophiles (Fig. 2A), overall indicating that hyper-piezopsychrophiles are autochthonous in deep, cold hadal trenches.

Updated definitions

The most widely shared definition of microbial preference for increased HP states that a microorganism is piezophilic when its μmax is observed at HPs >0.1 MPa. By setting the threshold to such a low value, this definition neglects the large variation in HP preference among described piezophiles, and the differential effects enhanced HP may impose on the vast diversity of microbial processes in nature. Assessing optimal growth rates by cultivation is required to identify the exact threshold level above which HP-adapted microorganisms clearly separate from those thriving at ambient pressure. In the present meta-analysis, the relevance of HP–T combinations first described by Yayanos [20] was updated to include all currently described piezophiles. The following updated definitions and perspectives on their application are proposed:

(1) HP–T relationship constrains μmax, and defines three functional groups based on T: piezopsychro-, piezomeso-, and piezothermophiles. These functional categories should be used to understand how piezophiles contribute to the functioning of deep-sea environments experiencing different HP–T combinations.

(2) Capture depth is a poor predictor of piezophilic traits, as piezosensitive and piezophilic groups are intermixed in the oceans.

3) A competitive advantage to piezophiles over piezosensitive is predicted to begin at 10 MPa and to consistently exist irrespective of T at HPs ≥20 MPa. Ecological modeling should specifically account for HP effects on biogeochemical processes beyond this point.

4) Hyper-piezopsychrophiles are autochthonous in hadal trenches. Their competitive advantage over piezopsychrophiles begins at HP ≥ 50 MPa. This threshold should be considered for ecological modeling of hadal trenches.