3.1. Classical Calorimetry and Bulk Cellular Thermometry

Cellular thermal measurement began with bulk approaches, in which the total heat generated by cell populations—such as cell suspensions, cultured cells, or tissue fragments—was quantified as heat flow using isothermal heat-conduction microcalorimetry, typically for populations comprising on the order of \(10^{5}\)–\(10^{6}\) cells. The underlying principle is to convert the temperature difference between a sample (cells) and a reference (e.g., culture medium) under isothermal conditions into an electrical signal, thereby enabling the continuous measurement of heat flow, or heat production rate, on the microwatt (μW) scale ²⁷. At this stage, the primary objective was to interpret cellular metabolic states through heat production. By around 1990, an analytical framework had been established in which heat flow from cultured cells was combined with oxygen consumption measurements (respirometry), and metabolic modes were characterized in terms of the heat-flow/oxygen-flow ratio (the calorimetric–respirometric ratio) ²⁸. In 1986, cultured adherent cells attached to microcarriers—small scaffold beads employed for the large-scale suspension culture of adherent cells—could be analyzed under stirred suspension while maintaining physiological conditions that closely approximated those of conventional cell culture ²⁹. This development made it possible to measure not only suspended cells but also cells retained under standard culture conditions. The temperature sensors used in these bulk measurements were electrical devices, such as thermocouples and thermistors. However, reports of cellular temperature measurements using comparable approaches appear to have remained relatively limited in the scientific literature until 2002.

3.2. The Shift Toward Single-Cell Measurement: Breakthroughs Enabled by Optical Methods

A major turning point in cellular thermal measurement came in 2012. In that year, Okabe et al. reported intracellular temperature mapping using an FPT in combination with fluorescence lifetime imaging ¹⁴. Sedlmeier et al. reported localized intracellular temperature measurements using upconversion nanoparticles that emit visible light under near-infrared excitation ¹⁵. Oyama et al. developed a method in which temperature-sensitive nanoparticles were introduced into organelles and used to monitor local temperature while moving within cells ¹⁶. Intracellular temperature mapping based on the fluorescence polarization anisotropy of GFP was also reported ³⁰. Collectively, these studies suggested that even within the same cell population, individual cells may exhibit distinct thermal behaviors and that temperature may also be spatially heterogeneous within a single cell. If heat production is inherently a local phenomenon, its essential characteristics may be obscured by population-averaged measurements. These findings highlighted the need for single-cell thermometry and marked a shift in the field from an emphasis on the “average” to the “local.”

This period was followed by reports of local intracellular thermometry based on quantum sensing using nitrogen-vacancy centers in nanodiamonds by Kucsko et al. ³¹, endoplasmic-reticulum-selective thermal imaging employing small-molecule fluorescent probes by Arai et al. ³², and efforts to directly measure organelle-specific heat production using organelle-targeted GFP-based sensors ³³. A key conceptual breakthrough of this era was that temperature was no longer measured through direct physical contact but was instead transduced into temperature-dependent optical signals—such as fluorescence intensity ratios, fluorescence lifetimes, and spectral shifts—for optical readout. This development opened the possibility of non-contact, high-spatial-resolution measurements of local temperature within living cells.

Optical approaches to intracellular thermometry can be broadly classified into thermoresponsive polymeric probes, inorganic nanoparticles, fluorescent proteins, and small-molecule fluorescent probes (organic dyes) (Fig. 1) ^13,14,34. These approaches have attracted considerable interest because they enable the non-contact, high-spatial-resolution visualization of temperature changes within single cells. At the same time, they differ substantially in their temperature-sensing mechanisms, intracellular delivery strategies, targeting capabilities, and quantitative reliability.

Thermoresponsive polymeric thermometers were among the representative tools that supported the early demonstrations of intracellular temperature mapping because they exploit temperature-dependent changes in fluorescence intensity, fluorescence lifetime, or ratiometric signals for temperature readout ^14,34,35. Designs that incorporate multiple functional units into a polymer backbone provide considerable flexibility for tuning temperature sensitivity and implementing internal referencing, making them well suited for wide-field imaging and FLIM ^14,35. At the same time, the influence of intracellular environmental factors other than temperature on the fluorescence response cannot always be completely excluded, and caution is therefore required when interpreting the measured signals ^34,35. In addition, heterogeneity in probe distribution and localization within the cytoplasm may contribute to uncertainty in the spatial comparison of absolute temperatures.

Representative inorganic nanoparticle thermometers include upconversion nanoparticles and nanodiamonds, which generally exhibit high photostability and are well suited for long-term observation ^15,31,34,36. These systems provide substantial flexibility because multiple physical quantities—such as emission wavelength ratios, luminescence lifetimes, and spin states—can serve as temperature indicators. Upconversion nanoparticles are particularly attractive because near-infrared excitation reduces background autofluorescence and light scattering ¹⁵. However, intracellular delivery pathways and localization depend on particle design and cellular uptake mechanisms, and in practice these particles may preferentially accumulate in intracellular compartments such as endosomes or lysosomes ^16,34. In addition, particle–cell interactions, biocompatibility, and the influence of non-thermal factors on the temperature readout must be carefully evaluated.

Fluorescent protein-based thermometers can be genetically expressed within cells, eliminating the need for exogenous probe delivery and making them well suited for long-term observation and lineage tracing ^30,33,34. Through signal peptides and fusion-protein design, they can also be targeted to specific organelles, such as mitochondria, the endoplasmic reticulum, and the nucleus, making them advantageous for organelle-selective thermal imaging. However, because fluorescence responses may also be influenced by biochemical and optical factors other than temperature, caution is required before directly interpreting fluorescence changes as absolute temperature changes. Indeed, the reliability of intracellular thermal readout using GFP-based thermometers remains a subject of ongoing debate.

Small-molecule fluorescent probes (organic dyes) offer a high degree of flexibility in molecular design, enabling the relatively facile tuning of temperature sensitivity, response kinetics, and targeting moieties ^32,34. Consequently, they are well suited for local temperature measurements targeted to specific organelles, and highly sensitive endoplasmic-reticulum-selective thermal probes have indeed been reported. In addition, small-molecule probe systems are comparatively easy to introduce exogenously and can be chemically modified to suit the intended application. On the other hand, photobleaching, cellular leakage, metabolic modification or degradation, and cross-sensitivity to environmental factors other than temperature may limit their applicability, particularly for long-term observation and absolute temperature quantification. As a result, although small-molecule probes are well suited for detecting local and relative temperature changes, rigorous quantitative thermometry requires careful calibration under the specific measurement conditions employed ^32,34.

3.3. A Major Point of Debate: The \(10^{5}\) Gap Issue

Any discussion of cellular thermometry must address the so-called “\(10^{5}\) gap issue.” This refers to the observation that temperature increases reported by single-cell measurements are approximately five orders of magnitude larger than those predicted by heat-conduction calculations. In other words, the experimentally measured values and theoretical estimates differ by a factor of approximately \(10^{5}\). In 2014, Baffou et al. argued that, based on simple heat-conduction calculations and scaling laws, large steady-state temperature gradients within a single cell arising solely from endogenous heat generation are difficult to justify on physical grounds ³⁷. This argument cast doubt on the physical consistency of the large local temperature changes reported at the time using fluorescent thermometers and raised the question of whether the observed signals genuinely reflected temperature. Since then, the debate has extended to encompass measurement principles, calibration strategies, theoretical models, and the understanding of intracellular heat transport. In 2015, Kiyonaka et al. responded by emphasizing the need for careful validation when interpreting temperature changes observed within single cells, while also supporting the plausibility of subcellular thermal changes detected using fluorescent temperature sensors ³⁸. Around the same time, Suzuki et al. explicitly framed the discrepancy between theoretical calculations and experimental measurements as the “\(10^{5}\) gap issue” and argued that it could largely be resolved by properly accounting for several factors: the cell type and the presence or type of stimulation (approximately one order of magnitude), the distance from the heat source used to evaluate the temperature rise (approximately one to two orders of magnitude), and the thermophysical properties of the cell, once these are better characterized (approximately one to two orders of magnitude) ³⁹. Baffou et al. subsequently responded, arguing that even if local intracellular heating were inferred from experimental measurements, such observations remained inconsistent with existing theoretical estimates ⁴⁰. The 2018 review by Okabe et al. further suggested that the straightforward application of macroscopic heat-diffusion models, which treat the cell interior as a homogeneous, continuous aqueous phase, may have important limitations ³⁴. They argued that these models should be re-examined by taking intracellular heterogeneity and the spatial scale of heat sources into account, and they identified the clarification of the chemical identity of intracellular heat sources and the development of mesoscopic nonequilibrium thermodynamics as important future challenges. The issue has not yet been fully resolved.

3.4. Contact-Based Approaches Using Microfabricated Sensors

Although optical methods are now widely associated with single-cell temperature measurement, the first realization of single-cell thermometry was in fact reported by Johannessen et al. in 2002. Advances in microfabrication technology during the 2000s enabled the miniaturization of calorimetry, thereby reducing sample volumes while increasing measurement sensitivity. Johannessen et al. developed a heat-conduction nanocalorimeter based on silicon microfabrication for pL-scale samples and reported the detection of small thermal events at the level of one to a few cells ⁴¹. Another approach employed an intracellular thermocouple probe consisting of a tungsten needle and a platinum film, with a tip diameter of approximately 500 nm, to directly measure temperature changes within single cells during drug stimulation. The reported temperature resolution was below 0.1 °C, and the response time was estimated by simulation to be approximately 400 ns ⁴². In addition, a GaN-nanomembrane-based optical microdevice was used to evaluate the thermal transport properties of cells by detecting temperature-dependent photoluminescence shifts, and thermal conductivities of several cancer cell lines were approximately 0.5–0.6 W m\(^{-1}\) K\(^{-1}\) ⁴³. Although several microscale temperature sensors capable of measuring small-volume biological samples, including biomolecules, have since been reported, published studies involving measurements on living cells remain extremely limited.

The authors have progressively advanced micro-temperature sensors based on microfabrication technology. By complementing the dominant optical approaches with high temperature resolution and rapid response, they have reported previously unobserved thermal phenomena in cells, as well as the quantitative evaluation of cellular thermophysical properties. In an early study published in 2012, a microsensor based on mechanical resonance was used to measure the temperature of a single suspended cell (Figs. 2(a)–(c)) ⁴⁴. Conventionally, mechanical resonant sensors are operated under vacuum to minimize vibrational damping and were therefore considered incompatible with live-cell measurements in aqueous environments. This limitation was overcome by integrating the resonant sensor into a microfluidic chip such that it extended through both a fluidic region containing the cells and a physically separated vacuum region. By adopting a doubly clamped beam structure, the system ultimately achieved a temperature resolution of 79 μK and a heat resolution of 1.90 nW ⁴⁵. An important biological finding of this study was that cells exhibited periodic, pulse-like heat generation even in the absence of chemical stimulation (Fig. 2(d)). While gradual temperature increases over several tens of minutes in response to chemical stimulation had already been documented during the era of bulk-calorimetry, such periodic, pulse-like heat generation had not previously been reported. Although this finding was initially met with skepticism, the same phenomenon was subsequently observed when the sensing element was replaced with a thermistor, supporting the conclusion that the measured signal originated from the cell ⁴⁶.

The target of measurement was subsequently extended to cultured cells. In a 2020 study, a device incorporating a VO\(_{2}\) microthermistor was employed to analyze the frequency spectrum of temperature signals from a single cultured cell ⁴⁷. This work was motivated by the earlier observation of periodic heat generation, and the frequency spectrum was found to vary with the ambient temperature. Specifically, when temperature signals acquired at 25 °C and 37 °C were compared, the signal intensity was greater at 37 °C, particularly in the frequency range below 2 Hz. Because 37 °C corresponds to the temperature range in which cells are highly active, these results suggested that intracellular biochemical processes may exhibit periodicity and that their dominant frequency components may lie below 2 Hz. Gradual temperature increases induced by chemical stimulation were also observed in this system, confirming its capability to detect heat generation from single cells. At that time, the system achieved a temperature resolution of 11 m°C and a response time of less than 0.1 s.

More recent studies have gone beyond the observation of temperature changes and have achieved the quantitative evaluation of thermophysical properties, such as the thermal conductivity and specific heat of single cells. In a 2023 Lab on a Chip paper, an on-chip-integrated microthermistor combined with localized heating by a focused infrared laser was employed to construct a measurement system in which thermal conductivity was derived from the steady-state temperature rise during heating, whereas specific heat was determined from the transient temperature-rise response (Figs. 3(a)–(f)) ⁴⁸. Thermal conductivity at the single-cell level had been reported in several earlier studies because it can be obtained from the steady-state temperature rise. Specific heat, by contrast, requires dynamic measurements, and its determination at the single-cell level was therefore demonstrated for the first time in that study. The study further showed that these thermophysical properties depend on environmental temperature and heating frequency (Fig. 4). It also suggested that cellular temperature signals may be controlled by external heating; that is, biochemical reactions may potentially be induced or regulated by heating applied within the same frequency band as the cell’s intrinsic heat generation. In addition to the studies introduced here, single-cell thermometry using microfabricated sensors has also been reported by other groups, and readers are referred to Reference ⁴⁹ for further details. Table 1 summarizes the representative cellular temperature measurement techniques discussed in this review, highlighting their differences in detection principle, temperature resolution, spatial resolution, and response time.

Another important question is why single-cell temperature measurement using microfabricated sensors did not attract widespread attention for a long time, despite having been realized earlier than thermosensitive nanoparticles. Initially, cellular temperature measurements using microfabricated devices focused on responses to known chemical stimuli and were positioned more as applications for validating device performance than as tools for discovering previously unknown phenomena or elucidating the physical principles underlying cellular behavior. As a result, the biochemical insights obtained from such measurements were relatively limited, and these studies were cited less frequently within the biophysics community. These studies subsequently helped establish microfabricated thermal sensors not only as demonstrations of device technology but also as powerful tools for investigating cellular thermal phenomena.