Membrane nanotubes (MNTs) are nanotubular cell-to-cell connections enabling cell-to-cell signaling and cargo transfer. The presence of local MNT bulges has been reported by several studies. However, a detailed characterization of MNT bulges concerning their geometrical properties has not been done yet. The aim of our study was to analyze MNT from cone-like photoreceptor cells (661 W) using scanning electron microscopy (SEM) in order to characterize MNT bulges, thereby increasing the knowledge of ultrastructural feature of MNTs.

Our SEM analysis of two MNTs with multiple bulges revealed that (i) MNT bulges are characterized by a statistically significant local increase in the MNT diameter (125% and 250% for both MNTs analyzed), (ii) the thickness and length of the MNT bulges correspond to dimensions of mitochondria, peroxisomes and exosomes, and (iii) the MNT bulges seem not to be randomly distributed on the MNTs but exhibit a preferred spacing with a different median value for each MNT. Our findings highlight that the ultrastructure of MNT exhibits interesting properties that need to be further investigated.


Scanning Electron Microscopy

Figure 1. Scanning electron microscopy (in lower secondary electron image mode) images of bulges of MNT between cone-like photoreceptor cells (661W).

(A, C) Two MNTs (MNT#1 and MNT#2) with bulges (indicated with red arrows). Scale bar of image A: 10 µm. Scale bar of image C: 1 µm. In figure C, a single bulge of a third MNT is indicated with a yellow arrow. The two MNTs are connected at one side with the cell and on the other side unconnected. The disconnection seems to be the result of the SEM staining process, possibly disrupting the state of the MNTs causing the breakage of long MNTs in some cases.

(B) Magnification of a MNT bulge selected from image A.

(D, E) Magnification of two MNT bulges selected from image C. In D the bulge is clearly visible where in E it is more difficult to recognize the bulge.

(F) Diagrammatic drawing of an MNT with two bulges and the four parameters determined for further analysis: MNT diameter (DMNT), diameter of the bulge (DB), distance between adjacent bulges (dBB) and the length of the bulges (LB). In the lower visualized MNT bulge in F an oval is shown (red) that was used to determine LB.

(G, H) MNT diameter and bulge diameter values, as well as, distance between adjacent bulges for both MNTs. Statistically significant (Bayer factor >10, Bayesian independent samples t-test) differences are indicated (*).

It should be noted that MNT#1 is approximately twice as wide as MNT#2. Smaller cargo in MNT#1 might therefore not be visible.


Nanotubular cell-to-cell connections termed “tunnelling nanotubes” or “membrane nanotubes” (MNTs) (the term used in this paper) enable diverse possibilities of cell-to-cell signaling and cargo transfer[1][2][3]. This includes the exchange of signal carriers (e.g. proteins), organelles (e.g. mitochondria), bacteria, viruses, exosomes, DNA, RNA, or electric long-range coupling.

The defining characteristics of MNTs are still being debated and the investigation of MNT properties is ongoing but recently a consensus of leading MNT researchers was published that MNTs can be defined as “tubular membrane connection between non-adjacent cells that allow direct intercellular communication, not necessarily gap junction-mediated”. MNTs also contain F-actin and/or tubulin, have a variable diameter of 50–800 nm and are open-ended[3]. It can be added that MNTs are filled with cytoplasm and have a lipid bilayer[9]. Some MNTs also contain microtubules[10] and have the gap junction protein Cx43 at the end[6]. In some cases, individual MNTs stick together to form a single, thicker, MNT[9].

As recently summarized by Rustom[11], MNTs are “intimately linked to the physiological state and pathological” state of cells and “represent a central joint element of diverse diseases, such as neurodegenerative disorders, diabetes or cancer”. Furthermore, MNTs seem to play an important role in long-range physical cell-to-cell signaling in multicellular organisms possibly enabling novel ways of physical signal transfer[12] and being involved in the functioning of neurosystems[13].
Studies have been published that report the existence of bulges, i.e. local increases of the diameter, of MNTs. This ultrastructural feature of MNTs was attributed to the presence of objects (vesicles or organelles) inside MNTs[6].


Our objective was to document and analyze bulges of MNTs using scanning electron microscopy (SEM) to increase the knowledge of ultrastructural feature of MNTs. To our knowledge, a detailed characterization of MNT bulges concerning their geometrical properties has not been done before.

Results & Discussion

By analyzing cell cultures of the cone-like photoreceptor cell line 661W via SEM, we found two MNTs (MNT#1 and MNT#2) that clearly showed multiple bulges along their axis (see Fig. 1A–E). MNT#1 had a total length of 61.60 µm and MNT#2 of 13.74 µm.

The MNT bulge diameter (DB) was statistically significant larger for both MNTs compared to the MNTs diameters (DMNT) determined on bulge-free parts of the MNTs (median, 95% confidence interval (CI)); DB#1 = 247.5 (220.0, 440.0) nm (n = 6) vs. DMNT#1 = 110.0 (55.0, 220.0) nm (n = 100), Bayes factor (BF) >100, effects size (Cohen’s d): d = -4.389; DB#2 = 85.0 (68.0, 136.0) nm (n = 4) vs. DMNT#2 = 68.0 (51.0, 102.0) nm (n = 100), BF = 23.4, d = 1.609) (see Fig. 1G). This analysis proofs that the visually observed bulges are real deformations of the MNT membrane. The analysis also shows that the larger the diameter of the MNTs is, the larger is the diameter of the MNT bulges (as quantified by the larger absolute differences of the medians (Δm) of DB values of MNT#1 (Δm = 137.5 nm, increase of 225%) compared to MNT#2 (Δm = 17.0 nm, increase of 125%)).

The distance between successive MNT bulges (dBB) was dBB = 7.527 (4.890, 9.451) µm for MNT#1 (n = 5) and dBB = 2.492 (1.644, 3.610) µm for MNT#2 (n = 3). The difference was statistically significant (BF = 11.27, d = 3.041) (see Fig. 1H).

The length of the MNT bulges (LB) was LB = 1.700 (1.536, 2.742) µm for MNT#1 (n = 6) and LB = 0.3495 (0.2730, 0.4260) µm for MNT#2 (n = 4). The difference was statistically significant (BF = 95.83, d = 3.879) (see Fig. 1H).

Our finding of the presence of bulges on MNTs is in agreement with previously published studies with other cell types[14][15][6][20]. Wittig et al.[6] documented examples of MNT bulges in the retinal pigment epithelial cells using differential interference contrast microscopy and SEM. The MNT bulge diameter was reported to be DB = 1 µm and the typical MNT diameter to be DMNT = 250 nm (with a range of 50–300 nm). These values are larger than the values we determined in our sample and study, i.e. the diameter of the two MNTs investigated in our study were 2.3 (MNT#1) and 3.6 (MNT#2) times smaller, and the bulge diameters were 4.04 (MNT#1) and 11.75 (MNT#2) times smaller, respectively.

This indicates the presence of a large heterogeneity of MNT morphology, confirming previous reports. Wittig et al. concluded that the MNT bulges suggest the presence of organelles in MNTs. Using the specific mitochondrial dye JC-1 they could label mitochondria inside MNTs. The presence of mitochondria inside MNT bulges was also reported by Patheja and Sahu[22]. In another study, Reichert et al.[20] reported also the presence of MNT bulges between cells (human primary CD34+ haematopoietic progenitor cells and leukaemic KG1a cells). The MNT bulge thickness was described to be smaller than 100 nm in diameter, in agreement with our finding concerning MNT#2.

With our study, we were the first to analyze LB and dBB values of MNT bulges. The range of LB values obtained (0.2730–2.742 µm) overlaps with the length distribution of mitochondria in cells (e.g. in retinal pigment epithelial ARPE-19 cells[23]: 0.4–74 µm, primary cortical neurons[24]: 1.27 ± 0.04 µm) as well as with the size distribution of peroxisomes (0.1–1 µm)[25]. The length of mitochondria is, however, a variable parameter that depends not only on the specific cell type but also on the state of the cell, i.e. cell cycle[26] or metabolic state[27].

That the location of MNT bulges was seemingly not random on the MNT is a novel finding of our study. The distance values between successive MNT bulges, as quantified by dBB, showed unimodal distributions for both MNTs investigated. The ratio dBB/LB based on the median values was 4.43 for MNT#1 and 7.13 for MNT#2, respectively; and the ratio dBB/DB based on the median values was 68.4 for MNT#1 and 25 for MNT#2, respectively.

This indicates that the periodicity of the bulges on an MNT seems to be independent of the MNT diameter as well as the length of the MNT bulges. Concerning the cause of the MNT bulge periodicity, no clear mechanism can be envisaged. Either it is a finding due to chance, or there is an underlying unexplored process that regulates the distance between successive bulges of MNTs. It could be indeed that it is only a finding due to chance since only a limited number of data points (i.e. distance values) were available for the analysis. Further studies are needed to investigate this aspect with a larger sample size.

What could be in general the cause of the MNT bulges observed? We think that three possible causes should be considered: (1) MNT bulges as artefacts caused by the cell staining process involving the SEM analysis, (2) MNT bulges as local deformations of the MNT without the presence of a cargo inside, or (3) MNT bulges as effects of an object inside the MNT that deforms the enclosing MNT locally.

Explanation (1) is unlikely since MNT bulges were also observed by optical microscopy[6], and MNT bulges are only occasionally observed (contrary to the expectation of the MNT bulge formation being a results of a general SEM staining artefact that should affect all or a large part of all MNTs present in the investigated culture). Explanation (2) is in principle possible but there are no nanomechanical processes known yet that result in an oval local deformation of an MNT without the involvement of an object inside the MNT causing the deformation.

Explanation (3) is most likely since (i) the presence of mitochondria inside MNT bulges was previously shown by several studies; (ii) the MNT bulge diameter values determined in our study (range: 68.0–440.0 nm) correspond to diameters of organelles like mitochondria (typical diameter: 100–1000 nm)[31][32][33], peroxisomes (typical diameter in RPE cells: 100–300 nm)[34], or exosomes (typical diameter: 30–100 nm)[35][36]; (iii) the length of the MNT bulges overlap with the length distributions of mitochondria and peroxisomes[23][34], and (iv) movements of MNT bulges were reported by other studies (speed: 0.16 µm/s[14], 20.7 ± 2.3 µm/h[37], 0.08 µm/s[19], 0.0259 ± 0.0079 µm/s[38], 0.033–0.059 µm/s[39], 0.045 ± 0.005 µm/s[21]) indicating the presence of a real object inside the MNT.


Our SEM analysis of two MNTs from cone-like photoreceptor cells of the cell line 661W with multiple bulges revealed that (i) bulges are characterized by a statistically significant local increase in MNT diameter (125% and 225% for both MNTs investigated); (ii) the thickness and length of the MNT bulges correspond to dimensions of mitochondria, peroxisomes and exosomes, and (iii) the MNT bulges seem not to be randomly distributed on the MNTs but exhibit preferred spacing with a different median value for each MNT. Our findings shed new light on the microstructure of MNTs and MNT bulges in particular.

Further studies should replicate and extend our findings with more MNTs and by applying SEM and fluorescence microscopy simultaneously by correlative light and electron microscopy (CLEM)[40] or even better 3D CLEM with serial blockface (SBF) SEM, as recently developed[41].


There are two main limitations in this study. First, only two MNTs could be identified in the SEM images taken from multiple sections of two cell cultures, resulting in a small number of MNT bulges that could be used for the statistical analysis of their geometrical properties. Secondly, we did not employ fluorescence microscopy to a nalyzewhether the MNT bulges were colocalized with the presence of fluorescence markers, e.g. JC-1 indicating the mitochondria.

This, however, could not be performed for technical reasons: performing SEM and fluorescence microscopy analysis of the same MNT consecutively with two instruments (as available in our lab) is nearly impossible due to change and disturbance of the cell/MNT state caused by the transportation between microscopes and the time-delay involved processing the sample for SEM analysis causing changes in the MNT state. A CLEM analysis would have been a solution.


Cell culture

Mouse cone-like photoreceptor cells 661W (University of Oklahoma, Oklahoma City, OK) were maintained and treated in DMEM (Gibco, Life Technologies, USA) supplemented with 10% FCS (Biochrom, Germany). The cells were grown at 37°C in a humidified environment with 5% CO2 and passaged every 3 to 4 days. For experimental procedures, cells were briefly rinsed with PBS (DPBS, Gibco, Life Technologies, USA). Afterwards, 1 ml of 0.25/0.02% (w/v) of trypsin in EDTA solution (Biochrom, Germany) was added to the cell culture and they were incubated at 37°C and with 5% CO2 for 5 min. Trypsin activity was neutralized by adding medium and centrifugation at 150 g for 5 min.

Scanning electron microscopy

For the scanning electron microscopy (SEM) analysis, cells with a cell number of 6×104 were grown on 12 mm round glass coverslips for 24 h. They were fixed with modified 1% glutaraldehyde in 100 mM a phosphate buffer (DPBS, Gibco, Life Technologies, USA). After several washes in PBS, the samples were postfixed in 1% osmium tetroxide/PBS for 2 h on ice, washed in water, en bloc contrasted in 1% uranyl acetate/water, and dehydrated in a graded series of ethanol (30%, 50%, 70%, 90%, 96%, 2×100% ethanol; pure ethanol on molecular sieve).

Finally, they were critical point dried using the Leica CPD 300 (Leica Microsystems, Vienna, Austria) and mounted on 12 mm aluminium stubs. Cells mounted on pieces of plastic were glued to conductive carbon pads and the edges of the plastic were surrounded with silver paint to reduce charging. Mounted samples were sputter coated with gold using the Baltec SCD 050 (Leica Microsystems, Vienna, Austria), and analyzed with a Jeol JSM 7500F cold field emission SEM (Jeol, Eching, Germany) at 8 mm working distance and 5 kV acceleration voltage using the lower secondary electron detector.

SEM image analysis has been performed in ImageJ (NIH). 78 SEM images were obtained and manually screened for MNT bulges in total. Four parameters were determined for further analysis of the MNT ultrastructure: the MNT diameter, DMNT, the diameter of the bulge, DB, the distance between adjacent bulges, dBB, and the length of the bulges LB. LB was determined by fitting an oval to the MNT bulge and determining the length from one end of the oval to the other.

Statistical analysis was conducted with JASP (https://jasp-stats.org). For comparison of distributions, the Bayesian independent samples t-test with a Cauchy scale of 0.5 was used. A Bayes factor of >10 was assigned as a threshold for a statistically significant difference of the distributions. Figure 1 was created using JASP and Adobe Illustrator.


The author would like to thank Dr. Thomas Kurth (TU Dresden, Center for Regenerative Therapies Dresden) for realization of electron microscopy and Simon Schaefer for proofreading the manuscript.

Conflict of interest

The authors declare no conflicts of interest.

Ethics Statement

Not applicable.

No fraudulence is committed in performing these experiments or during processing of the data. We understand that in the case of fraudulence, the study can be retracted by ScienceMatters.