Cephalopods are included in the Directive 2010/63/EU and, thus, regulated for their use in scientific research (Fiorito et al., 2014; Fiorito et al., 2015). Experiments included in this study were carried out in 2018 on tissue samples originating from wild animals. This study has been granted an ethical clearance for “label-free multiphoton microscopy for the investigation of the process of arm regeneration in Octopus vulgaris” by the institutional AWB (OBA: case 4/2021/ec AWB-SZN -28 June 2021).

This study was carried out on recently deceased Octopus vulgaris (N = 6; four males, two females, body weight: 194–402 g) collected from fishermen (Bay of Naples, Mediterranean Sea, Italy) during spring (seawater temperature range: 15–20°C). The animals were selected for the presence of one or more damaged arms in the phase of healing or regeneration (following stages reviewed in the study by Imperadore and Fiorito (2018)).

In the cases of octopuses still showing signs of life, the animals were euthanized (3.5% MgCl₂ in seawater, > 30 min), and death confirmed by transection of the dorsal aorta (Fiorito et al., 2015).

Quality of tissues was assessed through classical histological methods and was found suitable for immunohistochemistry and multiphoton microscopy imaging.

Damaged and corresponding contralateral uninjured arms (control) were harvested (∼3 cm in length) for a total number of 23 samples. For the control, the arm tip (i.e., the most distal part) and a piece of arm at around 50% of its length (proximal) were also collected.

The dissected samples were immediately processed, following the study by Imperadore et al. (2017). In brief, the tissues were fixed in 4% PFA in seawater (3 h), followed by PBS (pH 7.4) washes, and immersion in sucrose 30% (in PBS) until sinking. The samples were then embedded in freezing and blocking medium (OCT; Leica Biosystems) and stored at −80°C until use. Cryostat (Leica CM3050 S) sections (either 30 or 150 µm) were mounted on SuperFrost Plus glass slides.

Two additional control arm tips were harvested, fixed, and stored in PBS to image the whole mount sample.

The cryostat sections were air-dried for 30 min, rehydrated in PBS, and covered with a glass coverslip. Imaging was performed with an optical microscope Axio Examiner Z.1 coupled to a laser scanning module LSM 7 (Carl Zeiss AG, Jena, Germany) equipped with non-descanned detectors. An erbium fiber laser (Femto Fiber pro NIR from Toptica Photonics AG, Munich, Germany) provides excitation for TPEF and SHG by emitting at 781 nm with a pulse length of 1.2 ps and a maximum nominal power of 100 mW. The TPEF signal was acquired in the spectral range 500–550 nm, while the SHG signal was retrieved using a band-pass filter centered at 390 nm. CARS excitation needed a second laser source (i.e., the Femto Fiber pro TNIR from Toptica Photonics AG) which is tunable in the range 850–1,100 nm and has a pulse length of 0.8 ps. In all CARS experiments, the wavelength was set to 1,005 nm (emitted power 1.5 mW) in order to resonantly excite the symmetric stretching vibration of methylene groups at 2,850 cm⁻¹. CARS, TPEF, and SHG were simultaneously excited and acquired with a W Plan-Apochromat 20x/1.0 (Carl Zeiss AG) (for a schematic diagram of the system used for multiphoton microscopy, see Supplementary Figure S1).

For multimodal imaging of thicker slices (150 µm thickness) and whole arm tips, a Leica SP8 CARS microscope with SRS upgrade (special part request, Leica Microsystems GmbH, Mannheim, Germany) was used. A picoEmerald S Optical Parametric Oscillator (APE Angewandte Physik und Elektronik GmbH, Berlin, Germany) provides a Stokes beam at 1,031 nm and a tunable pump beam in the range of 720–970 nm. The two pulse trains (pulse duration 1–2 ps) were spatially and temporally overlapped. The images were acquired using a ×25 water objective (HCX IRAPO L ×25/NA 0.95/water, Leica Microsystems, Mannheim, Germany), and signals in the forward direction were collected using an air condenser (NA 0.4, Leica Microsystems, Mannheim, Germany). Forward CARS (2850cm⁻¹, CH₂-stretch vibration) was spectrally filtered by a short-pass filter SP750, a beam splitter BS560, and a band-pass filter BP670/125. SHG was detected in parallel and spectrally filtered by a short-pass filter SP750, a beam splitter BS560, and a band-pass filter BP465/170. Signals in the epi-direction were spectrally separated in SHG in the range from 400 to 510 nm and TPEF from 515 to 640 nm. All z-stacks were recorded with a voxel size of 0.2 µm × 0.2 µm × 3.0 µm. Z-stacks range from 100 to 130 µm in height.

The resulting multimodal RGB images are represented as follows: red channel = CARS, green channel = TPEF, and blue channel = SHG.

The images were processed with LAS X (Leica Microsystems, Mannheim, Germany) and Zen Blue Edition (Carl Zeiss, AG, Jena, Germany) software.

Following multiphoton imaging, the coverslip was carefully removed in PBS and slides used for immunohistochemistry or stained with hematoxylin and eosin (H&E). The H&E staining protocol consisted of a 2-min bath in Meyer hematoxylin followed by a 5-min step in tap water and 20 s in eosin.

IHC was performed as previously described (Imperadore et al., 2017). In brief, after blocking in normal goat serum (5% NGS, in PBT: PBS Tween 0.1%) for 1 h in RT, the sections were incubated overnight with primary antibody (i.e., anti–acetylated tubulin, SIGMA T6793, dilution 1:1,000; anti–phospho-Histone H3, Sigma H9908, dilution 1:600) in PBT and NGS 1% at 4°C. Following washes in PBT, the sections were incubated with secondary antibodies [1:250, Alexa Fluor goat anti-mouse IgG (H + L) 488 and Alexa Fluor goat anti-rat IgG (H + L) 594] for 1 h at room temperature. DAPI (14.3 μmol L⁻¹ in PBT) was used after IHC or on unstained sections to counterstain nuclei.

The sections following IHC protocol were mounted in PBS and imaged again for multiphoton microscopy; H&E sections were, instead, dehydrated in an ethanol series, cleared in xylene, coverslipped using DePex, and imaged using either Axio Examiner Z.1 (Carl Zeiss AG) equipped with the camera AxioCam or Axio Scope A1 (Carl Zeiss AG) equipped with the camera Canon DS126231. The images were processed with Zen Blue Edition (Carl Zeiss, AG, Jena, Germany) software.

The skin. Octopus skin contains various organs and elements (i.e., chromatophores, iridophores, leucophores, and papillae) that can be finely controlled to change the animal’s skin tone and texture, thereby providing the animals with extraordinary camouflaging and interspecific communication (Borrelli et al., 2006; How et al., 2017; for a review, see Packard and Hochberg, 1977). Chromatophores, in particular, are sacculus organs responsible for color change, owing to the presence of diverse pigment granules controlled by muscle bundles that are radially organized to open and close the sacculus (Messenger, 2001).

In the skin covering the arm (Figure 1A), chromatophores were easily identified as bright green spots (TPEF) distributed over the entire arm structure, which, in turn, appears in red (CARS) in multichannel images (Figures 1B,C, whole mount sample).

Higher-magnification imaging of sagittal thin sections of the arm allowed for identification of other distinctive elements of the skin and surrounding tissues. The epidermis appeared covered in microvilli, characterized by an intense CARS signal; mucous cells were identified as negative imprints in the epidermal layer (Figures 1D–F, asterisks); and close to chromatophores (Figures 1D–F, arrowheads), reflective elements appeared as round granules just underneath the epidermis (CARS, Figures 1D–F, arrows). Bright-field imaging (i.e., transmitted white light) of the same section highlighted the nuclear components of the epidermis (DAPI counterstain in blue in Figure 1F), confirming the identity of mucous cells by morphology and nuclear position.

The muscular tissue. A schematic drawing of the arm morphology in the transverse plane is included in Figure 2A to facilitate structural identification (Supplementary Figure S2A). The three muscle groups belonging to the intrinsic musculature of the arm were visualized in CARS (Figure 2B). In particular, the i) tightly packed transverse muscle bundles running perpendicular to the arm long axis, which elongate through structures called trabeculae, were visible among the ii) longitudinal muscles and the iii) three bundles of oblique muscles (see Supplementary Figure S2 for more details). Connective tissue sheaths, highlighted in SHG, appear to envelop the different muscle layers of the intrinsic musculature (Figure 2B).

CARS imaging also enabled identification of the intrinsic musculature of the sucker (data not shown) and the acetabulo-brachial muscles (Supplementary Figure S2), which connect the intrinsic muscles of the arm and the intrinsic muscles of the suckers.

The neural structures. The neural control for these sets of muscles is provided by six main nerve centers per arm, that is, a central axial nerve cord connected to four intramuscular nerve cords and to sucker ganglia. The axial nerve cord comprises two axonal tracts (dorsal) and several brachial ganglia (ventral), facing and innervating suckers (Figures 2A,B; Supplementary Figure S2). Running longitudinally to the axonal tracts, the main blood vessel supplying hemolymph to the arm (brachial artery) is shown by CARS and is surrounded by connective tissue (SHG) (Figures 2B,C; Supplementary Figure S2B).

Each brachial ganglion comprises an inner neuropil and an outer cellular layer (Figures 2B,D; Supplementary Figures S2B–D). The cellular layer contains many small and some big neurons, with nuclei ranging from less than 5–20 µm (Young 1963). These cells emit both in CARS and TPEF, with the latter mainly highlighting their cytoplasm, giving a strong signal of granular structures contained in the cells (Figure 2D; Supplementary Figure S2C) that could be partly due to lipofuscin. DAPI counterstaining confirmed these results and highlighted the presence of supporting cells around the neurons, which were not detected with multiphoton microscopy alone (see the dotted white rectangle in Figure 2D ). Axons in the intricate neuropil of the brachial ganglia are also highlighted in CARS and TPEF (Figure 2B; Supplementary Figures S2B,C). DAPI counterstaining and acetylated tubulin immunoreactivity confirmed these results (Supplementary Figure S2D).

Some of the nerves departing from the brachial ganglia are linked to the four intramuscular nerve cords (CARS; Figure 2B; Supplementary Figure S2B), and to the ganglion of the sucker (CARS and TPEF; Figures 2E,F). This ganglion comprises a central neuropil and neurons surrounding it (TPEF and DAPI) (Figures 2E,F).

Imaging of the arm tip allowed visualization of the abovementioned anatomical structures (a schematic drawing in the sagittal section is reported to facilitate structure identification, Figures 2G,L).

Compared to more aboral arm portions, in the tip, we observed a greater area occupied by the axial nerve cord, reducing the space for muscles; brachial ganglia get closer to each other (Figures 2H,I,N) as suckers get smaller and closer. Fibers from the brachial ganglia descend from the neuropil (CARS, Figure 2H, arrow), passing through the cellular layer (TPEF, Figure 2I).

Tissues and structures at the most distal part of the tip (delimited by a dotted line in Figures 2M,N) appear less organized and differentiated compared to all the other neighboring areas (Supplementary Figure S3). The tip appears characterized by a thick layer of connective tissue (SHG) (Figure 2M), which appears in between the epidermis and the muscular layer covering the nerve cord. CARS also highlighted the presence of numerous blood vessels in this zone (Figure 2M, arrowhead).

The wounded skin. The regenerative process of a damaged arm in O. vulgaris is always initiated by wound healing, with the dermis wound edges closing around the lesion. This process generally requires between 0 and 5 days, depending on several factors, such as temperature, animal age and sexual maturity, and health status (for a review, see Imperadore and Fiorito, 2018). To facilitate readers, the main phases of the healing process are sketched in Figures 3A–D.

The wounded dermis contracts and forms a rim that starts covering the wound to form a first protective layer for the exposed tissues (Figure 3A). The connective tissue within it appears involved in the process, narrowing around the muscular tissues of the stump (SHG, Figure 3A) and contributing to wound closure.

The central portion of the damaged arm (i.e., the internal muscles and axial nerve cord) remains exposed until a clot of agglutinated blood corpuscles start depositing over it (green dotted line in Figure 3A). Blood vessels, which are observed in great number in the arm stump (Figure 3A, arrowheads), represent the origin of these cells (Figures 3A′, see also inset in Figure 3A′).

The clot then increases in size, covering the whole exposed tissue in between the wound epithelium (Figure 3B) and forming a dense and fine network of interdigitated cells called primary blastema (Lange, 1920). The cells in this blastema appear full of dense granules highlighted in CARS and TPEF (Figure 3B′). A boundary line of connective tissue separates this blastema from the underlying and well-differentiated tissues (SHG, Figure 3B). The primary blastema finally never casts off, but rather is retained and eventually completely covered by the regenerating epithelium, the latter appearing highly vascularized (Figure 3C, arrowheads).

The muscular and neural tissues. At this stage, the damaged muscles and nerve tissues (i.e., the axonal tracts) show evident signs of degeneration (i.e., swelling and fragmentation). Degenerating tissues, highlighted in CARS and TPEF (Figure 3C′, asterisks) are not observed in control tissues.

The hemocytes. The muscular layers in the healed stump are invaded by many cells (Figure 3C), identified as hemocytes, which change their appearance once released from blood vessels. They indeed transform from circulating, round-shaped cells with a small cytoplasm and u-shaped nuclei (Figure 3C′′, dotted line with arrowhead) into amebocyte-like cells with a large and granular cytoplasm, intensely emitting in CARS and TPEF (Figure 3C′′, dotted line with asterisk). DAPI counterstaining confirmed the cellular nature of these structures (Figure 3C′′). Hemocytes released from the vessels around the nerve cord (Figure 3D, arrowheads) are first released into the connective tissue around it (Figure 3D, asterisk) and then invade all muscle layers below the wounded epithelium (Figure 3D, arrow).

The wounded skin. From the healed skin, a little knob appears, regenerating an arm from the dorsal side of the stump. The resulting arm is initially much thinner than the original stump (Figure 4A). The wound epithelium, narrowing around the original site of the lesion, is still visible and characterized by thick connective tissue (SHG, Figures 4A,C).

The muscular and neural tissues. Degeneration in this phase involves greater areas in the muscular tissues and the nerve cord of the stump (Figure 4A). Degeneration is particularly evident in the axonal tracts of the nerve cord, close to the original site of the lesion, where fibers appear swollen and broken into lumps (CARS and TPEF, Figure 4B). Degeneration can be followed along the axonal tract, with the number of fibers involved decreasing when farther from the site of the lesion. Distal to this site, fewer degenerative events are found using multiphoton imaging (Figure 4D).

Among the degenerating lumps in the axonal tract, regenerating fibers also appear (highlighted by strong CARS signal, Figure 4B) and can be followed providing innervation in the newly forming arm tip (Figure 4C, arrows). The regenerating tip is mostly occupied by the newly forming nerve cord (Figure 4C, arrows), and thin processes can be seen descending from it toward the ventral site, where new suckers will later form (Figure 4C, arrowhead). Suckers start to develop in the forming arm, close to the stump (Figure 4C).

The hemocytes. The tissues in the stump are invaded by cells whose cytoplasm is rich in small granules, strongly emitting in CARS and TPEF (Figures 4E–G). They mainly invade muscles around the nerve cord and the axonal tracts of the latter (see also Supplementary Figure S4A, arrowheads), but are never observed in the neuropil of the brachial ganglia. These cells also reach muscles in the regenerating arm tip, very close to the site of the lesion (Supplementary Figure S4A), but are not found distant to this site or in any other tissue of this new structure.

These cells appear to be released by blood vessels close to the injury site (Supplementary Video S1).

Areas invaded by these structures are also characterized by numerous proliferating cells (Supplementary Figure S4B).