Why adipose-derived stem cells?
Adult stem cells can be harvested from many sites such as lung tissue, umbilical cord blood, bone marrow, adipose tissue, and synovial tissue. These cells have been the focus of much regenerative research in the musculoskeletal tissue engineering field. The two types of adult stem cells that have been most utilized in tissue engineering are bone marrow-derived mesenchymal stem cells (BMSCs) and adipose-derived mesenchymal stem cells (ADSCs) . BMSCs are generally harvested through piercing into bone using a syringe, and aspirating a fraction of bone marrow . Bone marrow aspiration, however, creates great discomfort for the donor and may lead to chronic pain along with possible donor site morbidity .
ADSCs are a great alternative to BMSCs due to their ease of harvest and nearly identical differentiation potential . ADSCs are extracted during a minimally invasive liposuction procedure [125, 126], which is less painful and causes less donor site morbidity compared to bone marrow aspiration . Furthermore, the number of isolated stem cells from adipose tissue is higher than bone marrow. In bone marrow aspirate, BMSCs compose 0.002 % of cells present in the marrow aspirate, while ADSCs are about 2 % of the cell population in adipose tissue [127, 128]. ADSCs have proven their merit as a valuable tool for regenerative musculoskeletal engineering. Preliminary clinical studies, in vitro, and in vivo studies demonstrated that ADSCs can be differentiated into a desired cell lineage, and utilized in regenerative therapies. For instance, Thesleff et al. successfully used ADSCs to repair cranial defects and saw restoration to normal cranial integrity , and Mesimaki et al. had success repairing a maxillary defect using autologous ADSCs expanded in vitro . Yoshimura et al.  showed that ADSCs used in breast augmentation provided results that were superior to conventional lipoinjection.
The lineage commitment of ADSCs is usually controlled though soluble differentiation factors incorporated into the media or scaffold. However, it has been well proven that structural, chemical, and mechanical cues also affect the ADSCs differentiation and lineage commitment. For instance, Betre et al.  showed that elastin-like polypeptide hydrogels promoted chondrogenesis in ADSCs in the absence of differentiating factors. Flynn et al.  found that ADSCs cultured in non-adhesive hydrogels differentiated into adipocytes faster than ADSCs that were cultured in an adherent placenta-derived matrix, or a mixture of the two. Tseng et al.  showed that scaffold alignment caused cells to exhibit a tenocyte-like morphology. Rada et al.  showed that culturing ADSCs on hydrogel and gelatin scaffolds with differing mechanical properties significantly affected the cells expression of chondrogenic genes, morphology, and proliferation. Furthermore, it has been shown that physiologically relevant strain application can also affect the differentiation of ADSCs. Below is the concise review of the studies concentrated on mechanical strain-induced differentiation of ADSCs into various musculoskeletal tissue associated cell lineages.
Specific lineage commitment of ADSC under mechanical strain
Adipose stem cells have been used in multiple in vivo studies and have proven their value as regenerative tools; ADSCs have been used in vivo to repair bone [130, 136–138], muscle [13, 14, 139], cartilage [10, 140–142], and tendon . Only one of these studies, however, investigated the effects of strain on the effectiveness of the cells. Knag et al.  used a rotary bioreactor to apply some fluid stress to the cells before implantation, but the investigators did not note specifics on strain levels or rate, and somewhat attributed the noted increased regenerative potential to the more rapid nutrient exchange caused by the flowing media. These in vivo studies, especially the one by Mesimaki et al.  which used ADSCs to repair a maxillary bone defect in a human patient, indicate that ADSCs are soon to become an irreplaceable clinical tool. However, due to the relative difficulty of in vivo strain measurement and characterization, in vitro studies such as the following will be invaluable in in vivo application when understanding the effects of physiological strain.
Differentiation into osteogenic lineage
The differentiation of ADSCs towards an osteogenic lineage is the most-studied out of all the musculoskeletal lineages, however research is still preliminary. The following section outlines the currently completed studies on the osteogenic response of ADSCs to mechanical stimulation. Out of the following studies, all used uniaxial strain except a study by Tjabringa et al.  which utilized pulsitile fluid flow. All studies used a 1 Hz strain rate except for the study by Du et al.  which utilized a 0.5 Hz rate. In accordance with the previous section describing in vivo bone mechanobiology, many of the studies applied strains of 10 % which is above the commonly accepted whole bone levels of 0.1–0.2 %, but nonetheless acceptable as stated above. Three studies used strain levels of ~0.2 %, however, and saw significant increases in the osteogenic activity of the cells. This indicates that ADSCs are responsive to a large range of applied strains, including the low levels seen acting on whole bones in vivo. The precise effects of different strain, however, have not yet been determined, and therefore the following reports vary in application, though no evidence has been conclusive.
Ye et al.  studied the osteogenic differentiation of ADSCs under uniaxial loading. Human ADSCs and BMSCs (hADSCs and hBMSCs) were seeded on a polyethylene disk in osteogenic medium, and were subjected to tensile strain once they reached 80–90 % confluence for a total differentiation time of 14 days. A four-point bending device was used to apply strain on the magnitude of 0.24 % (2400 µε) at a frequency of 1 Hz with loading occurring for 2 h every day. ALP, mineralization, and PCR for RUNX2, BMP-2, ALP, and Osteocalcin (OC) were analyzed to determine osteogenic differentiation of hADSCs and hBMSCs. Strain significantly increased the mineral deposition and expression of ALP, BMP-2, and RUNX2 at 5 and 10 days in hADSCs. OC was only increased at day 10. However, hBMSCs had significantly more mineralization, and expression of BMP-2 at days 5 and 10 and ALP at day 10. This study showed that hBMSCs have a greater ability to differentiate into an osteogenic lineage than hADSCs. However, this study also showed that mechanically dynamic culture has a great effect on the differentiation of hADSCs towards an osteogenic lineage.
Yang et al.  studied the differentiation of ADSCs toward an osteogenic lineage under sustained or repeated uniaxial strain. ADSCs were harvested from rats, and osteoinduced in osteogenic media for 48 h before loading. Cells were maintained in osteogenic media, and stretched for either 10 days straight for 17 min or just 1 day for 6 h. All cells were stretched at 1 Hz and 0.2 % strain using uniaxial 4-point bending. Measurements were taken 2 h after end of last loading session. PCR for ALP, Osteocalcin (OCN), BMP-2, and RUNX-2 was performed and results showed upregulation (though not significant) of BMP-2 for day 7, and a slight peak for OCN at day 7. Only significant measurements were BMP-2 and RUNX2 for 6 h group. Cell alignment was observed for day 10, but not 6 h.
Wall et al.  studied the osteogenic differentiation of human ADSCs towards an osteogenic lineage under prolonged uniaxial strain. ADSCs were seeded into 70 % collagen gels and subjected to 14 days of 10 % uniaxial strain at 1 Hz for 4 h a day. This study showed that paladin—a gene that is upregulated as ADSCs differentiate towards an osteogenic lineage—was upregulated in response to cyclic strain.
Du et al.  cultured human ADSCs in differentiaion media with and without uniaxial cyclic strain (groups MC and C, respectively) to determine the effect loading had on the osteogenic differentiation of ADSCs. hADSCs were cultured in normal conditions for 48 h, then cultured in osteogenic media in static or strained culture. Cells experienced uniaxial strain by a 4-point bending apparatus of 0.2 % (2000 µε) at 0.5 Hz frequency 2 h a day, with a culture time of 7 days. ALP activity in group C induced saw a steady increase starting at day 3 and continuing to day 7. In group MC, ALP activity began to increase at day 2, but peaked at day 5 and day 7 saw a significant decrease. Gene expression of ALPL, COL 1, RUNX2, SPP1, and SPARC, as analyzed by PCR followed the exact same trends seen in ALP activity for both groups. On days 2, 3, and 5 group MC had higher expression of all genes and ALP, but on day 7, expression/activity dropped close to group C. This showed that mechanical strain can be useful for osteogenic differentiation of hADSCs along with chemical signals, but that the effects may decrease with time. Their results suggested that the decrease of differentiation signals may be due to aging of cells due to prolonged loading.
Tjabringa et al.  studied osteogenic differentiation of ADSCs under pulsatile fluid shear stress. hADSCs were cultured on polylysine coated glass slides and cultured overnight, the next day they were exposed to a 5 Hz pulsatile fluid flow (PFF) with a mean shear stress of 0.6 Pa, a pulse amplitude of 0.3 Pa, and a peak shear stress of 8.4 Pa/second. Samples of the media were taken at 0, 10, 30, and 60 min to measure NO production. PCR was run examining COX-1, COX-2, Runx-2, and osteopontin (OP) at 3 and 6 h after PFF. It was found that PFF significantly increased NO production at 60 min of PFF. PFF did not affect OP or COX-1, but increased COX-2 by 6- and 5-fold after 3 and 6 h respectively, and Runx2 1.3-fold at 3 h. These results prove that hADSCs have a bone cell-like response to mechanical loading even in the absence of chemical induction factors.
Hanson et al.  observed the osteogenic differentiation of two lines of ADSCs with differeing osteogenic potential under uniaxial strain. Two groups of hADSCs were initially assessed for deposition of calcium on tissue culture plastic in differentiation medium. One cell line showed approx. nine times as much calcium deposition as the other. The cells were then exposed to cyclic strain, and their proliferation, viability, and calcium deposition was measured again. The cells were seeded in bioflex culture plates and cultured for 5 days, until they reached 100 % confluence. Then 10 % uniaxial strain of 1 Hz and 4 h a day was applied to the cells for 2 weeks using a Flexcell tension plus system. Cells were either subjected to continuous 1 Hz strain, or rest-inserted strain (10 s between each cycle). No difference was shown between either the continuous or rest-inserted strain groups. Loading of either type increased calcium deposition, though the increase was significantly greater in the cell line with higher initial deposition on static culture plastic.
Diederichs et al.  observed the osteogenic differentiation of ADSCs under different lengths and repetitions of uniaxial strain regimes. Different lengths of application of strain were applied to hADSCs, ALP activity and expression of early and late osteogenic markers was investigated. Cell viability and collagen III expression (expressed under excessive strain) were observed to see if strain was excessive. Cells were induced in osteogenic differentiation medium on collagen I coated silicone plates for a week before application of uniaxial strain. Afterwards, cells had 15 min, 60 min, and 2 h strains applied to them at 1 Hz and 5 % by clamping and stretching of the silicone plates. Additionally, triple strain regimes were carried out with double the strain rest inserted e.g. 15 min strain–30 min rest–15 min strain–30 min rest–15 min strain. Cells were immediately harvested after application of strain. ALP activity was slightly decreased in repeated 15 min but increased in both 1 and 2 h repeated. Both the elongated and repeated strain were effective at enhancing differentiation as evidenced by upregulation of osteogenic genes, but repeated was always better than elongated at strain acclimation.
Charoenpanich et al.  measured osteogenic gene expression and calcium accretion of hADSCs was measured under the influence of mechanical strain, 3D culture, and osteogenic media. 24 h after seeding in 70 % collagen gels, cells were put in growth or osteogenic (50 mM ascorbic acid, 0.1 mM dexamethasone, and 10 mM b-glycerophosphate) media. Cells were loaded with 10 % strain at 1 Hz for 4 h/day in a stretching anchored collagen gel, after which microarray analysis was performed to determine gene upregulation. Mechanical strain was shown to increase the expression of PDLIM4 which is one of the top five upregulated genes shown to have polymorphisms often found with bone mineral density. Calcium accretion that was increased in the loaded cells was found to be controlled by different genetic pathways than soluble chemical factors. Additionally, 10 % strain resulted in upregulation of two crucial factors in bone regulation; proinflammatory cytokine regulators IL1RN and SOCS3, and angiogenic inducing factors FGF2, MMP2, and VEGF A. 3D culture also seems to use a different pathway than 2 days for osteogenic induction
Differentiation into myogenic lineage
Multiple studies have shown that ADSCs have potential in myogenic tissue engineering, and that mechanical strain can have a significant impact on their differentiation into muscle cells. The strain applied in the following experiments ranged from 5 to 15 %, with five studies applying uniaxial strain, and one applying biaxial strain. Strain was generally applied for 24 h a day for anywhere from 1 to 14 days, except one study where strain was applied for 1 h a day for 21 days. Only one study, by Amin et al.  tested for myogenic activity under strain without chemical induction, and found that mechanical strain was enough to significantly increase the myogenic activity of the ADSCs, though not as much as strain with chemical induction. Multiple studies showed that mechanical loading also enhanced the fusion of ADSCs with myocytes to form muscle fibers. Conversely, studies by Dugan et al.  and Lee et al.  showed that did not have a positive myogenic effect on the ADSCs. While not all the studies showed that mechanical strain aids in the differentiation of ADSCs, the majority indicate that mechanical strain can be a powerful tool for differentiation of ADSCs, especially along with chemical induction.
As previously stated, Amin et al.  studied the effect that equiaxial strain had on the differentiation of ADSCs. GATA4, a transcription factor that plays an important role in late embryonic heart development, was measured as a determinant of myogenic differentiation. Cells were seeded on 2D collagen membranes and exposed to 5-azacytidine for 24 h, then 10 % equiaxial cyclic strain was applied at 1 Hz for 24 h after removal of differentiating medium. Cells were cultured for 1, 4, and 7 days before assays were performed. GATA4 was upregulated most at day one by only mechanically stimulated cells, but for days 4 and 7—while still greater than control- it was surpassed by mechanically strained and chemically induced group.
Dugan et al.  studied the effects of cyclic uniaxial strain on ADSCs cultured with C2C12 myoblasts in myogenic media. Murine C2C12 myocytes and Human ADSCs were seeded on Flexcell flexible-bottom 6-well plates in a ratio of 1:5 and cultured for 24 h before myogenic media was added, and strain was applied. 12 % strain was applied continuously for 48 h at 1 Hz in pulses with 1 s duration and 1 s rest inserted between each application. After strain was applied, cells were cultured under static conditions for 5 days before being analyzed. However, after analyzing myogenic activity of the ADSCs using immunocytochemistry and gene expression, it was determined that strain had no apparent effect on the ADSCs myogenic activity either in monoculture, or in co-culture with C2C12 cells.
Lee et al.  studied effects that TGF-β1 and uniaxial strain had on the differentiation of ADSCs into smooth muscle. Cells were plated on collagen coated flexible plates, and cultured for 3 days before uniaxial mechanical strain by membrane stretching across a loading post. Cells experienced continuous 10 % strain, at 1 Hz for 7 days. Additionally, cells were cultured with TGF-β1 to induce smooth muscle differentiation. Cell proliferation was nearly identical in all cultures, and while TGF-β1 increased smooth muscle actin and calponin expression, strain did not appear to positively affect the differentiation of the cells. Strain did, however, align the cells on the plates.
Yilgor et al.  studied the effect that chemical induction and uniaxial strain had on human ADSC differentiation into a myogenic lineage. ADSCs were induced with myogenic induction medium 24 h after seeding on flexible membranes, after another 24 h, cells were washed and, media was replaced with growth media. Cells were exposed to 11 % uniaxial strain at 0.5 Hz for an hour a day during days 3–21 of culture using a Flexcell system. It was found that chemical induction was necessary for multi-nucleation of cells, as well as expression of muscle specific protein (desmin, myoD, and myosin heavy chain). Desmin and myoD were significantly greater in the dynamic culture conditions, unlike the static conditions. Myotube count and nuclei/myotube were significantly higher in dynamic culture.
Andersen et al.  studied the effects myogenic media, uniaxial strain, and co-culture with mouse myoblasts had on the myogenic differentiation of ADSCs. Cells were allowed to reach 90–95 % confluence on flexible-bottom plates coated with collagen with and without mouse myoblasts (with a ratio of 1:5). Mechanical 15 % uniaxial strain was applied at 0.5 Hz using rectangular pistons for 48 h in myogenic media. After stimulation, cells were cultured in differentiation media for differing amounts of time. ADSCs were infected with viral particles including GFP, to aid in monitoring ADSC differentiation. The culture conditions did not appear to increase the differentiation of ADSCs towards a myogenic lineage; however, it did greatly increase their rate of fusion with myoblasts and their alignment.
Bayati et al.  observed the effects of uniaxial loading on ADSCs with and without differentiation media. Gene expression and myosin synthesis was monitored in rat ADSCs after 3 days of chemical differentiation with 5-azacytidine, and loading on day 2. Cells were seeded on collagen-1 coated silicone membranes and cultured for 48 h, before addition of differentiation media. Addition of differentiation media occurred on day 1, and 10 % 1 Hz 24 h uniaxial loading was performed on day 2 using a clamping tensile device. The results showed that MyoD, myh2, and myog were significantly increased under strain in both control and myogenic differentiation media.
Park et al.  observed the myogenic differentiation of hADSCs exposed to many differentiation factors, along with uniaxial strain. Cells were seeded on PLCL scaffolds 3 days prior to stimulation, and then put in induction media, and stretched at 5 % at 1 Hz for 2 weeks using a uniaxial clamp device. Strain greatly increased the expression of a-SMA and MHC in certain myogenic media (retinoic acid, ascorbic acid, “smooth muscle induction media”) therefore, this study indicates that strain can be very beneficial to the induction of hADSCs to a myogenic lineage.
Differentiation into chondrogenic lineage
Few studies have been done on the role of mechanical loading on ADSC differentiation into chondrocytes. On the other hand, the ability of ADSCs to differentiate into a chondrogenic lineage through the use of chemical factors has been confirmed by many studies [151, 155, 157]. Out of the four studies that were found, only one used platen compression, while the other three utilized hydrostatic compression. The peak amplitudes of hydrostatic pressure applied ranged from 5 to 10 MPa, while 5 % compressive strain was applied under platen compression. All studies applied strain for 4 h a day and strain was applied from 7 to 35 days. These results show that under physiological loading conditions, ADSCs will differentiate effectively into chondrocytes even at longer time points.
Li et al.  studied the chondrogenic differentiation of rabbit ADSCs exposed to dynamic platen compression and insulin-like growth factor 1 (IGF-1) in chitosan/gelatin scaffolds. Cells were cultured in the scaffold for 2 days, and then they were subjected to 5 % compressive strain at 1 Hz for 4 h per day for 7 days. Compression loading was carried out by placing scaffolds between two platens in a compression bioreactor (BioDynamic ELF5110). Dynamic compression alone upregulated calcium signaling pathways and Sox-9 (which is known to induce chondrogenesis in BMSCs).
Carroll et al.  applied cyclic hydrostatic pressure to porcine ADSCs in agarose scaffolds to induce their chondrogenic differentiation. ADSCs were seeded in 2 % agarose gels, sealed in water-tight bags with 4 mL medium per construct. The bags were then placed in a water filled bioreactor which exposed them to 1 Hz cyclic hydrostatic pressure for 4 h a day, 5 days a week for 5 weeks, with a peak amplitude of 10 MPa. Sulfonated Glycosaminoglycan (sGAG) content, which is a key structural component of cartilage, was significantly increased in constructs exposed to cyclic hydrostatic pressure, which in turn increased the dynamic and equilibrium modulus of the constructs. Additionally, it was shown that when a chondrogenic growth factor (TGF-β3) was removed in some constructs after the first week, the loaded samples were able to retain collagen and sGAG content better than unloaded samples. Altogether, this study demonstrated the positive effects of hydrostatic pressure on both inducing and maintaining chondrogenic differentiation in ADSCs.
Correia et al.  investigated the effects that high and low cyclic hydrostatic pressure had on the differentiation of human ADSCs into a chondrogenic lineage. ADSCs encapsulated in gellan gum hydrogels and suspended in chondrogenic media were exposed to two different strain regimes as well as a static, unloaded control. The first group was exposed to 0.5 Hz cyclic pressure with a peak amplitude of 0.4 MPa for 4 h a day, 5 days a week for 3 weeks; the second was exposed at 0.5 Hz with a peak amplitude of 5 MPa for an identical length. The group exposed to 5 MPa pressure showed most GAG content from staining (Safranin O and Alcian Blue), while the 0.4 MPa group showed less than the 5 MPa group, but significantly more than the control group. However, gene expression of collagen II, aggrecan, and sox-9 indicated that 0.4 MPa was the most effective, though 5 MPa was still improved over control. These results show that ADSCs in vitro sense and respond in a chondrogenic-positive manner to cyclic hydrostatic pressure.
Puetzer et al.  showed that cyclic hydrostatic pressure even in the absence of chondrogenic differentiation factors was able to induce chondrogenic differentiation in ADSCs. ADSCs were encapsulated in 2 % agarose hydrogels, and exposed to cyclic pressure at 1 Hz with a peak amplitude of 7.5 MPa for 4 h a day for up to 21 days. At day 7, loaded cells exhibited better chondrogenic gene profiles as compared to control groups. However, at days 14 and 21, cells exhibited lower viability and decreased chondrogenic gene expression. This study indicated that cyclic loading can initiate chondrogenic differentiation of ADSCs even in the absence of chondrogenic differentiation factors.
Differentiation into tenogenic lineage
At the time of the writing of this article only one publication has shown the effects of mechanical loading on the tenogenic differentiation of ADSCs. In the study, Raabe, O. et al. exposed equine ADSCs were exposed to 21 % oxygen tension, as well as differentiation factors, and tensile strain and measured cell morphology and expression of tendon-relevant genes (collagen type 1 and 3, cartilage oligomeric protein, and scleraxis). Cells were seeded in collagen gels, and then compacted under 2.7 and 5.2 kPa around for 4 h until desired thickness was achieved. Cells were then put in a uniaxial linear stretching device for 21 days with 4 % strain for 2 h followed by 6 h rest. Authors concluded that differentiation of ADSCs in collagen seems to be best carried out with strain, and factors as evidenced by cell morphology and gene expression . Many other studies have also shown that ADSCs have the ability to differentiate into tenocytes with differentiation medium , and that strain assists in the differentiation of BMSCs into tenocytes [25, 27]. Therefore, further investigations into positive effects of strain on the tenogenic activity of ADSCs should be further studied.
Differentiation into adipogenic lineage
In musculoskeletal tissue engineering, it is important to make sure that cells seeded in scaffolds differentiate into the desired type. A major concern when using ADSCs is that they may differentiate into adipocytes. Confirming that ADSCs do not exhibit an adipogenic phenotype is crucial to the integrity of the implant site. Recent research has shown that mechanical loading can inhibit the differentiation of ADSCs into adipocytes, making it an even more attractive tool in musculoskeletal engineering.
Yang et al.  observed the adipogenic differentiation of adipogenically induced murine ADSCs Murine ADSCs were adipogenically induced for 3 days, then seeded on polyethylene plates, and after 6 h adipogenic media was replaced. Mechanical loading of 0.2 % (2000 µε) at 1 Hz for 2 or 6 h began immediately using a four point bending apparatus. ADSCs that had not been adipogenically-inducted underwent the same loading conditions. Loading significantly reduced the amount of oil droplet filled cells visible. Additionally, both 2 and 6 h significantly decreased PPAR-γ and increased Runx2 and Pref-1 in osteogenically induced cultures, as compared to their static counterparts. Control ADSCs, however, did not show a significant difference in gene expression compared to static un-induced when exposed to 2 or 6 h of strain. These results indicate that mechanical strain can hinder adipogenic differentiation, and that osteogenic and adipogenic lineages may be inversely related.
Huang et al.  seeded ADSCs on collagen-coated flexible bottom plates and allowed to attach for 16 h. Mechanical strain was then applied using a Flexcell 4000T tensile system with 0.5, 2, or 10 % at 0.5 Hz for 48 h in adipogenic or control media. This study found that 10 % loading applied to old mouse ADSCs increased their proliferative potential, and that 2 and 10 % strain levels inhibited adipogenic differentiation.
Li, G., et al.  exposed ADSCs to adipogenic media for 3 days, then seeded onto loading plates, after 6 h adipogenic media was added, and cells were subjected to 4-point mechanical loading of 0.2 % (2000 µε) at 1 Hz for 2 or 6 h. Oil red O was used to asses fat droplet formation 2 h after end of loading, and adipogenic genes, PPAR-c1 (a critical transactivator of adipogenesis) and APN (one of the adipocyte-predominant proteins), were examined by real-time PCR. Oil red O showed a significant reduction in fat droplet deposition in the 2 h loading compared to control, and in the 6 h group compared to the 2 h group. Both APN and PPAR-γ expression were significantly decreased for 2 and 6 h; APN was reduced to 80 % and PPAR-γ to 50 % for 2 h, and APN decreased to 20 % and PPAR-γ to 3 % for 6. This study found that mechanical compression significantly reduced the adipogenic differentiation of ADSCs even in adipogenic differentiation media.
In addition to studies that show mechanical strain as a useful tool in the differentiation of ADSCs into various musculoskeletal lineages, the results of these studies show that it can also inhibit adipogenesis. These results further solidify the need for mechanical loading in the musculoskeletal tissue engineering field. Table 2 compiles the results of representative studies utilized ADSCs along with various mechanical loading modalities to differentiate them into different linages.