Proc. Natl. Acad. Sci. USAVol. 77, No. 6, pp. 3455-3459, June 1980Cell Biology

Culture of hormone-dependent functional epithelial cellsfrom rat thyroids

(thyroglobulin/iodide transport/hormones/differentiation/aging)

F. S. AMBESI-IMPIOMBATO*, L. A. M. PARKSt, AND H. G. COONtf*Centro di Endocrinologia ed Oncologia Sperimentale del C.N.R.c/o Istituto di Patologia Generale, II Facolta' di Medicina, Universita' di Napoli, Naples, Italy;and tLaboratory of Cell Biology, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20205

Communicated by Rita Levi-Montalcini, March 4,1980

ABSTRACT Primary cultures of rat thyroid cells were madein medium supplemented with 0.1-0.5% calf serum and con-taining six hormones or growth factors: insulin, thyrotropin,transferrin, hydrocortisone, somatostatin, and glycyl-L-histi-dyl-L-lysine acetate. The FRTL strain was purified by successivecolonial isolations and was found to maintain highly differen-tiated features (secretion into the culture medium of physio-logical amounts of thyroglobulin and concentration of iodideby 100-fold). The FRTL strain has been observed for more than3 years in continuous culture. It has maintained the same bio-chemical and morphological characteristics that typified theprimary cultures of thyroid follicular cells immediately aftertheir enzymatic release from the rat thyroid. Thyroid epithelialcells that were grown under more conventional cell cultureconditions failed to retain these specialized characteristics. Weshow that maintenance in vitro of these specialized functionsof rat thyroid follicular cells is dependent on low serum con-centrations and supplementation with hormones in the primarycultures. Our observations indicate that this culture strategemmay be applicable to the general problem of maintenance ofdifferentiated characteristics in cultures of other epithelialcells.

In spite of many advances in cell culture there has been rela-tively little success in culturing differentiated epithelial cells.The majority of cultures that have been adapted to grow in vitroand have been cloned are mesenchymal cells or cells derivedfrom tumors in which it may be difficult to distinguish betweenthe epithelial and fibroblast-like phenotypes. Recently, how-ever, the number of amniote epithelial cells grown and studiedin vitro has been increasing as a consequence of a better defi-nition of the cellular requirements (1-6) and the availabilityof improved growth media. Most cultured cells, especially thoseof the fibroblast type, require the addition of relatively highamounts of serum (5-20%) for growth. The addition of serumis considered necessary to provide the cells with unknown"serum factors" needed for cell attachment to the substrate andfor cell division.The pioneering work of Sato (7), which has validated the

hypothesis that for many cultured cells serum serves as a sourceof hormones (many of which probably are still undefined), hasmade possible new precision in regulating the in vitro envi-ronment. This idea, which has proven successful with estab-lished cell lines (8) and tumor-derived cell lines, has served asthe basis for our approach to the study of normal thyroid cellstrains derived from primary culture. It has allowed us to growhighly differentiated thyroid follicular cell strains from the ratin continuous culture. When traditional culture techniques wereused, strains of epithelial thyroid follicle cells no longer ex-pressed important aspects of thyroid differentiation (production

of thyroglobulin and concentration of iodide), whereas cellstrains maintained continuously in medium supplemented with0.1-0.5% calf serum and hormones did retain these thyroid-specific functions.

MATERIALS AND METHODSTrypsinization and Primary Cultures. The cell dissociation

procedures used were similar to those now in standard use forthe isolation of primary cell lines (9, 10). Thyroid glands wereexcised from 5- to 6-week-old Fischer rats that had been killedby CO2 asphyxiation. Usually, the glands from three to six ratswere pooled and used for the primary cultures. All the proce-dures were performed under sterile conditions. The glands werefreed from adherent connective tissue, cut into small pieces, andwashed in Ca2+- and Mg2+-free Hanks' salt solution (11) bycentrifugation at approximately 500 X g. Enzymatic digestionwas carried out by adding to the washed pellet 3 ml of colla-genase (CLSPA, Worthington), 20 units per ml trypsin (1:300,ICN Pharmaceuticals), 0.75 mg per ml heat-inactivated di-alyzed chicken serum (GIBCO), 2% in Ca2+- and Mg2+-freeHanks' salt solution (CTC solution) (1).The tissue was incubated for 20 min at 37°C in a vigorously

shaking waterbath. Usually, two 20-min incubations and somevigorous pipetting were sufficient to reduce the whole tissuemass to a cell suspension. After a 2-min settling period (to allowlarge fragments to settle), the supernatants were collected,combined, washed in complete medium, and distributed into10-cm Falcon plastic tissue culture dishes at 104 to 105 cells perdish.

Passage of Cells. Secondary cultures were made by trans-ferring individual epithelial colonies with cloning cylinders.When uniform cell suspensions were desired, the CTC made2 mM in ethylene glycol bis(f3-aminoethyl ether)-N,N,-N',N'-tetraacetic acid (Sigma) which was added as a 0.2 Msolution in 1 M NaOH/Ca2+- and Mg2+-free Hanks' saline,1:1.

Culture Medium. The culture medium used was a modifi-cation (12) of Ham's F-12M (13) supplemented with 0.1-0.5%calf serum (GIBCO). Fetal calf serum was found to be unsuit-able for these cells. The culture medium and serum supplementwere terminally sterilized by filtration through 0.22-,um pre-washed Millipore filters. Just before use, this medium wassupplemented with a 60- or 100-fold concentrate of the sterilehormone mixture. The complete supplemented medium hasbeen designated mF12+6H to indicate that it is a modifiedF-12M formulation supplemented with the mixture of sixhormones (Table 1).

Abbreviations: mF12, modified F-12 medium (see Table 1); mF12+6H, mF12 supplemented with 0.5% calf serum and six hormones.t To whom reprint requests should be addressed.

3455

The publication costs of this article were defrayed in part by pagecharge payment. This article must therefore be hereby marked "ad-vertisement" in accordance with 18 U. S. C. §1734 solely to indicatethis fact.

Dow

nloa

ded

by g

uest

on

June

3, 2

021

3456 Cell Biology: Ambesi-Impiombato et al.

Table 1. Composition of the modified F-12 medium (mF12)Concentrationmg/

Component liter mM ComponentCalcium

422.0 2.0 pantothe-nate

42.0 0.2 Niacinamide7.8 0.06 Linoleic acid

26.2 0.2 Pyridoxine-HClThiamine.HCI

73.0 0.4 Riboflavin16.0 0.2 Folic acid9.0 0.06 Vitamin B-12

Thioctic acid10.0 0.06 myo-Inositol21.0 0.2 Ascorbic acid23.8 0.2 Choline-HCl

4.0

11.0

23.4

48.0

0.020.060.20.2

30.0 0.270.0 0.618.0 0.2

26.0 0.2

30.0 0.2

220.0 2.0

0.3 0.001

0.07 0.003

ThymidineHypoxanthine

NaClKCINa2HPO4-7H20

KH2PO4MgSO4.7H20MgCl246H20CaC12.2H20CuS04-5H20ZnSO4.7H20FeSO4-7H20

GlucoseNaHCO3Phenol redGlutamine

Concentrationmg/liter mM

0.50.040.090.060.2850.041.01.00.2

36.045.013.8

0.0010.00030.00030.00030.00080.00030.0020.00070.00090.20.2500.1

0.7 0.0034.0 0.030

7530.0 130.0305.0 4.0

250.0 0.968.0 0.5

104.0 0.4106.0 0.4165.0 1.1

0.002 0.0000080.144 0.00050.8 0.0028

2000.0 11.02500.0 29.7

1.25 0.0035292.0 2.0

Hormone Supplement for Rat Thyroid Cells. The finalhormone mixture was arrived at by testing various hormonesor factors at physiological or nearly physiological concentra-tions. This approach led to the total replacement of the serumrequirements for already established cell lines such as GH3 cells(8). The final concentrations of the hormones used were: insulin(Sigma), 10 ,ug/ml; hydrocortisone (Sigma), 10 nM; transferrin(Sigma), 5 ,ug/ml; glycyl-L-histidyl-L-lysine acetate (Calbio-chem), 10 ng/ml; somatostatin (Calbiochem), 10 ng/ml; andthyrotropin (Pituitary Hormone Distribution Program, NationalInstitute of Arthritis, Metabolism and Digestive Diseases), 10milliunits/ml. These hormones were made up as a single 60-to 100-fold concentrated stock solution in Ca2+- and Mg2+-freeHanks' saline, divided into aliquots, and frozen rapidly in liquidnitrogen (thyrotropin preparations may lose activity unless theyare frozen rapidly). After freezing, the stock solutions werestored at -80'C. No antibiotics were added. The medium wasrenewed twice weekly.

Iodide Uptake Assay. We adapted the method describedby Tong (9) for suspensions of freshly isolated cells. Between0.5 and 1.0 X 106 cpm of 125I per dish was added to the growthmedium in the presence of 1 gM cold iodide and 3 mM 1-methyl-2-mercaptoimidazole (to inhibit organification of io-dide). Cells were incubated for 2 hr. At the end of the incuba-tion, the medium was removed and its radioactivity was de-termined. Cells were then scraped with a rubber policeman,pelletted in pre-weighted tubes, and assayed for radioactivity.The C/M ratio (i.e., cpm/pl of medium:cpm/mg of cell pellet)was calculated.

Radioimmunoassay for Thyroglobulin. Thyroglobulin wasdetermined by P. R. Larsen by a double-antibody immu-noprecipitation radioimmunoassay technique. Details of thismethod and of 'the specificity of the antibodies have beenpublished (14).

RESULTSMost of the culture regimens and media tested produced eithernondividing cultures or dividing cultures of nonfunctionalepithelial or fibroblast-like cells. Two of the culture strategiesconsistently yielded epithelial cell strains but with oppositefunctional properties. The FRT cell strain was isolated in me-dium supplemented with 5% serum and dibutyryl cyclic AMPand then was "conditioned" by culture on thyroid fibroblastsfor 24 hr. The FRT cells were able to produce thyroglobulin inprimary cultures but after adaptation to growth in uncondi-tioned medium they lost the ability to concentrate iodide or toproduce thyroglobulin. The FRTL strain was isolated in me-dium supplemented with 0.5% calf serum and purified hor-mones. FRTL cells appear to represent differentiated thyroidfollicle cells growing in culture. Because our success or failurein producing functional cell cultures was determined by theinteraction between primary cells and the different media,these are described in detail below.

Primary cell suspensions were made by serial treatment withcollagenase and trypsin of three to five thyroid glands pooledfrom 6-week-old littermates of the NIH Fischer 344 inbredstrain of rats. After washing by centrifugation, the primary cellsuspensions were counted and distributed into tissue culturedishes at different densities in the experimental media.Medium with 2-10% Serum Supplement Only. Both the

epithelial and fibroblast-like cells from primary suspensions ofrat thyroid glands attached efficiently in mF12 supplementedwith 2-10% fetal calf serum. However, within 1 week the epi-thelial cells were usually found to have lysed selectively leavingonly the fibroblast-like cells in the cultures. Supplementationof mF12 with the same concentrations of calf serum yieldedcultures in which the epithelial component (initially >80% ofthe attached cell population) survived but did not divide.Cultures at high serum concentrations (2-10%) without otheradditives were eventually overgrown by fibroblast-like cells inall cases.

Epithelial cells grew in medium that was supplemented with5% calf serum and dibutyryl cyclic AMP (10 ttM) after it wasconditioned by thyroid fibroblasts. This regimen allowed theepithelial cells to divide and apparently restricted division ofthe fibroblast-like cells. These epithelial cells were later adaptedto grow in medium containing 5% serum without prior condi-tioning or added dibutyryl cyclic AMP. These cells were foundnot to produce detectable amounts of thyroglobulin or to con-centrate iodide (strain FRT in Tables 2 and 3), except when theywere cocultivated as primary or secondary cultures with thyroidfibroblasts.Medium Without Serum or Hormones. When primary rat

thyroid cells were cultured in mF12 with no serum or hormonesupplement, cell attachment was exceedingly rare. Within 3-4days, floating vesicles or follicles made up of initially undisso-ciated or reaggregated cells could still be found. However, nocell division or colony growth was ever observed.Medium with Hormones Alone or Low Serum Alone. In

the primary cultures, cell attachment was clearly enhanced inthe presence of the added hormones. Cell division and colonygrowth were evident among the epithelial cells immediatelyafter plating. In contrast, fibroblast-like cells were rarely ob-served to divide. Our attempts to transfer, by trypsinization,the primary epithelial cells grown in medium with hormone

L-Arginine.HC1 4

L-Histidine.HCI

L-IsoleucineL-LeucineL-Lysine.HC1

GlycineL-MethionineL-Phenylala-

nineL-SerineL-ThreonineL-Trypto-phan

L-TyrosineL-ValineL-CystineL-Asparagine.HCO

L-ProlineL-AlanineL-Aspartic

acidL-Glutamic

acid

Sodiumpyruvate

Putrescine-2HCI

Biotin

Proc. Natl. Acad. Sci. USA 77 (1980)D

ownl

oade

d by

gue

st o

n Ju

ne 3

, 202

1

Proc. Natl. Acad. Sci. USA 77 (1980) 3457

Table 2. Thyroglobulin production by cultured cellsThyroglobulin

Cells ng/ml* pg/cell/dayFRT

3458 Cell Biology: Ambesi-Impiombato et al.

a single cell, is probably the best method for achieving homo-geneity in cell strains of this kind that have very low platingefficiencies.The FRTL strain grows with a population doubling time of

5-7 days in mF12+6H at 370C. We calculate conservativelythat the FRTL strain has undergone more than 175 populationdoublings since its initiation in July 1976. These cells have re-mained diploid. Late in 1978, 25 metaphases were scanned; 23of them had 42 chromosomes and 2 had 41 (and probablyrepresented broken metaphases); all the chromosomes appearednormal in shape as judged by the quinacrine mustard fluo-rescent banding technique. These results were duplicated inthe fall of 1979.The morphological characteristics of the primary cells are

unchanged in the FRTL strain after more than 3 years of con-tinuous culture. By phase-contrast microscopy, individual cellsare small with a small nucleus containing one or two nucleoli.The FRTL cells are round and smaller than most cultured cellsbut, because they do not adhere broadly to the plastic support(due to the low serum concentration), the cytoplasm is oftenstretched in various directions, particularly immediately afterisolation. The cytoplasm of such isolated cells sometimes outlideslarge lumen-like structures (Fig. 2). Colonies often developfollicular lumina at their center, which tend to grow or to fuseinto larger "follicles" in time (Fig. 3). At high cell densities andafter several weeks in the same culture dish, the cells tend togrow on top of one another to form three-dimensional structures(often without a discernible lumen) rather than to expand ina monolayer to fill the empty spaces on the dish.

Thyroglobulin Production. The FRTL cells were found, byradioimmunoassay, to secrete thyroglobulin into the growthmedium. Tests were made first after 4 months and again at leastonce a year for the 3 years of their continuous culture. The su-pernatant medium (6 ml/2-4 X 106 cells per 100-mm plate)was removed for radioimmunoassay 3 days after the last com-plete medium exchange. Thyroglobulin was found in this me-dium at a concentration of 7.0.ug/ml (Table 2). Control mediafrom other cell strains including nonfunctional thyroid epi-

FIG. 2. Phase-contrast photomicrograph of isolated FRTL cellsin primary culture. The large vacuolated structures are typical ofprimary cultures of thyroid cells and may represent attempts toproduce follicle-like structures. The irregular shape of some of thecells when not part of a colony (bottom) is probably due to incompleteattachment to the plastic support and occurs only in low serum con-centrations. Bar indicates 100lOm.

FIG. 3. Phase-contrast photomicrograph of a portion of a largecolony of FRTL cells at the third passage. The tendency to formthree-dimensional follicle-like structures that increase in size withtime can be noted even in the absence of nonepithelial cells. Bar in-dicates 100,um.

thelial cells (strain FRT that were grown in medium with 5%serum but without hormones), thyroid fibroblasts, and liver andkidney cells all showed no thyroglobulin production.

Iodide Concentration. The C/M ratio was determined forseveral cell lines (Table 3). Fibroblast cells and nonfunctionalthyroid epithelial cells, as well as cells of nonthyroid origin, allhad a C/M close to 1.0, indicating that the equilibrium positionreached by the radioactive iodide added to the medium waslimited by a simple diffusion process. FRTL cells were able toconcentrate iodide to 100 times the external concentration.

DISCUSSIONOur observations indicate that the low-serum hormone-sup-plement approach to cell culture may be generally applicableto the culture of other specialized epithelial cells from kidney,liver, uterus, and lung. The obvious special advantage thatthyroid offers is that one specific tropic hormone for it is alreadywell known.

Perhaps one of the most valuable advantages of the low-serum strategy is that, from the first few days of primary cul-ture, the epithelial cells, not the fibroblasts, are selectively fa-vored in medium with 0.5% calf serum. The well-known pro-gressive selection in favor of fibroblast-like cells, which arebetter adapted to growth in conventional culture conditions(17), is thus avoided.

As our results show, it may not be easy to make unambiguousdecisions as to whether or not a growth factor is required. Inlow-cell-density experiments the degree of dissociation of thecells was found to be critical. In plates containing incompletelydissociated clusters of 25-50 cells, division may continue slowlyin these regions and after many weeks even restore or exceedthe original inoculum (as seen with curves 2 and 3 in Fig. 1). Wefound it difficult or impossible to establish an absolute re-quirement for glycylhistidyllysine and hydrocortisone at rel-atively high cell densities. With inocula >100 cells/mm2 eventhe requirements for thyrotropin and insulin and especiallythose for transferrin and somatostatin were slow to be expressed.Our results clearly show that the physiology of the cells may bedifferent under different conditions of cell density.

After more than 3 years of continuous culture, FRTL cells

Proc. Natl. Acad. Sci. USA 77 (1980)

Dow

nloa

ded

by g

uest

on

June

3, 2

021

Proc. Natl. Acad. Sci. USA 77 (1980) 3459

still retain a principal differentiated characteristic of folliclecells: dependence on thyrotropin for growth. When a gradedseries of thyrotropin concentrations were tried, we found thatcell growth and the persistence of the morphological and dif-ferentiated characteristics were dose dependent. Although aneffect could be seen at 0.5 milliunit of thyrotropin per ml, adramatic stimulation of growth and maintenance of normalmorphology was present at 5.0 and 10.0 milliunits/ml; con-centrations of 30, 50, or 100 milliunits/ml showed little furtherimprovement. When these experiments were repeated with amore highly purified preparation of thyrotropin the findingswere identical. This indicates that the growth-stimulating effectwas due to the thyrotropin and not to possible contaminantspresent in the less pure preparations that have been used rou-tinely as supplements in our cultures. Because of the sharp effectof thyrotropin on division in these cells it is possible that thy-rotropin deprivation might prove helpful in obtaining synch-rony of DNA synthesis or cell division.We consider that the maintenance of a strict dependence on

thyrotropin is strong evidence for the normality and the dif-ferentiated character of FRTL cell cultures. Recently it wasreported (18) that thyrotropin is not a growth factor for humanthyroid cells in culture. This conclusion was based on the findingthat conventional mass cultures of human thyroid tumorsshowed a dose-related decrease of [3H]thymidine incorporationin a 24-hr interval after administration of increasing doses ofbovine or human thyrotropin preparations. The functionalstatus of these comparatively short-term cultures was not re-ported. The data appear to fall short of establishing a role forthyrotropin in cultures of normal differentiated thyroid folliclecells. Our cultures of functional rat thyroid epithelial cells re-quired thyrotropin for sustained cell division and long-termsurvival at all cell densities.We have observed no evidence of senescence in vitro of these

FRTL cells or among other epithelial cell strains derived fromrat liver (BRL) (2). The FRTL cells have undergone at least 175population doublings without a discernible crisis or change inkaryotype, growth rate, or level of differentiated function.FRTL cells did not form tumors when 2 X 106 cells were in-jected into each of five Fischer rats that were subsequentlymonitored for more than a year. The cellular senescence phe-nomenon reported by Hayflick (19)-in which cultures ofhuman lung and skin fibroblasts ceased to divide after 50-100population doublings in vitro- does not seem to apply to epi-thelial cells of the rat. One of the considerations that has in-hibited investigators from using apparently normal primarycell strains as opposed to transformed cell lines for biochemicalstudies has been the fear that these cell strains would suddenlybecome "senescent" and stop growing, making it difficult tocomplete long-term studies. Our FRTL culture has beenmaintained free of heteroploidy or of any of the characteristicsof the transformed state without evidence of senescence. Toavoid a circular argument, we must deny that remaining ap-parently unchanged for 200 cell generations is in itself evidenceof transformation or of becoming an established cell line.During most of the 3 years of continuous culture of strain

FRTL, the cells were harvested and replated at 106/100-mmplate every 2-3 weeks. The cells from one dish were not pooledwith others before transfer, so that any change, such as aspontaneous transformation, that might have occurred in oneplate would be detected before it had spread to all the platesbeing carried. Periodically, samples of several lineages werefrozen and stored in liquid nitrogen. Twice a year, chromosome

counts were made and quinacrine mustard banded preparationswere scanned for obvious anomalies. During the second andthird years of FRTL serial passages, two spontaneous trans-formation events were detected and segregated from the mainlineage; during the third year, a translocation was found andeliminated by return to a previously frozen sample. One of thespontaneous transformations gave a fully functional trans-formed subline with a population doubling time of about 1 dayand a modal chromosome number of 67. The second appearedfibroblast-like and had lost both differentiated functions. Thetranslocation strain had nearly the same population doublingtime, morphology, and differentiated function as the mainlineage FRTL. With relatively simple precautions like passagewithout pooling and with periodic chromosome counts, it shouldbe possible to propagate nontransformed epithelial strains forat least 200 population doublings and perhaps without limit.We anticipate that the ability to culture and colonially purify

apparently normal differentiated thyroid cells in vitro willprove useful in a wide range of studies in cellular genetics, generegulation, pharmacology, and aging, apart from the obviousapplications in studies of thyroid endocrine physiology andbiochemistry. In particular, with the availability of diploidhormone-dependent endocrine cells in vitro, novel approachesto the study of the interactions between cells and hormones orhormone-like substances will become possible. This will cer-tainly lead to a more detailed understanding of the physiologicalinteractions between organs and hopefully will lead to a betterunderstanding of the integration among cell types and tissueswithin an organ.

Note Added in Proof. Low serum conditions are critical during earlypassages when selection of the more differentiated cells occurs.However, once isolated, a substrain of FRTL was grown for >1 yr inmF12+6H plus 5% serum, had a doubling time of 30-40 hr, and re-mained diploid and differentiated.

We acknowledge the excellent technical assistance of Ann Rahe andwe thank Dr. L. D. Kohn for the gift of purified thyrotropin, Dr. R.Larsen for the thyroglobulin radioimmunoassay, and Dr. G. H. Satofor helpful discussions and advice.1.

2.

3.

4.

5.

6.

.. 7.

8.

9.

10.

11L12.

13.

14.

15.

16.

17.

18.

19.

Coon, H. G. (1966) Proc. Natl. Acad. Sci. USA 55,66-73.Coon, H. G. (1962) Carnegie Inst. Washington Yearb. 67,419-421.Cahn, R. D. & Cahn, M. B. (1966) Proc. Natl. Acad. Sci. USA 55,106-114.Spooner, B. S. (1970) J. Cell Physiol. 75,33-41.Tashjian, A. H., Yasumura, Y., Levine, L., Sato, G. H. & Parker,M. L. (1968) Endocrinology 82,342-352.Green, H. & Rheinwald, J. G. (1975) Cell 6,331-343.Sato, G. H. (1975) in Biochemical Actions of Hormones, ed.Litwack, G. (Academic, New York), pp. 391-409.Hayashi, I. & Sato, G. H. (1976) Nature (London) 259, 132-134.Tong, W. (1974) Methods Enzymol. 32, 745-758.Winand, R. J. & Kohn, L. D. (1975) J. Biol. Chem. 250,6534-6540.Konigsberg, I. R. (1963) Science 140, 1273-1284.Coon, H. G. & Weiss, M. C. (1969) Proc. Natl. Acad. Sci. USA62,852-859.Ham, R. G. (1965) Proc. Natl. Acad. Sci. USA 53,288-293.Izumi, M. & Larsen, P. R. (1978) Endocrinology 103,96-100.Gospadarowicz, D. (1974) Nature (London) 249, 123-127.Cohen, S. (1962) J. Biol. Chem. 273, 1555-1562.Sato, G. H., Zaroff, L. & Mills, S. E. (1960) Proc. Natl. Acad. Sd.USA 46, 963-972.Westermark, B., Karlsson, F. A. & Walinder, 0. (1979) Proc. Natl.Acad. Sci. USA 76,2022-2026.Hayffick, L. (1965) Exp. Cell Res. 37, 614-636.

Cell Biology: Ambesi-Impiombato et al.

Dow

nloa

ded

by g

uest

on

June

3, 2

021