Proceedings of the XaLimmZ Acadm~ of Sciences Vol. 66, No. 1, pp. 16CLlG7, May 1970 Regulation of Axon Formation by Clonal Lines of a Neural Tumor Iv. W. Seeds,* A. G. Gilman, T. Amano, and M. W. Nirenberg L.\HOR.\TORY OF IlIOCIfKMICAL GENETICS, NITIOS.iL INSTITUTI'S OF HlilLTH, Hk:THESDA, MARYLAND Communicated February 6, 1.970 Abstract. Clonal lines of ncuroblastoma cells were found to estcnd or retract axons depending upon the concentration of serum. n'eurite extension was not inhibited by cycloheximide but was sensitive to colchicine or vinblastine, sug- gesting that neurite formation is dependent upon the assembly of microtubules or neurofilaments from preformed protein subunits. Clonal lines of neuroblastoma cells exhibit many properties of differentiated sympathetic neurons. Cells extend branched axons+ up to 3000 P in length,`-' possess membranes that arc electrically excitable,4 respond to acetylcholine,5 and also contain enzymes for the synthesis and metabolism of catecholaminesL~ 2, 6 and for the hydrolysis of acetylcholine.`~ 7 Cultures usually contain two types of cells quite different in morphology: round cells without processes and cells with axon-like processes. Although cell shape in vitro is notoriously variable and many varieties of cells are capable of extending processes, the distinctive neuronal appearance of neuroblast,oma processes agrees well with the neural characteristics demonstrated in biochemical and neurophysiological studies. The proportion of neuroblastoma cells with asons was found to vary greatly, depending upon the clone studied and the conditions of cell culture. The system is relatively simple from an experiment,al point of view and may be useful for studies of both the differentiation and the function of neurons. Materials and Methods. Neuroblastoma C-1300 was adapted for growth in vitro as previously described.4 Cells were first cloned as a colony from agar and were then cloned from single cells with the use of stainless steel cylinders. Cultures of neuroblastoma clone N-18 were grown in Falcon flasks or Petri dishes m Dulbecco's modification of Eagle's medium (DMEM) plus 10% fetal calf serum (except where indicated) at 37oC in an atmosphere of 10% COz and 90% air. Cells were evaluated for the presence of axon-like processes from phot,omicrographs (data of Figs. 4 and 5) or directly under the microscope. An initial cell concentration of approximately 3000-5000/cm2 proved to be convenient for counting cells and axons and was used except where indicated; 300-1200 cells were counted for each value. Repro- ducibility was approximately +15%. Trypsinized cells (500-1000) were counted with a hemocytometer and viability was estimated by exclusion of nigrosin. 3H-Proline and 3H-thymidine incorporations into protein and DNA, respectively, w-ere determined by addition of labeled and unlabeled compounds in Dulbecco's modification of Eagle's medium to plates to achieve the following concentrations: n-proline lop3 -11, 10 &i/ml, 2850 cpm/nmole; thymidine 3 x lop6 M, 5 ,&i/ml, 600 cpm/pmole. Cells 169 VOL. 66, 1970 BIOCHEMISTRY: SEEDS ET AL. 161 were incubated for 1 hr at 37oC in a 10% CO,-90% air atmosphere. Protein synthesis was determined by counting hot trichloroacetic acid-precipitable material on nitrocellu- lose filters. DNA synthesis was assayed by counting cold trichloroacetic acid-precipitable material on glass fiber filters (Millipore Co.). aH-Thymidine (10 Ci/mmole) and aH-proline (26.3 Ci/mmole) were obtained from Schwarz and were purified by paper chromatography prior to use. Bovine plasma protein fractions were obtained from Pentex (al-globulins, Cohn fraction IV1; crrglobulins, fraction IVn; p-globulins, fraction III; -y-globulins, fraction II; transferrin, 67y0 estimated purity; bovine serum albumin, crystalline). Results. We thought it likely that the proportion of neurobIastoma cells with axons and the length of axons might be related to the rate of cell division because cells are known to retract processes prior to cell division. Since serum is required for multiplication of neuroblastoma cells, the extent of cell division was restricted by incubating cells with relatively low concentrations of serum. Few cells possess processes in 10% fetal calf serum (Fig. 1A). Most cells are round and adhere to one another, forming clusters. When cells are incubat'ed without serum, axon-like processes are rapidly extended. Within 30 to 60 min, most cells possess processes 2.5-100 ,U in length (Fig. 1B) ; relatively long neurites are found after one day (Fig. 1C). After four days (Fig. 10) many cells possess axons up to 2000 p in length, and further elongation and arborization are apparent at, seven days (Fig. 1E). In most cases, two to four branched neurit'es extend from the body of a single cell. Usually a binary pattern of neurite branching was found (i.e., two neurites arise from each branch node). The relation between serum concentration and t'he proportion of cells with neurites is shown in Figure 2. Approximately l--57$ of the cells extend axons in 10% serum. J'Iost of the cells extend neurites in the presence of 0 to 1% serum; however, the rate of :~ppearancc of neurites is inversely related to serum concentration. The effect of cycloheximide upon axon formation was studied to determine \\hether axon outgrowth and migration are dependent upon protein synthesis (Fig. 3). In addition, cells were incubated with colchicine or vinblastine t'o invcstigatc the possibility that axon outgrowth is dependent upon the assembly of microtubule protein. Both alkaloids interact with microtubule prot,ein.sr g Cyclohcximidc was found to have litt,le effect on neurite outgrowth at concen- t,rntions up to 1.S X lop4 M. This concentration of cycloheximide inhibits the incorporation of proline into protein by more than 97%. However, vinblnstine and colchicine inhibit neurite out-growth completely at lo-' and lo+ Al, respectively. These results suggest that neurite formation does not re- quire tie nova protein synthesis but is dependent upon the assembIy of micro- tubules from performed protein subunits. In addition, $ectron microscopic observations demo&rate numerous microtubules, 240 A in diameter, and neurofilaments, 100 d in axons induced by growth in low serum. Additional factors influencing neurite outgrowth are shown in Table 1. Seuritcs are not cxtcnded by cells at 3" and outgrowth is greatly retarded at, 2-l" compared to 37'. The tcmpernturc of incubation thus markedly affects neurite outgrowth. As shown in Experiment 2, the serum factor(s) affecting neurite out'growth is not dialyznble and is active after incubation at 100" for 162 BIOCHEMISTRI': SEEDS ET :lL. 1'1coc. N. A. S. 15 min. Conditioned media (final serum concentration, l-2%) also inhibits neurite formation. Cells incubat,ed in Dulbccco's phospllate-buff ered saline solution plus glucose extend processes as ~~41 as those incubated in growth medium (Expt. 3). No processes are extended in the absence of Ca++ and iug++. VOL. w, 1970 BIOCHEMISTIZY: SEEDS ET AL. 163 FIG. Z.-Effect of serum concent'ra- lion upon rate of neurite formation. go Cells were grown in Dulbecco's modi- fication of Eagle's medium containing % 7 the indicated total concentration of an equa1 mixture of horse serum and fetal z s calf serum, and neurites were evaluated + 5 at, indicated times. A, No serum; 0, 0.37, serum; .Y, 1.0% serum; 0, 10% serum; V, data obtained at 1,2, and 8 hr in a separate experiment where serum was removed from cells previously grown in 10% serum for 24 hr. The effects of serum protein fractions updn neurite formation are shown in Table 2. Crystal- i line bovine serum albumin, t'ransferrin, and a bovine y-globulin fraction have little effect upon the formation of neurites. However, al-, ah-, and P-globulin fractions from bovine serum are inhibitory. ParGal inhibition is also observed at concentrations 10 to 100-fold lower than those shown. Chondroitin sulfate, a-lactalbumin, and p-lactoglobulin also are without effect (data not FIG. 3.-Effect of colchicine, vinblastine, and cycloheximide on shown). initial neurite formation. Cells An attempt was made t,o examine the relation between ncurite extension and cell division (Figs. ~~~.,"~",~$`",h,~~&`~~ "Tii,", 24 hr plates were rinsed with -U and B). During logarithmic growth in 10% Dulbecco's modification of serum, less than 3% of the cells possess neu- Eagle's medium and then in- rites. After incubat,ion for two days, the cul- cubated for 2 hr in this medium and the components indicated. ture media of some plates was replaced wit,h fresh media with 0.1% serum. Eighty percent of the cells extended neuriks during the next 24 hours. Further incubation led to a marked increase in neurite length; TABLE 1. Requirements for initial new-de jormalion. Percentage of Expt. no. Conditions cells with a~xons 1 Minus serum, 3oC 0.3 Minus serum, 24oC 9 Minus serum 37oC 10yO serum, :;7"C 73 1 Minus serum 70 1% serum 7 lyO serum, heat for 15 min at 100oC 15 1% serum, dialyzed 12 JIinus serum 76 Phosphate-buffered saline (minus serum and growth medium) 76 Ptlospllat~t)rlBered saline without C:I ++ and hlg + + (millus serum and growth medium) 8 Cellv were incubated 1 day in Dulbeceo's modifktion of Eagle's medium plus lOT$ fetal c:llf serum. Plates then were washed and fresh medium plus components listed above were added. Cultures then were incubated for 2 hr at 37'. 164 BIOCIIEI~IISTRY: SEEDS ET AL. Pnoc. N. A. S. TABLI~: 2. EJcct of serum jracfions on initia.1 neurilc jornmlion. Additions None 10yO fetal calf serum Albumin, crystalline Transferrin al-Globulin (IV-l) aa-Globulin (W-4) p-Globulin (III) T-Globulin (II) Percentage of cells with axons 63 2 74 s4 9 4 `21 68 Cells, incubated for 1 day in the presence of 10% fetal calf serum. were washed with Dulhecco's modification of Eagle's medium and then were incubated for 2 hr with this medium plus the com- ponents indicated (serum fractions were tested at 1 mg protein/ml of media, final concentration.) lroivever, the proportion of cells with neurites remained constant. On the fifth and seventh day, some cultures wzre stepped up from 0.1 to 10% serum; fewer cells with neurites were found after incubation. Additional results obtained by time-lapse cinematography show that neurites that' detach from the surface of the Petri dish are resorbed. However, many cells retain Iong, well-developed mu- rites. Since cells with relatively long neurites are uncommon during logarithmic growth, it seems probable t,hat some ncurites arc not' retracted under these con- ditions. The number of viable cells per plate is shown in Figure 4B. After a short lag, cells in 10% serum grew logarithmically; the population generation time mas I6 hr. The rate of cell multiplication decreased markedly when cells were shifted down from 10 to 0.1% serum. In 0.1% serum the number of cells per plate did not increase after two days of incubation. While the addition of 10% serum on the fifth day resulted in disappearance of neurites and cell multiplication, B CELLS 0 , , , , , 1 I 2 3 4 5 6 7 8 DAYS FIG. 4.-(A and B) Effect of serum on axon formation and cell multiplication. On suc- cessive days the number of viable and total cells were counted and photomicrographs were obtained for axon counts as described under Materials and Methods. Symbols represent the following: 0, 10yO fetal calf serum; A, on the second day the cultures indicat.ed in the figure were rinsed with Dulhecco's modification of Eagle's medium and then incubated in fresh medium plus 0.1% fetal calf serum; o , on t,he fifth day fetal calf serum was added where in- dicated (107,, final concentration); 0, on the seventh day fetal calf serum was added (lo'%, final concentration) to cultures as indicated. VOL. tie, 1970 BIOCHEMtSTItI': SEEDS ET BL. 165 the population generation times were 45 and 29 hours for the first and second generations, compared with 16 hr found with cells in logarithmic growth. Since the generation time of individual cells during logarithmic growth was also found to be 16-18 hours (by time-lapse cinematography), division by 35y0 of the cells could account for the prolonged population generation time. Morphological observations of such cultures support the possibility of a population of non- dividing cells with long axons. Alternatively, all cells may be capable of dividing with a prolonged generation time. The rates of DNA and protein synthesis were also determined in the above experiment (Table 3). In general, the rate of thymidine incorporation was `GABLE 3. EJccl of serum on sli-thymidine an.d 3fi-prokne incorporation. cpm (X 10 -3) Incorporated Growth Percentage per 105 Viable Cells condition of serum Day* 3H-thymidine aH-proline Logarithmic 10 l-4 12.5 z!z 1.6 3.5 zt 0.s St&p down 0.1 3-3 8.3 It 1.1 4.3 * 0.5 Sten Steb down 0.1 6-8 2.2 It 0.2 1.9 i 0.1 up 10 6 19.6 3.7 Step up 10 8 13.0 4.2 `Dav shown in Fies. 4A and i3. Experimental conditions are given in the legend to Fig. 4. 3H-Thymidine incorporation into nucleic acid and XII-proline incorporation into protein were determined :LS described under Materi& and Melhods. Average values k the stnndurd error of the me:m ine shown. Assays were performed in triplicate each day. similar to the rate of ccl1 multiplication; however, the initial rate of incorporat,ion after cells were shifted up t,o 10% serum on the sixth day was high compared with the second day of the experiment. It is possible that cells were syn- chronized to some extent with respect to position within the ceI1 cycle. The rate of proline incorporation into protein was rel- atively less affected by changes in serum con- centration, being approximately 50% after pro- NE"PlTE RETRACT ON I longed incubation in 0.1% serum. The effect of serum upon retrnct,ion of ncurites was examined in a separate experiment (Fig. 5). Cells were incubated for one day without serum; approximately S4y0 of the cell population then i 50- `,, "----y 0:: 1 z :I: `& 1 possessed short processes. The addition of `$ *,,- ' \- 1 serum resulted in a rapid decrease in the per- a IO' -. ,_ \c 10 Y. ccntage of cells with ncuritcs that was dependent L-L 0 , 2 3 4 upon serum concentration. Colchicine and vin- HOURS hlastinc also induced neurite retraction (data not Frc;. .j.--EtIect of serum on sl~o\~Il). neurite retraction. Cells were The'effcct of serum on process formation b> incubated 1 day in the presence other cell lines was also invcstignt,ed. In the of lOo$ fetal calf serum and then were incubated for anot)her dav prcscrrce of 10% surum, mouse L cells :md HcLn withortt, serrtm; then sertrm was cells flatten, spread, and form aonfluent mono- added iw indicated. Symbols rep- layers, in contrtlst to the behavior of the IICWO- reseiit. the following concentrn- tiolw: of serL,m: O , ruy;; v, blnstoma cells. L cells possessed processes in the 17~; ~0.37~; A, no serum. 166 I~IOCHEh~ISTRI': SEEDS ET AL. Paw. N. A. S. presence of 10% serum. Removal of serum from cultures of L or HeLa cells had no discernible effect upon processes. Colchicine (lO-`j M) inhibited the spread- ing of L or HeLa cells and the formation of processes by L cells. Thus L cells and HeLa cells differ from N-1s neuroblastoma cells in response to shifts in serum concentration. Discussion. Separate clones derived from the same tumor differ markedly in the ability to extend neurites. For example, one neuroblastoma clone extends processes infrequently, whereas the majority of cells from another clone extend processes soon after they are plated. r\ieuroblast,oma clone N-18 is particularly interesting because the proportion of cells with neurites was found to vary more than loo-fold depending upon environmental conditions. While I&be and Ruddle15 have selected for populations of neuroblast,oma cells with processes by elimination of dividing cells with fluorodeoxyuridine, we have devised a simple method for converting entire populations of round cells to cells with neurites that depends upon alteration of serum concentration. Serum probably affects process formation in several ways, for factors are present that influence the attachment of cells to plate, thus altering the balance be- t,ween process extension and retraction, in addition to stimulating cell division, nucleic acid, and protein synthesis.`0-13 As WeissI has emphasized, neurites migrate only on solid surfaces, thus the stability of interactions between cell and substratum is of great import'ance. The morphology of neuroblastoma cells is apparently derived by a process of selection. Neurites may explore an area 10,000 times that, occupied by the cell body and neurites forming the most stable set of attachments relative to perturbing forces are selected. Results obtained by time-lapse cinemat,ography show t'hat the rate of neurite migrat'ion from neuroblastoma cells is approximately 75-125 p/hr. However, migration is discontinuous. Removal of serum from cult,ures results in axon outgrowth which may be re- lated, at least in part, to a more stable interaction between cells and plate. How- ever, the restriction on the rate of cell division imposed by the absence of serum probably permits the uninterrupted synthesis of relatively long neurites, since neuroblastoma cells retract processes prior to mitosis. The relation between cell division and axon or dendrite extension may be of fundamental import'ance. Since 70-S5% of neuroblastoma cells ext'end pro- cesses within 60 min after removal of serum, it is clear that cells are capable of ext,ending processes during most of the cell cycle. One may hypothesize that axons and dendrites migrate from most normal neurons of the central nervous system when the neurons are in the G-l period of the cell cycle and are repressed with respect to cell division, since most neurons are diploid and do not divide. However, neurites may migrate from certain neurons, such as Purkinje cells, during the G-2 period, since these neurons are tetraploid. At least two modes of repressing neuron multiplication can thus be envisioned. Cycloheximide does not, inhibit, the initial outgrowth of neurons; thus, neurite synthesis is not dependent upon protein synthesis. Colchicine or vinblastine, which bind to microtubule protein,*! g completely inhibit neurite formation, VOL. 66, 1970 BIOCHEMISTRY: SEEDS ET AL. 167 implying that neurite synthesis is dependent upon the assembly of tnicrotubules or neurofilaments from preformed protein subunits. The formation of relatively long neurites probably is dependent upon t'he synthesis of additional microtubule protein subunits. These results are in accord with observations that pertain to flagellar regeneration.16 Olmsted et aL3 have shown that mouse neuroblastoma C-1300 contains micro- tubule protein in relatively high concentration and that neurit,es are birefringent. In addition to our observations, Schubert et aL2 previously demonstrated the presence of microtubules and neurofilaments in neuroblastoma cells by electron microscopy. It should be noted that cell mitosis is also dependent upon the assembly of microtubule subunits and is inhibited by colchicine and vinblastine. Since the termination of neuroblast multiplication either precedes or coincides with the initiation of axon formation, one wonders whether the sequence of events may relate, at least in some cases, to a mutual requirement for micro- tubule subunit assembly. We would like to acknowledge the t,echnicnl assistnnre of Sirs. R. Selinger and bliss E. Cutler. * Postdoctoral fellow of the National Science Foundation (48034). t Axons and neurites are used synonymously to designate any cellular extension greater than 25 p in length. 1 Augusti-Tocco, G., and G. Sato, these PROCEEDINGS, 44, 311 (1969). 2 Schubert, D., S. Humpbreys, C. Baroni, and M. Cohn, these PROCFXDIKGS, 64, 316 (1969). aOlmsted, J., K. Carlson, R. Klebe, F. Ruddle, and J. Rosenhaum, these PROCIIEDIKGS, 65, 129 (1970). `Nelson, P., W. Ruffner, and M. Nirenberg, these PROCEEDINGS, 64, 1004 (1969). 6 Peacock, J., and P. Nelson, manuscript in preparation. 6 Nirenberg, P., S. Wilson, N. Seeds, and M. Nirenberg, manuscript in preparation. 7 Blume, A., F. Gilbert, T. Amano, J. Farber, S. Wilson, 11. IXosenberg, and ?\l. Nirellberg, manuscript in preparation. B Weisenberg, R. C., G. G. Borisy, and E. W. Taylor, Biochcmislry, 7, 4466 (196X). o Marantz, R., M. Ventilla, and M. Shelanski, Science, 165, 498 (1!)6!)). lo Taylor, A. C., Exptl. Cell. Res., Suppl., 8, 1.54 (1961). I1 Puck, T., C. Waldren, and C. Jones, these PROCEEDINGS, 59, 102 (l!lSX). I2 Lieberman, I., and P. Ove, J. Biol. Chtwz., 233, 6:37 (195X). I3 Todaro, G., Cr. Lazar, and H. Green, J. Cell. Cmzp. Physlsiol., 66, X25 (196-i). I4 Weiss, P. A., Znt. Rev. Cytology, 7, 391 (1958). Is Klebe, R. J., and F. H. Ruddle, J. Cell. Rio/., 43, 69a (1969). I6 Rosenbaum, J., J. Moulder, and D. Ringo, J. Cell. Bid., 41, 600 (l!)F!)).