Canalization by Selection of de Novo Induced MutationsIstituto Pasteur Italia, Fondazione Cenci-Bolognetti and Dipartimento Di Biologia e Biotecnologie “C. Darwin,” Sapienza Università

HIGHLIGHTED ARTICLE| INVESTIGATION

Canalization by Selection of de NovoInduced Mutations

Laura Fanti,1,2 Lucia Piacentini,1,2 Ugo Cappucci, Assunta M. Casale, and Sergio Pimpinelli2

Istituto Pasteur Italia, Fondazione Cenci-Bolognetti and Dipartimento Di Biologia e Biotecnologie “C. Darwin,” Sapienza Universitàdi Roma, Rome 00185, Italy

ABSTRACT One of the most fascinating scientific problems, and a subject of intense debate, is that of the mechanisms of biologicalevolution. In this context, Waddington elaborated the concepts of “canalization and assimilation” to explain how an apparentlysomatic variant induced by stress could become heritable through the germline in Drosophila. He resolved this seemingly Lamarckianphenomenon by positing the existence of cryptic mutations that can be expressed and selected under stress. To investigate therelevance of such mechanisms, we performed experiments following the Waddington procedure, then isolated and fixed threephenotypic variants along with another induced mutation that was not preceded by any phenocopy. All the fixed mutations welooked at were actually generated de novo by DNA deletions or transposon insertions, highlighting a novel mechanism for theassimilation process. Our study shows that heat-shock stress produces both phenotypic variants and germline mutations, and suggestsan alternative explanation to that of Waddington for the apparent assimilation of an acquired character. The selection of the variants,under stress, for a number of generations allows for the coselection of newly induced corresponding germline mutations, making thephenotypic variants appear heritable.

KEYWORDS Drosophila; assimilation; canalization; heat shock; transposons

THE “canalization and assimilation” concepts were elabo-rated by Conrad Waddington (Waddington 1942; 1953;

1959) to offer a Darwinian explanation (Darwin 1859) for astriking case of the apparent inheritance of acquired char-acters in Drosophila. By treating Drosophila pupae withheat-shock for several generations, Waddington was ableto induce variations in somatic phenotypic traits, called phe-nocopies, that then became heritable through the germline.Since a phenocopy—a term introduced by Goldschmidt(1935)—is defined as a nonheritable phenotype that mimicsa genetic mutant, the Waddington results appeared to be aLamarkian phenomenon (Lamarck 1809).

To give aDarwinian explanation for this apparent paradox,Waddington hypothesized the existence of cryptic geneticvariants whose expression under normal conditions would

be prevented by robust developmental processes that hecalled “genetic canalization.” When environmental stressovercomes these buffering mechanisms, the cryptic variantsmay be expressed; their expression could then be fixedthrough generations by a “genetic assimilation” process.

In recent years, the results of various experiments haverevitalized the classical debate around the original Wadding-ton hypotheses. Several molecular explanations have beenproposed for this phenomenon, basedmainly on thebiologicalrole of the Hsp90 chaperone (Rutherford and Lindquist 1998;Queitsch et al. 2002; Sollars et al. 2003; Tariq et al. 2009;Specchia et al. 2010; Burga et al. 2011; Gangaraju et al. 2011;Piacentini et al. 2014).Mutations in this chaperone can in factinduce apparent somatic variants, which could then be fixedand become heritable. Two primary mechanisms have beenproposed: either Hsp90 is a molecular capacitor involved inthe buffering of preexisting cryptic mutations (Rutherfordand Lindquist 1998; Queitsch et al. 2002), or Hsp90 is amutator involved in transposon silencing; when Hsp90 itselfis mutated, transposons are activated and produce new mu-tations (Specchia et al. 2010). It has been reported thatmutations in the Hsp90 gene can induce, at a low frequency,several different developmental abnormalities depending

Copyright © 2017 by the Genetics Society of Americadoi: https://doi.org/10.1534/genetics.117.201079Manuscript received February 10, 2017; accepted for publication May 25, 2017;published Early Online May 31, 2017.Supplemental material is available online at www.genetics.org/lookup/suppl/doi:10.1534/genetics.117.201079/-/DC1.1These authors contributed equally to this work.2Corresponding authors: Dipartimento Di Biologia e Biotecnologie “C. Darwin,”Sapienza Università di Roma, Rome 00185, Italy. E-mail: [email protected]; [email protected]; [email protected]

Genetics, Vol. 206, 1995–2006 August 2017 1995

on genetic background (Rutherford and Lindquist 1998).This suggests that the functional impairment of Hsp90 ren-ders development pathways more sensitive to hidden geneticvariability. In this view, the chaperone acts as a prototype of amorphological capacitor (Rutherford and Lindquist 1998;Queitsch et al. 2002). Other data (Specchia et al. 2010) havesuggested an additional, if not alternative, hypothesis involv-ing the classic ideas of BarbaraMcClintock about the responseof the genome to physiological and environmental stressthrough transposon activation (McClintock 1984). Hsp90 isinvolved in the repression of transposons in the germline,whichwould suggest that thefixable phenotypic abnormalitiesobserved in Hsp90 mutants are probably due to de novo mu-tations induced by transposons (Specchia et al. 2010). Theapparent capacitor function of Hsp90 would thus be due toits mutagenic properties, through the activation of transpo-sons, rather than to the buffering of cryptic genetic variability.The theoretical implications of cryptic variability vs. induciblevariability are significant. Cryptic genetic variability wouldlead to an accumulation of unexpressed genetic variants thatdo not produce a phenotype, but instead acquire functionsunder special circumstances, such as environmental or geneticperturbations. However, it is difficult to imagine how this greataccumulation of neutral variants could bemaintained. Instead,inducible variability would posit the existence of mutagenicmechanisms that are silent under normal conditions but couldbe activated by environmental changes, thus creating a largeaccumulation of de novo variants with selectable phenotypes(Piacentini et al. 2014).

To examine the importance of cryptic genetic variation vs.de novo induced genetic variation in fixing phenotypic alter-ations produced by heat-shock stress, we performed a seriesof experiments following the Waddington procedure. Ourresults suggest strongly that the main mechanism underlyingthe apparent genetic assimilation of heat-shock-induced phe-nocopies is that of corresponding de novo mutations inducedby the same treatment.

Materials and Methods

Drosophila strains

Natural populations of Drosophila melanogaster were col-lected in two different areas of central Italy in 2012. Thebalanced and multi- or singly marked stocks used for geneticmapping and complementation tests have been kept in ourlaboratory for many years. w*; ry506 Sb1 P{D2-3}99B/TM2,rySC red1 (#1798) and FM7i/C(1)DX, y1 f1 (#5263), sepia1

(#1668), forked1 (#36), and b1 cact7 pr1/CyO (#34501)stocks were obtained from the Bloomington Stock Center.Cultures were maintained at 24� on standard cornmeal-sucrose-yeast-agar medium.

Isolation of genomic DNA

Genomic DNA was prepared either from 20–30 flies or fromsingle flies. Single flies were homogenized in 100 ml of ex-

traction buffer [120 mM Tris-HCl, pH 8.0, 60 mM EDTA,pH 8.0, 80 mM NaCl, 160 mM sucrose, 0.5% v/v SDS,200 mg/ml RNase (DNase-free)], and incubated at 65� for60 min. After cooling at room temperature for a few minutes,14 ml 8 M K-acetate was added. After 30 min on ice, thesamples were spun at 10,000 rpm for 15 min at 4�; DNAwas precipitated by adding 0.5 volume of isopropanol to thesupernatant, leaving for 10 min at room temperature, andspinning again for 10 min. The pellet was washed with 70%ethanol, dried and redissolved in 20 ml H2O. Genomic DNApurified from singleflies was amplified by Illustra Ready-To-GoGenomiPhi V3 DNA Amplification Kit (GE Healthcare). ForDNA extraction from multiple flies, the volume of the extrac-tion buffer was adjusted to 1 ml, and the volumes of the sub-sequent components were changed accordingly.

Primer design and PCR amplification

All PCR specific primers (18- to 25-mers with a minimum GCcontent of 50% and average Tm of 60�) were designed usingthe Invitrogen OligoPerfect designer web tool, and oligonu-cleotide sequences were screened using a BLAST search toconfirm the specificity. Supplemental Material, Table S1 con-tains a complete list of primers used in this study. PCR ampli-fications were performed in 13 PCR Buffer with 50–100 ngof genomic DNA, primers (0.4 mM), MgCl2 (1.5 mM),dNTPs (0.2 mM each), and 2U/rxn of PlatinumTM TaqDNA Polymerase (2U) (Invitrogen). The thermal profile forPCR amplifications was as follows: initial denaturation at 94�for 3 min, followed by 30 cycles of 94� for 30 sec, 58–60� for1 min, 72� for 2 min, and ending with a final extension at72� for 7 min. The PCR products were analyzed by agarosegel electrophoresis, purified to remove primers and excessnucleotides, and sequenced in two separate reactions, each usingoneof the PCRprimers as a sequencingprimer (Bio-FabResearch,Rome, Italy). Sanger sequences are available on request. Genediagrams were obtained using GSDS (Gene structure displayserver: http://gsds.cbi.pku.edu.cn/) (Hu et al. 2015).

Western blot analysis

For western blots, larvae were homogenized in SDS gel-loading buffer (60 mM Tris-HCl, pH 6.8, 10% glycerol, 2%SDS, 10 mM dithiothreitol, and 0.1% bromophenol blue) inthe presence of protease inhibitors (10 mM benzamidineHCl, 1 mMPMSF, 1 mg/ml phenanthroline, 10 mg/ml apro-tinin, 10 mg/ml leupeptin, and 10 mg/ml pepstatin A) andheated at 95� for 4 min. Insolubles were pelleted by centri-fugation before electrophoresis. Proteins fractionated by 10%SDS-PAGE were electroblotted onto Immobilion-P polyvinyl-difluoridemembranes (Bio-Rad Laboratories) in a buffer con-taining 10 mM 3-cyclohexylamino-1-propanesulfonic acid(CAPS) (Sigma-Aldrich), pH 11, and 20% methanol, in asemi-dry transfer apparatus (Amersham Biosciences). Thefilter was blocked with 5% nonfat dry milk in Tris-bufferedsaline with Tween 20 (TBST) buffer (20 mM Tris pH 7.5,150 mM NaCl, 0.1% Tween 20). After blocking, proteinswere probed with monoclonal antibodies against singed

1996 L. Fanti et al.

(DSHB Hybridoma Product sn 7C) (1:20), and a-tubulin(Sigma) (1:3000), and detected with a 1:10,000 dilution ofHorseradish Peroxidase (HRP)-conjugated goat anti-mouselinked protein A. The Enhanced Chemiluminescence kitwas purchased from GE Healthcare. Images were acquiredwith the ChemiDoc imaging system (Bio-Rad Laboratories).

RNA isolation and semiquantitative RT-PCR analysis

Total RNAwas isolated from larvae andadult heads accordingto the protocol supplied with Qiazol Reagent (Qiagen). Theconcentration and purity of the RNA samples were determinedusingaNanoDrop1000Spectrophotometer (ThermoScientific).A sample of 5 mg of total RNA was used as a template to syn-thesize cDNA using oligo dT, random hexamers, and Super-Script reverse transcriptase III (Invitrogen) according to themanufacturer’s protocol. For semiquantitative PCR amplifica-tion, 6 ml of the cDNA synthesized were used as template withPlatinum Taq (Invitrogen), according to the manufacturer’s rec-ommendations. The cycling conditions were as follow: 94� for3 min; 27 cycles of 30 sec denaturation at 94�, 30 sec at 58�and 60 sec at 72�. PCR was performed using the followingprimer pairs designed using NCBI/Primer-BLAST:

sn R(A,B,E,F) For 59-ACTGGAGTGCAGTTCGTGAG-39.sn R(A,B,E,F) Rev 59-TGCCAAACTGATCGACCGAA-39.sn RG For 59-CAAAGCGGTGAACAGTACGC-39.sn RG Rev 59-TGCCAAACTGATCGACCGAA-39.Rpl32 For 59-CCCAAGGGTATCGACAACAG-39.Rpl32 Rev 59-GACAATCTCCTTGCGCTTCT-39.

The final PCR products were electrophoresed on 1.5% aga-rose gels.

qRT-PCR analysis

TotalRNApurified fromcontrol andheat-shockedpupae fromPaliano and Carpineto Romano strains was reverse tran-scribed using oligo dT and SuperScript reverse transcriptaseIII (Invitrogen) according to themanufacturer’s protocol. TheqPCR reactions were carried out with QuantiFast SYBRGreenPCR Kit (Qiagen) according to manufacturer’s protocol. Forthe quantification of transposon transcripts, we used the 2DDCt

method by comparing the amount of transposon transcripts toRpl32 transcript. qRT-PCR experiments were performed intwo independent biological replicates each with three technicalreplicates. Statistical significance was determined by Mann-Whitney tests using GraphPad Prism Software. A P value ,0.05was considered statistically significant.

The primers used were:

Rpl32 For 59-GCGCACCAAGCACTTCATC-39.Rpl32 Rev 59-TTGGGCTTGCGCCATT-39.Gypsy For 59-CTTCACGTTCTGCGAGCGGTCT-39.Gypsy Rev 59-CGCTCGAAGGTTACCAGGTAGGTTC-39.Het-A For 59-ACTGCTGAAGCTCGGATTCC-39.Het-A Rev 59-TGTAGCCGGATTCGTCATATTTC-39.I-element For 59-CAATCACAACAACAAAATCC-39.I-element Rev 59-GGTGTTGGTGTGGTTGGTTG-39.

P-element For 59-CAAGCTTTGCGTACTCGCTTT-39.P-element Rev 59-TCAGCAGAGCTGTCATACTCGAA-39.micropia For 59-CGCTTTGAAACCGAAGAGAC-39.micropia Rev 59-CATACACCGCGTAACATTCG-39.roo For 59-CGTCTGCAATGTACTGGCTCT-39.roo Rev 59-CGGCACTCCACTAACTTCTCC-39.ZAM For 59-ACTTGACCTGGATACACTCACAAC-39.ZAM Rev 59-GAGTATTACGGCGACTAGGGATAC-39.

KP-element excision by D2-3 transposase

sncar (sncar/sncar;+/+) virgin females were crossed withmalescarrying the transposase D2-3 (w*/Y;ry506Sb1P{D2-3}99B/TM2, rySCred1). F1 males carrying both the singed mutationand the transgenic construct encoding P-element transpo-sase (stubble bristles, Sb1)(sncar/Y; ry506Sb1P{D2-3}99B/+)were mated with virgin females carrying an attached-X(C(1)DX,y1f1/Y;+/+), and the male progeny phenotypewas screened to isolate singed revertants. Finally, each malerevertant was backcrossed with C(1)DX virgin females toestablish a stock.

Data availability

Strains and Sanger sequences are available upon request.Table S1 contains the complete list of primers used in thisstudy.

Results

Phenocopy induction by heat-shock

Weused flies from twodifferent natural populations collectedin central Italy around the towns of Paliano and CarpinetoRomano. We chose natural populations to have a widerspectrumofgenetic variability compared to laboratory strains.As shown in Figure 1, for each population, 20 nonvirgin fe-males were allowed lay eggs to establish F1 stocks. Then,from the Paliano F1 offspring, 50 virgin females were crossedto 50males, and from theCarpinetoRomano offspring, 47 vir-gin females were crossed to 51 males; after egg deposition,we purified the genomic DNA from each of the F1 flies thatrepresent the parental generation in our experiments. The F2progeny from each population were subjected to 40� for 4 hrduring the first half of the pupal stage (from 10 to 40 hr afterpuparium formation). It has been reported that this period ismore sensitive for creating variants, andmore resistant to thepupal lethality after stress (Mitchell and Lipps 1978; Mitchelland Petersen 1982). In our heat-shocked populations, wealso found a low pupal mortality, but a complete larval lethal-ity. After eclosion, the treated flies were intercrossed, andtheir progeny was again treated by heat-shock at the pupalstage. The same protocol was repeated for several genera-tions until the emergence of a fair number of flies showingmorphological abnormalities resembling different classicalmutants. These flies were then selected and crossed to eachother in separate vials, and their progeny were again treated

Transposon Jumping in Canalization and Assimilation Concepts 1997

with heat-shock at the pupal stage. This procedure was re-peated until a selected phenotypic variant was apparentlygenetically fixed even in the absence of stress.

Both stressed strains also produced other types of pheno-typic abnormalities, such as two differently colored eyes, asingle affected wing or double genitalia, with significantfrequency (�25 and 12% per generation in the Paliano andCarpineto Romano populations, respectively) (Figure 2,Table 1, and Table 2).

For each strain, we followed and fixed phenotypic variantsthat corresponded to classical mutations at different loci, asdetermined by complementation tests. We isolated theX-linked bristle mutations forked (Figure 3A) and singed (Fig-ure 3B), respectively, in the Paliano and Carpineto Romanostrains. We also fixed the sepia phenotypic variant (Figure3C) in Paliano by isolating a corresponding autosomal sepiamutation; in Carpineto Romano, we fixed a phenotypic var-iant showing tumors in both larvae and adults by isolating thetumor-inducer cactus mutation (Figure 3D).

Wemade stable stocks for all thesemutations. The completedescription of the genes corresponding to the fixed mutants is

reported in FlyBase (http://flybase.org) (Attrill et al. 2016).We also want to stress that, through all the generations ofthe F1-established stocks used as controls, we did not observeany of the types of variants that were eventually fixed in thetreated strains. We observed only occasional abnormal vari-ants, such as the absence of one wing or a malformed antenna(�0.2% per generation). Among the heat-shocked pupae fromthe Paliano strain, some flies with a clear forked phenotypeappeared in different vials during the first few generations.From the first two generations, we collected four males withthe forked phenotype, and we crossed them to female fliesfrom classical forked1 strains in separate vials for 5 days; thesemales were then transferred to vials containing forked femalevariants to continue with the heat-shock treatments of theirpupae progeny. Such crosses (male forked variants 3 forked1

females) gave only progeny that would be expected for vari-ants that were actually phenocopies, namely, all forked malesand all wild type females. We also examined the progeny fromfemale variants showing the forked phenotype. If these forkedfemales had been truemutants, wewould expect to see severalforkedmales in F1 progeny, whether or not these females were

Figure 2 Phenotypic variants induced byheat-shock stress. (A) One haltere trans-formed to a wing. (B) Blistered wings. (C)Abnormal tergites. (D) Abnormal eye mor-phology and white color. (E) Male with twogenital apparatus. (F) Histogram showingthe total number of examined flies in boththe stressed (HS) and control (CTR) popula-tions, and the number of phenotypic vari-ants (red). In each generation the variantswith developmental abnormalities, not re-sembling heritable phenotypic mutants,were discarded and only wild type flieswere allowed to continue to mate.

Figure 1 Schematic representation of experimen-tal design. From each of the Paliano and CarpinetoRomano collections, 20 nonvirgin females wereallowed to lay eggs to establish F1 stocks. Then,from the F1 offspring, a number of virgin femaleswere crossed to a number of males. The F2 prog-eny from each population were subjected to 40�for 4 hr during the pupal stage; after eclosion, thetreated flies were intercrossed, and their progenywere again treated by heat shock at the pupalstage. The same protocol was repeated for severalgenerations and, in any generation, indicated asF*, where a substantial number of similar pheno-copies emerged (F* = fourth for sepia and sixth for

forked phenocopies in heat treated Paliano population; seventh for cactus and 12th for singed phenocopies in heat treated Carpineto Romanopopulation, as reported below), these were selected and intercrossed in separate vials, and the progeny were again treated by heat shock at the pupalstage (blue). This procedure was repeated for a number of generations until fixation of the phenotype. Meanwhile, we continued to stress the wild-typelines (green) of both populations looking for phenocopies resembling other classical mutants. We ended the treatments at the generations in which wefixed the last mutant from each population (the ninth generation in Paliano strain with sepia fixation and 12th generation in Carpineto Romano strainwith the fixation of cactus and the isolation of singed). Fn represents a number of consecutive treated generations.

1998 L. Fanti et al.

virgins; instead, we observed only wild-type males. Wethus concluded that the apparent forked females were alsophenocopies. At the sixth heat-shocked generation, theoriginal treated Paliano strain produced a vial with sixmales showing the forked phenotype. Since the forked muta-tion is located on the X chromosome, we deduced that thesemales were probably derived from a cluster, or a single forkedmutation, that arose in a grandfather’s germline. We testedthese males by crossing them with females carrying anattached-X compound chromosome. This cross producesmale progeny who receive the X chromosome directly fromtheir fathers. All the male progeny were in fact forked,indicating that the phenotypic variants indeed carried amutation at the corresponding forked gene, which wenamed forkedpal (fpal). Similarly, in the 12th generationof heat-shock stress in the Carpineto Romano strain, weobserved five males in a single vial showing the X-linkedsinged phenotype. We again crossed these males with fe-males carrying an attached-X, and all the progeny maleswere singed, indicating that these males were generatedby a mutation at the singed locus, which we called singedcar

(sncar). Since we did not observe singed phenocopies inprevious generations, this case corresponds to a classicmutagenic event.

In the Paliano strain, some flies with the sepia phenotypealso appeared in different vials during the first few genera-tions. We collected six male sepia variants, and crossed themto female flies from the classical mutant sepia1 strain in sepa-rate vials for 5 days; thesemaleswere then transferred to vialscontaining sepia female variants to continue with the heat-shock treatments of their pupae progeny. Also in this case,the crosses (male sepia variants 3 sepia1 females) gave onlyprogeny thatwould be expected for variants thatwere actuallyphenocopies, namely, all wild-type progeny. In the fourth heat-shocked generation of the original Paliano strain, however,12 flies showing the sepia phenotype were collected and inter-mated, and the pupae of their progeny were subjected to heat-shock. After eclosion, we found 11 wild type flies and 59 sepiaflies. We collected the sepia flies, and again their progenyweresubjected to heat-shock at the pupal stage. This procedure wasrepeated for another three generations, with increasing num-bers of flies showing the sepia phenotype (from 84 to 92%)until its fixation at the ninth generation (Figure 4A and Table3). We called this mutation sepiapal (sepal).

The Carpineto Romano strain also produced, in everygeneration, a variable number of flies showing melanotic

Figure 3 Fixed phenocopies from Paliano (A and C)and Carpineto Romano (B and D) natural populations.(A) Fly displaying the forked phenotype (right) hasshortened, gnarled and bent dorsal bristles comparedto the wild-type (left) (arrows). (B) Abnormal shape ofdorsal bristles (right) compared to the wild type (left)in the singed mutant (arrows). (C) Eye color pheno-types of wild-type (left) and sepia mutant (right). (D)Phenotype of hypomorphic cactus fixed mutation inthird instar larvae (left) and adult flies (right). Mela-notic nodules in the hemocoel are seen as black spots,as indicated by arrows.

Table 2 Frequency of morphological abnormalities observed inheat-stressed Carpineto Romano strain

Carpineto Romano

Generations Number of variants Total % Variants

1 34 447 7.612 85 847 1.043 96 1126 8.534 116 1143 10.155 146 1304 11.196 166 1381 12.027 156 1281 12.188 144 1194 12.069 148 1226 12.07

10 153 1239 12.3511 170 1164 14.6112 155 1146 13.53

Table 1 Frequency of morphological abnormalities observed inheat-stressed Paliano strain

Paliano

Generations Number of variants Total % Variants

1 81 343 23.622 180 744 24.193 212 813 26.084 222 951 23.345 261 1097 23.796 257 1047 24.557 240 881 27.248 205 855 23.989 277 1085 25.53

Transposon Jumping in Canalization and Assimilation Concepts 1999

tumors in the abdomen. Since melanotic tumors could becaused by mutations at several different genes, we could notprobe such variants by complementation tests. However, wecollected and intercrossed the variants in a vial for 5 daysbefore transferring them to another vial for the heat-shockprocedure. We did not observe any mutant phenotypes ineither F1 or F2 nonheat-treated offspring. These resultsstrongly suggested that the variants were just phenocopiesinduced by heat-shock. At the seventh heat-shocked genera-tion, the frequency of flies with tumors began to increase. Wecollected these flies, and subjected their progeny to heat-shockat the pupal stage.We repeated this procedure for a number ofgenerations, and, for each generation, we observed a gradualand significant increase in the frequency of the tumor pheno-type until its fixation at the 12th generation (Figure 4B andTable 4). Chromosome mapping and complementation teststold us that the phenotype was caused by a hypomorphicmutation at the cactus locus, which we have called cactuscar

(cactcar). Mutations of this gene are known to induce tumors(Sparrow 1978).

Molecular analysis of the fixed mutations

For all these mutations, we performed a molecular analysis ofthe corresponding loci by PCR amplification and Sanger se-quencing using appropriate overlapping primers (Table S1).We found that the forkedmutation was caused by a frameshift

deletion in the antepenultimate exon of the gene (Figure 5)that affects the coding of all the nine different transcripts.

The singed mutation is correlated with an insertion of theKP transposon (Lee et al. 1996) (Figure 6A). We were able toobtain revertants by the remobilization of the KP-elementusing an active transposase (Figure 6B). The KP insertionaffects the production of four of the five singed transcripts,except the sn-RG transcript (Figure 6, C and D).

The sepia mutation was caused by an extended deletionthat includes the first and almost the entire second exon ofthe sepia gene, and the entire second exon of the Gst03 ad-jacent gene (Figure 7, A and B).

Since these two genes share a strong homology, and thesame orientation, along chromosome 3, the deletion wasprobably generated by a pairing of the sepia gene with theGst03 gene followed by a heat-shock-induced recombination.

In the case of the cactus hypomorphic mutation, we foundan insertion of the micropia retrotransposon (Lankenau et al.1988) into the 39 UTR of the gene (Figure 8).

The fixed phenocopies were de novo induced byheat-shock treatment

To test whether these mutations were induced de novo by theheat-shock stress, or were already present in the original

Table 4 Frequency of cactus mutants collected from heat-stressedCarpineto Romano strain

Carpineto Romano

Generations Melanotic tumors Total % Melanotic tumors

1 3 447 0.672 7 847 0.833 7 1126 0.624 5 1143 0.445 9 1304 0.696 13 1381 0.947 67 1281 5.238 106 354 29.949 132 299 44.15

10 212 318 66.6711 249 303 82.1812 361 363 99.45

The numbers indicate the cactus variants seen in each generation from the total ofthe scored flies. From the eighth generation on, the bold numbers refer to the newstrain formed from the progeny of only the cactus variants collected from the pre-vious generation.

Table 3 Frequency of sepia variants collected from heat-stressedPaliano strain

Paliano

Generations Sepia Total % Sepia

1 5 343 1.462 5 744 0.673 4 813 0.494 12 951 1.265 59 70 84.296 93 110 84.547 127 145 87.598 171 186 91.949 201 201 100

The numbers indicate the sepia variants seen in each generation from the total ofthe scored flies. From the fifth generation on, the bold numbers refer to the newstrain formed from the progeny of only the sepia variants collected from the pre-vious generation.

Figure 4 (A) Frequency of flies showing the sepia pheno-type from the fifth heat-shocked generation to sepal fixa-tion in the ninth generation. (B) Frequency of cactus fliesfrom the eighth heat-shocked generation to cactus fixa-tion in the 12th generation.

2000 L. Fanti et al.

population, we used PCR to analyze all the flies collectedfrom the parental generation. We did not find any genomiclesions at the corresponding genes (Figure 9, Figure 10, Fig-ure 11, and Figure 12), thus excluding the existence of anycryptic variants underlying the fixed mutations.

Taken together, the data strongly suggest that the heat-shock stress caused both DNA deletions and the activation oftransposons followed by their mobilization. It has already

been shown that such stress produces DNA damage by theinduction of double-stranded breaks (DSBs) in G1 and G2cells, and by the arrest of replication fork progression inS-phase cells (Velichko et al. 2012).

We also tested for possible transposon activation byexamining the expression of several transposons, includingboth the P-element andmicropia, after heat shock. As shownin Figure 13, we found significantly more transcripts of all

Figure 5 Molecular characterizationof heat-shock-induced forked muta-tion. (A) The forkedpal mutation isassociated with a 4-nt frameshift de-letion in the antepenultimate exon ofthe gene as determined by Sanger se-quencing. Blue bars below the sche-matic representation the forked geneindicate the overlapped fragmentsamplified to study the structural orga-nization of forked. (B) The 4-nt dele-tion was verified by PCR analysis usingallele-specific primers for wild-typeforked (forked F_forked NO DEL R)that amplify only the wild type DNAbut not the forkedpal DNA, and allele-specific primers for the forkedpal mu-tant allele (forked F_forked DEL R) thatamplify only the forkedpal DNA butnot the forked wild-type DNA. Rpl32was used as internal control gene. (C)Partial sequence of the antepenulti-mate exon of the forked. The primersequences are highlighted in bold,and the arrows show the primer di-rection from 59 to 39. The 4 bp de-letion is underlined. wtpal = Palianowild type strain; fpal= forked Palianomutant.

Figure 6 Molecular characterizationof heat-induced singed mutation.(A) The singedmutation (sncar) is cor-related to a single copy KP-elementinsertion in the first intron of thesn-RG transcript; the same genomicposition corresponds to the 59-UTRregion of another four singed tran-scripts, sn-RA, sn-RB, sn-RE, and sn-RF. Blue bars below the schematicrepresentation of singed gene indi-cate the overlapping fragmentsamplified to study the structural orga-nization of singed gene. The primersindicated by the arrows allowed us tosequence and identify the KP transpo-son along with its insertion point. (B)PCR analysis of genomic DNA fromwtcar, sncar, and a singed revertant

strain (revcar). The genotyping demonstrates the precise excisions of KP-element. (C) Semiquantitative RT-PCR analysis shows that four sn transcripts aresignificantly affected by the KP-insertion in singedmutant larvae compared to the control wild-type larvae. cDNA was analyzed using two primer pairs: the firstdesigned to amplify the singed splice variants A, B, E, and F (lanes 1 and 2), and the second specific for sn-RG (lanes 3 and 4). The KP-insertion does not affectthe sn-RG transcript in larvae, where it is poorly expressed (C), or in adult heads (D), where it is highly expressed; Rpl32 was used as an internal control. (E)Western blot assay showing an incomplete but significant reduction of SINGED protein levels in sncar larvae compared to the control strain (HP1 was used asinternal control). wtcar = Carpineto Romano wild type strain; sncar = Carpineto Romano singed mutant.

Transposon Jumping in Canalization and Assimilation Concepts 2001

elements, except for ZAM and roo in the Carpineto Romanopopulation.

Discussion

Our results clearly show that heat-shock stress can induceDNA deletions and transpositions, and strongly suggest thatthemainmechanism inheat-shock is theproductionofdenovo

variants through its mutagenic properties. Although it is notpossible to exclude the existence of processes like thoseproposed by Waddington, we think that the prevalent effectof a temperature stress applied at the pupal stage for anumber of generations is to induce morphological alter-ations that can be fixed by the selection of de novo inducedcorrespondingmutations. The number of generations requiredfor the complete fixation of a specific phenocopy depends onthe number of genetic factors involved in its determination,their autosomal or sex chromosomal localization, their degreeof penetrance, and the time of their appearance. In addition,our selection procedures do not exclude the presence of non-virgin females, which could extend the time necessary forfixation. Thus, in cases such as that described by Wadding-ton, who observed that the assimilation of crossveinlessphenocopies was related to two mutations localized on twodifferent autosomes, a higher number of generations beforecomplete assimilation is clearly expected. We propose a com-plementary, if not alternative, model to that of Waddington(Figure 14A), that offers a mechanistic explanation for theapparent assimilation processes, which we call the “pseu-doassimilation model” (Figure 14B).

Intriguingly, this mechanism was already considered andrejected at the same time byWaddington (1953) himself withthe following sentence: “One could therefore at best supposethat the shock treatment increased the production of many dif-ferent types of mutations, tending to mimic all the relevantphenocopies. But since there are no considerations which forceus to postulate the occurrence of any new mutations at all, itdoes not seem necessary to pursue the argument any further.”Curiously, this sentence does not seem to take into accountprevious work that had already suggested a mutagenic ef-fect of repeated heat-shock treatment on Drosophila larvae,and, more strikingly, the formulation of a hypothesis similarto ours. That hypothesis, called “parallelism between somaticvariants and their fixation by induction of corresponding mu-tations” by Jollos (1931, 1933, 1934) is summarized by thefollowing sentence: “Heat treatment causes noninheritedvariations within the treated generation—Sooner or later,with repeated treatment, the same variations are producedas true mutations.”

Figure 7 Molecular characterization of heat-shock-induced sepia muta-tion. (A) PCR analysis of sepia genomic region, performed using outerprimers flanking the deleted area (se2 F_se2 R) and inner primers span-ning the deleted region (se1 F_se1 R), reveals that the sepiapal allele iscaused by a large deletion; in the sepal mutant, PCR with se2 primersamplifies a product of 1388 bp instead of the expected 2989 bp ampli-con; PCR with se1 primers instead amplifies a wild-type fragment of only854 bp in the Paliano wild-type, as expected. Rpl32 was used as aninternal control gene. (B) The deletion includes the first and almost theentire second exon of the sepia gene, and the entire second exon ofthe Gst03 adjacent gene (the 1601 bp deletion mutation is depictedby the shaded area). Blue bars below the schematic representation ofGst03 and sepia genes indicate the amplicons produced by PCR amplifi-cation of the Gst03 and sepia genome regions in the wild-type Palianostrain. wtpal = Paliano wild type strain; sepal = Paliano sepia mutant.

Figure 8 Molecular characterization ofheat-induced cactus mutation. (A) Thegenetic lesion responsible for the cactusallele is a full-length insertion of amicropia retrotransposon into the 39-UTR of the gene as demonstrated byPCR analysis (B). Blue bars below theschematic representation of cactus geneindicate the positions of overlappingfragments amplified to study the struc-tural organization of cactus gene. Thearrows indicate the oligonucleotidesthat allowed us to sequence and iden-tify the micropia retrotransposon, alongwith its insertion point.

2002 L. Fanti et al.

Figure 9 PCR analysis of forked gene corresponding tothe fixed phenocopies in the wild type Paliano parentalflies. (A) Schematic representation of forked gene show-ing the annealing positions of allele-specific PCR primers(arrows). (B) The forked PCR analysis, genomic DNA puri-fied from each single parental fly was amplified withboth allele-specific primers for wild-type forked (forkedF_forked NO DEL R) that amplify only the wild type DNAand not the forkedpal DNA, and allele-specific primers forforkedpal mutant allele (forked F_forked DEL R) that am-plify only the forkedpal DNA and not the forked wild-typeDNA. Rpl32 was used as internal control gene. In all sam-ples from the Paliano parental line, allele-specific PCR am-plifies the 400 bp target fragment only with specificprimers for the wild-type allele (+), which confirms theabsence of the molecular lesion responsible for thestress-induced forked phenotype. Allele specific primersfor wild-type and forkedpal are indicated by arrows in(A). The white boxes delimit the part of the gels contain-ing the control PCR reactions performed on genomic DNApurified from wild-type and mutant alleles.

Figure 10 PCR analysis of the genes corresponding to the singed fixedphenocopies in the wild type Carpineto Romano parental flies. (A) Sche-matic representation of singedcar allele showing the KP-element insertionand the position of PCR primers (arrows). (B) Genomic DNA purified fromCarpineto Romano parental flies was amplified using primers flanking(sn2.2 F_sn2.2 R) the KP-element insertion. We obtained a 1.5 kbp ampli-fied product in all parental flies as expected for a wild type genomic se-quence. (C) PCR analysis was performed using a forward primer designedon KP-element, and a reverse primer specific for singed 59 UTR sequence(KP-element F_sn R). These primers amplify only in presence of KP-elementinsertion (as shown in control PCR on sncar sample ). Rpl32 was used asinternal PCR control. The white boxes delimit the part of the gels containingthe control PCR reactions performed on genomic DNA purified from Car-pineto Romano wild-type strain and singedcar mutant allele, respectively.

Figure 11 PCR analysis of the sepia gene corresponding to the fixedphenocopies in the wild type Paliano parental flies. (A) Schematic rep-resentation of the deletion involving Gst03 and sepia genes. (B) PCRreactions were performed on genomic DNA purified from each singlefly of the Paliano parental generation using specific primers flankingthe gene regions altered in the sepia stress-induced mutation. se2outer primers, indicated by arrows in (A), amplify, in sepal, a productof 1388 bp instead of the expected 2989 bp wild-type amplicon. Thewhite boxes delimit the part of the gels containing the control PCRreactions performed on genomic DNA purified from wild-type andmutant alleles.

Transposon Jumping in Canalization and Assimilation Concepts 2003

The proposed mechanism for the fixation of an apparentphenocopy within a few generations needs an explanation,however. In principle, it is possible to envisage differentmechanisms for such a phenomenon. For example, hetero-zygotes for awild-type and a recessive allelicmutation couldproduce mutant phenocopies under heat shock due to ahaplo-insufficiency of the wild type gene. Such apparentphenocopies, called “heterocopies,” have already been de-scribed in heterozygotes for multiple wing hair (Petersenand Mitchell 1987) and forked (Mitchell and Petersen1985) recessive mutations. In such a case, a rapid fixationof this particular type of phenocopy could be expectedthrough the selection of homozygous mutants. Note thatthe heterocopies could be formally considered cryptic var-iants, but their fixation depends only on the selection ofhomozygote genotypes and cannot be ascribed to a sort ofcanalization and assimilation process. Another possibilityis that pseudoassimilation induced by stress depends onstrong modifications in the expression of specific loci,along with a hypermutability of the same loci in germ cells,and possibly also in somatic cells. We propose that underlying

these pseudoassimilation processes, epigenetic modificationsat specific loci produced by the heat-shock stress could play animportant role by altering gene expression, and thus giving riseto phenocopies. In this regard, there is evidence suggestingthat heat shock can induce epigenetic modifications that, insome cases, could be transgenerationally inherited [see forreview Wei et al. (2015) and Prokopuk et al. (2015)]. Forexample, in Drosophila, heat-shock treatment can produceheritable alterations of heterochromatin by the induced re-lease of dTAF-2, a factor required for heterochromatin assem-bly (Seong et al. 2011). In C. elegans, changes in theexpression profiles of heterochromatic arrays induced bystress can be transgenerationally transmitted (Klosin et al.2017). If the same epigenetic modifications that inducephenocopies make the corresponding loci more suscepti-ble to DNA alterations caused by transposon insertions orDNA rearrangements, it is possible to envisage a mecha-nism that increases the probability of pseudoassimilationof the phenocopies. Heat-shock stress could produce phe-nocopies during metamorphosis, and could increase themutability of the corresponding genes. This would permitthe rapid fixation of the phenocopies when coselected withthe newly arisen corresponding mutations. We call thistype of coselection “coincident selection.”

Figure 12 PCR analysis of the genes corresponding to the cactus fixedphenocopies in the wild Carpineto Romano parental flies. (A) Schematic di-agram showing the cactcar allele and the relative position of primers bindingsites (arrows). (B) PCR analysis, using specific primers (cact5.1.3 F_cact5.1.3 R)flanking micropia insertion, shows a 400 bp amplified product in all parentalflies as expected for a wild-type genomic sequence; conversely, in cactcar

control sample we obtain a 7 kb amplicon. (C) PCR analysis was performedusing a forward primer designed on micropia-element, and a reverse specificprimer for cactus 39 UTR sequence (micropia F5_cact5.1.3 R). These primersamplify only in presence of micropia-element insertion (as shown in controlPCR on cactcar sample); Rpl32 was used as internal PCR control. White boxesindicate the control PCR reactions performed on genomic DNA purified fromwild type and the cactuscar mutant allele.

Figure 13 Transposable element expression profiles from (A) Paliano and(B) Carpineto Romano control and heat-stressed pupae. Fold expressionlevels of transposon transcripts, determined by quantitative RT-PCR, areshown relative to Rpl32 expression (**P # 0.01).

2004 L. Fanti et al.

The evolutionary implications of such a mechanism areevident. In nature, environmental changes, when perceivedas a stress, might increase genetic variability by their muta-genic properties, and alsomight induce phenotypic variants,some of them beneficial, through individual phenotypicplasticity (Baldwin 1896; Piacentini et al. 2014). Such phe-notypic variants could be maintained over generations by apersistence of the stress, or by transgenerationally inheritedepigenetic mechanisms. The transgenerational persistenceof the phenotypic variants, along with a marked increase ofgenetic variability, should ultimately result in their fixationthrough the appearance of positively selected, correspond-ing germline mutations (Piacentini et al. 2014). Therefore,selection of advantageous phenotypes by the environmentdoes not necessarily imply a simultaneous selection of cor-responding underlying mutations. It is also possible toimagine the maintenance of a phenotypic variant withoutthe presence of a corresponding gene mutation, although,sooner or later, the induction and selection of such amutation will be required for its fixation. The differencesbetween the classical and this additional view lie inthe dynamics of the selection and fixation of favorablephenotypes. In other words, in the classical view, geneticmutations precede the corresponding phenotypes, whileaccording to the second view, phenotypes precede the cor-responding mutations, which could arise after a number ofgenerations, and then be fixed by a “coincident selection”process. In general terms, this model recalls the concept of“organic selection,” also defined as “the Baldwin effect,”simultaneously postulated in 1896 by Baldwin (1896),Lloyd Morgan (1896) and Osborn (1896), and alwaysconsidered of low significance among evolutionaryprocesses. The problem was the difficulty in imagining amechanism that could produce de novo mutations corre-sponding to the stress-induced phenotypes that were fixablein a relatively few generations, as found by Waddington.Considering the present data, along with the considerations

outlined above, and the growing body of evidence on theexistence of a mutagenic effect of stress in both prokaryotesand eukaryotes, the relevance of this type of mechanismin both adaptation and evolvability no longer appears he-retical (Galhardo et al. 2007; Mittelman 2013; Piacentiniet al. 2014).

Evolutionary phenomena similar to organic selection couldplay a major role in biological evolution, and should be in-troduced into current debate on evolutionary theory (Lalandet al. 2014). In conclusion, we are convinced that our findingscontribute an additional significant tool in the conceptualtoolbox for describing biological evolution.

Acknowledgments

We thank the editors and reviewers for their constructivesuggestions, which helped us improve the manuscript. Wethank Umberto Capozi for collecting the natural popula-tions, the Bloomington Drosophila Stock Center for flystocks, and the Developmental Studies Hybridoma Bankfor antibodies. Financial support for this research wasprovided by the Epigenomics Flagship Project EpiGen,Italian Ministry of Education and Research, NationalResearch Council.

Author contributions: S.P., L.F., and L.P. conceived theresearch. L.F. performed the genetic assimilation exper-iments. L.F., L.P., U.C., A.M.C., and S.P. performed thegenetic and molecular analysis of mutations. S.P. wrote thepaper. The authors declare no conflict of interest.

Literature Cited

Attrill, H., K. Falls, J. L. Goodman, G. H. Millburn, G. Antonazzoet al., 2016 FlyBase: establishing a gene group resource forDrosophila melanogaster. Nucleic Acids Res. 44(D1): D786–D792.

Figure 14 Models for the assimilation of a stress-inducedphenocopy. (A) According to Waddington, the phenocopyinduced by stress could be assimilated by the selection ofpreexisting cryptic mutations at one (monogenic deter-mination) or more (polygenic determination) genes. (B)According to our model, the phenocopy could be fixedby a coselection of a corresponding germline mutationde novo induced by stress. Note that this view is alsoadaptable to mutant phenotypes with a polygenic deter-mination. As discussed in the text, the assimilation pro-cess could take several generations after the mutationinduction and before the complete fixation of the mutantphenotype (red). Another important point is that themutational rate could be higher in stressed phenocopiesthan in stressed wild-type flies due to increased trans-poson activity in such variants.

Transposon Jumping in Canalization and Assimilation Concepts 2005

Baldwin, J. M., 1896 A new factor in evolution. Am. Nat. 30: 441–451.

Burga, A., M. O. Casanueva, and B. Lehner, 2011 Predicting mu-tation outcome from early stochastic variation in genetic inter-action partners. Nature 480: 250–253.

Darwin, C., 1859 On the Origin of Species by Means of NaturalSelection or Preservation of Favoured Races in the Struggle forLife. Murray, London.

Galhardo, R. S., P. J. Hastings, and S. M. Rosenberg, 2007 Mutation asa stress response and the regulation of evolvability. Crit. Rev. Biochem.Mol. Biol. 42: 399–435.

Gangaraju, V. K., H. Yin, M. M. Weiner, J. Wang, X. A. Huang et al.,2011 Drosophila piwi functions in Hsp90-mediated suppres-sion of phenotypic variation. Nat. Genet. 43: 153–158.

Goldschmidt, R., 1935 Gen und Ausseneigenschaft. I. Zeitschr.ind. Abstl. 69: 38–69.

Hu, B., J. Jin, A.-Y. Guo, H. Zhang, J. Luo et al., 2015 GSDS 2.0:an upgraded gene feature visualization server. Bioinformatics 31:1296–1297. Available at: http://bioinformatics.oxfordjournals.org/content/31/8/1296.full.

Jollos, V., 1931 Die experimentelle Auslösung von mutationen undihre Bedeutung für das mutationsproblem. Naturwissenschaften19: 171–177.

Jollos, V., 1933 Die Übereinstimmung der bei Drosophila mela-nogaster nach Hitzeeinwirkung entstehenden modifikationenund mutationen. Naturwissensch 21: 831–834.

Jollos, V., 1934 Inherited changes produced by heat-treatment inDrosophila melanogaster. Genetica 16: 476–494.

Klosin, A., E. Casas, C. Hidalgo-Carcedo, T. Vavouri, and B. Lehner,2017 Transgenerational transmission of environmental infor-mation in C. elegans. Science 356: 320–323.

Laland, K., T. Uller, M. Feldman, K. Sterelny, G. B. Müller et al.,2014 Does evolutionary theory need a rethink? Nature 514:161–164.

Lamarck, J. B., 1809 Philosophie Zoologique: ou Exposition desConsidérations Relative á l’histoire Naturelle des Animaux. Dentuet L’Auteur, Paris.

Lankenau, D. H., P. Huijser, E. Jansen, K. Miedema, and W. Hennig,1988 Micropia: a retrotransposon of Drosophila combiningstructural features of DNA viruses, retroviruses and non-viraltransposable elements. J. Mol. Biol. 204: 233–246.

Lee, C. C., Y. M. Mul, and D. C. Rio, 1996 The Drosophila P-elementKP repressor protein dimerizes and interacts with multiple sites onP-element DNA. Mol. Cell. Biol. 16: 5616–5622.

Lloyd Morgan, C., 1896 On modification and variation. Science 4:733–740.

McClintock, B., 1984 The significance of responses of the genometo challenge. Science 226: 792–801.

Mitchell, H. K., and L. S. Lipps, 1978 Heat-shock and phenocopyinduction in Drosophila. Cell 15: 907–918.

Mitchell, H. K., and N. S. Petersen, 1982 Developmental abnor-malities in Drosophila induced by heat-shock. Dev. Genet. 3: 91–102.

Mitchell, H. K., and N. S. Petersen, 1985 The recessive phenotypeof forked can be uncovered by heat-shock in Drosophila. Dev.Genet. 6: 93–100.

Mittelman, D., (Editor), 2013 Stress-Induced Mutagenesis.Springer, New York.

Osborn, H. F., 1896 A mode of evolution requiring neither naturalselection nor the inheritance of acquired characteristics. Trans.N. Y. Acad. Sci. 15: 141–148.

Petersen, N. S., and H. K. Mitchell, 1987 The induction of a mul-tiple wing hair phenocopy by heat-shock in mutant heterozy-gotes. Dev. Biol. 121: 335–341.

Piacentini, L., L. Fanti, V. Specchia, M. P. Bozzetti, M. Berloco et al.,2014 Transposons, environmental changes and heritable in-duced phenotypic variability. Chromosoma 123: 345–354.

Prokopuk, L., P. S. Western, and J. M. Stringer, 2015 Transgenerationalepigenetic inheritance: adaptation through the germline epi-genome? Epigenomics 7: 829–846.

Queitsch, C., T. A. Sangster, and S. Lindquist, 2002 Hsp90 as acapacitor of phenotypic variation. Nature 417: 618–624.

Rutherford, S. L., and S. Lindquist, 1998 HSP90 as a capacitor formorphological evolution. Nature 396: 336–342.

Seong, K.-Y., D. Li, H. Shimizu, R. Nakamura, and S. Ishii,2011 Inheritance of stress-induced ATF-2-dependent epi-genetic change. Cell 145: 1049–1061.

Sollars, V., X. Lu, L. Xiao, X. Wang, M. D. Garfinkel et al.,2003 Evidence for an epigenetic mechanism by whichHsp90 acts as a capacitor for morphological evolution.Nat. Genet. 33: 70–74.

Sparrow, J. C., 1978 Melanotic ’Tumours’, pp. 277–313 in The Ge-netics and Biology of Drosophila, Vol. 2b, edited by M. Ashburner,and T. R. F. Wright. Academic Press, London.

Specchia, V., L. Piacentini, P. Tritto, L. Fanti, R. D’Alessandro et al.,2010 HSP90 prevents phenotypic variation by suppressing themutagenic activity of transposons. Nature 463: 662–665.

Tariq, M., U. Nussbaumer, Y. Chen, C. Beisel, and R. Paro,2009 Trithorax requires Hsp90 for maintenance of activechromatin at sites of gene expression. Proc. Natl. Acad. Sci.USA 106: 1157–1162.

Velichko, A. K., N. V. Petrova, O. L. Kantidze, and S. V. Razin,2012 Dual effect of heat shock on DNA replication and ge-nome integrity. Mol. Biol. Cell 23(17): 3450–3460.

Waddington, C. H., 1942 Canalization of development and theinheritance of acquired characters. Nature 150: 563–565.

Waddington, C. H., 1953 Genetic assimilation of an acquiredcharacter. Evolution 7: 118–126.

Waddington, C. H., 1959 Canalization of development and ge-netic assimilation of acquired characters. Nature 183: 1654–1655.

Wei, Y., H. Schatten, and Q.-Y. Sun, 2015 Environmental epige-netic inheritance through gametes and implications for humanreproduction. Hum. Reprod. Update 21: 194–208.

Communicating editor: P. K. Geyer

2006 L. Fanti et al.