Experimental and field data support habitat expansion of the allopolyploid Arabidopsis kamchatica owing to parental legacy of heavy metal hyperaccumulation

Little empirical evidence is available whether allopolyploid species combine or merge adaptations of parental species. The allopolyploid species Arabidopsis kamchatica is a natural hybrid of the diploid parents A. halleri, a heavy metal hyperaccumulator, and A. lyrata, a non-hyperaccumulating species. Zinc and cadmium were measured in native soils and leaf tissues in natural populations, and in hydroponic cultures of A. kamchatica and A. halleri. Pyrosequencing was used to estimate homeolog expression ratios. Soils from human modified sites showed significantly higher Zn concentrations than non-modified sites. Leaf samples of A. kamchatica collected from 40 field localities had > 1,000 µg g-1 Zn in over half of the populations, with significantly higher amounts of Zn concentrations in plants from human modified sites. In addition, serpentine soils were found in two populations. Most genotypes accumulated >3000 µg g-1 of Zn in hydroponic culture with high variability among them. Genes involved in hyperaccumulation showed a bias in the halleri-derived homeolog. A. kamchatica has retained constitutive hyperaccumulation ability inherited from A. halleri. Our field and experimental data provides a compelling example in which the inheritance of genetic toolkits for soil adaptations likely contributed to the habitat expansion of an allopolyploid species.


78
The ecological advantages of whole genome duplication (WGD) and its implication on species habitats 79 has been speculated for decades (Ohno 1970;Stebbins 1971

99
Metal concentrations in soils are an important factor for environmental niches. Critical toxicity 100 of soils for plants has been defined using arbitrary thresholds, 100-300 µg g -1 for Zn and 6-8 µg g -1 for 101 Cd (Krämer 2010). Bert et al. (2002) proposed similar values of soil metals, in which soil with more 102 than 300 µg g -1 Zn or 2 µg g -1 Cd be classified as metalliferous (or metal-contaminated) soil according 103 to the French agricultural recommendation. Stein et al. (2017) classified metalliferous and non-104 metalliferous soils based on a multifactorial metal analysis rather than using threshold values.

105
Hyperaccumulation in plants has been defined to be > 3000 µg g -1 of Zn and >100 µg g -1 Cd in leaves 106 (Krämer 2010). The hyperaccumulator species A. halleri is living in both contaminated soils near 107 mines and non-contaminated soils, and plants from both types of soils can hyperaccumulate (Bert et al. 108 2002; Stein et al. 2017). Because hyperaccumulation is constitutive (species-wide) in A. halleri, it may 109 have evolved hyperaccumulation as a mechanism to extract high amounts of heavy metals from metal-110 deficient soils for chemical defense (Boyd 2007). It has been demonstrated experimentally that A. 111 4 halleri plants treated with Cd or Zn are more resistant to specialist and generalist insect herbivores 112 (Kazemi-Dinan et al. 2014. Zinc concentration in leaves of A. halleri above 1,000 µg g -1 was 113 shown to be effective in chemical defense. The other diploid parent, A. lyrata has known adaptations to 114 serpentine soils in some regions of the species distribution but this appears to be local adaptation 115 (Turner et al. 2010; Arnold et al. 2016) and it is not a constitutive trait. It is not known whether the 116 habitats of A. kamchatica encompass human contaminated sites. 117 Hyperaccumulation in A. kamchatica appears weakened compared to the A. halleri parent, but is 118 greatly increased compared to the A. lyrata parent. In experimental conditions, four natural genotypes 119 of A. kamchatica were shown to accumulate Zn in leaf tissues to about half of the A. halleri parent, but 120 10-100 times more than the non-hyperaccumulating A. lyrata parent ). It is not 121 surprising that the trait distribution in these four genotypes is between the two parents, considering the 122 divergence in hyperaccumulation between the diploid progenitors of A. kamchatica. However, in this 123 small sample size, little variation in Zn accumulation among genotypes was detected in leaf tissues.

124
While it is clear that some genotypes of A. kamchatica can accumulate substantial amounts of heavy 125 metals under experimental conditions, combining data from the laboratory and in natura is necessary 126 to study adaptation (Shimizu et al. 2011;Yamasaki et al. 2017

177
Arabidopsis kamchatica is a ruderal species which grows both in human-modified and natural habitats 178 (see definitions in next paragraph) but none of the samples in this study were found where mining 6 activities occurred. We sampled leaf tissues from 40 populations of A. kamchatica from Japan and 180 Alaska, USA to quantify Zn. We obtained soils of 37 of these localities and measured Zn concentration 181 of the soils. The concentration of Zn in soils among the sites ranged from 18.7 to 642.8 (average 150 182 µg g -1 , Fig. 1, Table S1-S3). The Zn concentrations in soils of 18 sites were above 100 µg g -1 , and five 183 sites were above 300 µg g -1 . This indicated that a considerable number of sites where A. kamchatica is 184 growing is are above the critical toxicity of 100-300 µg g -1 for Zn, while others are below the 185 thresholds.

186
Observations made during plant and soil collection were that several of the sites had some 187 evidence of human modification such as buildings, mountain lodges, fences or paved roads. To 188 examine whether A. kamchatica could adapt to naturally-and artificially-generated metalliferous soils, 189 we classified the localities into human-modified or non-modified sites. In human-modified sites, there  Table S4). Yet, many sites without human 196 modification also showed considerable amounts of Zn (seven sites with >100 µg g -1 Zn, one of them 197 >300 µg g -1 Zn), and are likely to reflect natural geology (Schlüchter et al. 1981). These data suggest 198 that the habitat of A. kamchatica encompasses localities with high Zn in soil, both by natural geology 199 and human modification, as well as sites with low Zn in either category.  Table S2). The 204 highest concentration of Zn was found in three plants at the Mt. Sirouma site (6,500 -6,661 µg g -1 ) in 205 Japan (2835 meters above sea level), a human modified site with the highest level of Zn in the soils 206 ( Fig. 1). We found that plants from human-modified habitats had on average higher quantities of Zn 207 (mean Zn concentration 1,815 µg g -1 ) than non-modified sites (mean Zn concentration 1,004 µg g -1 ) 208 and the difference was significant (p = 0.00136, Fig. 1, Tables S1-S3). In human-modified habitats 209 there was a significant positive correlation between leaf and soil concentrations of Zn (r = 0.68, p < 10 -210 5 ) suggesting that the increased concentrations of Zn in the soils at these sites increased the availability 211 of this metal for uptake by plants. By contrast, there was no correlation between leaf and soil 212 concentrations of Zn in non-modified habitats (r = 0.007, p = 0.95; Fig. S1). We found that A. 7 kamchatica accumulated high levels of Zn in the field and when there is greater availability of Zn in 214 the soils, plants tend to have higher levels of Zn in the leaves as we found in many of the modified 215 sites.

216
In the parental species A. halleri, collected from two sites in Russia and two sites in Japan, we 217 detected > 3,000 µg g -1 of Zn in leaves in all four populations, with the highest levels found in plants 218 from the Tada mine site in Japan (average of six replicates = 16,068 µg g -1 (Table S1)). Soils were 219 collected at the two mine sites in Japan (Tada and Omoidegawa (OMD)) (Table S4), and the Zn 220 concentration (1,317 -2,490 µg g -1 Zn) was several times higher than any other sites with A.  Table   228 S3). The Cd concentration in leaves in the five populations with the highest Cd ranged from 3.6 to 8.9 229 µg g -1 , while all other populations had less than 3 µg g -1 of Cd in leaves (Table S1). The maximum 230 amount of Cd in any single plant was 12.1 µg g -1 (Table S2, S3) and was found at the Mt. Siraiwa site, 231 which also had the highest Cd levels in soils. However, this level of Cd accumulation in leaves is much 232 lower than the threshold defined for Cd hyperaccumulation (>100 µg g -1 Cd) in leaves (Krämer 2010).

233
Unlike for Zn, there was no significant difference in Cd concentrations in the soils or leaf tissues 234 between the modified or non-modified sites (Table S3). Moreover, there was no correlation between 235 leaf and soil concentrations of Cd in the modified habitats, but there was a positive correlation in the 236 non-modified habitats (Fig. S1). We also found very little correlation between Cd and Zn leaf 237 accumulation (r = 0.08, p = 0.39), likely due to much less overall Cd in soils and leaves compared 238 with Zn. In summary, leaf and soil concentrations of Cd are below critical toxicity levels for plants 239 (Krämer 2010) and are negligible compared to Zn concentrations in the same populations.

240
For A. halleri, soil at the two Japanese mine populations contained 5.0 -6.9 µg g -1 of Cd (Table   241 S4), which is comparable to the critical toxicity defined by Krämer (2010) (6-8 µg g -1 ). Leaf Cd 242 concentration of the two Japanese and two Russian populations were measured, and one of them was 243 above the >100 µg g -1 threshold of hyperaccumulation (1,267 µg g -1 at Tada mine). This supports a     Table S5a). The shoot to root ratio of Zn accumulation was greater than one (i.e., higher Zn 296 concentrations in leaves than roots) for both genotypes of A. halleri, while all but one of the A. 297 kamchatica genotypes had a leaf to root ratio in Zn accumulation greater than one (the MAG genotype 298 had a leaf to root ratio greater than one (Table S5a)). There was no significant correlation between leaf 299 and root accumulation in A. kamchatica (r = -0.06, p = 0.37), indicating that genotypes with high metal 300 accumulation in leaves do not necessarily have lower accumulation in roots, and vice versa. Together, 301 these comparisons suggest that leaf tissues, rather than roots, are a sink for heavy metals in A. halleri, 302 and that the diploid species has a more efficient mechanism of root to shoot transport of Zn than A. 303 kamchatica.

304
We also included a synthesized allopolyploid generated by a cross between Asian A. halleri ssp.  Table S6). In contrast to natural A. kamchatica that experienced evolutionary 309 changes after polyploid speciation, the synthesized allopolyploid provides direct experimental evidence 310 that Zn hyperaccumulation in A. halleri can be inherited in A. kamchatica, but that the trait is 311 attenuated due to genome merging with the non-hyperaccumulator A. lyrata.

312
Next, we wanted to test whether plant genotypes responded similarly when the Zn treatment was 313 increased from 500 µM to 1,000 µM Zn. Using nine of the twenty A. kamchatica genotypes that were 314 used in the previous (500 µM) experiment, this time treated with 1,000 µM Zn, we found significant 315 increases in Zn accumulation in the leaves in all but two genotypes ( Figure S4) can accumulate large amounts of Zn even when exposed to relatively low concentrations of Zn in the 328 soil, but in our experiments above, A. halleri had a higher ratio of leaf to root Zn accumulation than 329 nearly all A. kamchatica genotypes. We therefore compared the relative efficiency of Zn transport 330 between A. halleri and A. kamchatica over 10-fold gradients of Zn treatment for 48 hours, so that a 331 rapid transport from roots to leaves upon exposure to various concentrations of Zn may be detected. In 332 this experiment, we tested Zn accumulation in leaves and roots of one A. halleri genotype and three A.

336
In leaf tissues, A. halleri accumulated > 2,000 µg g -1 when treated with only 10 µM of Zn. This 337 was a four-fold increase compared with the 1 µM treatment condition. There was no significant 338 difference in Zn accumulation in leaves of A. halleri between the 10 and 100 µM treatments, then 339 nearly 2-fold increase between the 100 and 1000 µM treatments ( Figure 4A). By contrast, A.

346
In root tissues, A. halleri showed little increase in Zn accumulation until the 1,000 µM treatment, 347 and Zn accumulation in roots did not exceed 5,000 µg g -1 in any treatment condition ( Figure 4B). The 348 shoot to root ratio of Zn concentrations in each of the three treatment conditions was ³ 1 for A. halleri 11 ( Figure 4C). For A. kamchatica, Zn concentrations in roots were significantly greater than A. halleri in 350 the three treatment conditions (with the exception of the MUR genotype at the 10 µM treatment). The 351 1,000 µM treatment resulted in the largest Zn accumulation in the roots, ³ 10,000 µg g -1 for all three A. 352 kamchatica genotypes ( Figure 4B), which was twice as high as A. halleri at this treatment. The high Zn 353 accumulation in the roots of A. kamchatica resulted in shoot to root ratios £ 0.5 at all treatments, less 354 than A. halleri at each Zn treatment condition ( Figure 4C). These results further show that A. halleri is 355 a more efficient hyperaccumulator of Zn, especially when exposed to low concentrations.   (Table S9). The one exceptional coefficient was found in the MTP3 gene x treatment x 378 tissue (root) which was highly significant (p = 0.0005). Whether this ratio change is due to significant 379 up-regulation of the L-origin copy, or down-regulation in the H-origin copy remains to be tested, but

391
Because the concentrations ranged in more than two orders of magnitudes (from 30 to > 6,000 µg g -1 ) 392 with no clear split into two classes of hyperaccumulating and non-hyperaccumulating genotypes (Table   393 S2), this suggests that the classification into these two categories may be too simple to study

397
Therefore, a similar level of Zn in leaves could also be sufficient for A. kamchatica to deter insects.

398
We did not find plants growing in sites with obvious contamination from mining as with A.     Table S5a). This is considerably higher than most natural genotypes which have 484 on average about 0.5 of the accumulation of Zn in leaves compared to A. halleri. Moreover, 485 accumulation of Zn in leaves of the synthetic polyploid was orders of magnitude higher than that of the 486 A. lyrata parent (see Fig. 1 in Paape et al. 2016), which clearly demonstrated that hyperaccumulation 487 can be retained following hybridization between the divergent parental species, but that the A. lyrata shoot to root ratio of Zn concentration was greater than or equal to one for the A. halleri genotype used 502 in our experiment, which is a similar result to a previous study that tested Zn accumulation over a 503 gradient of treatments using a European A. halleri genotype (Talke 2006

505
By contrast, the shoot to root ratio of Zn concentration for A. kamchatica was equal to or less 506 and one-half for the three genotypes used in our gradient experiment. Notably, the highest shoot to root 507 ratio was at the control condition (1 µM) for both A. halleri and A. kamchatica, which is consistent 508 with a Zn deficiency response due to constitutive expression of heavy metal transporters (Talke 2006; 509 Hanikenne et al. 2008). The hydroponic treatments demonstrated experimentally how plants can 510 accumulate high levels of Zn when exposed to both low and high levels of Zn, which is important for 511 understanding the relationship between soil concentrations and accumulation by plants in natural 512 conditions.

513
We suggest two reasons why an allopolyploid derived from two parents that were divergent for 514 heavy metal hyperaccumulation would show a reduction in the hyperaccumulation trait compared to 515 the diploid hyperaccumulator parent (A. halleri). First, a reduction or attenuation compared to the 516 diploid parents in total expression levels of metal transporters is expected due to allopolyploidization.  to increased copy number. The gene HMA3 also has a putative role in vacuolar sequestration of Zn 546 (Morel et al. 2008), similar to MTP1, but is only found in a single copy in both A. halleri and A. lyrata 547 (Paape et al. 2018). This gene also has very high allele specific expression in A. halleri (Talke 2006) 548 and strong H-origin bias in A. kamchatica, which must be due to cis-regulatory differences. Together, 549 these three genes (HMA3, HMA4, MTP1)

18
We suggest that the inheritance of hyperaccumulation from A. halleri conferred advantages 586 instantaneously at the polyploid speciation, estimated to be approximately 100,000 years ago (Paape et

654
We conducted three Zn treatment experiments. The first experiment used a supplement of 500 655 µM Zn added to the hydroponic solution for a period of 7 days. This experiment included 19 natural A. 656 kamchatica genotypes and one synthetic polyploid that was generated from A. halleri subsp. 657 gemmifera (w302) and A. lyrata subsp. petraea (Table S1). We also used two naturally collected A.